Series comparison

-[PULL 00/14] Block patches
+[Qemu-devel] [PULL v2 00/24] Block patches
-The following changes since commit 754f756cc4c6d9d14b7230c62b5bb20f9d655888:
+The following changes since commit 56f9e46b841c7be478ca038d8d4085d776ab4b0d:
-  Merge tag 'pull-target-arm-20220422-1' of https://git.linaro.org/people/pmaydell/qemu-arm into staging (2022-04-22 08:03:18 -0700)
+  Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2017-02-20' into staging (2017-02-20 17:42:47 +0000)
-are available in the Git repository at:
+are available in the git repository at:
-  https://gitlab.com/hreitz/qemu.git tags/pull-block-2022-04-25
+  git://github.com/stefanha/qemu.git tags/block-pull-request
-for you to fetch changes up to 348a0740afc5b313599533eb69bbb2b95d2f1bba:
+for you to fetch changes up to a7b91d35bab97a2d3e779d0c64c9b837b52a6cf7:
-  iotests/108: Fix when missing user_allow_other (2022-04-25 14:46:45 +0200)
+  coroutine-lock: make CoRwlock thread-safe and fair (2017-02-21 11:39:40 +0000)
 ----------------------------------------------------------------
-Block patches:
+Pull request
-- New @force parameter for blockdev-change-medium
-- Improvements to the iotests to help with debugging
+v2:
-- Fix iotest 108 on systems without user_allow_other in fuse.conf
+ * Rebased to resolve scsi conflicts
 ----------------------------------------------------------------
-Denis V. Lunev (1):
-  block: add 'force' parameter to 'blockdev-change-medium' command
-Hanna Reitz (1):
+Paolo Bonzini (24):
-  iotests/108: Fix when missing user_allow_other
+  block: move AioContext, QEMUTimer, main-loop to libqemuutil
   aio: introduce aio_co_schedule and aio_co_wake
   block-backend: allow blk_prw from coroutine context
   test-thread-pool: use generic AioContext infrastructure
   io: add methods to set I/O handlers on AioContext
   io: make qio_channel_yield aware of AioContexts
   nbd: convert to use qio_channel_yield
   coroutine-lock: reschedule coroutine on the AioContext it was running
     on
   blkdebug: reschedule coroutine on the AioContext it is running on
   qed: introduce qed_aio_start_io and qed_aio_next_io_cb
   aio: push aio_context_acquire/release down to dispatching
   block: explicitly acquire aiocontext in timers that need it
   block: explicitly acquire aiocontext in callbacks that need it
   block: explicitly acquire aiocontext in bottom halves that need it
   block: explicitly acquire aiocontext in aio callbacks that need it
   aio-posix: partially inline aio_dispatch into aio_poll
   async: remove unnecessary inc/dec pairs
   block: document fields protected by AioContext lock
   coroutine-lock: make CoMutex thread-safe
   coroutine-lock: add limited spinning to CoMutex
   test-aio-multithread: add performance comparison with thread-based
     mutexes
   coroutine-lock: place CoMutex before CoQueue in header
   coroutine-lock: add mutex argument to CoQueue APIs
   coroutine-lock: make CoRwlock thread-safe and fair
-John Snow (12):
+ Makefile.objs                       |   4 -
-  iotests: replace calls to log(qemu_io(...)) with qemu_io_log()
+ stubs/Makefile.objs                 |   1 +
-  iotests/163: Fix broken qemu-io invocation
+ tests/Makefile.include              |  19 +-
-  iotests: Don't check qemu_io() output for specific error strings
+ util/Makefile.objs                  |   6 +-
-  iotests/040: Don't check image pattern on zero-length image
+ block/nbd-client.h                  |   2 +-
-  iotests/040: Fix TestCommitWithFilters test
+ block/qed.h                         |   3 +
-  iotests: create generic qemu_tool() function
+ include/block/aio.h                 |  38 ++-
-  iotests: rebase qemu_io() on top of qemu_tool()
+ include/block/block_int.h           |  64 +++--
-  iotests/migration-permissions: use assertRaises() for qemu_io()
+ include/io/channel.h                |  72 +++++-
-    negative test
+ include/qemu/coroutine.h            |  84 ++++---
-  iotests/image-fleecing: switch to qemu_io()
+ include/qemu/coroutine_int.h        |  11 +-
-  iotests: remove qemu_io_pipe_and_status()
+ include/sysemu/block-backend.h      |  14 +-
-  iotests: remove qemu_io_silent() and qemu_io_silent_check().
+ tests/iothread.h                    |  25 ++
-  iotests: make qemu_io_log() check return codes by default
+ block/backup.c                      |   2 +-
+ block/blkdebug.c                    |   9 +-
- qapi/block.json                               |  6 ++
+ block/blkreplay.c                   |   2 +-
- block/qapi-sysemu.c                           |  3 +-
+ block/block-backend.c               |  13 +-
- monitor/hmp-cmds.c                            |  4 +-
+ block/curl.c                        |  44 +++-
- hmp-commands.hx                               | 11 ++-
+ block/gluster.c                     |   9 +-
- tests/qemu-iotests/030                        | 85 +++++++++++--------
+ block/io.c                          |  42 +---
- tests/qemu-iotests/040                        | 53 +++++++-----
+ block/iscsi.c                       |  15 +-
- tests/qemu-iotests/056                        |  2 +-
+ block/linux-aio.c                   |  10 +-
- tests/qemu-iotests/108                        |  2 +-
+ block/mirror.c                      |  12 +-
- tests/qemu-iotests/149                        |  6 +-
+ block/nbd-client.c                  | 119 +++++----
- tests/qemu-iotests/163                        |  5 +-
+ block/nfs.c                         |   9 +-
- tests/qemu-iotests/205                        |  4 +-
+ block/qcow2-cluster.c               |   4 +-
- tests/qemu-iotests/216                        | 12 +--
+ block/qed-cluster.c                 |   2 +
- tests/qemu-iotests/218                        |  5 +-
+ block/qed-table.c                   |  12 +-
- tests/qemu-iotests/224                        |  4 +-
+ block/qed.c                         |  58 +++--
- tests/qemu-iotests/242                        |  6 +-
+ block/sheepdog.c                    |  31 +--
- tests/qemu-iotests/245                        | 17 ++--
+ block/ssh.c                         |  29 +--
- tests/qemu-iotests/255                        |  4 +-
+ block/throttle-groups.c             |   4 +-
- tests/qemu-iotests/258                        | 11 ++-
+ block/win32-aio.c                   |   9 +-
- tests/qemu-iotests/298                        | 17 ++--
+ dma-helpers.c                       |   2 +
- tests/qemu-iotests/303                        |  4 +-
+ hw/9pfs/9p.c                        |   2 +-
- tests/qemu-iotests/310                        | 22 ++---
+ hw/block/virtio-blk.c               |  19 +-
- tests/qemu-iotests/iotests.py                 | 69 ++++++++-------
+ hw/scsi/scsi-bus.c                  |   2 +
- tests/qemu-iotests/tests/image-fleecing       | 30 ++++---
+ hw/scsi/scsi-disk.c                 |  15 ++
- .../qemu-iotests/tests/migration-permissions  | 28 +++---
+ hw/scsi/scsi-generic.c              |  20 +-
- .../tests/mirror-ready-cancel-error           |  2 +-
+ hw/scsi/virtio-scsi.c               |   7 +
- .../qemu-iotests/tests/nbd-reconnect-on-open  |  2 +-
+ io/channel-command.c                |  13 +
- .../qemu-iotests/tests/stream-error-on-reset  |  4 +-
+ io/channel-file.c                   |  11 +
- ui/cocoa.m                                    |  1 +
+ io/channel-socket.c                 |  16 +-
-files changed, 231 insertions(+), 188 deletions(-)
+ io/channel-tls.c                    |  12 +
  io/channel-watch.c                  |   6 +
  io/channel.c                        |  97 ++++++--
  nbd/client.c                        |   2 +-
  nbd/common.c                        |   9 +-
  nbd/server.c                        |  94 +++-----
  stubs/linux-aio.c                   |  32 +++
  stubs/set-fd-handler.c              |  11 -
  tests/iothread.c                    |  91 +++++++
  tests/test-aio-multithread.c        | 463 ++++++++++++++++++++++++++++++++++++
  tests/test-thread-pool.c            |  12 +-
  aio-posix.c => util/aio-posix.c     |  62 ++---
  aio-win32.c => util/aio-win32.c     |  30 +--
  util/aiocb.c                        |  55 +++++
  async.c => util/async.c             |  84 ++++++-
  iohandler.c => util/iohandler.c     |   0
  main-loop.c => util/main-loop.c     |   0
  util/qemu-coroutine-lock.c          | 254 ++++++++++++++++++--
  util/qemu-coroutine-sleep.c         |   2 +-
  util/qemu-coroutine.c               |   8 +
  qemu-timer.c => util/qemu-timer.c   |   0
  thread-pool.c => util/thread-pool.c |   8 +-
  trace-events                        |  11 -
  util/trace-events                   |  17 +-
 files changed, 1712 insertions(+), 533 deletions(-)
  create mode 100644 tests/iothread.h
  create mode 100644 stubs/linux-aio.c
  create mode 100644 tests/iothread.c
  create mode 100644 tests/test-aio-multithread.c
  rename aio-posix.c => util/aio-posix.c (94%)
  rename aio-win32.c => util/aio-win32.c (95%)
  create mode 100644 util/aiocb.c
  rename async.c => util/async.c (82%)
  rename iohandler.c => util/iohandler.c (100%)
  rename main-loop.c => util/main-loop.c (100%)
  rename qemu-timer.c => util/qemu-timer.c (100%)
  rename thread-pool.c => util/thread-pool.c (97%)
 --
-.35.1
+.9.3

-[PULL 12/14] iotests: remove qemu_io_silent() and qemu_io_silent_check().
+[Qemu-devel] [PULL v2 01/24] block: move AioContext, QEMUTimer, main-loop to libqemuutil
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-Like qemu-img, qemu-io returning 0 should be the norm and not the
+AioContext is fairly self contained, the only dependency is QEMUTimer but
-exception. Remove all calls to qemu_io_silent that just assert the
+that in turn doesn't need anything else.  So move them out of block-obj-y
-return code is zero (That's every last call, as it turns out), and
+to avoid introducing a dependency from io/ to block-obj-y.
-replace them with a normal qemu_io() call.
+main-loop and its dependency iohandler also need to be moved, because
-qemu_io_silent_check() appeared to have been unused already.
+later in this series io/ will call iohandler_get_aio_context.
-Signed-off-by: John Snow <jsnow@redhat.com>
+[Changed copyright "the QEMU team" to "other QEMU contributors" as
-Reviewed-by: Eric Blake <eblake@redhat.com>
+suggested by Daniel Berrange and agreed by Paolo.
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
+--Stefan]
-Message-Id: <20220418211504.943969-12-jsnow@redhat.com>
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Message-id: 20170213135235.12274-2-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qemu-iotests/216                        | 12 +++++-----
+ Makefile.objs                       |  4 ---
- tests/qemu-iotests/218                        |  5 ++---
+ stubs/Makefile.objs                 |  1 +
- tests/qemu-iotests/224                        |  4 ++--
+ tests/Makefile.include              | 11 ++++----
- tests/qemu-iotests/258                        | 11 +++++-----
+ util/Makefile.objs                  |  6 +++-
- tests/qemu-iotests/298                        | 17 ++++++--------
+ block/io.c                          | 29 -------------------
- tests/qemu-iotests/310                        | 22 +++++++++----------
+ stubs/linux-aio.c                   | 32 +++++++++++++++++++++
- tests/qemu-iotests/iotests.py                 | 16 --------------
+ stubs/set-fd-handler.c              | 11 --------
- tests/qemu-iotests/tests/image-fleecing       |  4 ++--
+ aio-posix.c => util/aio-posix.c     |  2 +-
- .../tests/mirror-ready-cancel-error           |  2 +-
+ aio-win32.c => util/aio-win32.c     |  0
- .../qemu-iotests/tests/stream-error-on-reset  |  4 ++--
+ util/aiocb.c                        | 55 +++++++++++++++++++++++++++++++++++++
-files changed, 38 insertions(+), 59 deletions(-)
+ async.c => util/async.c             |  3 +-
+ iohandler.c => util/iohandler.c     |  0
-diff --git a/tests/qemu-iotests/216 b/tests/qemu-iotests/216
+ main-loop.c => util/main-loop.c     |  0
-index XXXXXXX..XXXXXXX 100755
+ qemu-timer.c => util/qemu-timer.c   |  0
---- a/tests/qemu-iotests/216
+ thread-pool.c => util/thread-pool.c |  2 +-
-+++ b/tests/qemu-iotests/216
+ trace-events                        | 11 --------
-@@ -XXX,XX +XXX,XX @@
+ util/trace-events                   | 11 ++++++++
- # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
+files changed, 114 insertions(+), 64 deletions(-)
+ create mode 100644 stubs/linux-aio.c
- import iotests
+ rename aio-posix.c => util/aio-posix.c (99%)
--from iotests import log, qemu_img, qemu_io_silent
+ rename aio-win32.c => util/aio-win32.c (100%)
-+from iotests import log, qemu_img, qemu_io
+ create mode 100644 util/aiocb.c
+ rename async.c => util/async.c (99%)
- # Need backing file support
+ rename iohandler.c => util/iohandler.c (100%)
- iotests.script_initialize(supported_fmts=['qcow2', 'qcow', 'qed', 'vmdk'],
+ rename main-loop.c => util/main-loop.c (100%)
-@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
+ rename qemu-timer.c => util/qemu-timer.c (100%)
-     log('')
+ rename thread-pool.c => util/thread-pool.c (99%)
-     qemu_img('create', '-f', iotests.imgfmt, base_img_path, '64M')
+diff --git a/Makefile.objs b/Makefile.objs
--    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 0M 1M') == 0
+index XXXXXXX..XXXXXXX 100644
-+    qemu_io(base_img_path, '-c', 'write -P 1 0M 1M')
+--- a/Makefile.objs
-     qemu_img('create', '-f', iotests.imgfmt, '-b', base_img_path,
++++ b/Makefile.objs
-              '-F', iotests.imgfmt, top_img_path)
+@@ -XXX,XX +XXX,XX @@ chardev-obj-y = chardev/
--    assert qemu_io_silent(top_img_path,  '-c', 'write -P 2 1M 1M') == 0
+ #######################################################################
-+    qemu_io(top_img_path,  '-c', 'write -P 2 1M 1M')
+ # block-obj-y is code used by both qemu system emulation and qemu-img
-     log('Done')
+-block-obj-y = async.o thread-pool.o
+ block-obj-y += nbd/
-@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
+ block-obj-y += block.o blockjob.o
-     log('--- Checking COR result ---')
+-block-obj-y += main-loop.o iohandler.o qemu-timer.o
-     log('')
+-block-obj-$(CONFIG_POSIX) += aio-posix.o
+-block-obj-$(CONFIG_WIN32) += aio-win32.o
--    assert qemu_io_silent(base_img_path, '-c', 'discard 0 64M') == 0
+ block-obj-y += block/
--    assert qemu_io_silent(top_img_path,  '-c', 'read -P 1 0M 1M') == 0
+ block-obj-y += qemu-io-cmds.o
--    assert qemu_io_silent(top_img_path,  '-c', 'read -P 2 1M 1M') == 0
+ block-obj-$(CONFIG_REPLICATION) += replication.o
-+    qemu_io(base_img_path, '-c', 'discard 0 64M')
+diff --git a/stubs/Makefile.objs b/stubs/Makefile.objs
-+    qemu_io(top_img_path,  '-c', 'read -P 1 0M 1M')
+index XXXXXXX..XXXXXXX 100644
-+    qemu_io(top_img_path,  '-c', 'read -P 2 1M 1M')
+--- a/stubs/Makefile.objs
++++ b/stubs/Makefile.objs
-     log('Done')
+@@ -XXX,XX +XXX,XX @@ stub-obj-y += get-vm-name.o
-diff --git a/tests/qemu-iotests/218 b/tests/qemu-iotests/218
+ stub-obj-y += iothread.o
-index XXXXXXX..XXXXXXX 100755
+ stub-obj-y += iothread-lock.o
---- a/tests/qemu-iotests/218
+ stub-obj-y += is-daemonized.o
-+++ b/tests/qemu-iotests/218
++stub-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
-@@ -XXX,XX +XXX,XX @@
+ stub-obj-y += machine-init-done.o
- # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
+ stub-obj-y += migr-blocker.o
+ stub-obj-y += monitor.o
- import iotests
+diff --git a/tests/Makefile.include b/tests/Makefile.include
--from iotests import log, qemu_img, qemu_io_silent
+index XXXXXXX..XXXXXXX 100644
-+from iotests import log, qemu_img, qemu_io
+--- a/tests/Makefile.include
++++ b/tests/Makefile.include
- iotests.script_initialize(supported_fmts=['qcow2', 'raw'])
+@@ -XXX,XX +XXX,XX @@ check-unit-y += tests/test-visitor-serialization$(EXESUF)
+ check-unit-y += tests/test-iov$(EXESUF)
-@@ -XXX,XX +XXX,XX @@ with iotests.VM() as vm, \
+ gcov-files-test-iov-y = util/iov.c
-      iotests.FilePath('src.img') as src_img_path:
+ check-unit-y += tests/test-aio$(EXESUF)
++gcov-files-test-aio-y = util/async.c util/qemu-timer.o
-     qemu_img('create', '-f', iotests.imgfmt, src_img_path, '64M')
++gcov-files-test-aio-$(CONFIG_WIN32) += util/aio-win32.c
--    assert qemu_io_silent('-f', iotests.imgfmt, src_img_path,
++gcov-files-test-aio-$(CONFIG_POSIX) += util/aio-posix.c
--                          '-c', 'write -P 42 0M 64M') == 0
+ check-unit-y += tests/test-throttle$(EXESUF)
-+    qemu_io('-f', iotests.imgfmt, src_img_path, '-c', 'write -P 42 0M 64M')
+ gcov-files-test-aio-$(CONFIG_WIN32) = aio-win32.c
+ gcov-files-test-aio-$(CONFIG_POSIX) = aio-posix.c
-     vm.launch()
+@@ -XXX,XX +XXX,XX @@ tests/check-qjson$(EXESUF): tests/check-qjson.o $(test-util-obj-y)
+ tests/check-qom-interface$(EXESUF): tests/check-qom-interface.o $(test-qom-obj-y)
-diff --git a/tests/qemu-iotests/224 b/tests/qemu-iotests/224
+ tests/check-qom-proplist$(EXESUF): tests/check-qom-proplist.o $(test-qom-obj-y)
-index XXXXXXX..XXXXXXX 100755
---- a/tests/qemu-iotests/224
+-tests/test-char$(EXESUF): tests/test-char.o qemu-timer.o \
-+++ b/tests/qemu-iotests/224
+-    $(test-util-obj-y) $(qtest-obj-y) $(test-block-obj-y) $(chardev-obj-y)
-@@ -XXX,XX +XXX,XX @@
++tests/test-char$(EXESUF): tests/test-char.o $(test-util-obj-y) $(qtest-obj-y) $(test-io-obj-y) $(chardev-obj-y)
- # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
+ tests/test-coroutine$(EXESUF): tests/test-coroutine.o $(test-block-obj-y)
+ tests/test-aio$(EXESUF): tests/test-aio.o $(test-block-obj-y)
- import iotests
+ tests/test-throttle$(EXESUF): tests/test-throttle.o $(test-block-obj-y)
--from iotests import log, qemu_img, qemu_io_silent, filter_qmp_testfiles, \
+@@ -XXX,XX +XXX,XX @@ tests/test-vmstate$(EXESUF): tests/test-vmstate.o \
-+from iotests import log, qemu_img, qemu_io, filter_qmp_testfiles, \
+     migration/vmstate.o migration/qemu-file.o \
-                     filter_qmp_imgfmt
+         migration/qemu-file-channel.o migration/qjson.o \
- import json
+     $(test-io-obj-y)
+-tests/test-timed-average$(EXESUF): tests/test-timed-average.o qemu-timer.o \
-@@ -XXX,XX +XXX,XX @@ for filter_node_name in False, True:
+-    $(test-util-obj-y)
-                  '-F', iotests.imgfmt, top_img_path)
++tests/test-timed-average$(EXESUF): tests/test-timed-average.o $(test-util-obj-y)
+ tests/test-base64$(EXESUF): tests/test-base64.o \
-         # Something to commit
+     libqemuutil.a libqemustub.a
--        assert qemu_io_silent(mid_img_path, '-c', 'write -P 1 0 1M') == 0
+ tests/ptimer-test$(EXESUF): tests/ptimer-test.o tests/ptimer-test-stubs.o hw/core/ptimer.o libqemustub.a
-+        qemu_io(mid_img_path, '-c', 'write -P 1 0 1M')
+@@ -XXX,XX +XXX,XX @@ tests/usb-hcd-ehci-test$(EXESUF): tests/usb-hcd-ehci-test.o $(libqos-usb-obj-y)
+ tests/usb-hcd-xhci-test$(EXESUF): tests/usb-hcd-xhci-test.o $(libqos-usb-obj-y)
-         vm.launch()
+ tests/pc-cpu-test$(EXESUF): tests/pc-cpu-test.o
+ tests/postcopy-test$(EXESUF): tests/postcopy-test.o
-diff --git a/tests/qemu-iotests/258 b/tests/qemu-iotests/258
+-tests/vhost-user-test$(EXESUF): tests/vhost-user-test.o qemu-timer.o \
-index XXXXXXX..XXXXXXX 100755
++tests/vhost-user-test$(EXESUF): tests/vhost-user-test.o $(test-util-obj-y) \
---- a/tests/qemu-iotests/258
+     $(qtest-obj-y) $(test-io-obj-y) $(libqos-virtio-obj-y) $(libqos-pc-obj-y) \
-+++ b/tests/qemu-iotests/258
+     $(chardev-obj-y)
-@@ -XXX,XX +XXX,XX @@
+ tests/qemu-iotests/socket_scm_helper$(EXESUF): tests/qemu-iotests/socket_scm_helper.o
- # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
+diff --git a/util/Makefile.objs b/util/Makefile.objs
+index XXXXXXX..XXXXXXX 100644
- import iotests
+--- a/util/Makefile.objs
--from iotests import log, qemu_img, qemu_io_silent, \
++++ b/util/Makefile.objs
-+from iotests import log, qemu_img, qemu_io, \
+@@ -XXX,XX +XXX,XX @@
-         filter_qmp_testfiles, filter_qmp_imgfmt
+ util-obj-y = osdep.o cutils.o unicode.o qemu-timer-common.o
+ util-obj-y += bufferiszero.o
- # Returns a node for blockdev-add
+ util-obj-y += lockcnt.o
-@@ -XXX,XX +XXX,XX @@ def test_concurrent_finish(write_to_stream_node):
++util-obj-y += aiocb.o async.o thread-pool.o qemu-timer.o
-         if write_to_stream_node:
++util-obj-y += main-loop.o iohandler.o
-             # This is what (most of the time) makes commit finish
++util-obj-$(CONFIG_POSIX) += aio-posix.o
-             # earlier and then pull in stream
+ util-obj-$(CONFIG_POSIX) += compatfd.o
--            assert qemu_io_silent(node2_path,
+ util-obj-$(CONFIG_POSIX) += event_notifier-posix.o
--                                  '-c', 'write %iK 64K' % (65536 - 192),
+ util-obj-$(CONFIG_POSIX) += mmap-alloc.o
--                                  '-c', 'write %iK 64K' % (65536 -  64)) == 0
+ util-obj-$(CONFIG_POSIX) += oslib-posix.o
-+            qemu_io(node2_path,
+ util-obj-$(CONFIG_POSIX) += qemu-openpty.o
-+                    '-c', 'write %iK 64K' % (65536 - 192),
+ util-obj-$(CONFIG_POSIX) += qemu-thread-posix.o
-+                    '-c', 'write %iK 64K' % (65536 -  64))
+-util-obj-$(CONFIG_WIN32) += event_notifier-win32.o
+ util-obj-$(CONFIG_POSIX) += memfd.o
-             stream_throttle='tg'
++util-obj-$(CONFIG_WIN32) += aio-win32.o
-         else:
++util-obj-$(CONFIG_WIN32) += event_notifier-win32.o
-             # And this makes stream finish earlier
+ util-obj-$(CONFIG_WIN32) += oslib-win32.o
--            assert qemu_io_silent(node1_path,
+ util-obj-$(CONFIG_WIN32) += qemu-thread-win32.o
--                                  '-c', 'write %iK 64K' % (65536 - 64)) == 0
+ util-obj-y += envlist.o path.o module.o
-+            qemu_io(node1_path, '-c', 'write %iK 64K' % (65536 - 64))
+diff --git a/block/io.c b/block/io.c
+index XXXXXXX..XXXXXXX 100644
-             commit_throttle='tg'
+--- a/block/io.c
++++ b/block/io.c
-diff --git a/tests/qemu-iotests/298 b/tests/qemu-iotests/298
+@@ -XXX,XX +XXX,XX @@ BlockAIOCB *bdrv_aio_flush(BlockDriverState *bs,
-index XXXXXXX..XXXXXXX 100755
+     return &acb->common;
---- a/tests/qemu-iotests/298
+ }
-+++ b/tests/qemu-iotests/298
-@@ -XXX,XX +XXX,XX @@ class TestTruncate(iotests.QMPTestCase):
+-void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
-         os.remove(refdisk)
+-                   BlockCompletionFunc *cb, void *opaque)
+-{
-     def do_test(self, prealloc_mode, new_size):
+-    BlockAIOCB *acb;
--        ret = iotests.qemu_io_silent('--image-opts', '-c', 'write 0 10M', '-c',
+-
--                                     f'truncate -m {prealloc_mode} {new_size}',
+-    acb = g_malloc(aiocb_info->aiocb_size);
--                                     drive_opts)
+-    acb->aiocb_info = aiocb_info;
--        self.assertEqual(ret, 0)
+-    acb->bs = bs;
--
+-    acb->cb = cb;
--        ret = iotests.qemu_io_silent('-f', iotests.imgfmt, '-c', 'write 0 10M',
+-    acb->opaque = opaque;
--                                     '-c',
+-    acb->refcnt = 1;
--                                     f'truncate -m {prealloc_mode} {new_size}',
+-    return acb;
--                                     refdisk)
+-}
--        self.assertEqual(ret, 0)
+-
-+        iotests.qemu_io('--image-opts', '-c', 'write 0 10M', '-c',
+-void qemu_aio_ref(void *p)
-+                        f'truncate -m {prealloc_mode} {new_size}',
+-{
-+                        drive_opts)
+-    BlockAIOCB *acb = p;
-+
+-    acb->refcnt++;
-+        iotests.qemu_io('-f', iotests.imgfmt, '-c', 'write 0 10M',
+-}
-+                        '-c', f'truncate -m {prealloc_mode} {new_size}',
+-
-+                        refdisk)
+-void qemu_aio_unref(void *p)
+-{
-         stat = os.stat(disk)
+-    BlockAIOCB *acb = p;
-         refstat = os.stat(refdisk)
+-    assert(acb->refcnt > 0);
-diff --git a/tests/qemu-iotests/310 b/tests/qemu-iotests/310
+-    if (--acb->refcnt == 0) {
-index XXXXXXX..XXXXXXX 100755
+-        g_free(acb);
---- a/tests/qemu-iotests/310
+-    }
-+++ b/tests/qemu-iotests/310
+-}
 -
  /**************************************************************/
  /* Coroutine block device emulation */
 diff --git a/stubs/linux-aio.c b/stubs/linux-aio.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/stubs/linux-aio.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * Linux native AIO support.
 + *
 + * Copyright (C) 2009 IBM, Corp.
 + * Copyright (C) 2009 Red Hat, Inc.
 + *
 + * This work is licensed under the terms of the GNU GPL, version 2 or later.
 + * See the COPYING file in the top-level directory.
 + */
 +#include "qemu/osdep.h"
 +#include "block/aio.h"
 +#include "block/raw-aio.h"
 +
 +void laio_detach_aio_context(LinuxAioState *s, AioContext *old_context)
 +{
 +    abort();
 +}
 +
 +void laio_attach_aio_context(LinuxAioState *s, AioContext *new_context)
 +{
 +    abort();
 +}
 +
 +LinuxAioState *laio_init(void)
 +{
 +    abort();
 +}
 +
 +void laio_cleanup(LinuxAioState *s)
 +{
 +    abort();
 +}
 diff --git a/stubs/set-fd-handler.c b/stubs/set-fd-handler.c
 index XXXXXXX..XXXXXXX 100644
 --- a/stubs/set-fd-handler.c
 +++ b/stubs/set-fd-handler.c
@@ -XXX,XX +XXX,XX @@ void qemu_set_fd_handler(int fd,
  {
      abort();
  }
 -
 -void aio_set_fd_handler(AioContext *ctx,
 -                        int fd,
 -                        bool is_external,
 -                        IOHandler *io_read,
 -                        IOHandler *io_write,
 -                        AioPollFn *io_poll,
 -                        void *opaque)
 -{
 -    abort();
 -}
 diff --git a/aio-posix.c b/util/aio-posix.c
 similarity index 99%
 rename from aio-posix.c
 rename to util/aio-posix.c
 index XXXXXXX..XXXXXXX 100644
 --- a/aio-posix.c
 +++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@
  #include "qemu/rcu_queue.h"
  #include "qemu/sockets.h"
  #include "qemu/cutils.h"
 -#include "trace-root.h"
 +#include "trace.h"
  #ifdef CONFIG_EPOLL_CREATE1
  #include <sys/epoll.h>
  #endif
 diff --git a/aio-win32.c b/util/aio-win32.c
 similarity index 100%
 rename from aio-win32.c
 rename to util/aio-win32.c
 diff --git a/util/aiocb.c b/util/aiocb.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/util/aiocb.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * BlockAIOCB allocation
 + *
 + * Copyright (c) 2003-2017 Fabrice Bellard and other QEMU contributors
 + *
 + * Permission is hereby granted, free of charge, to any person obtaining a copy
 + * of this software and associated documentation files (the "Software"), to deal
 + * in the Software without restriction, including without limitation the rights
 + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 + * copies of the Software, and to permit persons to whom the Software is
 + * furnished to do so, subject to the following conditions:
 + *
 + * The above copyright notice and this permission notice shall be included in
 + * all copies or substantial portions of the Software.
 + *
 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 + * THE SOFTWARE.
 + */
 +
 +#include "qemu/osdep.h"
 +#include "block/aio.h"
 +
 +void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
 +                   BlockCompletionFunc *cb, void *opaque)
 +{
 +    BlockAIOCB *acb;
 +
 +    acb = g_malloc(aiocb_info->aiocb_size);
 +    acb->aiocb_info = aiocb_info;
 +    acb->bs = bs;
 +    acb->cb = cb;
 +    acb->opaque = opaque;
 +    acb->refcnt = 1;
 +    return acb;
 +}
 +
 +void qemu_aio_ref(void *p)
 +{
 +    BlockAIOCB *acb = p;
 +    acb->refcnt++;
 +}
 +
 +void qemu_aio_unref(void *p)
 +{
 +    BlockAIOCB *acb = p;
 +    assert(acb->refcnt > 0);
 +    if (--acb->refcnt == 0) {
 +        g_free(acb);
 +    }
 +}
 diff --git a/async.c b/util/async.c
 similarity index 99%
 rename from async.c
 rename to util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@
  /*
 - * QEMU System Emulator
 + * Data plane event loop
   *
   * Copyright (c) 2003-2008 Fabrice Bellard
 + * Copyright (c) 2009-2017 QEMU contributors
   *
   * Permission is hereby granted, free of charge, to any person obtaining a copy
   * of this software and associated documentation files (the "Software"), to deal
 diff --git a/iohandler.c b/util/iohandler.c
 similarity index 100%
 rename from iohandler.c
 rename to util/iohandler.c
 diff --git a/main-loop.c b/util/main-loop.c
 similarity index 100%
 rename from main-loop.c
 rename to util/main-loop.c
 diff --git a/qemu-timer.c b/util/qemu-timer.c
 similarity index 100%
 rename from qemu-timer.c
 rename to util/qemu-timer.c
 diff --git a/thread-pool.c b/util/thread-pool.c
 similarity index 99%
 rename from thread-pool.c
 rename to util/thread-pool.c
 index XXXXXXX..XXXXXXX 100644
 --- a/thread-pool.c
 +++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@
  #include "qemu/queue.h"
  #include "qemu/thread.h"
  #include "qemu/coroutine.h"
 -#include "trace-root.h"
 +#include "trace.h"
  #include "block/thread-pool.h"
  #include "qemu/main-loop.h"
 diff --git a/trace-events b/trace-events
 index XXXXXXX..XXXXXXX 100644
 --- a/trace-events
 +++ b/trace-events
 @@ -XXX,XX +XXX,XX @@
  #
+ # The <format-string> should be a sprintf()-compatible format string.
- import iotests
--from iotests import log, qemu_img, qemu_io_silent
+-# aio-posix.c
-+from iotests import log, qemu_img, qemu_io
+-run_poll_handlers_begin(void *ctx, int64_t max_ns) "ctx %p max_ns %"PRId64
+-run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
- # Need backing file support
+-poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
- iotests.script_initialize(supported_fmts=['qcow2'],
+-poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
-@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
+-
-     log('')
+-# thread-pool.c
+-thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
-     qemu_img('create', '-f', iotests.imgfmt, base_img_path, '64M')
+-thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
--    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 0M 1M') == 0
+-thread_pool_cancel(void *req, void *opaque) "req %p opaque %p"
--    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 3M 1M') == 0
+-
-+    qemu_io(base_img_path, '-c', 'write -P 1 0M 1M')
+ # ioport.c
-+    qemu_io(base_img_path, '-c', 'write -P 1 3M 1M')
+ cpu_in(unsigned int addr, char size, unsigned int val) "addr %#x(%c) value %u"
-     qemu_img('create', '-f', iotests.imgfmt, '-b', base_img_path,
+ cpu_out(unsigned int addr, char size, unsigned int val) "addr %#x(%c) value %u"
-              '-F', iotests.imgfmt, mid_img_path)
+diff --git a/util/trace-events b/util/trace-events
--    assert qemu_io_silent(mid_img_path, '-c', 'write -P 3 2M 1M') == 0
+index XXXXXXX..XXXXXXX 100644
--    assert qemu_io_silent(mid_img_path, '-c', 'write -P 3 4M 1M') == 0
+--- a/util/trace-events
-+    qemu_io(mid_img_path, '-c', 'write -P 3 2M 1M')
++++ b/util/trace-events
-+    qemu_io(mid_img_path, '-c', 'write -P 3 4M 1M')
+@@ -XXX,XX +XXX,XX @@
-     qemu_img('create', '-f', iotests.imgfmt, '-b', mid_img_path,
+ # See docs/tracing.txt for syntax documentation.
-              '-F', iotests.imgfmt, top_img_path)
--    assert qemu_io_silent(top_img_path, '-c', 'write -P 2 1M 1M') == 0
++# util/aio-posix.c
-+    qemu_io(top_img_path, '-c', 'write -P 2 1M 1M')
++run_poll_handlers_begin(void *ctx, int64_t max_ns) "ctx %p max_ns %"PRId64
++run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
- #      0 1 2 3 4
++poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
- # top    2
++poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
-@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
++
-     # Detach backing to check that we can read the data from the top level now
++# util/thread-pool.c
-     qemu_img('rebase', '-u', '-b', '', '-f', iotests.imgfmt, top_img_path)
++thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
++thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
--    assert qemu_io_silent(top_img_path, '-c', 'read -P 0 0 1M') == 0
++thread_pool_cancel(void *req, void *opaque) "req %p opaque %p"
--    assert qemu_io_silent(top_img_path, '-c', 'read -P 2 1M 1M') == 0
++
--    assert qemu_io_silent(top_img_path, '-c', 'read -P 3 2M 1M') == 0
+ # util/buffer.c
--    assert qemu_io_silent(top_img_path, '-c', 'read -P 0 3M 1M') == 0
+ buffer_resize(const char *buf, size_t olen, size_t len) "%s: old %zd, new %zd"
--    assert qemu_io_silent(top_img_path, '-c', 'read -P 3 4M 1M') == 0
+ buffer_move_empty(const char *buf, size_t len, const char *from) "%s: %zd bytes from %s"
 +    qemu_io(top_img_path, '-c', 'read -P 0 0 1M')
 +    qemu_io(top_img_path, '-c', 'read -P 2 1M 1M')
 +    qemu_io(top_img_path, '-c', 'read -P 3 2M 1M')
 +    qemu_io(top_img_path, '-c', 'read -P 0 3M 1M')
 +    qemu_io(top_img_path, '-c', 'read -P 3 4M 1M')
      log('Done')
 diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
 index XXXXXXX..XXXXXXX 100644
 --- a/tests/qemu-iotests/iotests.py
 +++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
      log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
      return result
 -def qemu_io_silent(*args):
 -    '''Run qemu-io and return the exit code, suppressing stdout'''
 -    args = qemu_io_wrap_args(args)
 -    result = subprocess.run(args, stdout=subprocess.DEVNULL, check=False)
 -    if result.returncode < 0:
 -        sys.stderr.write('qemu-io received signal %i: %s\n' %
 -                         (-result.returncode, ' '.join(args)))
 -    return result.returncode
 -
 -def qemu_io_silent_check(*args):
 -    '''Run qemu-io and return the true if subprocess returned 0'''
 -    args = qemu_io_wrap_args(args)
 -    result = subprocess.run(args, stdout=subprocess.DEVNULL,
 -                            stderr=subprocess.STDOUT, check=False)
 -    return result.returncode == 0
 -
  class QemuIoInteractive:
      def __init__(self, *args):
          self.args = qemu_io_wrap_args(args)
 diff --git a/tests/qemu-iotests/tests/image-fleecing b/tests/qemu-iotests/tests/image-fleecing
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/tests/image-fleecing
 +++ b/tests/qemu-iotests/tests/image-fleecing
@@ -XXX,XX +XXX,XX @@
  from subprocess import CalledProcessError
  import iotests
 -from iotests import log, qemu_img, qemu_io, qemu_io_silent
 +from iotests import log, qemu_img, qemu_io
  iotests.script_initialize(
      supported_fmts=['qcow2'],
@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
      for p in overwrite + remainder:
          cmd = 'read -P%s %s %s' % p
          log(cmd)
 -        assert qemu_io_silent(base_img_path, '-c', cmd) == 0
 +        qemu_io(base_img_path, '-c', cmd)
      log('')
      log('Done')
 diff --git a/tests/qemu-iotests/tests/mirror-ready-cancel-error b/tests/qemu-iotests/tests/mirror-ready-cancel-error
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/tests/mirror-ready-cancel-error
 +++ b/tests/qemu-iotests/tests/mirror-ready-cancel-error
@@ -XXX,XX +XXX,XX @@ class TestMirrorReadyCancelError(iotests.QMPTestCase):
          # Ensure that mirror will copy something before READY so the
          # target format layer will forward the pre-READY flush to its
          # file child
 -        assert iotests.qemu_io_silent('-c', 'write -P 1 0 64k', source) == 0
 +        iotests.qemu_io('-c', 'write -P 1 0 64k', source)
          self.vm = iotests.VM()
          self.vm.launch()
 diff --git a/tests/qemu-iotests/tests/stream-error-on-reset b/tests/qemu-iotests/tests/stream-error-on-reset
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/tests/stream-error-on-reset
 +++ b/tests/qemu-iotests/tests/stream-error-on-reset
@@ -XXX,XX +XXX,XX @@
  import os
  import iotests
 -from iotests import imgfmt, qemu_img_create, qemu_io_silent, QMPTestCase
 +from iotests import imgfmt, qemu_img_create, qemu_io, QMPTestCase
  image_size = 1 * 1024 * 1024
@@ -XXX,XX +XXX,XX @@ class TestStreamErrorOnReset(QMPTestCase):
          - top image is attached to a virtio-scsi device
          """
          qemu_img_create('-f', imgfmt, base, str(image_size))
 -        assert qemu_io_silent('-c', f'write 0 {data_size}', base) == 0
 +        qemu_io('-c', f'write 0 {data_size}', base)
          qemu_img_create('-f', imgfmt, top, str(image_size))
          self.vm = iotests.VM()
 --
-.35.1
+.9.3

-New patch
+[Qemu-devel] [PULL v2 02/24] aio: introduce aio_co_schedule and aio_co_wake
+From: Paolo Bonzini <pbonzini@redhat.com>
 aio_co_wake provides the infrastructure to start a coroutine on a "home"
 AioContext.  It will be used by CoMutex and CoQueue, so that coroutines
 don't jump from one context to another when they go to sleep on a
 mutex or waitqueue.  However, it can also be used as a more efficient
 alternative to one-shot bottom halves, and saves the effort of tracking
 which AioContext a coroutine is running on.
 aio_co_schedule is the part of aio_co_wake that starts a coroutine
 on a remove AioContext, but it is also useful to implement e.g.
 bdrv_set_aio_context callbacks.
 The implementation of aio_co_schedule is based on a lock-free
 multiple-producer, single-consumer queue.  The multiple producers use
 cmpxchg to add to a LIFO stack.  The consumer (a per-AioContext bottom
 half) grabs all items added so far, inverts the list to make it FIFO,
 and goes through it one item at a time until it's empty.  The data
 structure was inspired by OSv, which uses it in the very code we'll
 "port" to QEMU for the thread-safe CoMutex.
 Most of the new code is really tests.
 Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Message-id: 20170213135235.12274-3-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  tests/Makefile.include       |   8 +-
  include/block/aio.h          |  32 +++++++
  include/qemu/coroutine_int.h |  11 ++-
  tests/iothread.h             |  25 +++++
  tests/iothread.c             |  91 ++++++++++++++++++
  tests/test-aio-multithread.c | 213 +++++++++++++++++++++++++++++++++++++++++++
  util/async.c                 |  65 +++++++++++++
  util/qemu-coroutine.c        |   8 ++
  util/trace-events            |   4 +
 files changed, 453 insertions(+), 4 deletions(-)
  create mode 100644 tests/iothread.h
  create mode 100644 tests/iothread.c
  create mode 100644 tests/test-aio-multithread.c
 diff --git a/tests/Makefile.include b/tests/Makefile.include
 index XXXXXXX..XXXXXXX 100644
 --- a/tests/Makefile.include
 +++ b/tests/Makefile.include
@@ -XXX,XX +XXX,XX @@ check-unit-y += tests/test-aio$(EXESUF)
  gcov-files-test-aio-y = util/async.c util/qemu-timer.o
  gcov-files-test-aio-$(CONFIG_WIN32) += util/aio-win32.c
  gcov-files-test-aio-$(CONFIG_POSIX) += util/aio-posix.c
 +check-unit-y += tests/test-aio-multithread$(EXESUF)
 +gcov-files-test-aio-multithread-y = $(gcov-files-test-aio-y)
 +gcov-files-test-aio-multithread-y += util/qemu-coroutine.c tests/iothread.c
  check-unit-y += tests/test-throttle$(EXESUF)
 -gcov-files-test-aio-$(CONFIG_WIN32) = aio-win32.c
 -gcov-files-test-aio-$(CONFIG_POSIX) = aio-posix.c
  check-unit-y += tests/test-thread-pool$(EXESUF)
  gcov-files-test-thread-pool-y = thread-pool.c
  gcov-files-test-hbitmap-y = util/hbitmap.c
@@ -XXX,XX +XXX,XX @@ test-qapi-obj-y = tests/test-qapi-visit.o tests/test-qapi-types.o \
      $(test-qom-obj-y)
  test-crypto-obj-y = $(crypto-obj-y) $(test-qom-obj-y)
  test-io-obj-y = $(io-obj-y) $(test-crypto-obj-y)
 -test-block-obj-y = $(block-obj-y) $(test-io-obj-y)
 +test-block-obj-y = $(block-obj-y) $(test-io-obj-y) tests/iothread.o
  tests/check-qint$(EXESUF): tests/check-qint.o $(test-util-obj-y)
  tests/check-qstring$(EXESUF): tests/check-qstring.o $(test-util-obj-y)
@@ -XXX,XX +XXX,XX @@ tests/check-qom-proplist$(EXESUF): tests/check-qom-proplist.o $(test-qom-obj-y)
  tests/test-char$(EXESUF): tests/test-char.o $(test-util-obj-y) $(qtest-obj-y) $(test-io-obj-y) $(chardev-obj-y)
  tests/test-coroutine$(EXESUF): tests/test-coroutine.o $(test-block-obj-y)
  tests/test-aio$(EXESUF): tests/test-aio.o $(test-block-obj-y)
 +tests/test-aio-multithread$(EXESUF): tests/test-aio-multithread.o $(test-block-obj-y)
  tests/test-throttle$(EXESUF): tests/test-throttle.o $(test-block-obj-y)
  tests/test-blockjob$(EXESUF): tests/test-blockjob.o $(test-block-obj-y) $(test-util-obj-y)
  tests/test-blockjob-txn$(EXESUF): tests/test-blockjob-txn.o $(test-block-obj-y) $(test-util-obj-y)
 diff --git a/include/block/aio.h b/include/block/aio.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/block/aio.h
 +++ b/include/block/aio.h
@@ -XXX,XX +XXX,XX @@ typedef void QEMUBHFunc(void *opaque);
  typedef bool AioPollFn(void *opaque);
  typedef void IOHandler(void *opaque);
 +struct Coroutine;
  struct ThreadPool;
  struct LinuxAioState;
@@ -XXX,XX +XXX,XX @@ struct AioContext {
      bool notified;
      EventNotifier notifier;
 +    QSLIST_HEAD(, Coroutine) scheduled_coroutines;
 +    QEMUBH *co_schedule_bh;
 +
      /* Thread pool for performing work and receiving completion callbacks.
       * Has its own locking.
       */
@@ -XXX,XX +XXX,XX @@ static inline bool aio_node_check(AioContext *ctx, bool is_external)
  }
  /**
 + * aio_co_schedule:
 + * @ctx: the aio context
 + * @co: the coroutine
 + *
 + * Start a coroutine on a remote AioContext.
 + *
 + * The coroutine must not be entered by anyone else while aio_co_schedule()
 + * is active.  In addition the coroutine must have yielded unless ctx
 + * is the context in which the coroutine is running (i.e. the value of
 + * qemu_get_current_aio_context() from the coroutine itself).
 + */
 +void aio_co_schedule(AioContext *ctx, struct Coroutine *co);
 +
 +/**
 + * aio_co_wake:
 + * @co: the coroutine
 + *
 + * Restart a coroutine on the AioContext where it was running last, thus
 + * preventing coroutines from jumping from one context to another when they
 + * go to sleep.
 + *
 + * aio_co_wake may be executed either in coroutine or non-coroutine
 + * context.  The coroutine must not be entered by anyone else while
 + * aio_co_wake() is active.
 + */
 +void aio_co_wake(struct Coroutine *co);
 +
 +/**
   * Return the AioContext whose event loop runs in the current thread.
   *
   * If called from an IOThread this will be the IOThread's AioContext.  If
 diff --git a/include/qemu/coroutine_int.h b/include/qemu/coroutine_int.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/qemu/coroutine_int.h
 +++ b/include/qemu/coroutine_int.h
@@ -XXX,XX +XXX,XX @@ struct Coroutine {
      CoroutineEntry *entry;
      void *entry_arg;
      Coroutine *caller;
 +
 +    /* Only used when the coroutine has terminated.  */
      QSLIST_ENTRY(Coroutine) pool_next;
 +
      size_t locks_held;
 -    /* Coroutines that should be woken up when we yield or terminate */
 +    /* Coroutines that should be woken up when we yield or terminate.
 +     * Only used when the coroutine is running.
 +     */
      QSIMPLEQ_HEAD(, Coroutine) co_queue_wakeup;
 +
 +    /* Only used when the coroutine has yielded.  */
 +    AioContext *ctx;
      QSIMPLEQ_ENTRY(Coroutine) co_queue_next;
 +    QSLIST_ENTRY(Coroutine) co_scheduled_next;
  };
  Coroutine *qemu_coroutine_new(void);
 diff --git a/tests/iothread.h b/tests/iothread.h
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/tests/iothread.h
@@ -XXX,XX +XXX,XX @@
 +/*
 + * Event loop thread implementation for unit tests
 + *
 + * Copyright Red Hat Inc., 2013, 2016
 + *
 + * Authors:
 + *  Stefan Hajnoczi   <stefanha@redhat.com>
 + *  Paolo Bonzini     <pbonzini@redhat.com>
 + *
 + * This work is licensed under the terms of the GNU GPL, version 2 or later.
 + * See the COPYING file in the top-level directory.
 + */
 +#ifndef TEST_IOTHREAD_H
 +#define TEST_IOTHREAD_H
 +
 +#include "block/aio.h"
 +#include "qemu/thread.h"
 +
 +typedef struct IOThread IOThread;
 +
 +IOThread *iothread_new(void);
 +void iothread_join(IOThread *iothread);
 +AioContext *iothread_get_aio_context(IOThread *iothread);
 +
 +#endif
 diff --git a/tests/iothread.c b/tests/iothread.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/tests/iothread.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * Event loop thread implementation for unit tests
 + *
 + * Copyright Red Hat Inc., 2013, 2016
 + *
 + * Authors:
 + *  Stefan Hajnoczi   <stefanha@redhat.com>
 + *  Paolo Bonzini     <pbonzini@redhat.com>
 + *
 + * This work is licensed under the terms of the GNU GPL, version 2 or later.
 + * See the COPYING file in the top-level directory.
 + *
 + */
 +
 +#include "qemu/osdep.h"
 +#include "qapi/error.h"
 +#include "block/aio.h"
 +#include "qemu/main-loop.h"
 +#include "qemu/rcu.h"
 +#include "iothread.h"
 +
 +struct IOThread {
 +    AioContext *ctx;
 +
 +    QemuThread thread;
 +    QemuMutex init_done_lock;
 +    QemuCond init_done_cond;    /* is thread initialization done? */
 +    bool stopping;
 +};
 +
 +static __thread IOThread *my_iothread;
 +
 +AioContext *qemu_get_current_aio_context(void)
 +{
 +    return my_iothread ? my_iothread->ctx : qemu_get_aio_context();
 +}
 +
 +static void *iothread_run(void *opaque)
 +{
 +    IOThread *iothread = opaque;
 +
 +    rcu_register_thread();
 +
 +    my_iothread = iothread;
 +    qemu_mutex_lock(&iothread->init_done_lock);
 +    iothread->ctx = aio_context_new(&error_abort);
 +    qemu_cond_signal(&iothread->init_done_cond);
 +    qemu_mutex_unlock(&iothread->init_done_lock);
 +
 +    while (!atomic_read(&iothread->stopping)) {
 +        aio_poll(iothread->ctx, true);
 +    }
 +
 +    rcu_unregister_thread();
 +    return NULL;
 +}
 +
 +void iothread_join(IOThread *iothread)
 +{
 +    iothread->stopping = true;
 +    aio_notify(iothread->ctx);
 +    qemu_thread_join(&iothread->thread);
 +    qemu_cond_destroy(&iothread->init_done_cond);
 +    qemu_mutex_destroy(&iothread->init_done_lock);
 +    aio_context_unref(iothread->ctx);
 +    g_free(iothread);
 +}
 +
 +IOThread *iothread_new(void)
 +{
 +    IOThread *iothread = g_new0(IOThread, 1);
 +
 +    qemu_mutex_init(&iothread->init_done_lock);
 +    qemu_cond_init(&iothread->init_done_cond);
 +    qemu_thread_create(&iothread->thread, NULL, iothread_run,
 +                       iothread, QEMU_THREAD_JOINABLE);
 +
 +    /* Wait for initialization to complete */
 +    qemu_mutex_lock(&iothread->init_done_lock);
 +    while (iothread->ctx == NULL) {
 +        qemu_cond_wait(&iothread->init_done_cond,
 +                       &iothread->init_done_lock);
 +    }
 +    qemu_mutex_unlock(&iothread->init_done_lock);
 +    return iothread;
 +}
 +
 +AioContext *iothread_get_aio_context(IOThread *iothread)
 +{
 +    return iothread->ctx;
 +}
 diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/tests/test-aio-multithread.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * AioContext multithreading tests
 + *
 + * Copyright Red Hat, Inc. 2016
 + *
 + * Authors:
 + *  Paolo Bonzini    <pbonzini@redhat.com>
 + *
 + * This work is licensed under the terms of the GNU LGPL, version 2 or later.
 + * See the COPYING.LIB file in the top-level directory.
 + */
 +
 +#include "qemu/osdep.h"
 +#include <glib.h>
 +#include "block/aio.h"
 +#include "qapi/error.h"
 +#include "qemu/coroutine.h"
 +#include "qemu/thread.h"
 +#include "qemu/error-report.h"
 +#include "iothread.h"
 +
 +/* AioContext management */
 +
 +#define NUM_CONTEXTS 5
 +
 +static IOThread *threads[NUM_CONTEXTS];
 +static AioContext *ctx[NUM_CONTEXTS];
 +static __thread int id = -1;
 +
 +static QemuEvent done_event;
 +
 +/* Run a function synchronously on a remote iothread. */
 +
 +typedef struct CtxRunData {
 +    QEMUBHFunc *cb;
 +    void *arg;
 +} CtxRunData;
 +
 +static void ctx_run_bh_cb(void *opaque)
 +{
 +    CtxRunData *data = opaque;
 +
 +    data->cb(data->arg);
 +    qemu_event_set(&done_event);
 +}
 +
 +static void ctx_run(int i, QEMUBHFunc *cb, void *opaque)
 +{
 +    CtxRunData data = {
 +        .cb = cb,
 +        .arg = opaque
 +    };
 +
 +    qemu_event_reset(&done_event);
 +    aio_bh_schedule_oneshot(ctx[i], ctx_run_bh_cb, &data);
 +    qemu_event_wait(&done_event);
 +}
 +
 +/* Starting the iothreads. */
 +
 +static void set_id_cb(void *opaque)
 +{
 +    int *i = opaque;
 +
 +    id = *i;
 +}
 +
 +static void create_aio_contexts(void)
 +{
 +    int i;
 +
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        threads[i] = iothread_new();
 +        ctx[i] = iothread_get_aio_context(threads[i]);
 +    }
 +
 +    qemu_event_init(&done_event, false);
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        ctx_run(i, set_id_cb, &i);
 +    }
 +}
 +
 +/* Stopping the iothreads. */
 +
 +static void join_aio_contexts(void)
 +{
 +    int i;
 +
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        aio_context_ref(ctx[i]);
 +    }
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        iothread_join(threads[i]);
 +    }
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        aio_context_unref(ctx[i]);
 +    }
 +    qemu_event_destroy(&done_event);
 +}
 +
 +/* Basic test for the stuff above. */
 +
 +static void test_lifecycle(void)
 +{
 +    create_aio_contexts();
 +    join_aio_contexts();
 +}
 +
 +/* aio_co_schedule test.  */
 +
 +static Coroutine *to_schedule[NUM_CONTEXTS];
 +
 +static bool now_stopping;
 +
 +static int count_retry;
 +static int count_here;
 +static int count_other;
 +
 +static bool schedule_next(int n)
 +{
 +    Coroutine *co;
 +
 +    co = atomic_xchg(&to_schedule[n], NULL);
 +    if (!co) {
 +        atomic_inc(&count_retry);
 +        return false;
 +    }
 +
 +    if (n == id) {
 +        atomic_inc(&count_here);
 +    } else {
 +        atomic_inc(&count_other);
 +    }
 +
 +    aio_co_schedule(ctx[n], co);
 +    return true;
 +}
 +
 +static void finish_cb(void *opaque)
 +{
 +    schedule_next(id);
 +}
 +
 +static coroutine_fn void test_multi_co_schedule_entry(void *opaque)
 +{
 +    g_assert(to_schedule[id] == NULL);
 +    atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
 +
 +    while (!atomic_mb_read(&now_stopping)) {
 +        int n;
 +
 +        n = g_test_rand_int_range(0, NUM_CONTEXTS);
 +        schedule_next(n);
 +        qemu_coroutine_yield();
 +
 +        g_assert(to_schedule[id] == NULL);
 +        atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
 +    }
 +}
 +
 +
 +static void test_multi_co_schedule(int seconds)
 +{
 +    int i;
 +
 +    count_here = count_other = count_retry = 0;
 +    now_stopping = false;
 +
 +    create_aio_contexts();
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        Coroutine *co1 = qemu_coroutine_create(test_multi_co_schedule_entry, NULL);
 +        aio_co_schedule(ctx[i], co1);
 +    }
 +
 +    g_usleep(seconds * 1000000);
 +
 +    atomic_mb_set(&now_stopping, true);
 +    for (i = 0; i < NUM_CONTEXTS; i++) {
 +        ctx_run(i, finish_cb, NULL);
 +        to_schedule[i] = NULL;
 +    }
 +
 +    join_aio_contexts();
 +    g_test_message("scheduled %d, queued %d, retry %d, total %d\n",
 +                  count_other, count_here, count_retry,
 +                  count_here + count_other + count_retry);
 +}
 +
 +static void test_multi_co_schedule_1(void)
 +{
 +    test_multi_co_schedule(1);
 +}
 +
 +static void test_multi_co_schedule_10(void)
 +{
 +    test_multi_co_schedule(10);
 +}
 +
 +/* End of tests.  */
 +
 +int main(int argc, char **argv)
 +{
 +    init_clocks();
 +
 +    g_test_init(&argc, &argv, NULL);
 +    g_test_add_func("/aio/multi/lifecycle", test_lifecycle);
 +    if (g_test_quick()) {
 +        g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
 +    } else {
 +        g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
 +    }
 +    return g_test_run();
 +}
 diff --git a/util/async.c b/util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@
  #include "qemu/main-loop.h"
  #include "qemu/atomic.h"
  #include "block/raw-aio.h"
 +#include "qemu/coroutine_int.h"
 +#include "trace.h"
  /***********************************************************/
  /* bottom halves (can be seen as timers which expire ASAP) */
@@ -XXX,XX +XXX,XX @@ aio_ctx_finalize(GSource     *source)
      }
  #endif
 +    assert(QSLIST_EMPTY(&ctx->scheduled_coroutines));
 +    qemu_bh_delete(ctx->co_schedule_bh);
 +
      qemu_lockcnt_lock(&ctx->list_lock);
      assert(!qemu_lockcnt_count(&ctx->list_lock));
      while (ctx->first_bh) {
@@ -XXX,XX +XXX,XX @@ static bool event_notifier_poll(void *opaque)
      return atomic_read(&ctx->notified);
  }
 +static void co_schedule_bh_cb(void *opaque)
 +{
 +    AioContext *ctx = opaque;
 +    QSLIST_HEAD(, Coroutine) straight, reversed;
 +
 +    QSLIST_MOVE_ATOMIC(&reversed, &ctx->scheduled_coroutines);
 +    QSLIST_INIT(&straight);
 +
 +    while (!QSLIST_EMPTY(&reversed)) {
 +        Coroutine *co = QSLIST_FIRST(&reversed);
 +        QSLIST_REMOVE_HEAD(&reversed, co_scheduled_next);
 +        QSLIST_INSERT_HEAD(&straight, co, co_scheduled_next);
 +    }
 +
 +    while (!QSLIST_EMPTY(&straight)) {
 +        Coroutine *co = QSLIST_FIRST(&straight);
 +        QSLIST_REMOVE_HEAD(&straight, co_scheduled_next);
 +        trace_aio_co_schedule_bh_cb(ctx, co);
 +        qemu_coroutine_enter(co);
 +    }
 +}
 +
  AioContext *aio_context_new(Error **errp)
  {
      int ret;
@@ -XXX,XX +XXX,XX @@ AioContext *aio_context_new(Error **errp)
      }
      g_source_set_can_recurse(&ctx->source, true);
      qemu_lockcnt_init(&ctx->list_lock);
 +
 +    ctx->co_schedule_bh = aio_bh_new(ctx, co_schedule_bh_cb, ctx);
 +    QSLIST_INIT(&ctx->scheduled_coroutines);
 +
      aio_set_event_notifier(ctx, &ctx->notifier,
                             false,
                             (EventNotifierHandler *)
@@ -XXX,XX +XXX,XX @@ fail:
      return NULL;
  }
 +void aio_co_schedule(AioContext *ctx, Coroutine *co)
 +{
 +    trace_aio_co_schedule(ctx, co);
 +    QSLIST_INSERT_HEAD_ATOMIC(&ctx->scheduled_coroutines,
 +                              co, co_scheduled_next);
 +    qemu_bh_schedule(ctx->co_schedule_bh);
 +}
 +
 +void aio_co_wake(struct Coroutine *co)
 +{
 +    AioContext *ctx;
 +
 +    /* Read coroutine before co->ctx.  Matches smp_wmb in
 +     * qemu_coroutine_enter.
 +     */
 +    smp_read_barrier_depends();
 +    ctx = atomic_read(&co->ctx);
 +
 +    if (ctx != qemu_get_current_aio_context()) {
 +        aio_co_schedule(ctx, co);
 +        return;
 +    }
 +
 +    if (qemu_in_coroutine()) {
 +        Coroutine *self = qemu_coroutine_self();
 +        assert(self != co);
 +        QSIMPLEQ_INSERT_TAIL(&self->co_queue_wakeup, co, co_queue_next);
 +    } else {
 +        aio_context_acquire(ctx);
 +        qemu_coroutine_enter(co);
 +        aio_context_release(ctx);
 +    }
 +}
 +
  void aio_context_ref(AioContext *ctx)
  {
      g_source_ref(&ctx->source);
 diff --git a/util/qemu-coroutine.c b/util/qemu-coroutine.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/qemu-coroutine.c
 +++ b/util/qemu-coroutine.c
@@ -XXX,XX +XXX,XX @@
  #include "qemu/atomic.h"
  #include "qemu/coroutine.h"
  #include "qemu/coroutine_int.h"
 +#include "block/aio.h"
  enum {
      POOL_BATCH_SIZE = 64,
@@ -XXX,XX +XXX,XX @@ void qemu_coroutine_enter(Coroutine *co)
      }
      co->caller = self;
 +    co->ctx = qemu_get_current_aio_context();
 +
 +    /* Store co->ctx before anything that stores co.  Matches
 +     * barrier in aio_co_wake.
 +     */
 +    smp_wmb();
 +
      ret = qemu_coroutine_switch(self, co, COROUTINE_ENTER);
      qemu_co_queue_run_restart(co);
 diff --git a/util/trace-events b/util/trace-events
 index XXXXXXX..XXXXXXX 100644
 --- a/util/trace-events
 +++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@ run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
  poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
  poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
 +# util/async.c
 +aio_co_schedule(void *ctx, void *co) "ctx %p co %p"
 +aio_co_schedule_bh_cb(void *ctx, void *co) "ctx %p co %p"
 +
  # util/thread-pool.c
  thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
  thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
 --
 .9.3

-New patch
+[Qemu-devel] [PULL v2 03/24] block-backend: allow blk_prw from coroutine context
+From: Paolo Bonzini <pbonzini@redhat.com>
+qcow2_create2 calls this.  Do not run a nested event loop, as that
+breaks when aio_co_wake tries to queue the coroutine on the co_queue_wakeup
+list of the currently running one.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Message-id: 20170213135235.12274-4-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ block/block-backend.c | 12 ++++++++----
+file changed, 8 insertions(+), 4 deletions(-)
+diff --git a/block/block-backend.c b/block/block-backend.c
+index XXXXXXX..XXXXXXX 100644
+--- a/block/block-backend.c
++++ b/block/block-backend.c
+@@ -XXX,XX +XXX,XX @@ static int blk_prw(BlockBackend *blk, int64_t offset, uint8_t *buf,
+ {
+     QEMUIOVector qiov;
+     struct iovec iov;
+-    Coroutine *co;
+     BlkRwCo rwco;
+     iov = (struct iovec) {
+@@ -XXX,XX +XXX,XX @@ static int blk_prw(BlockBackend *blk, int64_t offset, uint8_t *buf,
+         .ret    = NOT_DONE,
+     };
+-    co = qemu_coroutine_create(co_entry, &rwco);
+-    qemu_coroutine_enter(co);
+-    BDRV_POLL_WHILE(blk_bs(blk), rwco.ret == NOT_DONE);
++    if (qemu_in_coroutine()) {
++        /* Fast-path if already in coroutine context */
++        co_entry(&rwco);
++    } else {
++        Coroutine *co = qemu_coroutine_create(co_entry, &rwco);
++        qemu_coroutine_enter(co);
++        BDRV_POLL_WHILE(blk_bs(blk), rwco.ret == NOT_DONE);
++    }
+     return rwco.ret;
+ }
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 04/24] test-thread-pool: use generic AioContext infrastructure
+From: Paolo Bonzini <pbonzini@redhat.com>
+Once the thread pool starts using aio_co_wake, it will also need
+qemu_get_current_aio_context().  Make test-thread-pool create
+an AioContext with qemu_init_main_loop, so that stubs/iothread.c
+and tests/iothread.c can provide the rest.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Message-id: 20170213135235.12274-5-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ tests/test-thread-pool.c | 12 +++---------
+file changed, 3 insertions(+), 9 deletions(-)
+diff --git a/tests/test-thread-pool.c b/tests/test-thread-pool.c
+index XXXXXXX..XXXXXXX 100644
+--- a/tests/test-thread-pool.c
++++ b/tests/test-thread-pool.c
+@@ -XXX,XX +XXX,XX @@
+ #include "qapi/error.h"
+ #include "qemu/timer.h"
+ #include "qemu/error-report.h"
++#include "qemu/main-loop.h"
+ static AioContext *ctx;
+ static ThreadPool *pool;
+@@ -XXX,XX +XXX,XX @@ static void test_cancel_async(void)
+ int main(int argc, char **argv)
+ {
+     int ret;
+-    Error *local_error = NULL;
+-    init_clocks();
+-
+-    ctx = aio_context_new(&local_error);
+-    if (!ctx) {
+-        error_reportf_err(local_error, "Failed to create AIO Context: ");
+-        exit(1);
+-    }
++    qemu_init_main_loop(&error_abort);
++    ctx = qemu_get_current_aio_context();
+     pool = aio_get_thread_pool(ctx);
+     g_test_init(&argc, &argv, NULL);
+@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
+     ret = g_test_run();
+-    aio_context_unref(ctx);
+     return ret;
+ }
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 05/24] io: add methods to set I/O handlers on AioContext
+From: Paolo Bonzini <pbonzini@redhat.com>
 This is in preparation for making qio_channel_yield work on
 AioContexts other than the main one.
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Message-id: 20170213135235.12274-6-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  include/io/channel.h | 25 +++++++++++++++++++++++++
  io/channel-command.c | 13 +++++++++++++
  io/channel-file.c    | 11 +++++++++++
  io/channel-socket.c  | 16 +++++++++++-----
  io/channel-tls.c     | 12 ++++++++++++
  io/channel-watch.c   |  6 ++++++
  io/channel.c         | 11 +++++++++++
 files changed, 89 insertions(+), 5 deletions(-)
 diff --git a/include/io/channel.h b/include/io/channel.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/io/channel.h
 +++ b/include/io/channel.h
@@ -XXX,XX +XXX,XX @@
  #include "qemu-common.h"
  #include "qom/object.h"
 +#include "block/aio.h"
  #define TYPE_QIO_CHANNEL "qio-channel"
  #define QIO_CHANNEL(obj)                                    \
@@ -XXX,XX +XXX,XX @@ struct QIOChannelClass {
                       off_t offset,
                       int whence,
                       Error **errp);
 +    void (*io_set_aio_fd_handler)(QIOChannel *ioc,
 +                                  AioContext *ctx,
 +                                  IOHandler *io_read,
 +                                  IOHandler *io_write,
 +                                  void *opaque);
  };
  /* General I/O handling functions */
@@ -XXX,XX +XXX,XX @@ void qio_channel_yield(QIOChannel *ioc,
  void qio_channel_wait(QIOChannel *ioc,
                        GIOCondition condition);
 +/**
 + * qio_channel_set_aio_fd_handler:
 + * @ioc: the channel object
 + * @ctx: the AioContext to set the handlers on
 + * @io_read: the read handler
 + * @io_write: the write handler
 + * @opaque: the opaque value passed to the handler
 + *
 + * This is used internally by qio_channel_yield().  It can
 + * be used by channel implementations to forward the handlers
 + * to another channel (e.g. from #QIOChannelTLS to the
 + * underlying socket).
 + */
 +void qio_channel_set_aio_fd_handler(QIOChannel *ioc,
 +                                    AioContext *ctx,
 +                                    IOHandler *io_read,
 +                                    IOHandler *io_write,
 +                                    void *opaque);
 +
  #endif /* QIO_CHANNEL_H */
 diff --git a/io/channel-command.c b/io/channel-command.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel-command.c
 +++ b/io/channel-command.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_command_close(QIOChannel *ioc,
  }
 +static void qio_channel_command_set_aio_fd_handler(QIOChannel *ioc,
 +                                                   AioContext *ctx,
 +                                                   IOHandler *io_read,
 +                                                   IOHandler *io_write,
 +                                                   void *opaque)
 +{
 +    QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
 +    aio_set_fd_handler(ctx, cioc->readfd, false, io_read, NULL, NULL, opaque);
 +    aio_set_fd_handler(ctx, cioc->writefd, false, NULL, io_write, NULL, opaque);
 +}
 +
 +
  static GSource *qio_channel_command_create_watch(QIOChannel *ioc,
                                                   GIOCondition condition)
  {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_command_class_init(ObjectClass *klass,
      ioc_klass->io_set_blocking = qio_channel_command_set_blocking;
      ioc_klass->io_close = qio_channel_command_close;
      ioc_klass->io_create_watch = qio_channel_command_create_watch;
 +    ioc_klass->io_set_aio_fd_handler = qio_channel_command_set_aio_fd_handler;
  }
  static const TypeInfo qio_channel_command_info = {
 diff --git a/io/channel-file.c b/io/channel-file.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel-file.c
 +++ b/io/channel-file.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_file_close(QIOChannel *ioc,
  }
 +static void qio_channel_file_set_aio_fd_handler(QIOChannel *ioc,
 +                                                AioContext *ctx,
 +                                                IOHandler *io_read,
 +                                                IOHandler *io_write,
 +                                                void *opaque)
 +{
 +    QIOChannelFile *fioc = QIO_CHANNEL_FILE(ioc);
 +    aio_set_fd_handler(ctx, fioc->fd, false, io_read, io_write, NULL, opaque);
 +}
 +
  static GSource *qio_channel_file_create_watch(QIOChannel *ioc,
                                                GIOCondition condition)
  {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_file_class_init(ObjectClass *klass,
      ioc_klass->io_seek = qio_channel_file_seek;
      ioc_klass->io_close = qio_channel_file_close;
      ioc_klass->io_create_watch = qio_channel_file_create_watch;
 +    ioc_klass->io_set_aio_fd_handler = qio_channel_file_set_aio_fd_handler;
  }
  static const TypeInfo qio_channel_file_info = {
 diff --git a/io/channel-socket.c b/io/channel-socket.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel-socket.c
 +++ b/io/channel-socket.c
@@ -XXX,XX +XXX,XX @@ qio_channel_socket_set_blocking(QIOChannel *ioc,
          qemu_set_block(sioc->fd);
      } else {
          qemu_set_nonblock(sioc->fd);
 -#ifdef WIN32
 -        WSAEventSelect(sioc->fd, ioc->event,
 -                       FD_READ | FD_ACCEPT | FD_CLOSE |
 -                       FD_CONNECT | FD_WRITE | FD_OOB);
 -#endif
      }
      return 0;
  }
@@ -XXX,XX +XXX,XX @@ qio_channel_socket_shutdown(QIOChannel *ioc,
      return 0;
  }
 +static void qio_channel_socket_set_aio_fd_handler(QIOChannel *ioc,
 +                                                  AioContext *ctx,
 +                                                  IOHandler *io_read,
 +                                                  IOHandler *io_write,
 +                                                  void *opaque)
 +{
 +    QIOChannelSocket *sioc = QIO_CHANNEL_SOCKET(ioc);
 +    aio_set_fd_handler(ctx, sioc->fd, false, io_read, io_write, NULL, opaque);
 +}
 +
  static GSource *qio_channel_socket_create_watch(QIOChannel *ioc,
                                                  GIOCondition condition)
  {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_socket_class_init(ObjectClass *klass,
      ioc_klass->io_set_cork = qio_channel_socket_set_cork;
      ioc_klass->io_set_delay = qio_channel_socket_set_delay;
      ioc_klass->io_create_watch = qio_channel_socket_create_watch;
 +    ioc_klass->io_set_aio_fd_handler = qio_channel_socket_set_aio_fd_handler;
  }
  static const TypeInfo qio_channel_socket_info = {
 diff --git a/io/channel-tls.c b/io/channel-tls.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel-tls.c
 +++ b/io/channel-tls.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_tls_close(QIOChannel *ioc,
      return qio_channel_close(tioc->master, errp);
  }
 +static void qio_channel_tls_set_aio_fd_handler(QIOChannel *ioc,
 +                                               AioContext *ctx,
 +                                               IOHandler *io_read,
 +                                               IOHandler *io_write,
 +                                               void *opaque)
 +{
 +    QIOChannelTLS *tioc = QIO_CHANNEL_TLS(ioc);
 +
 +    qio_channel_set_aio_fd_handler(tioc->master, ctx, io_read, io_write, opaque);
 +}
 +
  static GSource *qio_channel_tls_create_watch(QIOChannel *ioc,
                                               GIOCondition condition)
  {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_tls_class_init(ObjectClass *klass,
      ioc_klass->io_close = qio_channel_tls_close;
      ioc_klass->io_shutdown = qio_channel_tls_shutdown;
      ioc_klass->io_create_watch = qio_channel_tls_create_watch;
 +    ioc_klass->io_set_aio_fd_handler = qio_channel_tls_set_aio_fd_handler;
  }
  static const TypeInfo qio_channel_tls_info = {
 diff --git a/io/channel-watch.c b/io/channel-watch.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel-watch.c
 +++ b/io/channel-watch.c
@@ -XXX,XX +XXX,XX @@ GSource *qio_channel_create_socket_watch(QIOChannel *ioc,
      GSource *source;
      QIOChannelSocketSource *ssource;
 +#ifdef WIN32
 +    WSAEventSelect(socket, ioc->event,
 +                   FD_READ | FD_ACCEPT | FD_CLOSE |
 +                   FD_CONNECT | FD_WRITE | FD_OOB);
 +#endif
 +
      source = g_source_new(&qio_channel_socket_source_funcs,
                            sizeof(QIOChannelSocketSource));
      ssource = (QIOChannelSocketSource *)source;
 diff --git a/io/channel.c b/io/channel.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel.c
 +++ b/io/channel.c
@@ -XXX,XX +XXX,XX @@ GSource *qio_channel_create_watch(QIOChannel *ioc,
  }
 +void qio_channel_set_aio_fd_handler(QIOChannel *ioc,
 +                                    AioContext *ctx,
 +                                    IOHandler *io_read,
 +                                    IOHandler *io_write,
 +                                    void *opaque)
 +{
 +    QIOChannelClass *klass = QIO_CHANNEL_GET_CLASS(ioc);
 +
 +    klass->io_set_aio_fd_handler(ioc, ctx, io_read, io_write, opaque);
 +}
 +
  guint qio_channel_add_watch(QIOChannel *ioc,
                              GIOCondition condition,
                              QIOChannelFunc func,
 --
 .9.3

-[PULL 03/14] iotests/163: Fix broken qemu-io invocation
+[Qemu-devel] [PULL v2 06/24] io: make qio_channel_yield aware of AioContexts
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-The 'read' commands to qemu-io were malformed, and this invocation only
+Support separate coroutines for reading and writing, and place the
-worked by coincidence because the error messages were identical. Oops.
+read/write handlers on the AioContext that the QIOChannel is registered
+with.
-There's no point in checking the patterning of the reference image, so
-just check the empty image by itself instead.
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-(Note: as of this commit, nothing actually enforces that this command
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-completes successfully, but a forthcoming commit in this series will
+Reviewed-by: Fam Zheng <famz@redhat.com>
-enforce that qemu_io() must have a zero status code.)
+Message-id: 20170213135235.12274-7-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: John Snow <jsnow@redhat.com>
 Reviewed-by: Eric Blake <eblake@redhat.com>
 Reviewed-by: Hanna Reitz <hreitz@redhat.com>
 Message-Id: <20220418211504.943969-3-jsnow@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- tests/qemu-iotests/163 | 5 +----
+ include/io/channel.h | 47 ++++++++++++++++++++++++++--
-file changed, 1 insertion(+), 4 deletions(-)
+ io/channel.c         | 86 +++++++++++++++++++++++++++++++++++++++-------------
+files changed, 109 insertions(+), 24 deletions(-)
-diff --git a/tests/qemu-iotests/163 b/tests/qemu-iotests/163
-index XXXXXXX..XXXXXXX 100755
+diff --git a/include/io/channel.h b/include/io/channel.h
---- a/tests/qemu-iotests/163
+index XXXXXXX..XXXXXXX 100644
-+++ b/tests/qemu-iotests/163
+--- a/include/io/channel.h
-@@ -XXX,XX +XXX,XX @@ class ShrinkBaseClass(iotests.QMPTestCase):
++++ b/include/io/channel.h
-         qemu_img('resize',  '-f', iotests.imgfmt, '--shrink', test_img,
+@@ -XXX,XX +XXX,XX @@
-                  self.shrink_size)
+ #include "qemu-common.h"
--        self.assertEqual(
+ #include "qom/object.h"
--            qemu_io('-c', 'read -P 0x00 %s'%self.shrink_size, test_img),
++#include "qemu/coroutine.h"
--            qemu_io('-c', 'read -P 0x00 %s'%self.shrink_size, check_img),
+ #include "block/aio.h"
--            "Verifying image content")
-+        qemu_io('-c', f"read -P 0x00 0 {self.shrink_size}", test_img)
+ #define TYPE_QIO_CHANNEL "qio-channel"
+@@ -XXX,XX +XXX,XX @@ struct QIOChannel {
-         self.image_verify()
+     Object parent;
      unsigned int features; /* bitmask of QIOChannelFeatures */
      char *name;
 +    AioContext *ctx;
 +    Coroutine *read_coroutine;
 +    Coroutine *write_coroutine;
  #ifdef _WIN32
      HANDLE event; /* For use with GSource on Win32 */
  #endif
@@ -XXX,XX +XXX,XX @@ guint qio_channel_add_watch(QIOChannel *ioc,
  /**
 + * qio_channel_attach_aio_context:
 + * @ioc: the channel object
 + * @ctx: the #AioContext to set the handlers on
 + *
 + * Request that qio_channel_yield() sets I/O handlers on
 + * the given #AioContext.  If @ctx is %NULL, qio_channel_yield()
 + * uses QEMU's main thread event loop.
 + *
 + * You can move a #QIOChannel from one #AioContext to another even if
 + * I/O handlers are set for a coroutine.  However, #QIOChannel provides
 + * no synchronization between the calls to qio_channel_yield() and
 + * qio_channel_attach_aio_context().
 + *
 + * Therefore you should first call qio_channel_detach_aio_context()
 + * to ensure that the coroutine is not entered concurrently.  Then,
 + * while the coroutine has yielded, call qio_channel_attach_aio_context(),
 + * and then aio_co_schedule() to place the coroutine on the new
 + * #AioContext.  The calls to qio_channel_detach_aio_context()
 + * and qio_channel_attach_aio_context() should be protected with
 + * aio_context_acquire() and aio_context_release().
 + */
 +void qio_channel_attach_aio_context(QIOChannel *ioc,
 +                                    AioContext *ctx);
 +
 +/**
 + * qio_channel_detach_aio_context:
 + * @ioc: the channel object
 + *
 + * Disable any I/O handlers set by qio_channel_yield().  With the
 + * help of aio_co_schedule(), this allows moving a coroutine that was
 + * paused by qio_channel_yield() to another context.
 + */
 +void qio_channel_detach_aio_context(QIOChannel *ioc);
 +
 +/**
   * qio_channel_yield:
   * @ioc: the channel object
   * @condition: the I/O condition to wait for
   *
 - * Yields execution from the current coroutine until
 - * the condition indicated by @condition becomes
 - * available.
 + * Yields execution from the current coroutine until the condition
 + * indicated by @condition becomes available.  @condition must
 + * be either %G_IO_IN or %G_IO_OUT; it cannot contain both.  In
 + * addition, no two coroutine can be waiting on the same condition
 + * and channel at the same time.
   *
   * This must only be called from coroutine context
   */
 diff --git a/io/channel.c b/io/channel.c
 index XXXXXXX..XXXXXXX 100644
 --- a/io/channel.c
 +++ b/io/channel.c
@@ -XXX,XX +XXX,XX @@
  #include "qemu/osdep.h"
  #include "io/channel.h"
  #include "qapi/error.h"
 -#include "qemu/coroutine.h"
 +#include "qemu/main-loop.h"
  bool qio_channel_has_feature(QIOChannel *ioc,
                               QIOChannelFeature feature)
@@ -XXX,XX +XXX,XX @@ off_t qio_channel_io_seek(QIOChannel *ioc,
  }
 -typedef struct QIOChannelYieldData QIOChannelYieldData;
 -struct QIOChannelYieldData {
 -    QIOChannel *ioc;
 -    Coroutine *co;
 -};
 +static void qio_channel_set_aio_fd_handlers(QIOChannel *ioc);
 +static void qio_channel_restart_read(void *opaque)
 +{
 +    QIOChannel *ioc = opaque;
 +    Coroutine *co = ioc->read_coroutine;
 +
 +    ioc->read_coroutine = NULL;
 +    qio_channel_set_aio_fd_handlers(ioc);
 +    aio_co_wake(co);
 +}
 -static gboolean qio_channel_yield_enter(QIOChannel *ioc,
 -                                        GIOCondition condition,
 -                                        gpointer opaque)
 +static void qio_channel_restart_write(void *opaque)
  {
 -    QIOChannelYieldData *data = opaque;
 -    qemu_coroutine_enter(data->co);
 -    return FALSE;
 +    QIOChannel *ioc = opaque;
 +    Coroutine *co = ioc->write_coroutine;
 +
 +    ioc->write_coroutine = NULL;
 +    qio_channel_set_aio_fd_handlers(ioc);
 +    aio_co_wake(co);
  }
 +static void qio_channel_set_aio_fd_handlers(QIOChannel *ioc)
 +{
 +    IOHandler *rd_handler = NULL, *wr_handler = NULL;
 +    AioContext *ctx;
 +
 +    if (ioc->read_coroutine) {
 +        rd_handler = qio_channel_restart_read;
 +    }
 +    if (ioc->write_coroutine) {
 +        wr_handler = qio_channel_restart_write;
 +    }
 +
 +    ctx = ioc->ctx ? ioc->ctx : iohandler_get_aio_context();
 +    qio_channel_set_aio_fd_handler(ioc, ctx, rd_handler, wr_handler, ioc);
 +}
 +
 +void qio_channel_attach_aio_context(QIOChannel *ioc,
 +                                    AioContext *ctx)
 +{
 +    AioContext *old_ctx;
 +    if (ioc->ctx == ctx) {
 +        return;
 +    }
 +
 +    old_ctx = ioc->ctx ? ioc->ctx : iohandler_get_aio_context();
 +    qio_channel_set_aio_fd_handler(ioc, old_ctx, NULL, NULL, NULL);
 +    ioc->ctx = ctx;
 +    qio_channel_set_aio_fd_handlers(ioc);
 +}
 +
 +void qio_channel_detach_aio_context(QIOChannel *ioc)
 +{
 +    ioc->read_coroutine = NULL;
 +    ioc->write_coroutine = NULL;
 +    qio_channel_set_aio_fd_handlers(ioc);
 +    ioc->ctx = NULL;
 +}
  void coroutine_fn qio_channel_yield(QIOChannel *ioc,
                                      GIOCondition condition)
  {
 -    QIOChannelYieldData data;
 -
      assert(qemu_in_coroutine());
 -    data.ioc = ioc;
 -    data.co = qemu_coroutine_self();
 -    qio_channel_add_watch(ioc,
 -                          condition,
 -                          qio_channel_yield_enter,
 -                          &data,
 -                          NULL);
 +    if (condition == G_IO_IN) {
 +        assert(!ioc->read_coroutine);
 +        ioc->read_coroutine = qemu_coroutine_self();
 +    } else if (condition == G_IO_OUT) {
 +        assert(!ioc->write_coroutine);
 +        ioc->write_coroutine = qemu_coroutine_self();
 +    } else {
 +        abort();
 +    }
 +    qio_channel_set_aio_fd_handlers(ioc);
      qemu_coroutine_yield();
  }
 --
-.35.1
+.9.3

-New patch
+[Qemu-devel] [PULL v2 07/24] nbd: convert to use qio_channel_yield
+From: Paolo Bonzini <pbonzini@redhat.com>
+In the client, read the reply headers from a coroutine, switching the
+read side between the "read header" coroutine and the I/O coroutine that
+reads the body of the reply.
+In the server, if the server can read more requests it will create a new
+"read request" coroutine as soon as a request has been read.  Otherwise,
+the new coroutine is created in nbd_request_put.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Message-id: 20170213135235.12274-8-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ block/nbd-client.h |   2 +-
+ block/nbd-client.c | 117 ++++++++++++++++++++++++-----------------------------
+ nbd/client.c       |   2 +-
+ nbd/common.c       |   9 +----
+ nbd/server.c       |  94 +++++++++++++-----------------------------
+files changed, 83 insertions(+), 141 deletions(-)
+diff --git a/block/nbd-client.h b/block/nbd-client.h
+index XXXXXXX..XXXXXXX 100644
+--- a/block/nbd-client.h
++++ b/block/nbd-client.h
+@@ -XXX,XX +XXX,XX @@ typedef struct NBDClientSession {
+     CoMutex send_mutex;
+     CoQueue free_sema;
+-    Coroutine *send_coroutine;
++    Coroutine *read_reply_co;
+     int in_flight;
+     Coroutine *recv_coroutine[MAX_NBD_REQUESTS];
+diff --git a/block/nbd-client.c b/block/nbd-client.c
+index XXXXXXX..XXXXXXX 100644
+--- a/block/nbd-client.c
++++ b/block/nbd-client.c
+@@ -XXX,XX +XXX,XX @@
+ #define HANDLE_TO_INDEX(bs, handle) ((handle) ^ ((uint64_t)(intptr_t)bs))
+ #define INDEX_TO_HANDLE(bs, index)  ((index)  ^ ((uint64_t)(intptr_t)bs))
+-static void nbd_recv_coroutines_enter_all(NBDClientSession *s)
++static void nbd_recv_coroutines_enter_all(BlockDriverState *bs)
+ {
++    NBDClientSession *s = nbd_get_client_session(bs);
+     int i;
+     for (i = 0; i < MAX_NBD_REQUESTS; i++) {
+@@ -XXX,XX +XXX,XX @@ static void nbd_recv_coroutines_enter_all(NBDClientSession *s)
+             qemu_coroutine_enter(s->recv_coroutine[i]);
+         }
+     }
++    BDRV_POLL_WHILE(bs, s->read_reply_co);
+ }
+ static void nbd_teardown_connection(BlockDriverState *bs)
+@@ -XXX,XX +XXX,XX @@ static void nbd_teardown_connection(BlockDriverState *bs)
+     qio_channel_shutdown(client->ioc,
+                          QIO_CHANNEL_SHUTDOWN_BOTH,
+                          NULL);
+-    nbd_recv_coroutines_enter_all(client);
++    nbd_recv_coroutines_enter_all(bs);
+     nbd_client_detach_aio_context(bs);
+     object_unref(OBJECT(client->sioc));
+@@ -XXX,XX +XXX,XX @@ static void nbd_teardown_connection(BlockDriverState *bs)
+     client->ioc = NULL;
+ }
+-static void nbd_reply_ready(void *opaque)
++static coroutine_fn void nbd_read_reply_entry(void *opaque)
+ {
+-    BlockDriverState *bs = opaque;
+-    NBDClientSession *s = nbd_get_client_session(bs);
++    NBDClientSession *s = opaque;
+     uint64_t i;
+     int ret;
+-    if (!s->ioc) { /* Already closed */
+-        return;
+-    }
+-
+-    if (s->reply.handle == 0) {
+-        /* No reply already in flight.  Fetch a header.  It is possible
+-         * that another thread has done the same thing in parallel, so
+-         * the socket is not readable anymore.
+-         */
++    for (;;) {
++        assert(s->reply.handle == 0);
+         ret = nbd_receive_reply(s->ioc, &s->reply);
+-        if (ret == -EAGAIN) {
+-            return;
+-        }
+         if (ret < 0) {
+-            s->reply.handle = 0;
+-            goto fail;
++            break;
+         }
+-    }
+-    /* There's no need for a mutex on the receive side, because the
+-     * handler acts as a synchronization point and ensures that only
+-     * one coroutine is called until the reply finishes.  */
+-    i = HANDLE_TO_INDEX(s, s->reply.handle);
+-    if (i >= MAX_NBD_REQUESTS) {
+-        goto fail;
+-    }
++        /* There's no need for a mutex on the receive side, because the
++         * handler acts as a synchronization point and ensures that only
++         * one coroutine is called until the reply finishes.
++         */
++        i = HANDLE_TO_INDEX(s, s->reply.handle);
++        if (i >= MAX_NBD_REQUESTS || !s->recv_coroutine[i]) {
++            break;
++        }
+-    if (s->recv_coroutine[i]) {
+-        qemu_coroutine_enter(s->recv_coroutine[i]);
+-        return;
++        /* We're woken up by the recv_coroutine itself.  Note that there
++         * is no race between yielding and reentering read_reply_co.  This
++         * is because:
++         *
++         * - if recv_coroutine[i] runs on the same AioContext, it is only
++         *   entered after we yield
++         *
++         * - if recv_coroutine[i] runs on a different AioContext, reentering
++         *   read_reply_co happens through a bottom half, which can only
++         *   run after we yield.
++         */
++        aio_co_wake(s->recv_coroutine[i]);
++        qemu_coroutine_yield();
+     }
+-
+-fail:
+-    nbd_teardown_connection(bs);
+-}
+-
+-static void nbd_restart_write(void *opaque)
+-{
+-    BlockDriverState *bs = opaque;
+-
+-    qemu_coroutine_enter(nbd_get_client_session(bs)->send_coroutine);
++    s->read_reply_co = NULL;
+ }
+ static int nbd_co_send_request(BlockDriverState *bs,
+@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
+                                QEMUIOVector *qiov)
+ {
+     NBDClientSession *s = nbd_get_client_session(bs);
+-    AioContext *aio_context;
+     int rc, ret, i;
+     qemu_co_mutex_lock(&s->send_mutex);
+@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
+         return -EPIPE;
+     }
+-    s->send_coroutine = qemu_coroutine_self();
+-    aio_context = bdrv_get_aio_context(bs);
+-
+-    aio_set_fd_handler(aio_context, s->sioc->fd, false,
+-                       nbd_reply_ready, nbd_restart_write, NULL, bs);
+     if (qiov) {
+         qio_channel_set_cork(s->ioc, true);
+         rc = nbd_send_request(s->ioc, request);
+@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
+     } else {
+         rc = nbd_send_request(s->ioc, request);
+     }
+-    aio_set_fd_handler(aio_context, s->sioc->fd, false,
+-                       nbd_reply_ready, NULL, NULL, bs);
+-    s->send_coroutine = NULL;
+     qemu_co_mutex_unlock(&s->send_mutex);
+     return rc;
+ }
+@@ -XXX,XX +XXX,XX @@ static void nbd_co_receive_reply(NBDClientSession *s,
+ {
+     int ret;
+-    /* Wait until we're woken up by the read handler.  TODO: perhaps
+-     * peek at the next reply and avoid yielding if it's ours?  */
++    /* Wait until we're woken up by nbd_read_reply_entry.  */
+     qemu_coroutine_yield();
+     *reply = s->reply;
+     if (reply->handle != request->handle ||
+@@ -XXX,XX +XXX,XX @@ static void nbd_coroutine_start(NBDClientSession *s,
+     /* s->recv_coroutine[i] is set as soon as we get the send_lock.  */
+ }
+-static void nbd_coroutine_end(NBDClientSession *s,
++static void nbd_coroutine_end(BlockDriverState *bs,
+                               NBDRequest *request)
+ {
++    NBDClientSession *s = nbd_get_client_session(bs);
+     int i = HANDLE_TO_INDEX(s, request->handle);
++
+     s->recv_coroutine[i] = NULL;
+-    if (s->in_flight-- == MAX_NBD_REQUESTS) {
+-        qemu_co_queue_next(&s->free_sema);
++    s->in_flight--;
++    qemu_co_queue_next(&s->free_sema);
++
++    /* Kick the read_reply_co to get the next reply.  */
++    if (s->read_reply_co) {
++        aio_co_wake(s->read_reply_co);
+     }
+ }
+@@ -XXX,XX +XXX,XX @@ int nbd_client_co_preadv(BlockDriverState *bs, uint64_t offset,
+     } else {
+         nbd_co_receive_reply(client, &request, &reply, qiov);
+     }
+-    nbd_coroutine_end(client, &request);
++    nbd_coroutine_end(bs, &request);
+     return -reply.error;
+ }
+@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pwritev(BlockDriverState *bs, uint64_t offset,
+     } else {
+         nbd_co_receive_reply(client, &request, &reply, NULL);
+     }
+-    nbd_coroutine_end(client, &request);
++    nbd_coroutine_end(bs, &request);
+     return -reply.error;
+ }
+@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset,
+     } else {
+         nbd_co_receive_reply(client, &request, &reply, NULL);
+     }
+-    nbd_coroutine_end(client, &request);
++    nbd_coroutine_end(bs, &request);
+     return -reply.error;
+ }
+@@ -XXX,XX +XXX,XX @@ int nbd_client_co_flush(BlockDriverState *bs)
+     } else {
+         nbd_co_receive_reply(client, &request, &reply, NULL);
+     }
+-    nbd_coroutine_end(client, &request);
++    nbd_coroutine_end(bs, &request);
+     return -reply.error;
+ }
+@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pdiscard(BlockDriverState *bs, int64_t offset, int count)
+     } else {
+         nbd_co_receive_reply(client, &request, &reply, NULL);
+     }
+-    nbd_coroutine_end(client, &request);
++    nbd_coroutine_end(bs, &request);
+     return -reply.error;
+ }
+ void nbd_client_detach_aio_context(BlockDriverState *bs)
+ {
+-    aio_set_fd_handler(bdrv_get_aio_context(bs),
+-                       nbd_get_client_session(bs)->sioc->fd,
+-                       false, NULL, NULL, NULL, NULL);
++    NBDClientSession *client = nbd_get_client_session(bs);
++    qio_channel_detach_aio_context(QIO_CHANNEL(client->sioc));
+ }
+ void nbd_client_attach_aio_context(BlockDriverState *bs,
+                                    AioContext *new_context)
+ {
+-    aio_set_fd_handler(new_context, nbd_get_client_session(bs)->sioc->fd,
+-                       false, nbd_reply_ready, NULL, NULL, bs);
++    NBDClientSession *client = nbd_get_client_session(bs);
++    qio_channel_attach_aio_context(QIO_CHANNEL(client->sioc), new_context);
++    aio_co_schedule(new_context, client->read_reply_co);
+ }
+ void nbd_client_close(BlockDriverState *bs)
+@@ -XXX,XX +XXX,XX @@ int nbd_client_init(BlockDriverState *bs,
+     /* Now that we're connected, set the socket to be non-blocking and
+      * kick the reply mechanism.  */
+     qio_channel_set_blocking(QIO_CHANNEL(sioc), false, NULL);
+-
++    client->read_reply_co = qemu_coroutine_create(nbd_read_reply_entry, client);
+     nbd_client_attach_aio_context(bs, bdrv_get_aio_context(bs));
+     logout("Established connection with NBD server\n");
+diff --git a/nbd/client.c b/nbd/client.c
+index XXXXXXX..XXXXXXX 100644
+--- a/nbd/client.c
++++ b/nbd/client.c
+@@ -XXX,XX +XXX,XX @@ ssize_t nbd_receive_reply(QIOChannel *ioc, NBDReply *reply)
+     ssize_t ret;
+     ret = read_sync(ioc, buf, sizeof(buf));
+-    if (ret < 0) {
++    if (ret <= 0) {
+         return ret;
+     }
+diff --git a/nbd/common.c b/nbd/common.c
+index XXXXXXX..XXXXXXX 100644
+--- a/nbd/common.c
++++ b/nbd/common.c
+@@ -XXX,XX +XXX,XX @@ ssize_t nbd_wr_syncv(QIOChannel *ioc,
+         }
+         if (len == QIO_CHANNEL_ERR_BLOCK) {
+             if (qemu_in_coroutine()) {
+-                /* XXX figure out if we can create a variant on
+-                 * qio_channel_yield() that works with AIO contexts
+-                 * and consider using that in this branch */
+-                qemu_coroutine_yield();
+-            } else if (done) {
+-                /* XXX this is needed by nbd_reply_ready.  */
+-                qio_channel_wait(ioc,
+-                                 do_read ? G_IO_IN : G_IO_OUT);
++                qio_channel_yield(ioc, do_read ? G_IO_IN : G_IO_OUT);
+             } else {
+                 return -EAGAIN;
+             }
+diff --git a/nbd/server.c b/nbd/server.c
+index XXXXXXX..XXXXXXX 100644
+--- a/nbd/server.c
++++ b/nbd/server.c
+@@ -XXX,XX +XXX,XX @@ struct NBDClient {
+     CoMutex send_lock;
+     Coroutine *send_coroutine;
+-    bool can_read;
+-
+     QTAILQ_ENTRY(NBDClient) next;
+     int nb_requests;
+     bool closing;
+@@ -XXX,XX +XXX,XX @@ struct NBDClient {
+ /* That's all folks */
+-static void nbd_set_handlers(NBDClient *client);
+-static void nbd_unset_handlers(NBDClient *client);
+-static void nbd_update_can_read(NBDClient *client);
++static void nbd_client_receive_next_request(NBDClient *client);
+ static gboolean nbd_negotiate_continue(QIOChannel *ioc,
+                                        GIOCondition condition,
+@@ -XXX,XX +XXX,XX @@ void nbd_client_put(NBDClient *client)
+          */
+         assert(client->closing);
+-        nbd_unset_handlers(client);
++        qio_channel_detach_aio_context(client->ioc);
+         object_unref(OBJECT(client->sioc));
+         object_unref(OBJECT(client->ioc));
+         if (client->tlscreds) {
+@@ -XXX,XX +XXX,XX @@ static NBDRequestData *nbd_request_get(NBDClient *client)
+     assert(client->nb_requests <= MAX_NBD_REQUESTS - 1);
+     client->nb_requests++;
+-    nbd_update_can_read(client);
+     req = g_new0(NBDRequestData, 1);
+     nbd_client_get(client);
+@@ -XXX,XX +XXX,XX @@ static void nbd_request_put(NBDRequestData *req)
+     g_free(req);
+     client->nb_requests--;
+-    nbd_update_can_read(client);
++    nbd_client_receive_next_request(client);
++
+     nbd_client_put(client);
+ }
+@@ -XXX,XX +XXX,XX @@ static void blk_aio_attached(AioContext *ctx, void *opaque)
+     exp->ctx = ctx;
+     QTAILQ_FOREACH(client, &exp->clients, next) {
+-        nbd_set_handlers(client);
++        qio_channel_attach_aio_context(client->ioc, ctx);
++        if (client->recv_coroutine) {
++            aio_co_schedule(ctx, client->recv_coroutine);
++        }
++        if (client->send_coroutine) {
++            aio_co_schedule(ctx, client->send_coroutine);
++        }
+     }
+ }
+@@ -XXX,XX +XXX,XX @@ static void blk_aio_detach(void *opaque)
+     TRACE("Export %s: Detaching clients from AIO context %p\n", exp->name, exp->ctx);
+     QTAILQ_FOREACH(client, &exp->clients, next) {
+-        nbd_unset_handlers(client);
++        qio_channel_detach_aio_context(client->ioc);
+     }
+     exp->ctx = NULL;
+@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_send_reply(NBDRequestData *req, NBDReply *reply,
+     g_assert(qemu_in_coroutine());
+     qemu_co_mutex_lock(&client->send_lock);
+     client->send_coroutine = qemu_coroutine_self();
+-    nbd_set_handlers(client);
+     if (!len) {
+         rc = nbd_send_reply(client->ioc, reply);
+@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_send_reply(NBDRequestData *req, NBDReply *reply,
+     }
+     client->send_coroutine = NULL;
+-    nbd_set_handlers(client);
+     qemu_co_mutex_unlock(&client->send_lock);
+     return rc;
+ }
+@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_receive_request(NBDRequestData *req,
+     ssize_t rc;
+     g_assert(qemu_in_coroutine());
+-    client->recv_coroutine = qemu_coroutine_self();
+-    nbd_update_can_read(client);
+-
++    assert(client->recv_coroutine == qemu_coroutine_self());
+     rc = nbd_receive_request(client->ioc, request);
+     if (rc < 0) {
+         if (rc != -EAGAIN) {
+@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_receive_request(NBDRequestData *req,
+ out:
+     client->recv_coroutine = NULL;
+-    nbd_update_can_read(client);
++    nbd_client_receive_next_request(client);
+     return rc;
+ }
+-static void nbd_trip(void *opaque)
++/* Owns a reference to the NBDClient passed as opaque.  */
++static coroutine_fn void nbd_trip(void *opaque)
+ {
+     NBDClient *client = opaque;
+     NBDExport *exp = client->exp;
+     NBDRequestData *req;
+-    NBDRequest request;
++    NBDRequest request = { 0 };    /* GCC thinks it can be used uninitialized */
+     NBDReply reply;
+     ssize_t ret;
+     int flags;
+     TRACE("Reading request.");
+     if (client->closing) {
++        nbd_client_put(client);
+         return;
+     }
+@@ -XXX,XX +XXX,XX @@ static void nbd_trip(void *opaque)
+ done:
+     nbd_request_put(req);
++    nbd_client_put(client);
+     return;
+ out:
+     nbd_request_put(req);
+     client_close(client);
++    nbd_client_put(client);
+ }
+-static void nbd_read(void *opaque)
++static void nbd_client_receive_next_request(NBDClient *client)
+ {
+-    NBDClient *client = opaque;
+-
+-    if (client->recv_coroutine) {
+-        qemu_coroutine_enter(client->recv_coroutine);
+-    } else {
+-        qemu_coroutine_enter(qemu_coroutine_create(nbd_trip, client));
+-    }
+-}
+-
+-static void nbd_restart_write(void *opaque)
+-{
+-    NBDClient *client = opaque;
+-
+-    qemu_coroutine_enter(client->send_coroutine);
+-}
+-
+-static void nbd_set_handlers(NBDClient *client)
+-{
+-    if (client->exp && client->exp->ctx) {
+-        aio_set_fd_handler(client->exp->ctx, client->sioc->fd, true,
+-                           client->can_read ? nbd_read : NULL,
+-                           client->send_coroutine ? nbd_restart_write : NULL,
+-                           NULL, client);
+-    }
+-}
+-
+-static void nbd_unset_handlers(NBDClient *client)
+-{
+-    if (client->exp && client->exp->ctx) {
+-        aio_set_fd_handler(client->exp->ctx, client->sioc->fd, true, NULL,
+-                           NULL, NULL, NULL);
+-    }
+-}
+-
+-static void nbd_update_can_read(NBDClient *client)
+-{
+-    bool can_read = client->recv_coroutine ||
+-                    client->nb_requests < MAX_NBD_REQUESTS;
+-
+-    if (can_read != client->can_read) {
+-        client->can_read = can_read;
+-        nbd_set_handlers(client);
+-
+-        /* There is no need to invoke aio_notify(), since aio_set_fd_handler()
+-         * in nbd_set_handlers() will have taken care of that */
++    if (!client->recv_coroutine && client->nb_requests < MAX_NBD_REQUESTS) {
++        nbd_client_get(client);
++        client->recv_coroutine = qemu_coroutine_create(nbd_trip, client);
++        aio_co_schedule(client->exp->ctx, client->recv_coroutine);
+     }
+ }
+@@ -XXX,XX +XXX,XX @@ static coroutine_fn void nbd_co_client_start(void *opaque)
+         goto out;
+     }
+     qemu_co_mutex_init(&client->send_lock);
+-    nbd_set_handlers(client);
+     if (exp) {
+         QTAILQ_INSERT_TAIL(&exp->clients, client, next);
+     }
++
++    nbd_client_receive_next_request(client);
++
+ out:
+     g_free(data);
+ }
+@@ -XXX,XX +XXX,XX @@ void nbd_client_new(NBDExport *exp,
+     object_ref(OBJECT(client->sioc));
+     client->ioc = QIO_CHANNEL(sioc);
+     object_ref(OBJECT(client->ioc));
+-    client->can_read = true;
+     client->close = close_fn;
+     data->client = client;
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 08/24] coroutine-lock: reschedule coroutine on the AioContext it was running on
+From: Paolo Bonzini <pbonzini@redhat.com>
+As a small step towards the introduction of multiqueue, we want
+coroutines to remain on the same AioContext that started them,
+unless they are moved explicitly with e.g. aio_co_schedule.  This patch
+avoids that coroutines switch AioContext when they use a CoMutex.
+For now it does not make much of a difference, because the CoMutex
+is not thread-safe and the AioContext itself is used to protect the
+CoMutex from concurrent access.  However, this is going to change.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Message-id: 20170213135235.12274-9-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ util/qemu-coroutine-lock.c | 5 ++---
+ util/trace-events          | 1 -
+files changed, 2 insertions(+), 4 deletions(-)
+diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
+index XXXXXXX..XXXXXXX 100644
+--- a/util/qemu-coroutine-lock.c
++++ b/util/qemu-coroutine-lock.c
+@@ -XXX,XX +XXX,XX @@
+ #include "qemu/coroutine.h"
+ #include "qemu/coroutine_int.h"
+ #include "qemu/queue.h"
++#include "block/aio.h"
+ #include "trace.h"
+ void qemu_co_queue_init(CoQueue *queue)
+@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_run_restart(Coroutine *co)
+ static bool qemu_co_queue_do_restart(CoQueue *queue, bool single)
+ {
+-    Coroutine *self = qemu_coroutine_self();
+     Coroutine *next;
+     if (QSIMPLEQ_EMPTY(&queue->entries)) {
+@@ -XXX,XX +XXX,XX @@ static bool qemu_co_queue_do_restart(CoQueue *queue, bool single)
+     while ((next = QSIMPLEQ_FIRST(&queue->entries)) != NULL) {
+         QSIMPLEQ_REMOVE_HEAD(&queue->entries, co_queue_next);
+-        QSIMPLEQ_INSERT_TAIL(&self->co_queue_wakeup, next, co_queue_next);
+-        trace_qemu_co_queue_next(next);
++        aio_co_wake(next);
+         if (single) {
+             break;
+         }
+diff --git a/util/trace-events b/util/trace-events
+index XXXXXXX..XXXXXXX 100644
+--- a/util/trace-events
++++ b/util/trace-events
+@@ -XXX,XX +XXX,XX @@ qemu_coroutine_terminate(void *co) "self %p"
+ # util/qemu-coroutine-lock.c
+ qemu_co_queue_run_restart(void *co) "co %p"
+-qemu_co_queue_next(void *nxt) "next %p"
+ qemu_co_mutex_lock_entry(void *mutex, void *self) "mutex %p self %p"
+ qemu_co_mutex_lock_return(void *mutex, void *self) "mutex %p self %p"
+ qemu_co_mutex_unlock_entry(void *mutex, void *self) "mutex %p self %p"
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 09/24] blkdebug: reschedule coroutine on the AioContext it is running on
+From: Paolo Bonzini <pbonzini@redhat.com>
+Keep the coroutine on the same AioContext.  Without this change,
+there would be a race between yielding the coroutine and reentering it.
+While the race cannot happen now, because the code only runs from a single
+AioContext, this will change with multiqueue support in the block layer.
+While doing the change, replace custom bottom half with aio_co_schedule.
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Message-id: 20170213135235.12274-10-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ block/blkdebug.c | 9 +--------
+file changed, 1 insertion(+), 8 deletions(-)
+diff --git a/block/blkdebug.c b/block/blkdebug.c
+index XXXXXXX..XXXXXXX 100644
+--- a/block/blkdebug.c
++++ b/block/blkdebug.c
+@@ -XXX,XX +XXX,XX @@ out:
+     return ret;
+ }
+-static void error_callback_bh(void *opaque)
+-{
+-    Coroutine *co = opaque;
+-    qemu_coroutine_enter(co);
+-}
+-
+ static int inject_error(BlockDriverState *bs, BlkdebugRule *rule)
+ {
+     BDRVBlkdebugState *s = bs->opaque;
+@@ -XXX,XX +XXX,XX @@ static int inject_error(BlockDriverState *bs, BlkdebugRule *rule)
+     }
+     if (!immediately) {
+-        aio_bh_schedule_oneshot(bdrv_get_aio_context(bs), error_callback_bh,
+-                                qemu_coroutine_self());
++        aio_co_schedule(qemu_get_current_aio_context(), qemu_coroutine_self());
+         qemu_coroutine_yield();
+     }
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 10/24] qed: introduce qed_aio_start_io and qed_aio_next_io_cb
+From: Paolo Bonzini <pbonzini@redhat.com>
+qed_aio_start_io and qed_aio_next_io will not have to acquire/release
+the AioContext, while qed_aio_next_io_cb will.  Split the functionality
+and gain a little type-safety in the process.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Message-id: 20170213135235.12274-11-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ block/qed.c | 39 +++++++++++++++++++++++++--------------
+file changed, 25 insertions(+), 14 deletions(-)
+diff --git a/block/qed.c b/block/qed.c
+index XXXXXXX..XXXXXXX 100644
+--- a/block/qed.c
++++ b/block/qed.c
+@@ -XXX,XX +XXX,XX @@ static CachedL2Table *qed_new_l2_table(BDRVQEDState *s)
+     return l2_table;
+ }
+-static void qed_aio_next_io(void *opaque, int ret);
++static void qed_aio_next_io(QEDAIOCB *acb, int ret);
++
++static void qed_aio_start_io(QEDAIOCB *acb)
++{
++    qed_aio_next_io(acb, 0);
++}
++
++static void qed_aio_next_io_cb(void *opaque, int ret)
++{
++    QEDAIOCB *acb = opaque;
++
++    qed_aio_next_io(acb, ret);
++}
+ static void qed_plug_allocating_write_reqs(BDRVQEDState *s)
+ {
+@@ -XXX,XX +XXX,XX @@ static void qed_unplug_allocating_write_reqs(BDRVQEDState *s)
+     acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
+     if (acb) {
+-        qed_aio_next_io(acb, 0);
++        qed_aio_start_io(acb);
+     }
+ }
+@@ -XXX,XX +XXX,XX @@ static void qed_aio_complete(QEDAIOCB *acb, int ret)
+         QSIMPLEQ_REMOVE_HEAD(&s->allocating_write_reqs, next);
+         acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
+         if (acb) {
+-            qed_aio_next_io(acb, 0);
++            qed_aio_start_io(acb);
+         } else if (s->header.features & QED_F_NEED_CHECK) {
+             qed_start_need_check_timer(s);
+         }
+@@ -XXX,XX +XXX,XX @@ static void qed_commit_l2_update(void *opaque, int ret)
+     acb->request.l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
+     assert(acb->request.l2_table != NULL);
+-    qed_aio_next_io(opaque, ret);
++    qed_aio_next_io(acb, ret);
+ }
+ /**
+@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_l2_update(QEDAIOCB *acb, int ret, uint64_t offset)
+     if (need_alloc) {
+         /* Write out the whole new L2 table */
+         qed_write_l2_table(s, &acb->request, 0, s->table_nelems, true,
+-                            qed_aio_write_l1_update, acb);
++                           qed_aio_write_l1_update, acb);
+     } else {
+         /* Write out only the updated part of the L2 table */
+         qed_write_l2_table(s, &acb->request, index, acb->cur_nclusters, false,
+-                            qed_aio_next_io, acb);
++                           qed_aio_next_io_cb, acb);
+     }
+     return;
+@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_main(void *opaque, int ret)
+     }
+     if (acb->find_cluster_ret == QED_CLUSTER_FOUND) {
+-        next_fn = qed_aio_next_io;
++        next_fn = qed_aio_next_io_cb;
+     } else {
+         if (s->bs->backing) {
+             next_fn = qed_aio_write_flush_before_l2_update;
+@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_alloc(QEDAIOCB *acb, size_t len)
+     if (acb->flags & QED_AIOCB_ZERO) {
+         /* Skip ahead if the clusters are already zero */
+         if (acb->find_cluster_ret == QED_CLUSTER_ZERO) {
+-            qed_aio_next_io(acb, 0);
++            qed_aio_start_io(acb);
+             return;
+         }
+@@ -XXX,XX +XXX,XX @@ static void qed_aio_read_data(void *opaque, int ret,
+     /* Handle zero cluster and backing file reads */
+     if (ret == QED_CLUSTER_ZERO) {
+         qemu_iovec_memset(&acb->cur_qiov, 0, 0, acb->cur_qiov.size);
+-        qed_aio_next_io(acb, 0);
++        qed_aio_start_io(acb);
+         return;
+     } else if (ret != QED_CLUSTER_FOUND) {
+         qed_read_backing_file(s, acb->cur_pos, &acb->cur_qiov,
+-                              &acb->backing_qiov, qed_aio_next_io, acb);
++                              &acb->backing_qiov, qed_aio_next_io_cb, acb);
+         return;
+     }
+     BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
+     bdrv_aio_readv(bs->file, offset / BDRV_SECTOR_SIZE,
+                    &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
+-                   qed_aio_next_io, acb);
++                   qed_aio_next_io_cb, acb);
+     return;
+ err:
+@@ -XXX,XX +XXX,XX @@ err:
+ /**
+  * Begin next I/O or complete the request
+  */
+-static void qed_aio_next_io(void *opaque, int ret)
++static void qed_aio_next_io(QEDAIOCB *acb, int ret)
+ {
+-    QEDAIOCB *acb = opaque;
+     BDRVQEDState *s = acb_to_s(acb);
+     QEDFindClusterFunc *io_fn = (acb->flags & QED_AIOCB_WRITE) ?
+                                 qed_aio_write_data : qed_aio_read_data;
+@@ -XXX,XX +XXX,XX @@ static BlockAIOCB *qed_aio_setup(BlockDriverState *bs,
+     qemu_iovec_init(&acb->cur_qiov, qiov->niov);
+     /* Start request */
+-    qed_aio_next_io(acb, 0);
++    qed_aio_start_io(acb);
+     return &acb->common;
+ }
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 11/24] aio: push aio_context_acquire/release down to dispatching
+From: Paolo Bonzini <pbonzini@redhat.com>
+The AioContext data structures are now protected by list_lock and/or
+they are walked with FOREACH_RCU primitives.  There is no need anymore
+to acquire the AioContext for the entire duration of aio_dispatch.
+Instead, just acquire it before and after invoking the callbacks.
+The next step is then to push it further down.
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
+Message-id: 20170213135235.12274-12-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+---
+ util/aio-posix.c | 25 +++++++++++--------------
+ util/aio-win32.c | 15 +++++++--------
+ util/async.c     |  2 ++
+files changed, 20 insertions(+), 22 deletions(-)
+diff --git a/util/aio-posix.c b/util/aio-posix.c
+index XXXXXXX..XXXXXXX 100644
+--- a/util/aio-posix.c
++++ b/util/aio-posix.c
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
+             (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
+             aio_node_check(ctx, node->is_external) &&
+             node->io_read) {
++            aio_context_acquire(ctx);
+             node->io_read(node->opaque);
++            aio_context_release(ctx);
+             /* aio_notify() does not count as progress */
+             if (node->opaque != &ctx->notifier) {
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
+             (revents & (G_IO_OUT | G_IO_ERR)) &&
+             aio_node_check(ctx, node->is_external) &&
+             node->io_write) {
++            aio_context_acquire(ctx);
+             node->io_write(node->opaque);
++            aio_context_release(ctx);
+             progress = true;
+         }
+@@ -XXX,XX +XXX,XX @@ bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
+     }
+     /* Run our timers */
++    aio_context_acquire(ctx);
+     progress |= timerlistgroup_run_timers(&ctx->tlg);
++    aio_context_release(ctx);
+     return progress;
+ }
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+     int64_t timeout;
+     int64_t start = 0;
+-    aio_context_acquire(ctx);
+-    progress = false;
+-
+     /* aio_notify can avoid the expensive event_notifier_set if
+      * everything (file descriptors, bottom halves, timers) will
+      * be re-evaluated before the next blocking poll().  This is
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         start = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
+     }
+-    if (try_poll_mode(ctx, blocking)) {
+-        progress = true;
+-    } else {
++    aio_context_acquire(ctx);
++    progress = try_poll_mode(ctx, blocking);
++    aio_context_release(ctx);
++
++    if (!progress) {
+         assert(npfd == 0);
+         /* fill pollfds */
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         timeout = blocking ? aio_compute_timeout(ctx) : 0;
+         /* wait until next event */
+-        if (timeout) {
+-            aio_context_release(ctx);
+-        }
+         if (aio_epoll_check_poll(ctx, pollfds, npfd, timeout)) {
+             AioHandler epoll_handler;
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         } else  {
+             ret = qemu_poll_ns(pollfds, npfd, timeout);
+         }
+-        if (timeout) {
+-            aio_context_acquire(ctx);
+-        }
+     }
+     if (blocking) {
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         progress = true;
+     }
+-    aio_context_release(ctx);
+-
+     return progress;
+ }
+diff --git a/util/aio-win32.c b/util/aio-win32.c
+index XXXXXXX..XXXXXXX 100644
+--- a/util/aio-win32.c
++++ b/util/aio-win32.c
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
+             (revents || event_notifier_get_handle(node->e) == event) &&
+             node->io_notify) {
+             node->pfd.revents = 0;
++            aio_context_acquire(ctx);
+             node->io_notify(node->e);
++            aio_context_release(ctx);
+             /* aio_notify() does not count as progress */
+             if (node->e != &ctx->notifier) {
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
+             (node->io_read || node->io_write)) {
+             node->pfd.revents = 0;
+             if ((revents & G_IO_IN) && node->io_read) {
++                aio_context_acquire(ctx);
+                 node->io_read(node->opaque);
++                aio_context_release(ctx);
+                 progress = true;
+             }
+             if ((revents & G_IO_OUT) && node->io_write) {
++                aio_context_acquire(ctx);
+                 node->io_write(node->opaque);
++                aio_context_release(ctx);
+                 progress = true;
+             }
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+     int count;
+     int timeout;
+-    aio_context_acquire(ctx);
+     progress = false;
+     /* aio_notify can avoid the expensive event_notifier_set if
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         timeout = blocking && !have_select_revents
+             ? qemu_timeout_ns_to_ms(aio_compute_timeout(ctx)) : 0;
+-        if (timeout) {
+-            aio_context_release(ctx);
+-        }
+         ret = WaitForMultipleObjects(count, events, FALSE, timeout);
+         if (blocking) {
+             assert(first);
+             atomic_sub(&ctx->notify_me, 2);
+         }
+-        if (timeout) {
+-            aio_context_acquire(ctx);
+-        }
+         if (first) {
+             aio_notify_accept(ctx);
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
+         progress |= aio_dispatch_handlers(ctx, event);
+     } while (count > 0);
++    aio_context_acquire(ctx);
+     progress |= timerlistgroup_run_timers(&ctx->tlg);
+-
+     aio_context_release(ctx);
+     return progress;
+ }
+diff --git a/util/async.c b/util/async.c
+index XXXXXXX..XXXXXXX 100644
+--- a/util/async.c
++++ b/util/async.c
+@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
+                 ret = 1;
+             }
+             bh->idle = 0;
++            aio_context_acquire(ctx);
+             aio_bh_call(bh);
++            aio_context_release(ctx);
+         }
+         if (bh->deleted) {
+             deleted = true;
+--
+.9.3

-New patch
+[Qemu-devel] [PULL v2 12/24] block: explicitly acquire aiocontext in timers that need it
+From: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Message-id: 20170213135235.12274-13-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  block/qed.h                 |  3 +++
  block/curl.c                |  2 ++
  block/io.c                  |  5 +++++
  block/iscsi.c               |  8 ++++++--
  block/null.c                |  4 ++++
  block/qed.c                 | 12 ++++++++++++
  block/throttle-groups.c     |  2 ++
  util/aio-posix.c            |  2 --
  util/aio-win32.c            |  2 --
  util/qemu-coroutine-sleep.c |  2 +-
 files changed, 35 insertions(+), 7 deletions(-)
 diff --git a/block/qed.h b/block/qed.h
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed.h
 +++ b/block/qed.h
@@ -XXX,XX +XXX,XX @@ enum {
   */
  typedef void QEDFindClusterFunc(void *opaque, int ret, uint64_t offset, size_t len);
 +void qed_acquire(BDRVQEDState *s);
 +void qed_release(BDRVQEDState *s);
 +
  /**
   * Generic callback for chaining async callbacks
   */
 diff --git a/block/curl.c b/block/curl.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/curl.c
 +++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_multi_timeout_do(void *arg)
          return;
      }
 +    aio_context_acquire(s->aio_context);
      curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
      curl_multi_check_completion(s);
 +    aio_context_release(s->aio_context);
  #else
      abort();
  #endif
 diff --git a/block/io.c b/block/io.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/io.c
 +++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ void bdrv_aio_cancel(BlockAIOCB *acb)
          if (acb->aiocb_info->get_aio_context) {
              aio_poll(acb->aiocb_info->get_aio_context(acb), true);
          } else if (acb->bs) {
 +            /* qemu_aio_ref and qemu_aio_unref are not thread-safe, so
 +             * assert that we're not using an I/O thread.  Thread-safe
 +             * code should use bdrv_aio_cancel_async exclusively.
 +             */
 +            assert(bdrv_get_aio_context(acb->bs) == qemu_get_aio_context());
              aio_poll(bdrv_get_aio_context(acb->bs), true);
          } else {
              abort();
 diff --git a/block/iscsi.c b/block/iscsi.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/iscsi.c
 +++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void iscsi_retry_timer_expired(void *opaque)
      struct IscsiTask *iTask = opaque;
      iTask->complete = 1;
      if (iTask->co) {
 -        qemu_coroutine_enter(iTask->co);
 +        aio_co_wake(iTask->co);
      }
  }
@@ -XXX,XX +XXX,XX @@ static void iscsi_nop_timed_event(void *opaque)
  {
      IscsiLun *iscsilun = opaque;
 +    aio_context_acquire(iscsilun->aio_context);
      if (iscsi_get_nops_in_flight(iscsilun->iscsi) >= MAX_NOP_FAILURES) {
          error_report("iSCSI: NOP timeout. Reconnecting...");
          iscsilun->request_timed_out = true;
      } else if (iscsi_nop_out_async(iscsilun->iscsi, NULL, NULL, 0, NULL) != 0) {
          error_report("iSCSI: failed to sent NOP-Out. Disabling NOP messages.");
 -        return;
 +        goto out;
      }
      timer_mod(iscsilun->nop_timer, qemu_clock_get_ms(QEMU_CLOCK_REALTIME) + NOP_INTERVAL);
      iscsi_set_events(iscsilun);
 +
 +out:
 +    aio_context_release(iscsilun->aio_context);
  }
  static void iscsi_readcapacity_sync(IscsiLun *iscsilun, Error **errp)
 diff --git a/block/null.c b/block/null.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/null.c
 +++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static void null_bh_cb(void *opaque)
  static void null_timer_cb(void *opaque)
  {
      NullAIOCB *acb = opaque;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 +
 +    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, 0);
 +    aio_context_release(ctx);
      timer_deinit(&acb->timer);
      qemu_aio_unref(acb);
  }
 diff --git a/block/qed.c b/block/qed.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed.c
 +++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_need_check_timer_cb(void *opaque)
      trace_qed_need_check_timer_cb(s);
 +    qed_acquire(s);
      qed_plug_allocating_write_reqs(s);
      /* Ensure writes are on disk before clearing flag */
      bdrv_aio_flush(s->bs->file->bs, qed_clear_need_check, s);
 +    qed_release(s);
 +}
 +
 +void qed_acquire(BDRVQEDState *s)
 +{
 +    aio_context_acquire(bdrv_get_aio_context(s->bs));
 +}
 +
 +void qed_release(BDRVQEDState *s)
 +{
 +    aio_context_release(bdrv_get_aio_context(s->bs));
  }
  static void qed_start_need_check_timer(BDRVQEDState *s)
 diff --git a/block/throttle-groups.c b/block/throttle-groups.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/throttle-groups.c
 +++ b/block/throttle-groups.c
@@ -XXX,XX +XXX,XX @@ static void timer_cb(BlockBackend *blk, bool is_write)
      qemu_mutex_unlock(&tg->lock);
      /* Run the request that was waiting for this timer */
 +    aio_context_acquire(blk_get_aio_context(blk));
      empty_queue = !qemu_co_enter_next(&blkp->throttled_reqs[is_write]);
 +    aio_context_release(blk_get_aio_context(blk));
      /* If the request queue was empty then we have to take care of
       * scheduling the next one */
 diff --git a/util/aio-posix.c b/util/aio-posix.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-posix.c
 +++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
      }
      /* Run our timers */
 -    aio_context_acquire(ctx);
      progress |= timerlistgroup_run_timers(&ctx->tlg);
 -    aio_context_release(ctx);
      return progress;
  }
 diff --git a/util/aio-win32.c b/util/aio-win32.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-win32.c
 +++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
          progress |= aio_dispatch_handlers(ctx, event);
      } while (count > 0);
 -    aio_context_acquire(ctx);
      progress |= timerlistgroup_run_timers(&ctx->tlg);
 -    aio_context_release(ctx);
      return progress;
  }
 diff --git a/util/qemu-coroutine-sleep.c b/util/qemu-coroutine-sleep.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/qemu-coroutine-sleep.c
 +++ b/util/qemu-coroutine-sleep.c
@@ -XXX,XX +XXX,XX @@ static void co_sleep_cb(void *opaque)
  {
      CoSleepCB *sleep_cb = opaque;
 -    qemu_coroutine_enter(sleep_cb->co);
 +    aio_co_wake(sleep_cb->co);
  }
  void coroutine_fn co_aio_sleep_ns(AioContext *ctx, QEMUClockType type,
 --
 .9.3

-[PULL 01/14] block: add 'force' parameter to 'blockdev-change-medium' command
+[Qemu-devel] [PULL v2 13/24] block: explicitly acquire aiocontext in callbacks that need it
-From: "Denis V. Lunev" <den@openvz.org>
+From: Paolo Bonzini <pbonzini@redhat.com>
-'blockdev-change-medium' is a convinient wrapper for the following
+This covers both file descriptor callbacks and polling callbacks,
-sequence of commands:
+since they execute related code.
  * blockdev-open-tray
  * blockdev-remove-medium
  * blockdev-insert-medium
  * blockdev-close-tray
 and should be used f.e. to change ISO image inside the CD-ROM tray.
 Though the guest could lock the tray and some linux guests like
 CentOS 8.5 actually does that. In this case the execution if this
 command results in the error like the following:
   Device 'scsi0-0-1-0' is locked and force was not specified,
   wait for tray to open and try again.
-This situation is could be resolved 'blockdev-open-tray' by passing
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-flag 'force' inside. Thus is seems reasonable to add the same
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-capability for 'blockdev-change-medium' too.
+Reviewed-by: Fam Zheng <famz@redhat.com>
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Message-id: 20170213135235.12274-14-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  block/curl.c          | 16 +++++++++++++---
  block/iscsi.c         |  4 ++++
  block/linux-aio.c     |  4 ++++
  block/nfs.c           |  6 ++++++
  block/sheepdog.c      | 29 +++++++++++++++--------------
  block/ssh.c           | 29 +++++++++--------------------
  block/win32-aio.c     | 10 ++++++----
  hw/block/virtio-blk.c |  5 ++++-
  hw/scsi/virtio-scsi.c |  7 +++++++
  util/aio-posix.c      |  7 -------
  util/aio-win32.c      |  6 ------
 files changed, 68 insertions(+), 55 deletions(-)
-Signed-off-by: Denis V. Lunev <den@openvz.org>
+diff --git a/block/curl.c b/block/curl.c
-Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@openvz.org>
+index XXXXXXX..XXXXXXX 100644
-Acked-by: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
+--- a/block/curl.c
-CC: Kevin Wolf <kwolf@redhat.com>
++++ b/block/curl.c
-CC: Hanna Reitz <hreitz@redhat.com>
+@@ -XXX,XX +XXX,XX @@ static void curl_multi_check_completion(BDRVCURLState *s)
-CC: Eric Blake <eblake@redhat.com>
+     }
-CC: Markus Armbruster <armbru@redhat.com>
+ }
-Message-Id: <20220412221846.280723-1-den@openvz.org>
-Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
+-static void curl_multi_do(void *arg)
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
++static void curl_multi_do_locked(CURLState *s)
----
+ {
- qapi/block.json     |  6 ++++++
+-    CURLState *s = (CURLState *)arg;
- block/qapi-sysemu.c |  3 ++-
+     CURLSocket *socket, *next_socket;
- monitor/hmp-cmds.c  |  4 +++-
+     int running;
- hmp-commands.hx     | 11 +++++++----
+     int r;
- ui/cocoa.m          |  1 +
+@@ -XXX,XX +XXX,XX @@ static void curl_multi_do(void *arg)
-files changed, 19 insertions(+), 6 deletions(-)
+     }
+ }
-diff --git a/qapi/block.json b/qapi/block.json
-index XXXXXXX..XXXXXXX 100644
++static void curl_multi_do(void *arg)
---- a/qapi/block.json
++{
-+++ b/qapi/block.json
++    CURLState *s = (CURLState *)arg;
-@@ -XXX,XX +XXX,XX @@
++
- # @read-only-mode: change the read-only mode of the device; defaults
++    aio_context_acquire(s->s->aio_context);
- #                  to 'retain'
++    curl_multi_do_locked(s);
- #
++    aio_context_release(s->s->aio_context);
-+# @force: if false (the default), an eject request through blockdev-open-tray
++}
-+#         will be sent to the guest if it has locked the tray (and the tray
++
-+#         will not be opened immediately); if true, the tray will be opened
+ static void curl_multi_read(void *arg)
-+#         regardless of whether it is locked. (since 7.1)
+ {
-+#
+     CURLState *s = (CURLState *)arg;
- # Features:
- # @deprecated: Member @device is deprecated.  Use @id instead.
+-    curl_multi_do(arg);
- #
++    aio_context_acquire(s->s->aio_context);
-@@ -XXX,XX +XXX,XX @@
++    curl_multi_do_locked(s);
-             '*id': 'str',
+     curl_multi_check_completion(s->s);
-             'filename': 'str',
++    aio_context_release(s->s->aio_context);
-             '*format': 'str',
+ }
-+            '*force': 'bool',
-             '*read-only-mode': 'BlockdevChangeReadOnlyMode' } }
+ static void curl_multi_timeout_do(void *arg)
+diff --git a/block/iscsi.c b/block/iscsi.c
+index XXXXXXX..XXXXXXX 100644
-diff --git a/block/qapi-sysemu.c b/block/qapi-sysemu.c
+--- a/block/iscsi.c
-index XXXXXXX..XXXXXXX 100644
++++ b/block/iscsi.c
---- a/block/qapi-sysemu.c
+@@ -XXX,XX +XXX,XX @@ iscsi_process_read(void *arg)
-+++ b/block/qapi-sysemu.c
+     IscsiLun *iscsilun = arg;
-@@ -XXX,XX +XXX,XX @@ void qmp_blockdev_change_medium(bool has_device, const char *device,
+     struct iscsi_context *iscsi = iscsilun->iscsi;
-                                 bool has_id, const char *id,
-                                 const char *filename,
++    aio_context_acquire(iscsilun->aio_context);
-                                 bool has_format, const char *format,
+     iscsi_service(iscsi, POLLIN);
-+                                bool has_force, bool force,
+     iscsi_set_events(iscsilun);
-                                 bool has_read_only,
++    aio_context_release(iscsilun->aio_context);
-                                 BlockdevChangeReadOnlyMode read_only,
+ }
-                                 Error **errp)
-@@ -XXX,XX +XXX,XX @@ void qmp_blockdev_change_medium(bool has_device, const char *device,
+ static void
+@@ -XXX,XX +XXX,XX @@ iscsi_process_write(void *arg)
-     rc = do_open_tray(has_device ? device : NULL,
+     IscsiLun *iscsilun = arg;
-                       has_id ? id : NULL,
+     struct iscsi_context *iscsi = iscsilun->iscsi;
--                      false, &err);
-+                      force, &err);
++    aio_context_acquire(iscsilun->aio_context);
-     if (rc && rc != -ENOSYS) {
+     iscsi_service(iscsi, POLLOUT);
-         error_propagate(errp, err);
+     iscsi_set_events(iscsilun);
-         goto fail;
++    aio_context_release(iscsilun->aio_context);
-diff --git a/monitor/hmp-cmds.c b/monitor/hmp-cmds.c
+ }
-index XXXXXXX..XXXXXXX 100644
---- a/monitor/hmp-cmds.c
+ static int64_t sector_lun2qemu(int64_t sector, IscsiLun *iscsilun)
-+++ b/monitor/hmp-cmds.c
+diff --git a/block/linux-aio.c b/block/linux-aio.c
-@@ -XXX,XX +XXX,XX @@ void hmp_change(Monitor *mon, const QDict *qdict)
+index XXXXXXX..XXXXXXX 100644
-     const char *target = qdict_get_str(qdict, "target");
+--- a/block/linux-aio.c
-     const char *arg = qdict_get_try_str(qdict, "arg");
++++ b/block/linux-aio.c
-     const char *read_only = qdict_get_try_str(qdict, "read-only-mode");
+@@ -XXX,XX +XXX,XX @@ static void qemu_laio_completion_cb(EventNotifier *e)
-+    bool force = qdict_get_try_bool(qdict, "force", false);
+     LinuxAioState *s = container_of(e, LinuxAioState, e);
-     BlockdevChangeReadOnlyMode read_only_mode = 0;
-     Error *err = NULL;
+     if (event_notifier_test_and_clear(&s->e)) {
++        aio_context_acquire(s->aio_context);
-@@ -XXX,XX +XXX,XX @@ void hmp_change(Monitor *mon, const QDict *qdict)
+         qemu_laio_process_completions_and_submit(s);
 +        aio_context_release(s->aio_context);
      }
  }
@@ -XXX,XX +XXX,XX @@ static bool qemu_laio_poll_cb(void *opaque)
          return false;
      }
 +    aio_context_acquire(s->aio_context);
      qemu_laio_process_completions_and_submit(s);
 +    aio_context_release(s->aio_context);
      return true;
  }
 diff --git a/block/nfs.c b/block/nfs.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/nfs.c
 +++ b/block/nfs.c
@@ -XXX,XX +XXX,XX @@ static void nfs_set_events(NFSClient *client)
  static void nfs_process_read(void *arg)
  {
      NFSClient *client = arg;
 +
 +    aio_context_acquire(client->aio_context);
      nfs_service(client->context, POLLIN);
      nfs_set_events(client);
 +    aio_context_release(client->aio_context);
  }
  static void nfs_process_write(void *arg)
  {
      NFSClient *client = arg;
 +
 +    aio_context_acquire(client->aio_context);
      nfs_service(client->context, POLLOUT);
      nfs_set_events(client);
 +    aio_context_release(client->aio_context);
  }
  static void nfs_co_init_task(BlockDriverState *bs, NFSRPC *task)
 diff --git a/block/sheepdog.c b/block/sheepdog.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/sheepdog.c
 +++ b/block/sheepdog.c
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int send_co_req(int sockfd, SheepdogReq *hdr, void *data,
      return ret;
  }
 -static void restart_co_req(void *opaque)
 -{
 -    Coroutine *co = opaque;
 -
 -    qemu_coroutine_enter(co);
 -}
 -
  typedef struct SheepdogReqCo {
      int sockfd;
      BlockDriverState *bs;
@@ -XXX,XX +XXX,XX @@ typedef struct SheepdogReqCo {
      unsigned int *rlen;
      int ret;
      bool finished;
 +    Coroutine *co;
  } SheepdogReqCo;
 +static void restart_co_req(void *opaque)
 +{
 +    SheepdogReqCo *srco = opaque;
 +
 +    aio_co_wake(srco->co);
 +}
 +
  static coroutine_fn void do_co_req(void *opaque)
  {
      int ret;
 -    Coroutine *co;
      SheepdogReqCo *srco = opaque;
      int sockfd = srco->sockfd;
      SheepdogReq *hdr = srco->hdr;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void do_co_req(void *opaque)
      unsigned int *wlen = srco->wlen;
      unsigned int *rlen = srco->rlen;
 -    co = qemu_coroutine_self();
 +    srco->co = qemu_coroutine_self();
      aio_set_fd_handler(srco->aio_context, sockfd, false,
 -                       NULL, restart_co_req, NULL, co);
 +                       NULL, restart_co_req, NULL, srco);
      ret = send_co_req(sockfd, hdr, data, wlen);
      if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void do_co_req(void *opaque)
      }
      aio_set_fd_handler(srco->aio_context, sockfd, false,
 -                       restart_co_req, NULL, NULL, co);
 +                       restart_co_req, NULL, NULL, srco);
      ret = qemu_co_recv(sockfd, hdr, sizeof(*hdr));
      if (ret != sizeof(*hdr)) {
@@ -XXX,XX +XXX,XX @@ out:
      aio_set_fd_handler(srco->aio_context, sockfd, false,
                         NULL, NULL, NULL, NULL);
 +    srco->co = NULL;
      srco->ret = ret;
      srco->finished = true;
      if (srco->bs) {
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn aio_read_response(void *opaque)
           * We've finished all requests which belong to the AIOCB, so
           * we can switch back to sd_co_readv/writev now.
           */
 -        qemu_coroutine_enter(acb->coroutine);
 +        aio_co_wake(acb->coroutine);
      }
      return;
@@ -XXX,XX +XXX,XX @@ static void co_read_response(void *opaque)
          s->co_recv = qemu_coroutine_create(aio_read_response, opaque);
      }
 -    qemu_coroutine_enter(s->co_recv);
 +    aio_co_wake(s->co_recv);
  }
  static void co_write_request(void *opaque)
  {
      BDRVSheepdogState *s = opaque;
 -    qemu_coroutine_enter(s->co_send);
 +    aio_co_wake(s->co_send);
  }
  /*
 diff --git a/block/ssh.c b/block/ssh.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/ssh.c
 +++ b/block/ssh.c
@@ -XXX,XX +XXX,XX @@ static void restart_coroutine(void *opaque)
      DPRINTF("co=%p", co);
 -    qemu_coroutine_enter(co);
 +    aio_co_wake(co);
  }
 -static coroutine_fn void set_fd_handler(BDRVSSHState *s, BlockDriverState *bs)
 +/* A non-blocking call returned EAGAIN, so yield, ensuring the
 + * handlers are set up so that we'll be rescheduled when there is an
 + * interesting event on the socket.
 + */
 +static coroutine_fn void co_yield(BDRVSSHState *s, BlockDriverState *bs)
  {
      int r;
      IOHandler *rd_handler = NULL, *wr_handler = NULL;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void set_fd_handler(BDRVSSHState *s, BlockDriverState *bs)
      aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
                         false, rd_handler, wr_handler, NULL, co);
 -}
 -
 -static coroutine_fn void clear_fd_handler(BDRVSSHState *s,
 -                                          BlockDriverState *bs)
 -{
 -    DPRINTF("s->sock=%d", s->sock);
 -    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
 -                       false, NULL, NULL, NULL, NULL);
 -}
 -
 -/* A non-blocking call returned EAGAIN, so yield, ensuring the
 - * handlers are set up so that we'll be rescheduled when there is an
 - * interesting event on the socket.
 - */
 -static coroutine_fn void co_yield(BDRVSSHState *s, BlockDriverState *bs)
 -{
 -    set_fd_handler(s, bs);
      qemu_coroutine_yield();
 -    clear_fd_handler(s, bs);
 +    DPRINTF("s->sock=%d - back", s->sock);
 +    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock, false,
 +                       NULL, NULL, NULL, NULL);
  }
  /* SFTP has a function `libssh2_sftp_seek64' which seeks to a position
 diff --git a/block/win32-aio.c b/block/win32-aio.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/win32-aio.c
 +++ b/block/win32-aio.c
@@ -XXX,XX +XXX,XX @@ struct QEMUWin32AIOState {
      HANDLE hIOCP;
      EventNotifier e;
      int count;
 -    bool is_aio_context_attached;
 +    AioContext *aio_ctx;
  };
  typedef struct QEMUWin32AIOCB {
@@ -XXX,XX +XXX,XX @@ static void win32_aio_process_completion(QEMUWin32AIOState *s,
      }
 +    aio_context_acquire(s->aio_ctx);
      waiocb->common.cb(waiocb->common.opaque, ret);
 +    aio_context_release(s->aio_ctx);
      qemu_aio_unref(waiocb);
  }
@@ -XXX,XX +XXX,XX @@ void win32_aio_detach_aio_context(QEMUWin32AIOState *aio,
                                    AioContext *old_context)
  {
      aio_set_event_notifier(old_context, &aio->e, false, NULL, NULL);
 -    aio->is_aio_context_attached = false;
 +    aio->aio_ctx = NULL;
  }
  void win32_aio_attach_aio_context(QEMUWin32AIOState *aio,
                                    AioContext *new_context)
  {
 -    aio->is_aio_context_attached = true;
 +    aio->aio_ctx = new_context;
      aio_set_event_notifier(new_context, &aio->e, false,
                             win32_aio_completion_cb, NULL);
  }
@@ -XXX,XX +XXX,XX @@ out_free_state:
  void win32_aio_cleanup(QEMUWin32AIOState *aio)
  {
 -    assert(!aio->is_aio_context_attached);
 +    assert(!aio->aio_ctx);
      CloseHandle(aio->hIOCP);
      event_notifier_cleanup(&aio->e);
      g_free(aio);
 diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/block/virtio-blk.c
 +++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_ioctl_complete(void *opaque, int status)
  {
      VirtIOBlockIoctlReq *ioctl_req = opaque;
      VirtIOBlockReq *req = ioctl_req->req;
 -    VirtIODevice *vdev = VIRTIO_DEVICE(req->dev);
 +    VirtIOBlock *s = req->dev;
 +    VirtIODevice *vdev = VIRTIO_DEVICE(s);
      struct virtio_scsi_inhdr *scsi;
      struct sg_io_hdr *hdr;
@@ -XXX,XX +XXX,XX @@ bool virtio_blk_handle_vq(VirtIOBlock *s, VirtQueue *vq)
      MultiReqBuffer mrb = {};
      bool progress = false;
 +    aio_context_acquire(blk_get_aio_context(s->blk));
      blk_io_plug(s->blk);
      do {
@@ -XXX,XX +XXX,XX @@ bool virtio_blk_handle_vq(VirtIOBlock *s, VirtQueue *vq)
      }
      blk_io_unplug(s->blk);
 +    aio_context_release(blk_get_aio_context(s->blk));
      return progress;
  }
 diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/scsi/virtio-scsi.c
 +++ b/hw/scsi/virtio-scsi.c
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_ctrl_vq(VirtIOSCSI *s, VirtQueue *vq)
      VirtIOSCSIReq *req;
      bool progress = false;
 +    virtio_scsi_acquire(s);
      while ((req = virtio_scsi_pop_req(s, vq))) {
          progress = true;
          virtio_scsi_handle_ctrl_req(s, req);
      }
 +    virtio_scsi_release(s);
      return progress;
  }
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_cmd_vq(VirtIOSCSI *s, VirtQueue *vq)
      QTAILQ_HEAD(, VirtIOSCSIReq) reqs = QTAILQ_HEAD_INITIALIZER(reqs);
 +    virtio_scsi_acquire(s);
      do {
          virtio_queue_set_notification(vq, 0);
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_cmd_vq(VirtIOSCSI *s, VirtQueue *vq)
      QTAILQ_FOREACH_SAFE(req, &reqs, next, next) {
          virtio_scsi_handle_cmd_req_submit(s, req);
      }
 +    virtio_scsi_release(s);
      return progress;
  }
@@ -XXX,XX +XXX,XX @@ out:
  bool virtio_scsi_handle_event_vq(VirtIOSCSI *s, VirtQueue *vq)
  {
 +    virtio_scsi_acquire(s);
      if (s->events_dropped) {
          virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
 +        virtio_scsi_release(s);
          return true;
      }
 +    virtio_scsi_release(s);
      return false;
  }
 diff --git a/util/aio-posix.c b/util/aio-posix.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-posix.c
 +++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
              (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
              aio_node_check(ctx, node->is_external) &&
              node->io_read) {
 -            aio_context_acquire(ctx);
              node->io_read(node->opaque);
 -            aio_context_release(ctx);
              /* aio_notify() does not count as progress */
              if (node->opaque != &ctx->notifier) {
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
              (revents & (G_IO_OUT | G_IO_ERR)) &&
              aio_node_check(ctx, node->is_external) &&
              node->io_write) {
 -            aio_context_acquire(ctx);
              node->io_write(node->opaque);
 -            aio_context_release(ctx);
              progress = true;
          }
-         qmp_blockdev_change_medium(true, device, false, NULL, target,
+@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
--                                   !!arg, arg, !!read_only, read_only_mode,
+         start = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
-+                                   !!arg, arg, true, force,
+     }
-+                                   !!read_only, read_only_mode,
-                                    &err);
+-    aio_context_acquire(ctx);
-     }
+     progress = try_poll_mode(ctx, blocking);
+-    aio_context_release(ctx);
-diff --git a/hmp-commands.hx b/hmp-commands.hx
+-
-index XXXXXXX..XXXXXXX 100644
+     if (!progress) {
---- a/hmp-commands.hx
+         assert(npfd == 0);
-+++ b/hmp-commands.hx
-@@ -XXX,XX +XXX,XX @@ ERST
+diff --git a/util/aio-win32.c b/util/aio-win32.c
+index XXXXXXX..XXXXXXX 100644
-     {
+--- a/util/aio-win32.c
-         .name       = "change",
++++ b/util/aio-win32.c
--        .args_type  = "device:B,target:F,arg:s?,read-only-mode:s?",
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
--        .params     = "device filename [format [read-only-mode]]",
+             (revents || event_notifier_get_handle(node->e) == event) &&
--        .help       = "change a removable medium, optional format",
+             node->io_notify) {
-+        .args_type  = "device:B,force:-f,target:F,arg:s?,read-only-mode:s?",
+             node->pfd.revents = 0;
-+        .params     = "device [-f] filename [format [read-only-mode]]",
+-            aio_context_acquire(ctx);
-+        .help       = "change a removable medium, optional format, use -f to force the operation",
+             node->io_notify(node->e);
-         .cmd        = hmp_change,
+-            aio_context_release(ctx);
-     },
+             /* aio_notify() does not count as progress */
-@@ -XXX,XX +XXX,XX @@ SRST
+             if (node->e != &ctx->notifier) {
- ``change`` *device* *setting*
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
-   Change the configuration of a device.
+             (node->io_read || node->io_write)) {
+             node->pfd.revents = 0;
--  ``change`` *diskdevice* *filename* [*format* [*read-only-mode*]]
+             if ((revents & G_IO_IN) && node->io_read) {
-+  ``change`` *diskdevice* [-f] *filename* [*format* [*read-only-mode*]]
+-                aio_context_acquire(ctx);
-     Change the medium for a removable disk device to point to *filename*. eg::
+                 node->io_read(node->opaque);
+-                aio_context_release(ctx);
-       (qemu) change ide1-cd0 /path/to/some.iso
+                 progress = true;
+             }
-+    ``-f``
+             if ((revents & G_IO_OUT) && node->io_write) {
-+      forces the operation even if the guest has locked the tray.
+-                aio_context_acquire(ctx);
-+
+                 node->io_write(node->opaque);
-     *format* is optional.
+-                aio_context_release(ctx);
+                 progress = true;
-     *read-only-mode* may be used to change the read-only status of the device.
+             }
-diff --git a/ui/cocoa.m b/ui/cocoa.m
 index XXXXXXX..XXXXXXX 100644
 --- a/ui/cocoa.m
 +++ b/ui/cocoa.m
@@ -XXX,XX +XXX,XX @@ - (void)changeDeviceMedia:(id)sender
                                         [file cStringUsingEncoding:
                                                   NSASCIIStringEncoding],
                                         true, "raw",
 +                                       true, false,
                                         false, 0,
                                         &err);
          });
 --
-.35.1
+.9.3

-[PULL 14/14] iotests/108: Fix when missing user_allow_other
+[Qemu-devel] [PULL v2 14/24] block: explicitly acquire aiocontext in bottom halves that need it
-FUSE exports' allow-other option defaults to "auto", which means that it
+From: Paolo Bonzini <pbonzini@redhat.com>
 will try passing allow_other as a mount option, and fall back to not
 using it when an error occurs.  We make no effort to hide fusermount's
 error message (because it would be difficult, and because users might
 want to know about the fallback occurring), and so when allow_other does
 not work (primarily when /etc/fuse.conf does not contain
 user_allow_other), this error message will appear and break the
 reference output.
-We do not need allow_other here, though, so we can just pass
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-allow-other=off to fix that.
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Message-id: 20170213135235.12274-15-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  block/archipelago.c   |  3 +++
  block/blkreplay.c     |  2 +-
  block/block-backend.c |  6 ++++++
  block/curl.c          | 26 ++++++++++++++++++--------
  block/gluster.c       |  9 +--------
  block/io.c            |  6 +++++-
  block/iscsi.c         |  6 +++++-
  block/linux-aio.c     | 15 +++++++++------
  block/nfs.c           |  3 ++-
  block/null.c          |  4 ++++
  block/qed.c           |  3 +++
  block/rbd.c           |  4 ++++
  dma-helpers.c         |  2 ++
  hw/block/virtio-blk.c |  2 ++
  hw/scsi/scsi-bus.c    |  2 ++
  util/async.c          |  4 ++--
  util/thread-pool.c    |  2 ++
 files changed, 71 insertions(+), 28 deletions(-)
-Reported-by: Markus Armbruster <armbru@redhat.com>
+diff --git a/block/archipelago.c b/block/archipelago.c
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
+index XXXXXXX..XXXXXXX 100644
-Message-Id: <20220421142435.569600-1-hreitz@redhat.com>
+--- a/block/archipelago.c
-Tested-by: Markus Armbruster <armbru@redhat.com>
++++ b/block/archipelago.c
-Tested-by: Eric Blake <eblake@redhat.com>
+@@ -XXX,XX +XXX,XX @@ static void qemu_archipelago_complete_aio(void *opaque)
----
+ {
- tests/qemu-iotests/108 | 2 +-
+     AIORequestData *reqdata = (AIORequestData *) opaque;
-file changed, 1 insertion(+), 1 deletion(-)
+     ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
 +    AioContext *ctx = bdrv_get_aio_context(aio_cb->common.bs);
 +    aio_context_acquire(ctx);
      aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
 +    aio_context_release(ctx);
      aio_cb->status = 0;
      qemu_aio_unref(aio_cb);
 diff --git a/block/blkreplay.c b/block/blkreplay.c
 index XXXXXXX..XXXXXXX 100755
 --- a/block/blkreplay.c
 +++ b/block/blkreplay.c
@@ -XXX,XX +XXX,XX @@ static int64_t blkreplay_getlength(BlockDriverState *bs)
  static void blkreplay_bh_cb(void *opaque)
  {
      Request *req = opaque;
 -    qemu_coroutine_enter(req->co);
 +    aio_co_wake(req->co);
      qemu_bh_delete(req->bh);
      g_free(req);
  }
 diff --git a/block/block-backend.c b/block/block-backend.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/block-backend.c
 +++ b/block/block-backend.c
@@ -XXX,XX +XXX,XX @@ int blk_make_zero(BlockBackend *blk, BdrvRequestFlags flags)
  static void error_callback_bh(void *opaque)
  {
      struct BlockBackendAIOCB *acb = opaque;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      bdrv_dec_in_flight(acb->common.bs);
 +    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, acb->ret);
 +    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
@@ -XXX,XX +XXX,XX @@ static void blk_aio_complete(BlkAioEmAIOCB *acb)
  static void blk_aio_complete_bh(void *opaque)
  {
      BlkAioEmAIOCB *acb = opaque;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      assert(acb->has_returned);
 +    aio_context_acquire(ctx);
      blk_aio_complete(acb);
 +    aio_context_release(ctx);
  }
  static BlockAIOCB *blk_aio_prwv(BlockBackend *blk, int64_t offset, int bytes,
 diff --git a/block/curl.c b/block/curl.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/curl.c
 +++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
  {
      CURLState *state;
      int running;
 +    int ret = -EINPROGRESS;
      CURLAIOCB *acb = p;
 -    BDRVCURLState *s = acb->common.bs->opaque;
 +    BlockDriverState *bs = acb->common.bs;
 +    BDRVCURLState *s = bs->opaque;
 +    AioContext *ctx = bdrv_get_aio_context(bs);
      size_t start = acb->sector_num * BDRV_SECTOR_SIZE;
      size_t end;
 +    aio_context_acquire(ctx);
 +
      // In case we have the requested data already (e.g. read-ahead),
      // we can just call the callback and be done.
      switch (curl_find_buf(s, start, acb->nb_sectors * BDRV_SECTOR_SIZE, acb)) {
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
              qemu_aio_unref(acb);
              // fall through
          case FIND_RET_WAIT:
 -            return;
 +            goto out;
          default:
              break;
      }
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
      // No cache found, so let's start a new request
      state = curl_init_state(acb->common.bs, s);
      if (!state) {
 -        acb->common.cb(acb->common.opaque, -EIO);
 -        qemu_aio_unref(acb);
 -        return;
 +        ret = -EIO;
 +        goto out;
      }
      acb->start = 0;
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
      state->orig_buf = g_try_malloc(state->buf_len);
      if (state->buf_len && state->orig_buf == NULL) {
          curl_clean_state(state);
 -        acb->common.cb(acb->common.opaque, -ENOMEM);
 -        qemu_aio_unref(acb);
 -        return;
 +        ret = -ENOMEM;
 +        goto out;
      }
      state->acb[0] = acb;
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
      /* Tell curl it needs to kick things off */
      curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
 +
 +out:
 +    if (ret != -EINPROGRESS) {
 +        acb->common.cb(acb->common.opaque, ret);
 +        qemu_aio_unref(acb);
 +    }
 +    aio_context_release(ctx);
  }
  static BlockAIOCB *curl_aio_readv(BlockDriverState *bs,
 diff --git a/block/gluster.c b/block/gluster.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/gluster.c
 +++ b/block/gluster.c
@@ -XXX,XX +XXX,XX @@ static struct glfs *qemu_gluster_init(BlockdevOptionsGluster *gconf,
      return qemu_gluster_glfs_init(gconf, errp);
  }
 -static void qemu_gluster_complete_aio(void *opaque)
 -{
 -    GlusterAIOCB *acb = (GlusterAIOCB *)opaque;
 -
 -    qemu_coroutine_enter(acb->coroutine);
 -}
 -
  /*
   * AIO callback routine called from GlusterFS thread.
   */
@@ -XXX,XX +XXX,XX @@ static void gluster_finish_aiocb(struct glfs_fd *fd, ssize_t ret, void *arg)
          acb->ret = -EIO; /* Partial read/write - fail it */
      }
 -    aio_bh_schedule_oneshot(acb->aio_context, qemu_gluster_complete_aio, acb);
 +    aio_co_schedule(acb->aio_context, acb->coroutine);
  }
  static void qemu_gluster_parse_flags(int bdrv_flags, int *open_flags)
 diff --git a/block/io.c b/block/io.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/io.c
 +++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque)
      bdrv_dec_in_flight(bs);
      bdrv_drained_begin(bs);
      data->done = true;
 -    qemu_coroutine_enter(co);
 +    aio_co_wake(co);
  }
  static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs)
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_complete(BlockAIOCBCoroutine *acb)
  static void bdrv_co_em_bh(void *opaque)
  {
      BlockAIOCBCoroutine *acb = opaque;
 +    BlockDriverState *bs = acb->common.bs;
 +    AioContext *ctx = bdrv_get_aio_context(bs);
      assert(!acb->need_bh);
 +    aio_context_acquire(ctx);
      bdrv_co_complete(acb);
 +    aio_context_release(ctx);
  }
  static void bdrv_co_maybe_schedule_bh(BlockAIOCBCoroutine *acb)
 diff --git a/block/iscsi.c b/block/iscsi.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/iscsi.c
 +++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void
  iscsi_bh_cb(void *p)
  {
      IscsiAIOCB *acb = p;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      qemu_bh_delete(acb->bh);
      g_free(acb->buf);
      acb->buf = NULL;
 +    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, acb->status);
 +    aio_context_release(ctx);
      if (acb->task != NULL) {
          scsi_free_scsi_task(acb->task);
@@ -XXX,XX +XXX,XX @@ iscsi_schedule_bh(IscsiAIOCB *acb)
  static void iscsi_co_generic_bh_cb(void *opaque)
  {
      struct IscsiTask *iTask = opaque;
 +
      iTask->complete = 1;
 -    qemu_coroutine_enter(iTask->co);
 +    aio_co_wake(iTask->co);
  }
  static void iscsi_retry_timer_expired(void *opaque)
 diff --git a/block/linux-aio.c b/block/linux-aio.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/linux-aio.c
 +++ b/block/linux-aio.c
@@ -XXX,XX +XXX,XX @@ struct LinuxAioState {
      io_context_t ctx;
      EventNotifier e;
 -    /* io queue for submit at batch */
 +    /* io queue for submit at batch.  Protected by AioContext lock. */
      LaioQueue io_q;
 -    /* I/O completion processing */
 +    /* I/O completion processing.  Only runs in I/O thread.  */
      QEMUBH *completion_bh;
      int event_idx;
      int event_max;
@@ -XXX,XX +XXX,XX @@ static inline ssize_t io_event_ret(struct io_event *ev)
   */
  static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
  {
 +    LinuxAioState *s = laiocb->ctx;
      int ret;
      ret = laiocb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
      }
      laiocb->ret = ret;
 +    aio_context_acquire(s->aio_context);
      if (laiocb->co) {
          /* If the coroutine is already entered it must be in ioq_submit() and
           * will notice laio->ret has been filled in when it eventually runs
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
          laiocb->common.cb(laiocb->common.opaque, ret);
          qemu_aio_unref(laiocb);
      }
 +    aio_context_release(s->aio_context);
  }
  /**
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completions(LinuxAioState *s)
  static void qemu_laio_process_completions_and_submit(LinuxAioState *s)
  {
      qemu_laio_process_completions(s);
 +
 +    aio_context_acquire(s->aio_context);
      if (!s->io_q.plugged && !QSIMPLEQ_EMPTY(&s->io_q.pending)) {
          ioq_submit(s);
      }
 +    aio_context_release(s->aio_context);
  }
  static void qemu_laio_completion_bh(void *opaque)
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_completion_cb(EventNotifier *e)
      LinuxAioState *s = container_of(e, LinuxAioState, e);
      if (event_notifier_test_and_clear(&s->e)) {
 -        aio_context_acquire(s->aio_context);
          qemu_laio_process_completions_and_submit(s);
 -        aio_context_release(s->aio_context);
      }
  }
@@ -XXX,XX +XXX,XX @@ static bool qemu_laio_poll_cb(void *opaque)
          return false;
      }
 -    aio_context_acquire(s->aio_context);
      qemu_laio_process_completions_and_submit(s);
 -    aio_context_release(s->aio_context);
      return true;
  }
@@ -XXX,XX +XXX,XX @@ void laio_detach_aio_context(LinuxAioState *s, AioContext *old_context)
  {
      aio_set_event_notifier(old_context, &s->e, false, NULL, NULL);
      qemu_bh_delete(s->completion_bh);
 +    s->aio_context = NULL;
  }
  void laio_attach_aio_context(LinuxAioState *s, AioContext *new_context)
 diff --git a/block/nfs.c b/block/nfs.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/nfs.c
 +++ b/block/nfs.c
@@ -XXX,XX +XXX,XX @@ static void nfs_co_init_task(BlockDriverState *bs, NFSRPC *task)
  static void nfs_co_generic_bh_cb(void *opaque)
  {
      NFSRPC *task = opaque;
 +
      task->complete = 1;
 -    qemu_coroutine_enter(task->co);
 +    aio_co_wake(task->co);
  }
  static void
 diff --git a/block/null.c b/block/null.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/null.c
 +++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static const AIOCBInfo null_aiocb_info = {
  static void null_bh_cb(void *opaque)
  {
      NullAIOCB *acb = opaque;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 +
 +    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, 0);
 +    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
 diff --git a/block/qed.c b/block/qed.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed.c
 +++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_update_l2_table(BDRVQEDState *s, QEDTable *table, int index,
  static void qed_aio_complete_bh(void *opaque)
  {
      QEDAIOCB *acb = opaque;
 +    BDRVQEDState *s = acb_to_s(acb);
      BlockCompletionFunc *cb = acb->common.cb;
      void *user_opaque = acb->common.opaque;
      int ret = acb->bh_ret;
@@ -XXX,XX +XXX,XX @@ static void qed_aio_complete_bh(void *opaque)
      qemu_aio_unref(acb);
      /* Invoke callback */
 +    qed_acquire(s);
      cb(user_opaque, ret);
 +    qed_release(s);
  }
  static void qed_aio_complete(QEDAIOCB *acb, int ret)
 diff --git a/block/rbd.c b/block/rbd.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/rbd.c
 +++ b/block/rbd.c
@@ -XXX,XX +XXX,XX @@ shutdown:
  static void qemu_rbd_complete_aio(RADOSCB *rcb)
  {
      RBDAIOCB *acb = rcb->acb;
 +    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      int64_t r;
      r = rcb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
          qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
      }
      qemu_vfree(acb->bounce);
 +
 +    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
 +    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
 diff --git a/dma-helpers.c b/dma-helpers.c
 index XXXXXXX..XXXXXXX 100644
 --- a/dma-helpers.c
 +++ b/dma-helpers.c
@@ -XXX,XX +XXX,XX @@ static void dma_blk_cb(void *opaque, int ret)
                                  QEMU_ALIGN_DOWN(dbs->iov.size, dbs->align));
      }
 +    aio_context_acquire(dbs->ctx);
      dbs->acb = dbs->io_func(dbs->offset, &dbs->iov,
                              dma_blk_cb, dbs, dbs->io_func_opaque);
 +    aio_context_release(dbs->ctx);
      assert(dbs->acb);
  }
 diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/block/virtio-blk.c
 +++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_dma_restart_bh(void *opaque)
      s->rq = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
      while (req) {
          VirtIOBlockReq *next = req->next;
          if (virtio_blk_handle_request(req, &mrb)) {
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_dma_restart_bh(void *opaque)
      if (mrb.num_reqs) {
          virtio_blk_submit_multireq(s->blk, &mrb);
      }
 +    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
  }
  static void virtio_blk_dma_restart_cb(void *opaque, int running,
 diff --git a/hw/scsi/scsi-bus.c b/hw/scsi/scsi-bus.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/scsi/scsi-bus.c
 +++ b/hw/scsi/scsi-bus.c
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_restart_bh(void *opaque)
      qemu_bh_delete(s->bh);
      s->bh = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->conf.blk));
      QTAILQ_FOREACH_SAFE(req, &s->requests, next, next) {
          scsi_req_ref(req);
          if (req->retry) {
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_restart_bh(void *opaque)
          }
          scsi_req_unref(req);
      }
 +    aio_context_release(blk_get_aio_context(s->conf.blk));
  }
  void scsi_req_retry(SCSIRequest *req)
 diff --git a/util/async.c b/util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
                  ret = 1;
              }
              bh->idle = 0;
 -            aio_context_acquire(ctx);
              aio_bh_call(bh);
 -            aio_context_release(ctx);
          }
          if (bh->deleted) {
              deleted = true;
@@ -XXX,XX +XXX,XX @@ static void co_schedule_bh_cb(void *opaque)
          Coroutine *co = QSLIST_FIRST(&straight);
          QSLIST_REMOVE_HEAD(&straight, co_scheduled_next);
          trace_aio_co_schedule_bh_cb(ctx, co);
 +        aio_context_acquire(ctx);
          qemu_coroutine_enter(co);
 +        aio_context_release(ctx);
      }
  }
 diff --git a/util/thread-pool.c b/util/thread-pool.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/thread-pool.c
 +++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@ static void thread_pool_completion_bh(void *opaque)
      ThreadPool *pool = opaque;
      ThreadPoolElement *elem, *next;
 +    aio_context_acquire(pool->ctx);
  restart:
      QLIST_FOREACH_SAFE(elem, &pool->head, all, next) {
          if (elem->state != THREAD_DONE) {
@@ -XXX,XX +XXX,XX @@ restart:
              qemu_aio_unref(elem);
          }
      }
 +    aio_context_release(pool->ctx);
  }
  static void thread_pool_cancel(BlockAIOCB *acb)
 --
 .9.3
-diff --git a/tests/qemu-iotests/108 b/tests/qemu-iotests/108
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/108
 +++ b/tests/qemu-iotests/108
@@ -XXX,XX +XXX,XX @@ else
      $QSD \
          --blockdev file,node-name=export-node,filename="$TEST_IMG" \
 -        --export fuse,id=fuse-export,node-name=export-node,mountpoint="$export_mp",writable=on,growable=off \
 +        --export fuse,id=fuse-export,node-name=export-node,mountpoint="$export_mp",writable=on,growable=off,allow-other=off \
          --pidfile "$TEST_DIR/qsd.pid" \
          --daemonize
  fi
 --
 .35.1

-[PULL 13/14] iotests: make qemu_io_log() check return codes by default
+[Qemu-devel] [PULL v2 15/24] block: explicitly acquire aiocontext in aio callbacks that need it
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-Just like qemu_img_log(), upgrade qemu_io_log() to enforce a return code
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-of zero by default.
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Message-id: 20170213135235.12274-16-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  block/archipelago.c    |  3 ---
  block/block-backend.c  |  7 -------
  block/curl.c           |  2 +-
  block/io.c             |  6 +-----
  block/iscsi.c          |  3 ---
  block/linux-aio.c      |  5 +----
  block/mirror.c         | 12 +++++++++---
  block/null.c           |  8 --------
  block/qed-cluster.c    |  2 ++
  block/qed-table.c      | 12 ++++++++++--
  block/qed.c            |  4 ++--
  block/rbd.c            |  4 ----
  block/win32-aio.c      |  3 ---
  hw/block/virtio-blk.c  | 12 +++++++++++-
  hw/scsi/scsi-disk.c    | 15 +++++++++++++++
  hw/scsi/scsi-generic.c | 20 +++++++++++++++++---
  util/thread-pool.c     |  4 +++-
 files changed, 72 insertions(+), 50 deletions(-)
-Tests that use qemu_io_log(): 242 245 255 274 303 307 nbd-reconnect-on-open
+diff --git a/block/archipelago.c b/block/archipelago.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/archipelago.c
 +++ b/block/archipelago.c
@@ -XXX,XX +XXX,XX @@ static void qemu_archipelago_complete_aio(void *opaque)
  {
      AIORequestData *reqdata = (AIORequestData *) opaque;
      ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
 -    AioContext *ctx = bdrv_get_aio_context(aio_cb->common.bs);
 -    aio_context_acquire(ctx);
      aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
 -    aio_context_release(ctx);
      aio_cb->status = 0;
      qemu_aio_unref(aio_cb);
 diff --git a/block/block-backend.c b/block/block-backend.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/block-backend.c
 +++ b/block/block-backend.c
@@ -XXX,XX +XXX,XX @@ int blk_make_zero(BlockBackend *blk, BdrvRequestFlags flags)
  static void error_callback_bh(void *opaque)
  {
      struct BlockBackendAIOCB *acb = opaque;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      bdrv_dec_in_flight(acb->common.bs);
 -    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, acb->ret);
 -    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
@@ -XXX,XX +XXX,XX @@ static void blk_aio_complete(BlkAioEmAIOCB *acb)
  static void blk_aio_complete_bh(void *opaque)
  {
      BlkAioEmAIOCB *acb = opaque;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 -
      assert(acb->has_returned);
 -    aio_context_acquire(ctx);
      blk_aio_complete(acb);
 -    aio_context_release(ctx);
  }
  static BlockAIOCB *blk_aio_prwv(BlockBackend *blk, int64_t offset, int bytes,
 diff --git a/block/curl.c b/block/curl.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/curl.c
 +++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
      curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
  out:
 +    aio_context_release(ctx);
      if (ret != -EINPROGRESS) {
          acb->common.cb(acb->common.opaque, ret);
          qemu_aio_unref(acb);
      }
 -    aio_context_release(ctx);
  }
  static BlockAIOCB *curl_aio_readv(BlockDriverState *bs,
 diff --git a/block/io.c b/block/io.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/io.c
 +++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_io_em_complete(void *opaque, int ret)
      CoroutineIOCompletion *co = opaque;
      co->ret = ret;
 -    qemu_coroutine_enter(co->coroutine);
 +    aio_co_wake(co->coroutine);
  }
  static int coroutine_fn bdrv_driver_preadv(BlockDriverState *bs,
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_complete(BlockAIOCBCoroutine *acb)
  static void bdrv_co_em_bh(void *opaque)
  {
      BlockAIOCBCoroutine *acb = opaque;
 -    BlockDriverState *bs = acb->common.bs;
 -    AioContext *ctx = bdrv_get_aio_context(bs);
      assert(!acb->need_bh);
 -    aio_context_acquire(ctx);
      bdrv_co_complete(acb);
 -    aio_context_release(ctx);
  }
  static void bdrv_co_maybe_schedule_bh(BlockAIOCBCoroutine *acb)
 diff --git a/block/iscsi.c b/block/iscsi.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/iscsi.c
 +++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void
  iscsi_bh_cb(void *p)
  {
      IscsiAIOCB *acb = p;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      qemu_bh_delete(acb->bh);
      g_free(acb->buf);
      acb->buf = NULL;
 -    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, acb->status);
 -    aio_context_release(ctx);
      if (acb->task != NULL) {
          scsi_free_scsi_task(acb->task);
 diff --git a/block/linux-aio.c b/block/linux-aio.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/linux-aio.c
 +++ b/block/linux-aio.c
@@ -XXX,XX +XXX,XX @@ static inline ssize_t io_event_ret(struct io_event *ev)
   */
  static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
  {
 -    LinuxAioState *s = laiocb->ctx;
      int ret;
      ret = laiocb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
      }
      laiocb->ret = ret;
 -    aio_context_acquire(s->aio_context);
      if (laiocb->co) {
          /* If the coroutine is already entered it must be in ioq_submit() and
           * will notice laio->ret has been filled in when it eventually runs
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
           * that!
           */
          if (!qemu_coroutine_entered(laiocb->co)) {
 -            qemu_coroutine_enter(laiocb->co);
 +            aio_co_wake(laiocb->co);
          }
      } else {
          laiocb->common.cb(laiocb->common.opaque, ret);
          qemu_aio_unref(laiocb);
      }
 -    aio_context_release(s->aio_context);
  }
  /**
 diff --git a/block/mirror.c b/block/mirror.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/mirror.c
 +++ b/block/mirror.c
@@ -XXX,XX +XXX,XX @@ static void mirror_write_complete(void *opaque, int ret)
  {
      MirrorOp *op = opaque;
      MirrorBlockJob *s = op->s;
 +
 +    aio_context_acquire(blk_get_aio_context(s->common.blk));
      if (ret < 0) {
          BlockErrorAction action;
@@ -XXX,XX +XXX,XX @@ static void mirror_write_complete(void *opaque, int ret)
          }
      }
      mirror_iteration_done(op, ret);
 +    aio_context_release(blk_get_aio_context(s->common.blk));
  }
  static void mirror_read_complete(void *opaque, int ret)
  {
      MirrorOp *op = opaque;
      MirrorBlockJob *s = op->s;
 +
 +    aio_context_acquire(blk_get_aio_context(s->common.blk));
      if (ret < 0) {
          BlockErrorAction action;
@@ -XXX,XX +XXX,XX @@ static void mirror_read_complete(void *opaque, int ret)
          }
          mirror_iteration_done(op, ret);
 -        return;
 +    } else {
 +        blk_aio_pwritev(s->target, op->sector_num * BDRV_SECTOR_SIZE, &op->qiov,
 +                        0, mirror_write_complete, op);
      }
 -    blk_aio_pwritev(s->target, op->sector_num * BDRV_SECTOR_SIZE, &op->qiov,
 -                    0, mirror_write_complete, op);
 +    aio_context_release(blk_get_aio_context(s->common.blk));
  }
  static inline void mirror_clip_sectors(MirrorBlockJob *s,
 diff --git a/block/null.c b/block/null.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/null.c
 +++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static const AIOCBInfo null_aiocb_info = {
  static void null_bh_cb(void *opaque)
  {
      NullAIOCB *acb = opaque;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 -
 -    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, 0);
 -    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
  static void null_timer_cb(void *opaque)
  {
      NullAIOCB *acb = opaque;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 -
 -    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, 0);
 -    aio_context_release(ctx);
      timer_deinit(&acb->timer);
      qemu_aio_unref(acb);
  }
 diff --git a/block/qed-cluster.c b/block/qed-cluster.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed-cluster.c
 +++ b/block/qed-cluster.c
@@ -XXX,XX +XXX,XX @@ static void qed_find_cluster_cb(void *opaque, int ret)
      unsigned int index;
      unsigned int n;
 +    qed_acquire(s);
      if (ret) {
          goto out;
      }
@@ -XXX,XX +XXX,XX @@ static void qed_find_cluster_cb(void *opaque, int ret)
  out:
      find_cluster_cb->cb(find_cluster_cb->opaque, ret, offset, len);
 +    qed_release(s);
      g_free(find_cluster_cb);
  }
 diff --git a/block/qed-table.c b/block/qed-table.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed-table.c
 +++ b/block/qed-table.c
@@ -XXX,XX +XXX,XX @@ static void qed_read_table_cb(void *opaque, int ret)
  {
      QEDReadTableCB *read_table_cb = opaque;
      QEDTable *table = read_table_cb->table;
 +    BDRVQEDState *s = read_table_cb->s;
      int noffsets = read_table_cb->qiov.size / sizeof(uint64_t);
      int i;
@@ -XXX,XX +XXX,XX @@ static void qed_read_table_cb(void *opaque, int ret)
      }
      /* Byteswap offsets */
 +    qed_acquire(s);
      for (i = 0; i < noffsets; i++) {
          table->offsets[i] = le64_to_cpu(table->offsets[i]);
      }
 +    qed_release(s);
  out:
      /* Completion */
 -    trace_qed_read_table_cb(read_table_cb->s, read_table_cb->table, ret);
 +    trace_qed_read_table_cb(s, read_table_cb->table, ret);
      gencb_complete(&read_table_cb->gencb, ret);
  }
@@ -XXX,XX +XXX,XX @@ typedef struct {
  static void qed_write_table_cb(void *opaque, int ret)
  {
      QEDWriteTableCB *write_table_cb = opaque;
 +    BDRVQEDState *s = write_table_cb->s;
 -    trace_qed_write_table_cb(write_table_cb->s,
 +    trace_qed_write_table_cb(s,
                               write_table_cb->orig_table,
                               write_table_cb->flush,
                               ret);
@@ -XXX,XX +XXX,XX @@ static void qed_write_table_cb(void *opaque, int ret)
      if (write_table_cb->flush) {
          /* We still need to flush first */
          write_table_cb->flush = false;
 +        qed_acquire(s);
          bdrv_aio_flush(write_table_cb->s->bs, qed_write_table_cb,
                         write_table_cb);
 +        qed_release(s);
          return;
      }
@@ -XXX,XX +XXX,XX @@ static void qed_read_l2_table_cb(void *opaque, int ret)
      CachedL2Table *l2_table = request->l2_table;
      uint64_t l2_offset = read_l2_table_cb->l2_offset;
 +    qed_acquire(s);
      if (ret) {
          /* can't trust loaded L2 table anymore */
          qed_unref_l2_cache_entry(l2_table);
@@ -XXX,XX +XXX,XX @@ static void qed_read_l2_table_cb(void *opaque, int ret)
          request->l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
          assert(request->l2_table != NULL);
      }
 +    qed_release(s);
      gencb_complete(&read_l2_table_cb->gencb, ret);
  }
 diff --git a/block/qed.c b/block/qed.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/qed.c
 +++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_is_allocated_cb(void *opaque, int ret, uint64_t offset, size_t l
      }
      if (cb->co) {
 -        qemu_coroutine_enter(cb->co);
 +        aio_co_wake(cb->co);
      }
  }
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qed_co_pwrite_zeroes_cb(void *opaque, int ret)
      cb->done = true;
      cb->ret = ret;
      if (cb->co) {
 -        qemu_coroutine_enter(cb->co);
 +        aio_co_wake(cb->co);
      }
  }
 diff --git a/block/rbd.c b/block/rbd.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/rbd.c
 +++ b/block/rbd.c
@@ -XXX,XX +XXX,XX @@ shutdown:
  static void qemu_rbd_complete_aio(RADOSCB *rcb)
  {
      RBDAIOCB *acb = rcb->acb;
 -    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
      int64_t r;
      r = rcb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
          qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
      }
      qemu_vfree(acb->bounce);
 -
 -    aio_context_acquire(ctx);
      acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
 -    aio_context_release(ctx);
      qemu_aio_unref(acb);
  }
 diff --git a/block/win32-aio.c b/block/win32-aio.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/win32-aio.c
 +++ b/block/win32-aio.c
@@ -XXX,XX +XXX,XX @@ static void win32_aio_process_completion(QEMUWin32AIOState *s,
          qemu_vfree(waiocb->buf);
      }
 -
 -    aio_context_acquire(s->aio_ctx);
      waiocb->common.cb(waiocb->common.opaque, ret);
 -    aio_context_release(s->aio_ctx);
      qemu_aio_unref(waiocb);
  }
 diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/block/virtio-blk.c
 +++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
  static void virtio_blk_rw_complete(void *opaque, int ret)
  {
      VirtIOBlockReq *next = opaque;
 +    VirtIOBlock *s = next->dev;
 +    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
      while (next) {
          VirtIOBlockReq *req = next;
          next = req->mr_next;
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_rw_complete(void *opaque, int ret)
          block_acct_done(blk_get_stats(req->dev->blk), &req->acct);
          virtio_blk_free_request(req);
      }
 +    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
  }
  static void virtio_blk_flush_complete(void *opaque, int ret)
  {
      VirtIOBlockReq *req = opaque;
 +    VirtIOBlock *s = req->dev;
 +    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
      if (ret) {
          if (virtio_blk_handle_rw_error(req, -ret, 0)) {
 -            return;
 +            goto out;
          }
      }
      virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
      block_acct_done(blk_get_stats(req->dev->blk), &req->acct);
      virtio_blk_free_request(req);
 +
 +out:
 +    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
  }
  #ifdef __linux__
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_ioctl_complete(void *opaque, int status)
      virtio_stl_p(vdev, &scsi->data_len, hdr->dxfer_len);
  out:
 +    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
      virtio_blk_req_complete(req, status);
      virtio_blk_free_request(req);
 +    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
      g_free(ioctl_req);
  }
 diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/scsi/scsi-disk.c
 +++ b/hw/scsi/scsi-disk.c
@@ -XXX,XX +XXX,XX @@ static void scsi_aio_complete(void *opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (scsi_disk_req_check_error(r, ret, true)) {
          goto done;
      }
@@ -XXX,XX +XXX,XX @@ static void scsi_aio_complete(void *opaque, int ret)
      scsi_req_complete(&r->req, GOOD);
  done:
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
      scsi_req_unref(&r->req);
  }
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_complete(void *opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (ret < 0) {
          block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
      } else {
          block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
      }
      scsi_dma_complete_noio(r, ret);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  static void scsi_read_complete(void * opaque, int ret)
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (scsi_disk_req_check_error(r, ret, true)) {
          goto done;
      }
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
  done:
      scsi_req_unref(&r->req);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  /* Actually issue a read to the block device.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_do_read_cb(void *opaque, int ret)
      assert (r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (ret < 0) {
          block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
      } else {
          block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
      }
      scsi_do_read(opaque, ret);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  /* Read more data from scsi device into buffer.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
      assert (r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (ret < 0) {
          block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
      } else {
          block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
      }
      scsi_write_complete_noio(r, ret);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  static void scsi_write_data(SCSIRequest *req)
@@ -XXX,XX +XXX,XX @@ static void scsi_unmap_complete(void *opaque, int ret)
  {
      UnmapCBData *data = opaque;
      SCSIDiskReq *r = data->r;
 +    SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      scsi_unmap_complete_noio(data, ret);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  static void scsi_disk_emulate_unmap(SCSIDiskReq *r, uint8_t *inbuf)
@@ -XXX,XX +XXX,XX @@ static void scsi_write_same_complete(void *opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
      if (scsi_disk_req_check_error(r, ret, true)) {
          goto done;
      }
@@ -XXX,XX +XXX,XX @@ done:
      scsi_req_unref(&r->req);
      qemu_vfree(data->iov.iov_base);
      g_free(data);
 +    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
  }
  static void scsi_disk_emulate_write_same(SCSIDiskReq *r, uint8_t *inbuf)
 diff --git a/hw/scsi/scsi-generic.c b/hw/scsi/scsi-generic.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/scsi/scsi-generic.c
 +++ b/hw/scsi/scsi-generic.c
@@ -XXX,XX +XXX,XX @@ done:
  static void scsi_command_complete(void *opaque, int ret)
  {
      SCSIGenericReq *r = (SCSIGenericReq *)opaque;
 +    SCSIDevice *s = r->req.dev;
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +
 +    aio_context_acquire(blk_get_aio_context(s->conf.blk));
      scsi_command_complete_noio(r, ret);
 +    aio_context_release(blk_get_aio_context(s->conf.blk));
  }
  static int execute_command(BlockBackend *blk,
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->conf.blk));
 +
      if (ret || r->req.io_canceled) {
          scsi_command_complete_noio(r, ret);
 -        return;
 +        goto done;
      }
      len = r->io_header.dxfer_len - r->io_header.resid;
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
      r->len = -1;
      if (len == 0) {
          scsi_command_complete_noio(r, 0);
 -        return;
 +        goto done;
      }
      /* Snoop READ CAPACITY output to set the blocksize.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
      }
      scsi_req_data(&r->req, len);
      scsi_req_unref(&r->req);
 +
 +done:
 +    aio_context_release(blk_get_aio_context(s->conf.blk));
  }
  /* Read more data from scsi device into buffer.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
      assert(r->req.aiocb != NULL);
      r->req.aiocb = NULL;
 +    aio_context_acquire(blk_get_aio_context(s->conf.blk));
 +
      if (ret || r->req.io_canceled) {
          scsi_command_complete_noio(r, ret);
 -        return;
 +        goto done;
      }
      if (r->req.cmd.buf[0] == MODE_SELECT && r->req.cmd.buf[4] == 12 &&
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
      }
      scsi_command_complete_noio(r, ret);
 +
 +done:
 +    aio_context_release(blk_get_aio_context(s->conf.blk));
  }
  /* Write data to a scsi device.  Returns nonzero on failure.
 diff --git a/util/thread-pool.c b/util/thread-pool.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/thread-pool.c
 +++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@ restart:
               */
              qemu_bh_schedule(pool->completion_bh);
 +            aio_context_release(pool->ctx);
              elem->common.cb(elem->common.opaque, elem->ret);
 +            aio_context_acquire(pool->ctx);
              qemu_aio_unref(elem);
              goto restart;
          } else {
@@ -XXX,XX +XXX,XX @@ static void thread_pool_co_cb(void *opaque, int ret)
      ThreadPoolCo *co = opaque;
      co->ret = ret;
 -    qemu_coroutine_enter(co->co);
 +    aio_co_wake(co->co);
  }
  int coroutine_fn thread_pool_submit_co(ThreadPool *pool, ThreadPoolFunc *func,
 --
 .9.3
-Signed-off-by: John Snow <jsnow@redhat.com>
-Reviewed-by: Eric Blake <eblake@redhat.com>
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
-Message-Id: <20220418211504.943969-13-jsnow@redhat.com>
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
----
- tests/qemu-iotests/iotests.py                  | 5 +++--
- tests/qemu-iotests/tests/nbd-reconnect-on-open | 2 +-
-files changed, 4 insertions(+), 3 deletions(-)
-diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
-index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/iotests.py
-+++ b/tests/qemu-iotests/iotests.py
-@@ -XXX,XX +XXX,XX @@ def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
-     return qemu_tool(*qemu_io_wrap_args(args),
-                      check=check, combine_stdio=combine_stdio)
--def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
--    result = qemu_io(*args, check=False)
-+def qemu_io_log(*args: str, check: bool = True
-+                ) -> 'subprocess.CompletedProcess[str]':
-+    result = qemu_io(*args, check=check)
-     log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
-     return result
-diff --git a/tests/qemu-iotests/tests/nbd-reconnect-on-open b/tests/qemu-iotests/tests/nbd-reconnect-on-open
-index XXXXXXX..XXXXXXX 100755
---- a/tests/qemu-iotests/tests/nbd-reconnect-on-open
-+++ b/tests/qemu-iotests/tests/nbd-reconnect-on-open
-@@ -XXX,XX +XXX,XX @@ def check_fail_to_connect(open_timeout):
-     log(f'Check fail to connect with {open_timeout} seconds of timeout')
-     start_t = time.time()
--    qemu_io_log(*create_args(open_timeout))
-+    qemu_io_log(*create_args(open_timeout), check=False)
-     delta_t = time.time() - start_t
-     max_delta = open_timeout + 0.2
---
-.35.1

-[PULL 06/14] iotests/040: Fix TestCommitWithFilters test
+[Qemu-devel] [PULL v2 16/24] aio-posix: partially inline aio_dispatch into aio_poll
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-Without this change, asserting that qemu_io always returns 0 causes this
+This patch prepares for the removal of unnecessary lockcnt inc/dec pairs.
-test to fail in a way we happened not to be catching previously:
+Extract the dispatching loop for file descriptor handlers into a new
 function aio_dispatch_handlers, and then inline aio_dispatch into
 aio_poll.
- qemu.utils.VerboseProcessError: Command
+aio_dispatch can now become void.
   '('/home/jsnow/src/qemu/bin/git/tests/qemu-iotests/../../qemu-io',
   '--cache', 'writeback', '--aio', 'threads', '-f', 'qcow2', '-c',
   'read -P 4 3M 1M',
   '/home/jsnow/src/qemu/bin/git/tests/qemu-iotests/scratch/3.img')'
   returned non-zero exit status 1.
   ┏━ output ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
   ┃ qemu-io: can't open device
   ┃ /home/jsnow/src/qemu/bin/git/tests/qemu-iotests/scratch/3.img:
   ┃ Could not open backing file: Could not open backing file: Throttle
   ┃ group 'tg' does not exist
   ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-The commit jobs changes the backing file string stored in the image file
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-header belonging to the node above the commit’s top node to point to the
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-commit target (the base node).  QEMU tries to be as accurate as
+Reviewed-by: Fam Zheng <famz@redhat.com>
-possible, and so in these test cases will include the filter that is
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
-part of the block graph in that backing file string (by virtue of making
+Message-id: 20170213135235.12274-17-pbonzini@redhat.com
-it a json:{} description of the post-commit subgraph).  This makes
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
-little sense outside of QEMU, though: Specifically, the throttle node in
+---
-that subgraph will dearly miss its supposedly associated throttle group
+ include/block/aio.h |  6 +-----
-object.
+ util/aio-posix.c    | 44 ++++++++++++++------------------------------
  util/aio-win32.c    | 13 ++++---------
  util/async.c        |  2 +-
 files changed, 20 insertions(+), 45 deletions(-)
-When starting the commit job, we can specify a custom backing file
+diff --git a/include/block/aio.h b/include/block/aio.h
-string to write into said image file, so let’s use that feature to write
+index XXXXXXX..XXXXXXX 100644
-the plain filename of the backing chain’s next actual image file there.
+--- a/include/block/aio.h
++++ b/include/block/aio.h
-Explicitly provide the backing file so that opening the file outside of
+@@ -XXX,XX +XXX,XX @@ bool aio_pending(AioContext *ctx);
-QEMU (Where we will not have throttle groups) will succeed.
+ /* Dispatch any pending callbacks from the GSource attached to the AioContext.
+  *
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
+  * This is used internally in the implementation of the GSource.
-Signed-off-by: John Snow <jsnow@redhat.com>
+- *
-Reviewed-by: Eric Blake <eblake@redhat.com>
+- * @dispatch_fds: true to process fds, false to skip them
-Message-Id: <20220418211504.943969-6-jsnow@redhat.com>
+- *                (can be used as an optimization by callers that know there
----
+- *                are no fds ready)
- tests/qemu-iotests/040 | 6 ++++--
+  */
-file changed, 4 insertions(+), 2 deletions(-)
+-bool aio_dispatch(AioContext *ctx, bool dispatch_fds);
++void aio_dispatch(AioContext *ctx);
-diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
-index XXXXXXX..XXXXXXX 100755
+ /* Progress in completing AIO work to occur.  This can issue new pending
---- a/tests/qemu-iotests/040
+  * aio as a result of executing I/O completion or bh callbacks.
-+++ b/tests/qemu-iotests/040
+diff --git a/util/aio-posix.c b/util/aio-posix.c
-@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
+index XXXXXXX..XXXXXXX 100644
-                              job_id='commit',
+--- a/util/aio-posix.c
-                              device='top-filter',
++++ b/util/aio-posix.c
-                              top_node='cow-2',
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
--                             base_node='cow-1')
+     AioHandler *node, *tmp;
-+                             base_node='cow-1',
+     bool progress = false;
-+                             backing_file=self.img1)
-         self.assert_qmp(result, 'return', {})
+-    /*
-         self.wait_until_completed(drive='commit')
+-     * We have to walk very carefully in case aio_set_fd_handler is
+-     * called while we're walking.
-@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
+-     */
-                              job_id='commit',
+-    qemu_lockcnt_inc(&ctx->list_lock);
-                              device='top-filter',
+-
-                              top_node='cow-1',
+     QLIST_FOREACH_SAFE_RCU(node, &ctx->aio_handlers, node, tmp) {
--                             base_node='cow-0')
+         int revents;
-+                             base_node='cow-0',
-+                             backing_file=self.img0)
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
-         self.assert_qmp(result, 'return', {})
+         }
-         self.wait_until_completed(drive='commit')
+     }
 -    qemu_lockcnt_dec(&ctx->list_lock);
      return progress;
  }
 -/*
 - * Note that dispatch_fds == false has the side-effect of post-poning the
 - * freeing of deleted handlers.
 - */
 -bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
 +void aio_dispatch(AioContext *ctx)
  {
 -    bool progress;
 +    aio_bh_poll(ctx);
 -    /*
 -     * If there are callbacks left that have been queued, we need to call them.
 -     * Do not call select in this case, because it is possible that the caller
 -     * does not need a complete flush (as is the case for aio_poll loops).
 -     */
 -    progress = aio_bh_poll(ctx);
 +    qemu_lockcnt_inc(&ctx->list_lock);
 +    aio_dispatch_handlers(ctx);
 +    qemu_lockcnt_dec(&ctx->list_lock);
 -    if (dispatch_fds) {
 -        progress |= aio_dispatch_handlers(ctx);
 -    }
 -
 -    /* Run our timers */
 -    progress |= timerlistgroup_run_timers(&ctx->tlg);
 -
 -    return progress;
 +    timerlistgroup_run_timers(&ctx->tlg);
  }
  /* These thread-local variables are used only in a small part of aio_poll
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
      npfd = 0;
      qemu_lockcnt_dec(&ctx->list_lock);
 -    /* Run dispatch even if there were no readable fds to run timers */
 -    if (aio_dispatch(ctx, ret > 0)) {
 -        progress = true;
 +    progress |= aio_bh_poll(ctx);
 +
 +    if (ret > 0) {
 +        qemu_lockcnt_inc(&ctx->list_lock);
 +        progress |= aio_dispatch_handlers(ctx);
 +        qemu_lockcnt_dec(&ctx->list_lock);
      }
 +    progress |= timerlistgroup_run_timers(&ctx->tlg);
 +
      return progress;
  }
 diff --git a/util/aio-win32.c b/util/aio-win32.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-win32.c
 +++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
      return progress;
  }
 -bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
 +void aio_dispatch(AioContext *ctx)
  {
 -    bool progress;
 -
 -    progress = aio_bh_poll(ctx);
 -    if (dispatch_fds) {
 -        progress |= aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
 -    }
 -    progress |= timerlistgroup_run_timers(&ctx->tlg);
 -    return progress;
 +    aio_bh_poll(ctx);
 +    aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
 +    timerlistgroup_run_timers(&ctx->tlg);
  }
  bool aio_poll(AioContext *ctx, bool blocking)
 diff --git a/util/async.c b/util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ aio_ctx_dispatch(GSource     *source,
      AioContext *ctx = (AioContext *) source;
      assert(callback == NULL);
 -    aio_dispatch(ctx, true);
 +    aio_dispatch(ctx);
      return true;
  }
 --
-.35.1
+.9.3

-[PULL 11/14] iotests: remove qemu_io_pipe_and_status()
+[Qemu-devel] [PULL v2 17/24] async: remove unnecessary inc/dec pairs
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-I know we just added it, sorry. This is done in favor of qemu_io() which
+Pull the increment/decrement pair out of aio_bh_poll and into the
-*also* returns the console output and status, but with more robust error
+callers.
 handling on failure.
-Signed-off-by: John Snow <jsnow@redhat.com>
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
-Reviewed-by: Eric Blake <eblake@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
-Message-Id: <20220418211504.943969-11-jsnow@redhat.com>
+Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
+Message-id: 20170213135235.12274-18-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qemu-iotests/iotests.py | 3 ---
+ util/aio-posix.c |  8 +++-----
-file changed, 3 deletions(-)
+ util/aio-win32.c |  8 ++++----
  util/async.c     | 12 ++++++------
 files changed, 13 insertions(+), 15 deletions(-)
-diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
+diff --git a/util/aio-posix.c b/util/aio-posix.c
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/iotests.py
+--- a/util/aio-posix.c
-+++ b/tests/qemu-iotests/iotests.py
++++ b/util/aio-posix.c
-@@ -XXX,XX +XXX,XX @@ def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
+@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
-     return qemu_tool(*qemu_io_wrap_args(args),
-                      check=check, combine_stdio=combine_stdio)
+ void aio_dispatch(AioContext *ctx)
+ {
--def qemu_io_pipe_and_status(*args):
++    qemu_lockcnt_inc(&ctx->list_lock);
--    return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))
+     aio_bh_poll(ctx);
 -
- def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
+-    qemu_lockcnt_inc(&ctx->list_lock);
-     result = qemu_io(*args, check=False)
+     aio_dispatch_handlers(ctx);
-     log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
+     qemu_lockcnt_dec(&ctx->list_lock);
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
      }
      npfd = 0;
 -    qemu_lockcnt_dec(&ctx->list_lock);
      progress |= aio_bh_poll(ctx);
      if (ret > 0) {
 -        qemu_lockcnt_inc(&ctx->list_lock);
          progress |= aio_dispatch_handlers(ctx);
 -        qemu_lockcnt_dec(&ctx->list_lock);
      }
 +    qemu_lockcnt_dec(&ctx->list_lock);
 +
      progress |= timerlistgroup_run_timers(&ctx->tlg);
      return progress;
 diff --git a/util/aio-win32.c b/util/aio-win32.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-win32.c
 +++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
      bool progress = false;
      AioHandler *tmp;
 -    qemu_lockcnt_inc(&ctx->list_lock);
 -
      /*
       * We have to walk very carefully in case aio_set_fd_handler is
       * called while we're walking.
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
          }
      }
 -    qemu_lockcnt_dec(&ctx->list_lock);
      return progress;
  }
  void aio_dispatch(AioContext *ctx)
  {
 +    qemu_lockcnt_inc(&ctx->list_lock);
      aio_bh_poll(ctx);
      aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
 +    qemu_lockcnt_dec(&ctx->list_lock);
      timerlistgroup_run_timers(&ctx->tlg);
  }
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
          }
      }
 -    qemu_lockcnt_dec(&ctx->list_lock);
      first = true;
      /* ctx->notifier is always registered.  */
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
          progress |= aio_dispatch_handlers(ctx, event);
      } while (count > 0);
 +    qemu_lockcnt_dec(&ctx->list_lock);
 +
      progress |= timerlistgroup_run_timers(&ctx->tlg);
      return progress;
  }
 diff --git a/util/async.c b/util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ void aio_bh_call(QEMUBH *bh)
      bh->cb(bh->opaque);
  }
 -/* Multiple occurrences of aio_bh_poll cannot be called concurrently */
 +/* Multiple occurrences of aio_bh_poll cannot be called concurrently.
 + * The count in ctx->list_lock is incremented before the call, and is
 + * not affected by the call.
 + */
  int aio_bh_poll(AioContext *ctx)
  {
      QEMUBH *bh, **bhp, *next;
      int ret;
      bool deleted = false;
 -    qemu_lockcnt_inc(&ctx->list_lock);
 -
      ret = 0;
      for (bh = atomic_rcu_read(&ctx->first_bh); bh; bh = next) {
          next = atomic_rcu_read(&bh->next);
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
      /* remove deleted bhs */
      if (!deleted) {
 -        qemu_lockcnt_dec(&ctx->list_lock);
          return ret;
      }
 -    if (qemu_lockcnt_dec_and_lock(&ctx->list_lock)) {
 +    if (qemu_lockcnt_dec_if_lock(&ctx->list_lock)) {
          bhp = &ctx->first_bh;
          while (*bhp) {
              bh = *bhp;
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
                  bhp = &bh->next;
              }
          }
 -        qemu_lockcnt_unlock(&ctx->list_lock);
 +        qemu_lockcnt_inc_and_unlock(&ctx->list_lock);
      }
      return ret;
  }
 --
-.35.1
+.9.3

-[PULL 10/14] iotests/image-fleecing: switch to qemu_io()
+[Qemu-devel] [PULL v2 18/24] block: document fields protected by AioContext lock
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-This test expects failure ... but only sometimes. When? Why?
+Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
 Reviewed-by: Fam Zheng <famz@redhat.com>
 Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
 Message-id: 20170213135235.12274-19-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
  include/block/block_int.h      | 64 +++++++++++++++++++++++++-----------------
  include/sysemu/block-backend.h | 14 ++++++---
 files changed, 49 insertions(+), 29 deletions(-)
-It's for reads of a region not defined by a bitmap. Adjust the test to
+diff --git a/include/block/block_int.h b/include/block/block_int.h
-be more explicit about what it expects to fail and why.
+index XXXXXXX..XXXXXXX 100644
 --- a/include/block/block_int.h
 +++ b/include/block/block_int.h
@@ -XXX,XX +XXX,XX @@ struct BdrvChild {
   * copied as well.
   */
  struct BlockDriverState {
 -    int64_t total_sectors; /* if we are reading a disk image, give its
 -                              size in sectors */
 +    /* Protected by big QEMU lock or read-only after opening.  No special
 +     * locking needed during I/O...
 +     */
      int open_flags; /* flags used to open the file, re-used for re-open */
      bool read_only; /* if true, the media is read only */
      bool encrypted; /* if true, the media is encrypted */
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
      bool sg;        /* if true, the device is a /dev/sg* */
      bool probed;    /* if true, format was probed rather than specified */
 -    int copy_on_read; /* if nonzero, copy read backing sectors into image.
 -                         note this is a reference count */
 -
 -    CoQueue flush_queue;            /* Serializing flush queue */
 -    bool active_flush_req;          /* Flush request in flight? */
 -    unsigned int write_gen;         /* Current data generation */
 -    unsigned int flushed_gen;       /* Flushed write generation */
 -
      BlockDriver *drv; /* NULL means no media */
      void *opaque;
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
      BdrvChild *backing;
      BdrvChild *file;
 -    /* Callback before write request is processed */
 -    NotifierWithReturnList before_write_notifiers;
 -
 -    /* number of in-flight requests; overall and serialising */
 -    unsigned int in_flight;
 -    unsigned int serialising_in_flight;
 -
 -    bool wakeup;
 -
 -    /* Offset after the highest byte written to */
 -    uint64_t wr_highest_offset;
 -
      /* I/O Limits */
      BlockLimits bl;
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
      QTAILQ_ENTRY(BlockDriverState) bs_list;
      /* element of the list of monitor-owned BDS */
      QTAILQ_ENTRY(BlockDriverState) monitor_list;
 -    QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
      int refcnt;
 -    QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
 -
      /* operation blockers */
      QLIST_HEAD(, BdrvOpBlocker) op_blockers[BLOCK_OP_TYPE_MAX];
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
      /* The error object in use for blocking operations on backing_hd */
      Error *backing_blocker;
 +    /* Protected by AioContext lock */
 +
 +    /* If true, copy read backing sectors into image.  Can be >1 if more
 +     * than one client has requested copy-on-read.
 +     */
 +    int copy_on_read;
 +
 +    /* If we are reading a disk image, give its size in sectors.
 +     * Generally read-only; it is written to by load_vmstate and save_vmstate,
 +     * but the block layer is quiescent during those.
 +     */
 +    int64_t total_sectors;
 +
 +    /* Callback before write request is processed */
 +    NotifierWithReturnList before_write_notifiers;
 +
 +    /* number of in-flight requests; overall and serialising */
 +    unsigned int in_flight;
 +    unsigned int serialising_in_flight;
 +
 +    bool wakeup;
 +
 +    /* Offset after the highest byte written to */
 +    uint64_t wr_highest_offset;
 +
      /* threshold limit for writes, in bytes. "High water mark". */
      uint64_t write_threshold_offset;
      NotifierWithReturn write_threshold_notifier;
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
      /* counter for nested bdrv_io_plug */
      unsigned io_plugged;
 +    QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
 +    CoQueue flush_queue;                  /* Serializing flush queue */
 +    bool active_flush_req;                /* Flush request in flight? */
 +    unsigned int write_gen;               /* Current data generation */
 +    unsigned int flushed_gen;             /* Flushed write generation */
 +
 +    QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
 +
 +    /* do we need to tell the quest if we have a volatile write cache? */
 +    int enable_write_cache;
 +
      int quiesce_counter;
  };
 diff --git a/include/sysemu/block-backend.h b/include/sysemu/block-backend.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/sysemu/block-backend.h
 +++ b/include/sysemu/block-backend.h
@@ -XXX,XX +XXX,XX @@ typedef struct BlockDevOps {
   * fields that must be public. This is in particular for QLIST_ENTRY() and
   * friends so that BlockBackends can be kept in lists outside block-backend.c */
  typedef struct BlockBackendPublic {
 -    /* I/O throttling.
 -     * throttle_state tells us if this BlockBackend has I/O limits configured.
 -     * io_limits_disabled tells us if they are currently being enforced */
 +    /* I/O throttling has its own locking, but also some fields are
 +     * protected by the AioContext lock.
 +     */
 +
 +    /* Protected by AioContext lock.  */
      CoQueue      throttled_reqs[2];
 +
 +    /* Nonzero if the I/O limits are currently being ignored; generally
 +     * it is zero.  */
      unsigned int io_limits_disabled;
      /* The following fields are protected by the ThrottleGroup lock.
 -     * See the ThrottleGroup documentation for details. */
 +     * See the ThrottleGroup documentation for details.
 +     * throttle_state tells us if I/O limits are configured. */
      ThrottleState *throttle_state;
      ThrottleTimers throttle_timers;
      unsigned       pending_reqs[2];
 --
 .9.3
-Signed-off-by: John Snow <jsnow@redhat.com>
-Reviewed-by: Eric Blake <eblake@redhat.com>
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
-Message-Id: <20220418211504.943969-10-jsnow@redhat.com>
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
----
- tests/qemu-iotests/tests/image-fleecing | 28 +++++++++++++++++--------
-file changed, 19 insertions(+), 9 deletions(-)
-diff --git a/tests/qemu-iotests/tests/image-fleecing b/tests/qemu-iotests/tests/image-fleecing
-index XXXXXXX..XXXXXXX 100755
---- a/tests/qemu-iotests/tests/image-fleecing
-+++ b/tests/qemu-iotests/tests/image-fleecing
-@@ -XXX,XX +XXX,XX @@
- #
- # Creator/Owner: John Snow <jsnow@redhat.com>
-+from subprocess import CalledProcessError
-+
- import iotests
--from iotests import log, qemu_img, qemu_io, qemu_io_silent, \
--    qemu_io_pipe_and_status
-+from iotests import log, qemu_img, qemu_io, qemu_io_silent
- iotests.script_initialize(
-     supported_fmts=['qcow2'],
-@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
-         for p in patterns + zeroes:
-             cmd = 'read -P%s %s %s' % p
-             log(cmd)
--            out, ret = qemu_io_pipe_and_status('-r', '-f', 'raw', '-c', cmd,
--                                               nbd_uri)
--            if ret != 0:
--                print(out)
-+
-+            try:
-+                qemu_io('-r', '-f', 'raw', '-c', cmd, nbd_uri)
-+            except CalledProcessError as exc:
-+                if bitmap and p in zeroes:
-+                    log(exc.stdout)
-+                else:
-+                    raise
-     log('')
-     log('--- Testing COW ---')
-@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
-             args += [target_img_path]
-         else:
-             args += ['-f', 'raw', nbd_uri]
--        out, ret = qemu_io_pipe_and_status(*args)
--        if ret != 0:
--            print(out)
-+
-+        try:
-+            qemu_io(*args)
-+        except CalledProcessError as exc:
-+            if bitmap and p in zeroes:
-+                log(exc.stdout)
-+            else:
-+                raise
-     log('')
-     log('--- Cleanup ---')
---
-.35.1

-[PULL 09/14] iotests/migration-permissions: use assertRaises() for qemu_io() negative test
+[Qemu-devel] [PULL v2 19/24] coroutine-lock: make CoMutex thread-safe
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-Modify this test to use assertRaises for its negative testing of
+This uses the lock-free mutex described in the paper '"Blocking without
-qemu_io. If the exception raised does not match the one we tell it to
+Locking", or LFTHREADS: A lock-free thread library' by Gidenstam and
-expect, we get *that* exception unhandled. If we get no exception, we
+Papatriantafilou.  The same technique is used in OSv, and in fact
-get a unittest assertion failure and the provided emsg printed to
+the code is essentially a conversion to C of OSv's code.
-screen.
+[Added missing coroutine_fn in tests/test-aio-multithread.c.
-If we get the CalledProcessError exception but the output is not what we
+--Stefan]
-expect, we re-raise the original CalledProcessError.
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Tidy.
+Reviewed-by: Fam Zheng <famz@redhat.com>
+Message-id: 20170213181244.16297-2-pbonzini@redhat.com
-(Note: Yes, you can reference "with" objects after that block ends; it
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 just means that ctx.__exit__(...) will have been called on it. It does
 not *actually* go out of scope. unittests expects you to want to inspect
 the Exception object, so they leave it defined post-exit.)
 Signed-off-by: John Snow <jsnow@redhat.com>
 Reviewed-by: Eric Blake <eblake@redhat.com>
 Tested-by: Eric Blake <eblake@redhat.com>
 Reviewed-by: Hanna Reitz <hreitz@redhat.com>
 Message-Id: <20220418211504.943969-9-jsnow@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- .../qemu-iotests/tests/migration-permissions  | 28 +++++++++----------
+ include/qemu/coroutine.h     |  17 ++++-
-file changed, 14 insertions(+), 14 deletions(-)
+ tests/test-aio-multithread.c |  86 ++++++++++++++++++++++++
+ util/qemu-coroutine-lock.c   | 155 ++++++++++++++++++++++++++++++++++++++++---
-diff --git a/tests/qemu-iotests/tests/migration-permissions b/tests/qemu-iotests/tests/migration-permissions
+ util/trace-events            |   1 +
-index XXXXXXX..XXXXXXX 100755
+files changed, 246 insertions(+), 13 deletions(-)
---- a/tests/qemu-iotests/tests/migration-permissions
-+++ b/tests/qemu-iotests/tests/migration-permissions
+diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/qemu/coroutine.h
 +++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue);
  /**
   * Provides a mutex that can be used to synchronise coroutines
   */
 +struct CoWaitRecord;
  typedef struct CoMutex {
 -    bool locked;
 +    /* Count of pending lockers; 0 for a free mutex, 1 for an
 +     * uncontended mutex.
 +     */
 +    unsigned locked;
 +
 +    /* A queue of waiters.  Elements are added atomically in front of
 +     * from_push.  to_pop is only populated, and popped from, by whoever
 +     * is in charge of the next wakeup.  This can be an unlocker or,
 +     * through the handoff protocol, a locker that is about to go to sleep.
 +     */
 +    QSLIST_HEAD(, CoWaitRecord) from_push, to_pop;
 +
 +    unsigned handoff, sequence;
 +
      Coroutine *holder;
 -    CoQueue queue;
  } CoMutex;
  /**
 diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
 index XXXXXXX..XXXXXXX 100644
 --- a/tests/test-aio-multithread.c
 +++ b/tests/test-aio-multithread.c
@@ -XXX,XX +XXX,XX @@ static void test_multi_co_schedule_10(void)
      test_multi_co_schedule(10);
  }
 +/* CoMutex thread-safety.  */
 +
 +static uint32_t atomic_counter;
 +static uint32_t running;
 +static uint32_t counter;
 +static CoMutex comutex;
 +
 +static void coroutine_fn test_multi_co_mutex_entry(void *opaque)
 +{
 +    while (!atomic_mb_read(&now_stopping)) {
 +        qemu_co_mutex_lock(&comutex);
 +        counter++;
 +        qemu_co_mutex_unlock(&comutex);
 +
 +        /* Increase atomic_counter *after* releasing the mutex.  Otherwise
 +         * there is a chance (it happens about 1 in 3 runs) that the iothread
 +         * exits before the coroutine is woken up, causing a spurious
 +         * assertion failure.
 +         */
 +        atomic_inc(&atomic_counter);
 +    }
 +    atomic_dec(&running);
 +}
 +
 +static void test_multi_co_mutex(int threads, int seconds)
 +{
 +    int i;
 +
 +    qemu_co_mutex_init(&comutex);
 +    counter = 0;
 +    atomic_counter = 0;
 +    now_stopping = false;
 +
 +    create_aio_contexts();
 +    assert(threads <= NUM_CONTEXTS);
 +    running = threads;
 +    for (i = 0; i < threads; i++) {
 +        Coroutine *co1 = qemu_coroutine_create(test_multi_co_mutex_entry, NULL);
 +        aio_co_schedule(ctx[i], co1);
 +    }
 +
 +    g_usleep(seconds * 1000000);
 +
 +    atomic_mb_set(&now_stopping, true);
 +    while (running > 0) {
 +        g_usleep(100000);
 +    }
 +
 +    join_aio_contexts();
 +    g_test_message("%d iterations/second\n", counter / seconds);
 +    g_assert_cmpint(counter, ==, atomic_counter);
 +}
 +
 +/* Testing with NUM_CONTEXTS threads focuses on the queue.  The mutex however
 + * is too contended (and the threads spend too much time in aio_poll)
 + * to actually stress the handoff protocol.
 + */
 +static void test_multi_co_mutex_1(void)
 +{
 +    test_multi_co_mutex(NUM_CONTEXTS, 1);
 +}
 +
 +static void test_multi_co_mutex_10(void)
 +{
 +    test_multi_co_mutex(NUM_CONTEXTS, 10);
 +}
 +
 +/* Testing with fewer threads stresses the handoff protocol too.  Still, the
 + * case where the locker _can_ pick up a handoff is very rare, happening
 + * about 10 times in 1 million, so increase the runtime a bit compared to
 + * other "quick" testcases that only run for 1 second.
 + */
 +static void test_multi_co_mutex_2_3(void)
 +{
 +    test_multi_co_mutex(2, 3);
 +}
 +
 +static void test_multi_co_mutex_2_30(void)
 +{
 +    test_multi_co_mutex(2, 30);
 +}
 +
  /* End of tests.  */
  int main(int argc, char **argv)
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
      g_test_add_func("/aio/multi/lifecycle", test_lifecycle);
      if (g_test_quick()) {
          g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
 +        g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_1);
 +        g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_3);
      } else {
          g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
 +        g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_10);
 +        g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_30);
      }
      return g_test_run();
  }
 diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/qemu-coroutine-lock.c
 +++ b/util/qemu-coroutine-lock.c
 @@ -XXX,XX +XXX,XX @@
- #
+  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
- import os
+  * THE SOFTWARE.
-+from subprocess import CalledProcessError
++ *
-+
++ * The lock-free mutex implementation is based on OSv
- import iotests
++ * (core/lfmutex.cc, include/lockfree/mutex.hh).
- from iotests import imgfmt, qemu_img_create, qemu_io
++ * Copyright (C) 2013 Cloudius Systems, Ltd.
+  */
-@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
-     def test_post_migration_permissions(self):
+ #include "qemu/osdep.h"
-         # Try to access the image R/W, which should fail because virtio-blk
+@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue)
-         # has not been configured with share-rw=on
+     return QSIMPLEQ_FIRST(&queue->entries) == NULL;
--        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
+ }
--        if not log.strip():
--            print('ERROR (pre-migration): qemu-io should not be able to '
++/* The wait records are handled with a multiple-producer, single-consumer
--                  'access this image, but it reported no error')
++ * lock-free queue.  There cannot be two concurrent pop_waiter() calls
--        else:
++ * because pop_waiter() can only be called while mutex->handoff is zero.
--            # This is the expected output
++ * This can happen in three cases:
--            assert 'Is another process using the image' in log
++ * - in qemu_co_mutex_unlock, before the hand-off protocol has started.
-+        emsg = ('ERROR (pre-migration): qemu-io should not be able to '
++ *   In this case, qemu_co_mutex_lock will see mutex->handoff == 0 and
-+                'access this image, but it reported no error')
++ *   not take part in the handoff.
-+        with self.assertRaises(CalledProcessError, msg=emsg) as ctx:
++ * - in qemu_co_mutex_lock, if it steals the hand-off responsibility from
-+            qemu_io('-f', imgfmt, '-c', 'quit', test_img)
++ *   qemu_co_mutex_unlock.  In this case, qemu_co_mutex_unlock will fail
-+        if 'Is another process using the image' not in ctx.exception.stdout:
++ *   the cmpxchg (it will see either 0 or the next sequence value) and
-+            raise ctx.exception
++ *   exit.  The next hand-off cannot begin until qemu_co_mutex_lock has
++ *   woken up someone.
-         # Now migrate the VM
++ * - in qemu_co_mutex_unlock, if it takes the hand-off token itself.
-         self.vm_s.qmp('migrate', uri=f'unix:{mig_sock}')
++ *   In this case another iteration starts with mutex->handoff == 0;
-@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
++ *   a concurrent qemu_co_mutex_lock will fail the cmpxchg, and
++ *   qemu_co_mutex_unlock will go back to case (1).
-         # Try the same qemu-io access again, verifying that the WRITE
++ *
-         # permission remains unshared
++ * The following functions manage this queue.
--        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
++ */
--        if not log.strip():
++typedef struct CoWaitRecord {
--            print('ERROR (post-migration): qemu-io should not be able to '
++    Coroutine *co;
--                  'access this image, but it reported no error')
++    QSLIST_ENTRY(CoWaitRecord) next;
--        else:
++} CoWaitRecord;
--            # This is the expected output
++
--            assert 'Is another process using the image' in log
++static void push_waiter(CoMutex *mutex, CoWaitRecord *w)
-+        emsg = ('ERROR (post-migration): qemu-io should not be able to '
++{
-+                'access this image, but it reported no error')
++    w->co = qemu_coroutine_self();
-+        with self.assertRaises(CalledProcessError, msg=emsg) as ctx:
++    QSLIST_INSERT_HEAD_ATOMIC(&mutex->from_push, w, next);
-+            qemu_io('-f', imgfmt, '-c', 'quit', test_img)
++}
-+        if 'Is another process using the image' not in ctx.exception.stdout:
++
-+            raise ctx.exception
++static void move_waiters(CoMutex *mutex)
++{
++    QSLIST_HEAD(, CoWaitRecord) reversed;
- if __name__ == '__main__':
++    QSLIST_MOVE_ATOMIC(&reversed, &mutex->from_push);
 +    while (!QSLIST_EMPTY(&reversed)) {
 +        CoWaitRecord *w = QSLIST_FIRST(&reversed);
 +        QSLIST_REMOVE_HEAD(&reversed, next);
 +        QSLIST_INSERT_HEAD(&mutex->to_pop, w, next);
 +    }
 +}
 +
 +static CoWaitRecord *pop_waiter(CoMutex *mutex)
 +{
 +    CoWaitRecord *w;
 +
 +    if (QSLIST_EMPTY(&mutex->to_pop)) {
 +        move_waiters(mutex);
 +        if (QSLIST_EMPTY(&mutex->to_pop)) {
 +            return NULL;
 +        }
 +    }
 +    w = QSLIST_FIRST(&mutex->to_pop);
 +    QSLIST_REMOVE_HEAD(&mutex->to_pop, next);
 +    return w;
 +}
 +
 +static bool has_waiters(CoMutex *mutex)
 +{
 +    return QSLIST_EMPTY(&mutex->to_pop) || QSLIST_EMPTY(&mutex->from_push);
 +}
 +
  void qemu_co_mutex_init(CoMutex *mutex)
  {
      memset(mutex, 0, sizeof(*mutex));
 -    qemu_co_queue_init(&mutex->queue);
  }
 -void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
 +static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
  {
      Coroutine *self = qemu_coroutine_self();
 +    CoWaitRecord w;
 +    unsigned old_handoff;
      trace_qemu_co_mutex_lock_entry(mutex, self);
 +    w.co = self;
 +    push_waiter(mutex, &w);
 -    while (mutex->locked) {
 -        qemu_co_queue_wait(&mutex->queue);
 +    /* This is the "Responsibility Hand-Off" protocol; a lock() picks from
 +     * a concurrent unlock() the responsibility of waking somebody up.
 +     */
 +    old_handoff = atomic_mb_read(&mutex->handoff);
 +    if (old_handoff &&
 +        has_waiters(mutex) &&
 +        atomic_cmpxchg(&mutex->handoff, old_handoff, 0) == old_handoff) {
 +        /* There can be no concurrent pops, because there can be only
 +         * one active handoff at a time.
 +         */
 +        CoWaitRecord *to_wake = pop_waiter(mutex);
 +        Coroutine *co = to_wake->co;
 +        if (co == self) {
 +            /* We got the lock ourselves!  */
 +            assert(to_wake == &w);
 +            return;
 +        }
 +
 +        aio_co_wake(co);
      }
 -    mutex->locked = true;
 -    mutex->holder = self;
 -    self->locks_held++;
 -
 +    qemu_coroutine_yield();
      trace_qemu_co_mutex_lock_return(mutex, self);
  }
 +void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
 +{
 +    Coroutine *self = qemu_coroutine_self();
 +
 +    if (atomic_fetch_inc(&mutex->locked) == 0) {
 +        /* Uncontended.  */
 +        trace_qemu_co_mutex_lock_uncontended(mutex, self);
 +    } else {
 +        qemu_co_mutex_lock_slowpath(mutex);
 +    }
 +    mutex->holder = self;
 +    self->locks_held++;
 +}
 +
  void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
  {
      Coroutine *self = qemu_coroutine_self();
      trace_qemu_co_mutex_unlock_entry(mutex, self);
 -    assert(mutex->locked == true);
 +    assert(mutex->locked);
      assert(mutex->holder == self);
      assert(qemu_in_coroutine());
 -    mutex->locked = false;
      mutex->holder = NULL;
      self->locks_held--;
 -    qemu_co_queue_next(&mutex->queue);
 +    if (atomic_fetch_dec(&mutex->locked) == 1) {
 +        /* No waiting qemu_co_mutex_lock().  Pfew, that was easy!  */
 +        return;
 +    }
 +
 +    for (;;) {
 +        CoWaitRecord *to_wake = pop_waiter(mutex);
 +        unsigned our_handoff;
 +
 +        if (to_wake) {
 +            Coroutine *co = to_wake->co;
 +            aio_co_wake(co);
 +            break;
 +        }
 +
 +        /* Some concurrent lock() is in progress (we know this because
 +         * mutex->locked was >1) but it hasn't yet put itself on the wait
 +         * queue.  Pick a sequence number for the handoff protocol (not 0).
 +         */
 +        if (++mutex->sequence == 0) {
 +            mutex->sequence = 1;
 +        }
 +
 +        our_handoff = mutex->sequence;
 +        atomic_mb_set(&mutex->handoff, our_handoff);
 +        if (!has_waiters(mutex)) {
 +            /* The concurrent lock has not added itself yet, so it
 +             * will be able to pick our handoff.
 +             */
 +            break;
 +        }
 +
 +        /* Try to do the handoff protocol ourselves; if somebody else has
 +         * already taken it, however, we're done and they're responsible.
 +         */
 +        if (atomic_cmpxchg(&mutex->handoff, our_handoff, 0) != our_handoff) {
 +            break;
 +        }
 +    }
      trace_qemu_co_mutex_unlock_return(mutex, self);
  }
 diff --git a/util/trace-events b/util/trace-events
 index XXXXXXX..XXXXXXX 100644
 --- a/util/trace-events
 +++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@ qemu_coroutine_terminate(void *co) "self %p"
  # util/qemu-coroutine-lock.c
  qemu_co_queue_run_restart(void *co) "co %p"
 +qemu_co_mutex_lock_uncontended(void *mutex, void *self) "mutex %p self %p"
  qemu_co_mutex_lock_entry(void *mutex, void *self) "mutex %p self %p"
  qemu_co_mutex_lock_return(void *mutex, void *self) "mutex %p self %p"
  qemu_co_mutex_unlock_entry(void *mutex, void *self) "mutex %p self %p"
 --
-.35.1
+.9.3

-[PULL 05/14] iotests/040: Don't check image pattern on zero-length image
+[Qemu-devel] [PULL v2 20/24] coroutine-lock: add limited spinning to CoMutex
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-qemu-io fails on read/write beyond end-of-file on raw images, so skip
+Running a very small critical section on pthread_mutex_t and CoMutex
-these invocations when running the zero-length image tests.
+shows that pthread_mutex_t is much faster because it doesn't actually
 go to sleep.  What happens is that the critical section is shorter
 than the latency of entering the kernel and thus FUTEX_WAIT always
 fails.  With CoMutex there is no such latency but you still want to
 avoid wait and wakeup.  So introduce it artificially.
-Signed-off-by: John Snow <jsnow@redhat.com>
+This only works with one waiters; because CoMutex is fair, it will
-Reviewed-by: Eric Blake <eblake@redhat.com>
+always have more waits and wakeups than a pthread_mutex_t.
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
-Message-Id: <20220418211504.943969-5-jsnow@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Signed-off-by: Hanna Reitz <hreitz@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
 Message-id: 20170213181244.16297-3-pbonzini@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qemu-iotests/040 | 14 ++++++++++++--
+ include/qemu/coroutine.h   |  5 +++++
-file changed, 12 insertions(+), 2 deletions(-)
+ util/qemu-coroutine-lock.c | 51 ++++++++++++++++++++++++++++++++++++++++------
  util/qemu-coroutine.c      |  2 +-
 files changed, 51 insertions(+), 7 deletions(-)
-diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
+diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
-index XXXXXXX..XXXXXXX 100755
+index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/040
+--- a/include/qemu/coroutine.h
-+++ b/tests/qemu-iotests/040
++++ b/include/qemu/coroutine.h
-@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
+@@ -XXX,XX +XXX,XX @@ typedef struct CoMutex {
-         qemu_img('create', '-f', iotests.imgfmt,
+      */
-                  '-o', 'backing_file=%s' % mid_img,
+     unsigned locked;
-                  '-F', iotests.imgfmt, test_img)
--        qemu_io('-f', 'raw', '-c', 'write -P 0xab 0 524288', backing_img)
++    /* Context that is holding the lock.  Useful to avoid spinning
--        qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xef 524288 524288', mid_img)
++     * when two coroutines on the same AioContext try to get the lock. :)
-+        if self.image_len:
++     */
-+            qemu_io('-f', 'raw', '-c', 'write -P 0xab 0 524288', backing_img)
++    AioContext *ctx;
-+            qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xef 524288 524288',
++
-+                    mid_img)
+     /* A queue of waiters.  Elements are added atomically in front of
-         self.vm = iotests.VM().add_drive(test_img, "node-name=top,backing.node-name=mid,backing.backing.node-name=base", interface="none")
+      * from_push.  to_pop is only populated, and popped from, by whoever
-         self.vm.add_device('virtio-scsi')
+      * is in charge of the next wakeup.  This can be an unlocker or,
-         self.vm.add_device("scsi-hd,id=scsi0,drive=drive0")
+diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
-@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
+index XXXXXXX..XXXXXXX 100644
+--- a/util/qemu-coroutine-lock.c
-     def test_commit(self):
++++ b/util/qemu-coroutine-lock.c
-         self.run_commit_test(mid_img, backing_img)
+@@ -XXX,XX +XXX,XX @@
-+        if not self.image_len:
+ #include "qemu-common.h"
-+            return
+ #include "qemu/coroutine.h"
-         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+ #include "qemu/coroutine_int.h"
-         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
++#include "qemu/processor.h"
+ #include "qemu/queue.h"
-     def test_commit_node(self):
+ #include "block/aio.h"
-         self.run_commit_test("mid", "base", node_names=True)
+ #include "trace.h"
-+        if not self.image_len:
+@@ -XXX,XX +XXX,XX @@ void qemu_co_mutex_init(CoMutex *mutex)
-+            return
+     memset(mutex, 0, sizeof(*mutex));
-         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+ }
-         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+-static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
-@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
++static void coroutine_fn qemu_co_mutex_wake(CoMutex *mutex, Coroutine *co)
++{
-     def test_top_is_active(self):
++    /* Read co before co->ctx; pairs with smp_wmb() in
-         self.run_commit_test(test_img, backing_img, need_ready=True)
++     * qemu_coroutine_enter().
-+        if not self.image_len:
++     */
-+            return
++    smp_read_barrier_depends();
-         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
++    mutex->ctx = co->ctx;
-         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
++    aio_co_wake(co);
++}
-     def test_top_is_default_active(self):
++
-         self.run_default_commit_test()
++static void coroutine_fn qemu_co_mutex_lock_slowpath(AioContext *ctx,
-+        if not self.image_len:
++                                                     CoMutex *mutex)
-+            return
+ {
-         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+     Coroutine *self = qemu_coroutine_self();
-         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+     CoWaitRecord w;
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
          if (co == self) {
              /* We got the lock ourselves!  */
              assert(to_wake == &w);
 +            mutex->ctx = ctx;
              return;
          }
 -        aio_co_wake(co);
 +        qemu_co_mutex_wake(mutex, co);
      }
      qemu_coroutine_yield();
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
  void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
  {
 +    AioContext *ctx = qemu_get_current_aio_context();
      Coroutine *self = qemu_coroutine_self();
 +    int waiters, i;
 -    if (atomic_fetch_inc(&mutex->locked) == 0) {
 +    /* Running a very small critical section on pthread_mutex_t and CoMutex
 +     * shows that pthread_mutex_t is much faster because it doesn't actually
 +     * go to sleep.  What happens is that the critical section is shorter
 +     * than the latency of entering the kernel and thus FUTEX_WAIT always
 +     * fails.  With CoMutex there is no such latency but you still want to
 +     * avoid wait and wakeup.  So introduce it artificially.
 +     */
 +    i = 0;
 +retry_fast_path:
 +    waiters = atomic_cmpxchg(&mutex->locked, 0, 1);
 +    if (waiters != 0) {
 +        while (waiters == 1 && ++i < 1000) {
 +            if (atomic_read(&mutex->ctx) == ctx) {
 +                break;
 +            }
 +            if (atomic_read(&mutex->locked) == 0) {
 +                goto retry_fast_path;
 +            }
 +            cpu_relax();
 +        }
 +        waiters = atomic_fetch_inc(&mutex->locked);
 +    }
 +
 +    if (waiters == 0) {
          /* Uncontended.  */
          trace_qemu_co_mutex_lock_uncontended(mutex, self);
 +        mutex->ctx = ctx;
      } else {
 -        qemu_co_mutex_lock_slowpath(mutex);
 +        qemu_co_mutex_lock_slowpath(ctx, mutex);
      }
      mutex->holder = self;
      self->locks_held++;
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
      assert(mutex->holder == self);
      assert(qemu_in_coroutine());
 +    mutex->ctx = NULL;
      mutex->holder = NULL;
      self->locks_held--;
      if (atomic_fetch_dec(&mutex->locked) == 1) {
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
          unsigned our_handoff;
          if (to_wake) {
 -            Coroutine *co = to_wake->co;
 -            aio_co_wake(co);
 +            qemu_co_mutex_wake(mutex, to_wake->co);
              break;
          }
 diff --git a/util/qemu-coroutine.c b/util/qemu-coroutine.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/qemu-coroutine.c
 +++ b/util/qemu-coroutine.c
@@ -XXX,XX +XXX,XX @@ void qemu_coroutine_enter(Coroutine *co)
      co->ctx = qemu_get_current_aio_context();
      /* Store co->ctx before anything that stores co.  Matches
 -     * barrier in aio_co_wake.
 +     * barrier in aio_co_wake and qemu_co_mutex_wake.
       */
      smp_wmb();
 --
-.35.1
+.9.3

-[PULL 08/14] iotests: rebase qemu_io() on top of qemu_tool()
+[Qemu-devel] [PULL v2 21/24] test-aio-multithread: add performance comparison with thread-based mutexes
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-Rework qemu_io() to be analogous to qemu_img(); a function that requires
+Add two implementations of the same benchmark as the previous patch,
-a return code of zero by default unless disabled explicitly.
+but using pthreads.  One uses a normal QemuMutex, the other is Linux
+only and implements a fair mutex based on MCS locks and futexes.
-Tests that use qemu_io():
+This shows that the slower performance of the 5-thread case is due to
-040 041 044 055 056 093 124 129 132 136 148 149 151 152 163 165 205
+the fairness of CoMutex, rather than to coroutines.  If fairness does
-219 236 245 248 254 255 257 260 264 280 298 300 302 304
+not matter, as is the case with two threads, CoMutex can actually be
-image-fleecing migrate-bitmaps-postcopy-test migrate-bitmaps-test
+faster than pthreads.
-migrate-during-backup migration-permissions
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Test that use qemu_io_log():
+Reviewed-by: Fam Zheng <famz@redhat.com>
-245 255 274 303 307 nbd-reconnect-on-open
+Message-id: 20170213181244.16297-4-pbonzini@redhat.com
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 Copy-pastables for testing/verification:
 ./check -qcow2 030 040 041 044 055 056 124 129 132 151 152 163 165 209 \
 236 242 245 248 254 255 257 260 264 274 \
 298 300 302 303 304 307 image-fleecing \
                migrate-bitmaps-postcopy-test migrate-bitmaps-test \
                migrate-during-backup nbd-reconnect-on-open
 ./check -raw 093 136 148 migration-permissions
 ./check -nbd 205
 # ./configure configure --disable-gnutls --enable-gcrypt
 # this ALSO requires passwordless sudo.
 ./check -luks 149
 # Just the tests that were edited in this commit:
 ./check -qcow2 030 040 242 245
 ./check -raw migration-permissions
 ./check -nbd 205
 ./check -luks 149
 Signed-off-by: John Snow <jsnow@redhat.com>
 Message-Id: <20220418211504.943969-8-jsnow@redhat.com>
 Reviewed-by: Eric Blake <eblake@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- tests/qemu-iotests/030                        | 85 +++++++++++--------
+ tests/test-aio-multithread.c | 164 +++++++++++++++++++++++++++++++++++++++++++
- tests/qemu-iotests/149                        |  6 +-
+file changed, 164 insertions(+)
- tests/qemu-iotests/205                        |  4 +-
- tests/qemu-iotests/245                        | 17 ++--
+diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
  tests/qemu-iotests/iotests.py                 | 19 +++--
  .../qemu-iotests/tests/migration-permissions  |  4 +-
 files changed, 81 insertions(+), 54 deletions(-)
 diff --git a/tests/qemu-iotests/030 b/tests/qemu-iotests/030
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/030
 +++ b/tests/qemu-iotests/030
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
 -                         'image file map does not match backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
 +            'image file map does not match backing file after streaming')
      def test_stream_intermediate(self):
          self.assert_no_active_block_jobs()
 -        self.assertNotEqual(qemu_io('-f', 'raw', '-rU', '-c', 'map', backing_img),
 -                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', mid_img),
 -                            'image file map matches backing file before streaming')
 +        self.assertNotEqual(
 +            qemu_io('-f', 'raw', '-rU', '-c', 'map', backing_img).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', mid_img).stdout,
 +            'image file map matches backing file before streaming')
          result = self.vm.qmp('block-stream', device='mid', job_id='stream-mid')
          self.assert_qmp(result, 'return', {})
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img),
 -                         'image file map does not match backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img).stdout,
 +            'image file map does not match backing file after streaming')
      def test_stream_pause(self):
          self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
 -                         'image file map does not match backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
 +            'image file map does not match backing file after streaming')
      def test_stream_no_op(self):
          self.assert_no_active_block_jobs()
          # The image map is empty before the operation
 -        empty_map = qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', test_img)
 +        empty_map = qemu_io(
 +            '-f', iotests.imgfmt, '-rU', '-c', 'map', test_img).stdout
          # This is a no-op: no data should ever be copied from the base image
          result = self.vm.qmp('block-stream', device='drive0', base=mid_img)
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
 -                         empty_map, 'image file map changed after a no-op')
 +        self.assertEqual(
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
 +            empty_map, 'image file map changed after a no-op')
      def test_stream_partial(self):
          self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
 -                         'image file map does not match backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
 +            'image file map does not match backing file after streaming')
      def test_device_not_found(self):
          result = self.vm.qmp('block-stream', device='nonexistent')
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
          # Check that the maps don't match before the streaming operations
          for i in range(2, self.num_imgs, 2):
 -            self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i]),
 -                                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i-1]),
 -                                'image file map matches backing file before streaming')
 +            self.assertNotEqual(
 +                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i]).stdout,
 +                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i-1]).stdout,
 +                'image file map matches backing file before streaming')
          # Create all streaming jobs
          pending_jobs = []
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
          # Check that all maps match now
          for i in range(2, self.num_imgs, 2):
 -            self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i]),
 -                             qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i-1]),
 -                             'image file map does not match backing file after streaming')
 +            self.assertEqual(
 +                qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i]).stdout,
 +                qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i-1]).stdout,
 +                'image file map does not match backing file after streaming')
      # Test that it's not possible to perform two block-stream
      # operations if there are nodes involved in both.
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
      def test_stream_base_node_name(self):
          self.assert_no_active_block_jobs()
 -        self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[4]),
 -                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[3]),
 -                            'image file map matches backing file before streaming')
 +        self.assertNotEqual(
 +            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[4]).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[3]).stdout,
 +            'image file map matches backing file before streaming')
          # Error: the base node does not exist
          result = self.vm.qmp('block-stream', device='node4', base_node='none', job_id='stream')
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[4]),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[3]),
 -                         'image file map matches backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[4]).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[3]).stdout,
 +            'image file map matches backing file after streaming')
  class TestQuorum(iotests.QMPTestCase):
      num_children = 3
@@ -XXX,XX +XXX,XX @@ class TestQuorum(iotests.QMPTestCase):
              os.remove(img)
      def test_stream_quorum(self):
 -        self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.children[0]),
 -                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.backing[0]),
 -                            'image file map matches backing file before streaming')
 +        self.assertNotEqual(
 +            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.children[0]).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.backing[0]).stdout,
 +            'image file map matches backing file before streaming')
          self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestQuorum(iotests.QMPTestCase):
          self.assert_no_active_block_jobs()
          self.vm.shutdown()
 -        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.children[0]),
 -                         qemu_io('-f', iotests.imgfmt, '-c', 'map', self.backing[0]),
 -                         'image file map does not match backing file after streaming')
 +        self.assertEqual(
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.children[0]).stdout,
 +            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.backing[0]).stdout,
 +            'image file map does not match backing file after streaming')
  class TestSmallerBackingFile(iotests.QMPTestCase):
      backing_len = 1 * 1024 * 1024 # MB
 diff --git a/tests/qemu-iotests/149 b/tests/qemu-iotests/149
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/149
 +++ b/tests/qemu-iotests/149
@@ -XXX,XX +XXX,XX @@ def qemu_io_write_pattern(config, pattern, offset_mb, size_mb, dev=False):
      args = ["-c", "write -P 0x%x %dM %dM" % (pattern, offset_mb, size_mb)]
      args.extend(qemu_io_image_args(config, dev))
      iotests.log("qemu-io " + " ".join(args), filters=[iotests.filter_test_dir])
 -    iotests.log(check_cipher_support(config, iotests.qemu_io(*args)),
 +    output = iotests.qemu_io(*args, check=False).stdout
 +    iotests.log(check_cipher_support(config, output),
                  filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
@@ -XXX,XX +XXX,XX @@ def qemu_io_read_pattern(config, pattern, offset_mb, size_mb, dev=False):
      args = ["-c", "read -P 0x%x %dM %dM" % (pattern, offset_mb, size_mb)]
      args.extend(qemu_io_image_args(config, dev))
      iotests.log("qemu-io " + " ".join(args), filters=[iotests.filter_test_dir])
 -    iotests.log(check_cipher_support(config, iotests.qemu_io(*args)),
 +    output = iotests.qemu_io(*args, check=False).stdout
 +    iotests.log(check_cipher_support(config, output),
                  filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
 diff --git a/tests/qemu-iotests/205 b/tests/qemu-iotests/205
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/205
 +++ b/tests/qemu-iotests/205
@@ -XXX,XX +XXX,XX @@ class TestNbdServerRemove(iotests.QMPTestCase):
      def do_test_connect_after_remove(self, mode=None):
          args = ('-r', '-f', 'raw', '-c', 'read 0 512', nbd_uri)
 -        self.assertReadOk(qemu_io(*args))
 +        self.assertReadOk(qemu_io(*args).stdout)
          result = self.remove_export('exp', mode)
          self.assert_qmp(result, 'return', {})
          self.assertExportNotFound('exp')
 -        self.assertConnectFailed(qemu_io(*args))
 +        self.assertConnectFailed(qemu_io(*args, check=False).stdout)
      def test_connect_after_remove_default(self):
          self.do_test_connect_after_remove()
 diff --git a/tests/qemu-iotests/245 b/tests/qemu-iotests/245
 index XXXXXXX..XXXXXXX 100755
 --- a/tests/qemu-iotests/245
 +++ b/tests/qemu-iotests/245
@@ -XXX,XX +XXX,XX @@
  # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  #
 +import copy
 +import json
  import os
  import re
 +from subprocess import CalledProcessError
 +
  import iotests
 -import copy
 -import json
  from iotests import qemu_img, qemu_io
  hd_path = [
@@ -XXX,XX +XXX,XX @@ class TestBlockdevReopen(iotests.QMPTestCase):
      # Reopen an image several times changing some of its options
      def test_reopen(self):
 -        # Check whether the filesystem supports O_DIRECT
 -        if 'O_DIRECT' in qemu_io('-f', 'raw', '-t', 'none', '-c', 'quit', hd_path[0]):
 -            supports_direct = False
 -        else:
 +        try:
 +            qemu_io('-f', 'raw', '-t', 'none', '-c', 'quit', hd_path[0])
              supports_direct = True
 +        except CalledProcessError as exc:
 +            if 'O_DIRECT' in exc.stdout:
 +                supports_direct = False
 +            else:
 +                raise
          # Open the hd1 image passing all backing options
          opts = hd_opts(1)
 diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/iotests.py
+--- a/tests/test-aio-multithread.c
-+++ b/tests/qemu-iotests/iotests.py
++++ b/tests/test-aio-multithread.c
-@@ -XXX,XX +XXX,XX @@ def qemu_io_wrap_args(args: Sequence[str]) -> List[str]:
+@@ -XXX,XX +XXX,XX @@ static void test_multi_co_mutex_2_30(void)
- def qemu_io_popen(*args):
+     test_multi_co_mutex(2, 30);
-     return qemu_tool_popen(qemu_io_wrap_args(args))
+ }
--def qemu_io(*args):
++/* Same test with fair mutexes, for performance comparison.  */
--    '''Run qemu-io and return the stdout data'''
++
--    return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))[0]
++#ifdef CONFIG_LINUX
-+def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
++#include "qemu/futex.h"
-+            ) -> 'subprocess.CompletedProcess[str]':
++
-+    """
++/* The nodes for the mutex reside in this structure (on which we try to avoid
-+    Run QEMU_IO_PROG and return the status code and console output.
++ * false sharing).  The head of the mutex is in the "mutex_head" variable.
-+
++ */
-+    This function always prepends either QEMU_IO_OPTIONS or
++static struct {
-+    QEMU_IO_OPTIONS_NO_FMT.
++    int next, locked;
-+    """
++    int padding[14];
-+    return qemu_tool(*qemu_io_wrap_args(args),
++} nodes[NUM_CONTEXTS] __attribute__((__aligned__(64)));
-+                     check=check, combine_stdio=combine_stdio)
++
++static int mutex_head = -1;
- def qemu_io_pipe_and_status(*args):
++
-     return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))
++static void mcs_mutex_lock(void)
++{
--def qemu_io_log(*args):
++    int prev;
--    result = qemu_io(*args)
++
--    log(result, filters=[filter_testfiles, filter_qemu_io])
++    nodes[id].next = -1;
-+def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
++    nodes[id].locked = 1;
-+    result = qemu_io(*args, check=False)
++    prev = atomic_xchg(&mutex_head, id);
-+    log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
++    if (prev != -1) {
-     return result
++        atomic_set(&nodes[prev].next, id);
++        qemu_futex_wait(&nodes[id].locked, 1);
- def qemu_io_silent(*args):
++    }
-diff --git a/tests/qemu-iotests/tests/migration-permissions b/tests/qemu-iotests/tests/migration-permissions
++}
-index XXXXXXX..XXXXXXX 100755
++
---- a/tests/qemu-iotests/tests/migration-permissions
++static void mcs_mutex_unlock(void)
-+++ b/tests/qemu-iotests/tests/migration-permissions
++{
-@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
++    int next;
-     def test_post_migration_permissions(self):
++    if (nodes[id].next == -1) {
-         # Try to access the image R/W, which should fail because virtio-blk
++        if (atomic_read(&mutex_head) == id &&
-         # has not been configured with share-rw=on
++            atomic_cmpxchg(&mutex_head, id, -1) == id) {
--        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img)
++            /* Last item in the list, exit.  */
-+        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
++            return;
-         if not log.strip():
++        }
-             print('ERROR (pre-migration): qemu-io should not be able to '
++        while (atomic_read(&nodes[id].next) == -1) {
-                   'access this image, but it reported no error')
++            /* mcs_mutex_lock did the xchg, but has not updated
-@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
++             * nodes[prev].next yet.
++             */
-         # Try the same qemu-io access again, verifying that the WRITE
++        }
-         # permission remains unshared
++    }
--        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img)
++
-+        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
++    /* Wake up the next in line.  */
-         if not log.strip():
++    next = nodes[id].next;
-             print('ERROR (post-migration): qemu-io should not be able to '
++    nodes[next].locked = 0;
-                   'access this image, but it reported no error')
++    qemu_futex_wake(&nodes[next].locked, 1);
 +}
 +
 +static void test_multi_fair_mutex_entry(void *opaque)
 +{
 +    while (!atomic_mb_read(&now_stopping)) {
 +        mcs_mutex_lock();
 +        counter++;
 +        mcs_mutex_unlock();
 +        atomic_inc(&atomic_counter);
 +    }
 +    atomic_dec(&running);
 +}
 +
 +static void test_multi_fair_mutex(int threads, int seconds)
 +{
 +    int i;
 +
 +    assert(mutex_head == -1);
 +    counter = 0;
 +    atomic_counter = 0;
 +    now_stopping = false;
 +
 +    create_aio_contexts();
 +    assert(threads <= NUM_CONTEXTS);
 +    running = threads;
 +    for (i = 0; i < threads; i++) {
 +        Coroutine *co1 = qemu_coroutine_create(test_multi_fair_mutex_entry, NULL);
 +        aio_co_schedule(ctx[i], co1);
 +    }
 +
 +    g_usleep(seconds * 1000000);
 +
 +    atomic_mb_set(&now_stopping, true);
 +    while (running > 0) {
 +        g_usleep(100000);
 +    }
 +
 +    join_aio_contexts();
 +    g_test_message("%d iterations/second\n", counter / seconds);
 +    g_assert_cmpint(counter, ==, atomic_counter);
 +}
 +
 +static void test_multi_fair_mutex_1(void)
 +{
 +    test_multi_fair_mutex(NUM_CONTEXTS, 1);
 +}
 +
 +static void test_multi_fair_mutex_10(void)
 +{
 +    test_multi_fair_mutex(NUM_CONTEXTS, 10);
 +}
 +#endif
 +
 +/* Same test with pthread mutexes, for performance comparison and
 + * portability.  */
 +
 +static QemuMutex mutex;
 +
 +static void test_multi_mutex_entry(void *opaque)
 +{
 +    while (!atomic_mb_read(&now_stopping)) {
 +        qemu_mutex_lock(&mutex);
 +        counter++;
 +        qemu_mutex_unlock(&mutex);
 +        atomic_inc(&atomic_counter);
 +    }
 +    atomic_dec(&running);
 +}
 +
 +static void test_multi_mutex(int threads, int seconds)
 +{
 +    int i;
 +
 +    qemu_mutex_init(&mutex);
 +    counter = 0;
 +    atomic_counter = 0;
 +    now_stopping = false;
 +
 +    create_aio_contexts();
 +    assert(threads <= NUM_CONTEXTS);
 +    running = threads;
 +    for (i = 0; i < threads; i++) {
 +        Coroutine *co1 = qemu_coroutine_create(test_multi_mutex_entry, NULL);
 +        aio_co_schedule(ctx[i], co1);
 +    }
 +
 +    g_usleep(seconds * 1000000);
 +
 +    atomic_mb_set(&now_stopping, true);
 +    while (running > 0) {
 +        g_usleep(100000);
 +    }
 +
 +    join_aio_contexts();
 +    g_test_message("%d iterations/second\n", counter / seconds);
 +    g_assert_cmpint(counter, ==, atomic_counter);
 +}
 +
 +static void test_multi_mutex_1(void)
 +{
 +    test_multi_mutex(NUM_CONTEXTS, 1);
 +}
 +
 +static void test_multi_mutex_10(void)
 +{
 +    test_multi_mutex(NUM_CONTEXTS, 10);
 +}
 +
  /* End of tests.  */
  int main(int argc, char **argv)
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
          g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
          g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_1);
          g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_3);
 +#ifdef CONFIG_LINUX
 +        g_test_add_func("/aio/multi/mutex/mcs", test_multi_fair_mutex_1);
 +#endif
 +        g_test_add_func("/aio/multi/mutex/pthread", test_multi_mutex_1);
      } else {
          g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
          g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_10);
          g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_30);
 +#ifdef CONFIG_LINUX
 +        g_test_add_func("/aio/multi/mutex/mcs", test_multi_fair_mutex_10);
 +#endif
 +        g_test_add_func("/aio/multi/mutex/pthread", test_multi_mutex_10);
      }
      return g_test_run();
  }
 --
-.35.1
+.9.3

-[PULL 07/14] iotests: create generic qemu_tool() function
+[Qemu-devel] [PULL v2 22/24] coroutine-lock: place CoMutex before CoQueue in header
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-reimplement qemu_img() in terms of qemu_tool() in preparation for doing
+This will avoid forward references in the next patch.  It is also
-the same with qemu_io().
+more logical because CoQueue is not anymore the basic primitive.
-Signed-off-by: John Snow <jsnow@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Reviewed-by: Eric Blake <eblake@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
+Message-id: 20170213181244.16297-5-pbonzini@redhat.com
-Message-Id: <20220418211504.943969-7-jsnow@redhat.com>
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- tests/qemu-iotests/iotests.py | 32 +++++++++++++++++++++-----------
+ include/qemu/coroutine.h | 89 ++++++++++++++++++++++++------------------------
-file changed, 21 insertions(+), 11 deletions(-)
+file changed, 44 insertions(+), 45 deletions(-)
-diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
+diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/iotests.py
+--- a/include/qemu/coroutine.h
-+++ b/tests/qemu-iotests/iotests.py
++++ b/include/qemu/coroutine.h
-@@ -XXX,XX +XXX,XX @@ def qemu_img_create_prepare_args(args: List[str]) -> List[str]:
+@@ -XXX,XX +XXX,XX @@ bool qemu_in_coroutine(void);
+  */
-     return result
+ bool qemu_coroutine_entered(Coroutine *co);
 -def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
 -             ) -> 'subprocess.CompletedProcess[str]':
 -    """
 -    Run qemu_img and return the status code and console output.
 -    This function always prepends QEMU_IMG_OPTIONS and may further alter
 -    the args for 'create' commands.
 +def qemu_tool(*args: str, check: bool = True, combine_stdio: bool = True
 +              ) -> 'subprocess.CompletedProcess[str]':
 +    """
 +    Run a qemu tool and return its status code and console output.
 -    :param args: command-line arguments to qemu-img.
 +    :param args: full command line to run.
      :param check: Enforce a return code of zero.
      :param combine_stdio: set to False to keep stdout/stderr separated.
@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
          properties. If streams are not combined, it will also have a
          stderr property.
      """
 -    full_args = qemu_img_args + qemu_img_create_prepare_args(list(args))
 -
-     subp = subprocess.run(
+-/**
--        full_args,
+- * CoQueues are a mechanism to queue coroutines in order to continue executing
-+        args,
+- * them later. They provide the fundamental primitives on which coroutine locks
-         stdout=subprocess.PIPE,
+- * are built.
-         stderr=subprocess.STDOUT if combine_stdio else subprocess.PIPE,
+- */
-         universal_newlines=True,
+-typedef struct CoQueue {
-@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
+-    QSIMPLEQ_HEAD(, Coroutine) entries;
+-} CoQueue;
-     if check and subp.returncode or (subp.returncode < 0):
+-
-         raise VerboseProcessError(
+-/**
--            subp.returncode, full_args,
+- * Initialise a CoQueue. This must be called before any other operation is used
-+            subp.returncode, args,
+- * on the CoQueue.
-             output=subp.stdout,
+- */
-             stderr=subp.stderr,
+-void qemu_co_queue_init(CoQueue *queue);
-         )
+-
-@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
+-/**
-     return subp
+- * Adds the current coroutine to the CoQueue and transfers control to the
+- * caller of the coroutine.
+- */
-+def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
+-void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
-+             ) -> 'subprocess.CompletedProcess[str]':
+-
-+    """
+-/**
-+    Run QEMU_IMG_PROG and return its status code and console output.
+- * Restarts the next coroutine in the CoQueue and removes it from the queue.
 - *
 - * Returns true if a coroutine was restarted, false if the queue is empty.
 - */
 -bool coroutine_fn qemu_co_queue_next(CoQueue *queue);
 -
 -/**
 - * Restarts all coroutines in the CoQueue and leaves the queue empty.
 - */
 -void coroutine_fn qemu_co_queue_restart_all(CoQueue *queue);
 -
 -/**
 - * Enter the next coroutine in the queue
 - */
 -bool qemu_co_enter_next(CoQueue *queue);
 -
 -/**
 - * Checks if the CoQueue is empty.
 - */
 -bool qemu_co_queue_empty(CoQueue *queue);
 -
 -
  /**
   * Provides a mutex that can be used to synchronise coroutines
   */
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex);
   */
  void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex);
 +
-+    This function always prepends QEMU_IMG_OPTIONS and may further alter
++/**
-+    the args for 'create' commands.
++ * CoQueues are a mechanism to queue coroutines in order to continue executing
 + * them later.
 + */
 +typedef struct CoQueue {
 +    QSIMPLEQ_HEAD(, Coroutine) entries;
 +} CoQueue;
 +
-+    See `qemu_tool()` for greater detail.
++/**
-+    """
++ * Initialise a CoQueue. This must be called before any other operation is used
-+    full_args = qemu_img_args + qemu_img_create_prepare_args(list(args))
++ * on the CoQueue.
-+    return qemu_tool(*full_args, check=check, combine_stdio=combine_stdio)
++ */
 +void qemu_co_queue_init(CoQueue *queue);
 +
 +/**
 + * Adds the current coroutine to the CoQueue and transfers control to the
 + * caller of the coroutine.
 + */
 +void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
 +
 +/**
 + * Restarts the next coroutine in the CoQueue and removes it from the queue.
 + *
 + * Returns true if a coroutine was restarted, false if the queue is empty.
 + */
 +bool coroutine_fn qemu_co_queue_next(CoQueue *queue);
 +
 +/**
 + * Restarts all coroutines in the CoQueue and leaves the queue empty.
 + */
 +void coroutine_fn qemu_co_queue_restart_all(CoQueue *queue);
 +
 +/**
 + * Enter the next coroutine in the queue
 + */
 +bool qemu_co_enter_next(CoQueue *queue);
 +
 +/**
 + * Checks if the CoQueue is empty.
 + */
 +bool qemu_co_queue_empty(CoQueue *queue);
 +
 +
- def ordered_qmp(qmsg, conv_keys=True):
+ typedef struct CoRwlock {
-     # Dictionaries are not ordered prior to 3.6, therefore:
+     bool writer;
-     if isinstance(qmsg, list):
+     int reader;
 --
-.35.1
+.9.3

-[PULL 04/14] iotests: Don't check qemu_io() output for specific error strings
+[Qemu-devel] [PULL v2 23/24] coroutine-lock: add mutex argument to CoQueue APIs
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-A forthcoming commit updates qemu_io() to raise an exception on non-zero
+All that CoQueue needs in order to become thread-safe is help
-return by default, and changes its return type.
+from an external mutex.  Add this to the API.
-In preparation, simplify some calls to qemu_io() that assert that
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-specific error message strings do not appear in qemu-io's
+Reviewed-by: Fam Zheng <famz@redhat.com>
-output. Asserting that all of these calls return a status code of zero
+Message-id: 20170213181244.16297-6-pbonzini@redhat.com
-will be a more robust way to guard against failure.
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: John Snow <jsnow@redhat.com>
 Reviewed-by: Eric Blake <eblake@redhat.com>
 Reviewed-by: Hanna Reitz <hreitz@redhat.com>
 Message-Id: <20220418211504.943969-4-jsnow@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- tests/qemu-iotests/040 | 33 ++++++++++++++++-----------------
+ include/qemu/coroutine.h   |  8 +++++---
- tests/qemu-iotests/056 |  2 +-
+ block/backup.c             |  2 +-
-files changed, 17 insertions(+), 18 deletions(-)
+ block/io.c                 |  4 ++--
+ block/nbd-client.c         |  2 +-
-diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
+ block/qcow2-cluster.c      |  4 +---
-index XXXXXXX..XXXXXXX 100755
+ block/sheepdog.c           |  2 +-
---- a/tests/qemu-iotests/040
+ block/throttle-groups.c    |  2 +-
-+++ b/tests/qemu-iotests/040
+ hw/9pfs/9p.c               |  2 +-
-@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
+ util/qemu-coroutine-lock.c | 24 +++++++++++++++++++++---
+files changed, 34 insertions(+), 16 deletions(-)
-     def test_commit(self):
-         self.run_commit_test(mid_img, backing_img)
+diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
+index XXXXXXX..XXXXXXX 100644
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+--- a/include/qemu/coroutine.h
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
++++ b/include/qemu/coroutine.h
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex);
-     def test_commit_node(self):
+ /**
-         self.run_commit_test("mid", "base", node_names=True)
+  * CoQueues are a mechanism to queue coroutines in order to continue executing
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
+- * them later.
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
++ * them later.  They are similar to condition variables, but they need help
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
++ * from an external mutex in order to maintain thread-safety.
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+  */
+ typedef struct CoQueue {
-     @iotests.skip_if_unsupported(['throttle'])
+     QSIMPLEQ_HEAD(, Coroutine) entries;
-     def test_commit_with_filter_and_quit(self):
+@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_init(CoQueue *queue);
-@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
+ /**
-     def test_top_is_active(self):
+  * Adds the current coroutine to the CoQueue and transfers control to the
-         self.run_commit_test(test_img, backing_img, need_ready=True)
+- * caller of the coroutine.
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
++ * caller of the coroutine.  The mutex is unlocked during the wait and
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
++ * locked again afterwards.
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+  */
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+-void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
++void coroutine_fn qemu_co_queue_wait(CoQueue *queue, CoMutex *mutex);
-     def test_top_is_default_active(self):
-         self.run_default_commit_test()
+ /**
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
+  * Restarts the next coroutine in the CoQueue and removes it from the queue.
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+diff --git a/block/backup.c b/block/backup.c
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+index XXXXXXX..XXXXXXX 100644
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
+--- a/block/backup.c
++++ b/block/backup.c
-     def test_top_and_base_reversed(self):
+@@ -XXX,XX +XXX,XX @@ static void coroutine_fn wait_for_overlapping_requests(BackupBlockJob *job,
-         self.assert_no_active_block_jobs()
+         retry = false;
-@@ -XXX,XX +XXX,XX @@ class TestRelativePaths(ImageCommitTestCase):
+         QLIST_FOREACH(req, &job->inflight_reqs, list) {
+             if (end > req->start && start < req->end) {
-     def test_commit(self):
+-                qemu_co_queue_wait(&req->wait_queue);
-         self.run_commit_test(self.mid_img, self.backing_img)
++                qemu_co_queue_wait(&req->wait_queue, NULL);
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs).find("verification failed"))
+                 retry = true;
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs).find("verification failed"))
+                 break;
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs)
+             }
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs)
+diff --git a/block/io.c b/block/io.c
+index XXXXXXX..XXXXXXX 100644
-     def test_device_not_found(self):
+--- a/block/io.c
-         result = self.vm.qmp('block-commit', device='nonexistent', top='%s' % self.mid_img)
++++ b/block/io.c
-@@ -XXX,XX +XXX,XX @@ class TestRelativePaths(ImageCommitTestCase):
+@@ -XXX,XX +XXX,XX @@ static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
+                  * (instead of producing a deadlock in the former case). */
-     def test_top_is_active(self):
+                 if (!req->waiting_for) {
-         self.run_commit_test(self.test_img, self.backing_img)
+                     self->waiting_for = req;
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs).find("verification failed"))
+-                    qemu_co_queue_wait(&req->wait_queue);
--        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs).find("verification failed"))
++                    qemu_co_queue_wait(&req->wait_queue, NULL);
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs)
+                     self->waiting_for = NULL;
-+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs)
+                     retry = true;
+                     waited = true;
-     def test_top_and_base_reversed(self):
+@@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
-         self.assert_no_active_block_jobs()
-@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
+     /* Wait until any previous flushes are completed */
+     while (bs->active_flush_req) {
-     def do_test_io(self, read_or_write):
+-        qemu_co_queue_wait(&bs->flush_queue);
-         for index, pattern_file in enumerate(self.pattern_files):
++        qemu_co_queue_wait(&bs->flush_queue, NULL);
--            result = qemu_io('-f', iotests.imgfmt,
+     }
--                             '-c',
--                             f'{read_or_write} -P {index + 1} {index}M 1M',
+     bs->active_flush_req = true;
--                             pattern_file)
+diff --git a/block/nbd-client.c b/block/nbd-client.c
--            self.assertFalse('Pattern verification failed' in result)
+index XXXXXXX..XXXXXXX 100644
-+            qemu_io('-f', iotests.imgfmt,
+--- a/block/nbd-client.c
-+                    '-c',
++++ b/block/nbd-client.c
-+                    f'{read_or_write} -P {index + 1} {index}M 1M',
+@@ -XXX,XX +XXX,XX @@ static void nbd_coroutine_start(NBDClientSession *s,
-+                    pattern_file)
+     /* Poor man semaphore.  The free_sema is locked when no other request
+      * can be accepted, and unlocked after receiving one reply.  */
-     @iotests.skip_if_unsupported(['throttle'])
+     if (s->in_flight == MAX_NBD_REQUESTS) {
-     def setUp(self):
+-        qemu_co_queue_wait(&s->free_sema);
-diff --git a/tests/qemu-iotests/056 b/tests/qemu-iotests/056
++        qemu_co_queue_wait(&s->free_sema, NULL);
-index XXXXXXX..XXXXXXX 100755
+         assert(s->in_flight < MAX_NBD_REQUESTS);
---- a/tests/qemu-iotests/056
+     }
-+++ b/tests/qemu-iotests/056
+     s->in_flight++;
-@@ -XXX,XX +XXX,XX @@ class TestSyncModesNoneAndTop(iotests.QMPTestCase):
+diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
+index XXXXXXX..XXXXXXX 100644
-         self.vm.shutdown()
+--- a/block/qcow2-cluster.c
-         time.sleep(1)
++++ b/block/qcow2-cluster.c
--        self.assertEqual(-1, qemu_io('-c', 'read -P0x41 0 512', target_img).find("verification failed"))
+@@ -XXX,XX +XXX,XX @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
-+        qemu_io('-c', 'read -P0x41 0 512', target_img)
+             if (bytes == 0) {
+                 /* Wait for the dependency to complete. We need to recheck
- class TestBeforeWriteNotifier(iotests.QMPTestCase):
+                  * the free/allocated clusters when we continue. */
-     def setUp(self):
+-                qemu_co_mutex_unlock(&s->lock);
 -                qemu_co_queue_wait(&old_alloc->dependent_requests);
 -                qemu_co_mutex_lock(&s->lock);
 +                qemu_co_queue_wait(&old_alloc->dependent_requests, &s->lock);
                  return -EAGAIN;
              }
          }
 diff --git a/block/sheepdog.c b/block/sheepdog.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/sheepdog.c
 +++ b/block/sheepdog.c
@@ -XXX,XX +XXX,XX @@ static void wait_for_overlapping_aiocb(BDRVSheepdogState *s, SheepdogAIOCB *acb)
  retry:
      QLIST_FOREACH(cb, &s->inflight_aiocb_head, aiocb_siblings) {
          if (AIOCBOverlapping(acb, cb)) {
 -            qemu_co_queue_wait(&s->overlapping_queue);
 +            qemu_co_queue_wait(&s->overlapping_queue, NULL);
              goto retry;
          }
      }
 diff --git a/block/throttle-groups.c b/block/throttle-groups.c
 index XXXXXXX..XXXXXXX 100644
 --- a/block/throttle-groups.c
 +++ b/block/throttle-groups.c
@@ -XXX,XX +XXX,XX @@ void coroutine_fn throttle_group_co_io_limits_intercept(BlockBackend *blk,
      if (must_wait || blkp->pending_reqs[is_write]) {
          blkp->pending_reqs[is_write]++;
          qemu_mutex_unlock(&tg->lock);
 -        qemu_co_queue_wait(&blkp->throttled_reqs[is_write]);
 +        qemu_co_queue_wait(&blkp->throttled_reqs[is_write], NULL);
          qemu_mutex_lock(&tg->lock);
          blkp->pending_reqs[is_write]--;
      }
 diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
 index XXXXXXX..XXXXXXX 100644
 --- a/hw/9pfs/9p.c
 +++ b/hw/9pfs/9p.c
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn v9fs_flush(void *opaque)
          /*
           * Wait for pdu to complete.
           */
 -        qemu_co_queue_wait(&cancel_pdu->complete);
 +        qemu_co_queue_wait(&cancel_pdu->complete, NULL);
          cancel_pdu->cancelled = 0;
          pdu_free(cancel_pdu);
      }
 diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/qemu-coroutine-lock.c
 +++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_init(CoQueue *queue)
      QSIMPLEQ_INIT(&queue->entries);
  }
 -void coroutine_fn qemu_co_queue_wait(CoQueue *queue)
 +void coroutine_fn qemu_co_queue_wait(CoQueue *queue, CoMutex *mutex)
  {
      Coroutine *self = qemu_coroutine_self();
      QSIMPLEQ_INSERT_TAIL(&queue->entries, self, co_queue_next);
 +
 +    if (mutex) {
 +        qemu_co_mutex_unlock(mutex);
 +    }
 +
 +    /* There is no race condition here.  Other threads will call
 +     * aio_co_schedule on our AioContext, which can reenter this
 +     * coroutine but only after this yield and after the main loop
 +     * has gone through the next iteration.
 +     */
      qemu_coroutine_yield();
      assert(qemu_in_coroutine());
 +
 +    /* TODO: OSv implements wait morphing here, where the wakeup
 +     * primitive automatically places the woken coroutine on the
 +     * mutex's queue.  This avoids the thundering herd effect.
 +     */
 +    if (mutex) {
 +        qemu_co_mutex_lock(mutex);
 +    }
  }
  /**
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_rdlock(CoRwlock *lock)
      Coroutine *self = qemu_coroutine_self();
      while (lock->writer) {
 -        qemu_co_queue_wait(&lock->queue);
 +        qemu_co_queue_wait(&lock->queue, NULL);
      }
      lock->reader++;
      self->locks_held++;
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_wrlock(CoRwlock *lock)
      Coroutine *self = qemu_coroutine_self();
      while (lock->writer || lock->reader) {
 -        qemu_co_queue_wait(&lock->queue);
 +        qemu_co_queue_wait(&lock->queue, NULL);
      }
      lock->writer = true;
      self->locks_held++;
 --
-.35.1
+.9.3

-[PULL 02/14] iotests: replace calls to log(qemu_io(...)) with qemu_io_log()
+[Qemu-devel] [PULL v2 24/24] coroutine-lock: make CoRwlock thread-safe and fair
-From: John Snow <jsnow@redhat.com>
+From: Paolo Bonzini <pbonzini@redhat.com>
-This makes these callsites a little simpler, but the real motivation is
+This adds a CoMutex around the existing CoQueue.  Because the write-side
-a forthcoming commit will change the return type of qemu_io(), so removing
+can just take CoMutex, the old "writer" field is not necessary anymore.
-users of the return value now is helpful.
+Instead of removing it altogether, count the number of pending writers
 during a read-side critical section and forbid further readers from
 entering.
-Signed-off-by: John Snow <jsnow@redhat.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
-Reviewed-by: Eric Blake <eblake@redhat.com>
+Reviewed-by: Fam Zheng <famz@redhat.com>
-Reviewed-by: Hanna Reitz <hreitz@redhat.com>
+Message-id: 20170213181244.16297-7-pbonzini@redhat.com
-Message-Id: <20220418211504.943969-2-jsnow@redhat.com>
+Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Hanna Reitz <hreitz@redhat.com>
 ---
- tests/qemu-iotests/242 | 6 +++---
+ include/qemu/coroutine.h   |  3 ++-
- tests/qemu-iotests/255 | 4 +---
+ util/qemu-coroutine-lock.c | 35 ++++++++++++++++++++++++-----------
- tests/qemu-iotests/303 | 4 ++--
+files changed, 26 insertions(+), 12 deletions(-)
 files changed, 6 insertions(+), 8 deletions(-)
-diff --git a/tests/qemu-iotests/242 b/tests/qemu-iotests/242
+diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
-index XXXXXXX..XXXXXXX 100755
+index XXXXXXX..XXXXXXX 100644
---- a/tests/qemu-iotests/242
+--- a/include/qemu/coroutine.h
-+++ b/tests/qemu-iotests/242
++++ b/include/qemu/coroutine.h
-@@ -XXX,XX +XXX,XX @@
+@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue);
- import iotests
- import json
- import struct
+ typedef struct CoRwlock {
--from iotests import qemu_img_create, qemu_io, qemu_img_info, \
+-    bool writer;
--    file_path, img_info_log, log, filter_qemu_io
++    int pending_writer;
-+from iotests import qemu_img_create, qemu_io_log, qemu_img_info, \
+     int reader;
-+    file_path, img_info_log, log
++    CoMutex mutex;
+     CoQueue queue;
- iotests.script_initialize(supported_fmts=['qcow2'],
+ } CoRwlock;
-                           supported_protocols=['file'],
-@@ -XXX,XX +XXX,XX @@ def add_bitmap(bitmap_number, persistent, disabled):
+diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
+index XXXXXXX..XXXXXXX 100644
- def write_to_disk(offset, size):
+--- a/util/qemu-coroutine-lock.c
-     write = 'write {} {}'.format(offset, size)
++++ b/util/qemu-coroutine-lock.c
--    log(qemu_io('-c', write, disk), filters=[filter_qemu_io])
+@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_init(CoRwlock *lock)
-+    qemu_io_log('-c', write, disk)
+ {
+     memset(lock, 0, sizeof(*lock));
+     qemu_co_queue_init(&lock->queue);
- def toggle_flag(offset):
++    qemu_co_mutex_init(&lock->mutex);
-diff --git a/tests/qemu-iotests/255 b/tests/qemu-iotests/255
+ }
-index XXXXXXX..XXXXXXX 100755
---- a/tests/qemu-iotests/255
+ void qemu_co_rwlock_rdlock(CoRwlock *lock)
-+++ b/tests/qemu-iotests/255
+ {
-@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('src.qcow2') as src_path, \
+     Coroutine *self = qemu_coroutine_self();
-     iotests.qemu_img_create('-f', iotests.imgfmt, src_path, size_str)
-     iotests.qemu_img_create('-f', iotests.imgfmt, dst_path, size_str)
+-    while (lock->writer) {
+-        qemu_co_queue_wait(&lock->queue, NULL);
--    iotests.log(iotests.qemu_io('-f', iotests.imgfmt, '-c', 'write 0 1M',
++    qemu_co_mutex_lock(&lock->mutex);
--                                src_path),
++    /* For fairness, wait if a writer is in line.  */
--                filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
++    while (lock->pending_writer) {
-+    iotests.qemu_io_log('-f', iotests.imgfmt, '-c', 'write 0 1M', src_path),
++        qemu_co_queue_wait(&lock->queue, &lock->mutex);
+     }
-     vm.add_object('throttle-group,x-bps-read=4096,id=throttle0')
+     lock->reader++;
++    qemu_co_mutex_unlock(&lock->mutex);
-diff --git a/tests/qemu-iotests/303 b/tests/qemu-iotests/303
++
-index XXXXXXX..XXXXXXX 100755
++    /* The rest of the read-side critical section is run without the mutex.  */
---- a/tests/qemu-iotests/303
+     self->locks_held++;
-+++ b/tests/qemu-iotests/303
+ }
-@@ -XXX,XX +XXX,XX @@
+@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_unlock(CoRwlock *lock)
- import iotests
+     Coroutine *self = qemu_coroutine_self();
- import subprocess
--from iotests import qemu_img_create, qemu_io, file_path, log, filter_qemu_io, \
+     assert(qemu_in_coroutine());
-+from iotests import qemu_img_create, qemu_io_log, file_path, log, \
+-    if (lock->writer) {
-         verify_qcow2_zstd_compression
+-        lock->writer = false;
++    if (!lock->reader) {
- iotests.script_initialize(supported_fmts=['qcow2'],
++        /* The critical section started in qemu_co_rwlock_wrlock.  */
-@@ -XXX,XX +XXX,XX @@ def create_bitmap(bitmap_number, disabled):
+         qemu_co_queue_restart_all(&lock->queue);
+     } else {
- def write_to_disk(offset, size):
++        self->locks_held--;
-     write = f'write {offset} {size}'
++
--    log(qemu_io('-c', write, disk), filters=[filter_qemu_io])
++        qemu_co_mutex_lock(&lock->mutex);
-+    qemu_io_log('-c', write, disk)
+         lock->reader--;
+         assert(lock->reader >= 0);
+         /* Wakeup only one waiting writer */
- def add_bitmap(num, begin, end, disabled):
+@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_unlock(CoRwlock *lock)
              qemu_co_queue_next(&lock->queue);
          }
      }
 -    self->locks_held--;
 +    qemu_co_mutex_unlock(&lock->mutex);
  }
  void qemu_co_rwlock_wrlock(CoRwlock *lock)
  {
 -    Coroutine *self = qemu_coroutine_self();
 -
 -    while (lock->writer || lock->reader) {
 -        qemu_co_queue_wait(&lock->queue, NULL);
 +    qemu_co_mutex_lock(&lock->mutex);
 +    lock->pending_writer++;
 +    while (lock->reader) {
 +        qemu_co_queue_wait(&lock->queue, &lock->mutex);
      }
 -    lock->writer = true;
 -    self->locks_held++;
 +    lock->pending_writer--;
 +
 +    /* The rest of the write-side critical section is run with
 +     * the mutex taken, so that lock->reader remains zero.
 +     * There is no need to update self->locks_held.
 +     */
  }
 --
-.35.1
+.9.3

The following changes since commit 754f756cc4c6d9d14b7230c62b5bb20f9d655888:

Merge tag 'pull-target-arm-20220422-1' of https://git.linaro.org/people/pmaydell/qemu-arm into staging (2022-04-22 08:03:18 -0700)

are available in the Git repository at:

https://gitlab.com/hreitz/qemu.git tags/pull-block-2022-04-25

for you to fetch changes up to 348a0740afc5b313599533eb69bbb2b95d2f1bba:

iotests/108: Fix when missing user_allow_other (2022-04-25 14:46:45 +0200)

----------------------------------------------------------------
Block patches:
- New @force parameter for blockdev-change-medium
- Improvements to the iotests to help with debugging
- Fix iotest 108 on systems without user_allow_other in fuse.conf

----------------------------------------------------------------
Denis V. Lunev (1):
  block: add 'force' parameter to 'blockdev-change-medium' command

Hanna Reitz (1):
  iotests/108: Fix when missing user_allow_other

John Snow (12):
  iotests: replace calls to log(qemu_io(...)) with qemu_io_log()
  iotests/163: Fix broken qemu-io invocation
  iotests: Don't check qemu_io() output for specific error strings
  iotests/040: Don't check image pattern on zero-length image
  iotests/040: Fix TestCommitWithFilters test
  iotests: create generic qemu_tool() function
  iotests: rebase qemu_io() on top of qemu_tool()
  iotests/migration-permissions: use assertRaises() for qemu_io()
    negative test
  iotests/image-fleecing: switch to qemu_io()
  iotests: remove qemu_io_pipe_and_status()
  iotests: remove qemu_io_silent() and qemu_io_silent_check().
  iotests: make qemu_io_log() check return codes by default

-- 
2.35.1

From: "Denis V. Lunev" <den@openvz.org>

'blockdev-change-medium' is a convinient wrapper for the following
sequence of commands:
 * blockdev-open-tray
 * blockdev-remove-medium
 * blockdev-insert-medium
 * blockdev-close-tray
and should be used f.e. to change ISO image inside the CD-ROM tray.
Though the guest could lock the tray and some linux guests like
CentOS 8.5 actually does that. In this case the execution if this
command results in the error like the following:
  Device 'scsi0-0-1-0' is locked and force was not specified,
  wait for tray to open and try again.

This situation is could be resolved 'blockdev-open-tray' by passing
flag 'force' inside. Thus is seems reasonable to add the same
capability for 'blockdev-change-medium' too.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@openvz.org>
Acked-by: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Hanna Reitz <hreitz@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
Message-Id: <20220412221846.280723-1-den@openvz.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
 qapi/block.json     |  6 ++++++
 block/qapi-sysemu.c |  3 ++-
 monitor/hmp-cmds.c  |  4 +++-
 hmp-commands.hx     | 11 +++++++----
 ui/cocoa.m          |  1 +
 5 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/qapi/block.json b/qapi/block.json
index XXXXXXX..XXXXXXX 100644
--- a/qapi/block.json
+++ b/qapi/block.json
@@ -XXX,XX +XXX,XX @@
 # @read-only-mode: change the read-only mode of the device; defaults
 #                  to 'retain'
 #
+# @force: if false (the default), an eject request through blockdev-open-tray
+#         will be sent to the guest if it has locked the tray (and the tray
+#         will not be opened immediately); if true, the tray will be opened
+#         regardless of whether it is locked. (since 7.1)
+#
 # Features:
 # @deprecated: Member @device is deprecated.  Use @id instead.
 #
@@ -XXX,XX +XXX,XX @@
             '*id': 'str',
             'filename': 'str',
             '*format': 'str',
+            '*force': 'bool',
             '*read-only-mode': 'BlockdevChangeReadOnlyMode' } }
 
 
diff --git a/block/qapi-sysemu.c b/block/qapi-sysemu.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qapi-sysemu.c
+++ b/block/qapi-sysemu.c
@@ -XXX,XX +XXX,XX @@ void qmp_blockdev_change_medium(bool has_device, const char *device,
                                 bool has_id, const char *id,
                                 const char *filename,
                                 bool has_format, const char *format,
+                                bool has_force, bool force,
                                 bool has_read_only,
                                 BlockdevChangeReadOnlyMode read_only,
                                 Error **errp)
@@ -XXX,XX +XXX,XX @@ void qmp_blockdev_change_medium(bool has_device, const char *device,
 
     rc = do_open_tray(has_device ? device : NULL,
                       has_id ? id : NULL,
-                      false, &err);
+                      force, &err);
     if (rc && rc != -ENOSYS) {
         error_propagate(errp, err);
         goto fail;
diff --git a/monitor/hmp-cmds.c b/monitor/hmp-cmds.c
index XXXXXXX..XXXXXXX 100644
--- a/monitor/hmp-cmds.c
+++ b/monitor/hmp-cmds.c
@@ -XXX,XX +XXX,XX @@ void hmp_change(Monitor *mon, const QDict *qdict)
     const char *target = qdict_get_str(qdict, "target");
     const char *arg = qdict_get_try_str(qdict, "arg");
     const char *read_only = qdict_get_try_str(qdict, "read-only-mode");
+    bool force = qdict_get_try_bool(qdict, "force", false);
     BlockdevChangeReadOnlyMode read_only_mode = 0;
     Error *err = NULL;
 
@@ -XXX,XX +XXX,XX @@ void hmp_change(Monitor *mon, const QDict *qdict)
         }
 
         qmp_blockdev_change_medium(true, device, false, NULL, target,
-                                   !!arg, arg, !!read_only, read_only_mode,
+                                   !!arg, arg, true, force,
+                                   !!read_only, read_only_mode,
                                    &err);
     }
 
diff --git a/hmp-commands.hx b/hmp-commands.hx
index XXXXXXX..XXXXXXX 100644
--- a/hmp-commands.hx
+++ b/hmp-commands.hx
@@ -XXX,XX +XXX,XX @@ ERST
 
     {
         .name       = "change",
-        .args_type  = "device:B,target:F,arg:s?,read-only-mode:s?",
-        .params     = "device filename [format [read-only-mode]]",
-        .help       = "change a removable medium, optional format",
+        .args_type  = "device:B,force:-f,target:F,arg:s?,read-only-mode:s?",
+        .params     = "device [-f] filename [format [read-only-mode]]",
+        .help       = "change a removable medium, optional format, use -f to force the operation",
         .cmd        = hmp_change,
     },
 
@@ -XXX,XX +XXX,XX @@ SRST
 ``change`` *device* *setting*
   Change the configuration of a device.
 
-  ``change`` *diskdevice* *filename* [*format* [*read-only-mode*]]
+  ``change`` *diskdevice* [-f] *filename* [*format* [*read-only-mode*]]
     Change the medium for a removable disk device to point to *filename*. eg::
 
       (qemu) change ide1-cd0 /path/to/some.iso
 
+    ``-f``
+      forces the operation even if the guest has locked the tray.
+
     *format* is optional.
 
     *read-only-mode* may be used to change the read-only status of the device.
diff --git a/ui/cocoa.m b/ui/cocoa.m
index XXXXXXX..XXXXXXX 100644
--- a/ui/cocoa.m
+++ b/ui/cocoa.m
@@ -XXX,XX +XXX,XX @@ - (void)changeDeviceMedia:(id)sender
                                        [file cStringUsingEncoding:
                                                  NSASCIIStringEncoding],
                                        true, "raw",
+                                       true, false,
                                        false, 0,
                                        &err);
         });
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

This makes these callsites a little simpler, but the real motivation is
a forthcoming commit will change the return type of qemu_io(), so removing
users of the return value now is helpful.

diff --git a/tests/qemu-iotests/242 b/tests/qemu-iotests/242
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/242
+++ b/tests/qemu-iotests/242
@@ -XXX,XX +XXX,XX @@
 import iotests
 import json
 import struct
-from iotests import qemu_img_create, qemu_io, qemu_img_info, \
-    file_path, img_info_log, log, filter_qemu_io
+from iotests import qemu_img_create, qemu_io_log, qemu_img_info, \
+    file_path, img_info_log, log
 
 iotests.script_initialize(supported_fmts=['qcow2'],
                           supported_protocols=['file'],
@@ -XXX,XX +XXX,XX @@ def add_bitmap(bitmap_number, persistent, disabled):
 
 def write_to_disk(offset, size):
     write = 'write {} {}'.format(offset, size)
-    log(qemu_io('-c', write, disk), filters=[filter_qemu_io])
+    qemu_io_log('-c', write, disk)
 
 
 def toggle_flag(offset):
diff --git a/tests/qemu-iotests/255 b/tests/qemu-iotests/255
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/255
+++ b/tests/qemu-iotests/255
@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('src.qcow2') as src_path, \
     iotests.qemu_img_create('-f', iotests.imgfmt, src_path, size_str)
     iotests.qemu_img_create('-f', iotests.imgfmt, dst_path, size_str)
 
-    iotests.log(iotests.qemu_io('-f', iotests.imgfmt, '-c', 'write 0 1M',
-                                src_path),
-                filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
+    iotests.qemu_io_log('-f', iotests.imgfmt, '-c', 'write 0 1M', src_path),
 
     vm.add_object('throttle-group,x-bps-read=4096,id=throttle0')
 
diff --git a/tests/qemu-iotests/303 b/tests/qemu-iotests/303
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/303
+++ b/tests/qemu-iotests/303
@@ -XXX,XX +XXX,XX @@
 
 import iotests
 import subprocess
-from iotests import qemu_img_create, qemu_io, file_path, log, filter_qemu_io, \
+from iotests import qemu_img_create, qemu_io_log, file_path, log, \
         verify_qcow2_zstd_compression
 
 iotests.script_initialize(supported_fmts=['qcow2'],
@@ -XXX,XX +XXX,XX @@ def create_bitmap(bitmap_number, disabled):
 
 def write_to_disk(offset, size):
     write = f'write {offset} {size}'
-    log(qemu_io('-c', write, disk), filters=[filter_qemu_io])
+    qemu_io_log('-c', write, disk)
 
 
 def add_bitmap(num, begin, end, disabled):
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

The 'read' commands to qemu-io were malformed, and this invocation only
worked by coincidence because the error messages were identical. Oops.

There's no point in checking the patterning of the reference image, so
just check the empty image by itself instead.

(Note: as of this commit, nothing actually enforces that this command
completes successfully, but a forthcoming commit in this series will
enforce that qemu_io() must have a zero status code.)

diff --git a/tests/qemu-iotests/163 b/tests/qemu-iotests/163
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/163
+++ b/tests/qemu-iotests/163
@@ -XXX,XX +XXX,XX @@ class ShrinkBaseClass(iotests.QMPTestCase):
         qemu_img('resize',  '-f', iotests.imgfmt, '--shrink', test_img,
                  self.shrink_size)
 
-        self.assertEqual(
-            qemu_io('-c', 'read -P 0x00 %s'%self.shrink_size, test_img),
-            qemu_io('-c', 'read -P 0x00 %s'%self.shrink_size, check_img),
-            "Verifying image content")
+        qemu_io('-c', f"read -P 0x00 0 {self.shrink_size}", test_img)
 
         self.image_verify()
 
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

A forthcoming commit updates qemu_io() to raise an exception on non-zero
return by default, and changes its return type.

In preparation, simplify some calls to qemu_io() that assert that
specific error message strings do not appear in qemu-io's
output. Asserting that all of these calls return a status code of zero
will be a more robust way to guard against failure.

diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/040
+++ b/tests/qemu-iotests/040
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
 
     def test_commit(self):
         self.run_commit_test(mid_img, backing_img)
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     def test_commit_node(self):
         self.run_commit_test("mid", "base", node_names=True)
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     @iotests.skip_if_unsupported(['throttle'])
     def test_commit_with_filter_and_quit(self):
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
 
     def test_top_is_active(self):
         self.run_commit_test(test_img, backing_img, need_ready=True)
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     def test_top_is_default_active(self):
         self.run_default_commit_test()
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     def test_top_and_base_reversed(self):
         self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestRelativePaths(ImageCommitTestCase):
 
     def test_commit(self):
         self.run_commit_test(self.mid_img, self.backing_img)
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs)
 
     def test_device_not_found(self):
         result = self.vm.qmp('block-commit', device='nonexistent', top='%s' % self.mid_img)
@@ -XXX,XX +XXX,XX @@ class TestRelativePaths(ImageCommitTestCase):
 
     def test_top_is_active(self):
         self.run_commit_test(self.test_img, self.backing_img)
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs).find("verification failed"))
-        self.assertEqual(-1, qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs).find("verification failed"))
+        qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', self.backing_img_abs)
+        qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', self.backing_img_abs)
 
     def test_top_and_base_reversed(self):
         self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
 
     def do_test_io(self, read_or_write):
         for index, pattern_file in enumerate(self.pattern_files):
-            result = qemu_io('-f', iotests.imgfmt,
-                             '-c',
-                             f'{read_or_write} -P {index + 1} {index}M 1M',
-                             pattern_file)
-            self.assertFalse('Pattern verification failed' in result)
+            qemu_io('-f', iotests.imgfmt,
+                    '-c',
+                    f'{read_or_write} -P {index + 1} {index}M 1M',
+                    pattern_file)
 
     @iotests.skip_if_unsupported(['throttle'])
     def setUp(self):
diff --git a/tests/qemu-iotests/056 b/tests/qemu-iotests/056
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/056
+++ b/tests/qemu-iotests/056
@@ -XXX,XX +XXX,XX @@ class TestSyncModesNoneAndTop(iotests.QMPTestCase):
 
         self.vm.shutdown()
         time.sleep(1)
-        self.assertEqual(-1, qemu_io('-c', 'read -P0x41 0 512', target_img).find("verification failed"))
+        qemu_io('-c', 'read -P0x41 0 512', target_img)
 
 class TestBeforeWriteNotifier(iotests.QMPTestCase):
     def setUp(self):
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

qemu-io fails on read/write beyond end-of-file on raw images, so skip
these invocations when running the zero-length image tests.

diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/040
+++ b/tests/qemu-iotests/040
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
         qemu_img('create', '-f', iotests.imgfmt,
                  '-o', 'backing_file=%s' % mid_img,
                  '-F', iotests.imgfmt, test_img)
-        qemu_io('-f', 'raw', '-c', 'write -P 0xab 0 524288', backing_img)
-        qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xef 524288 524288', mid_img)
+        if self.image_len:
+            qemu_io('-f', 'raw', '-c', 'write -P 0xab 0 524288', backing_img)
+            qemu_io('-f', iotests.imgfmt, '-c', 'write -P 0xef 524288 524288',
+                    mid_img)
         self.vm = iotests.VM().add_drive(test_img, "node-name=top,backing.node-name=mid,backing.backing.node-name=base", interface="none")
         self.vm.add_device('virtio-scsi')
         self.vm.add_device("scsi-hd,id=scsi0,drive=drive0")
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
 
     def test_commit(self):
         self.run_commit_test(mid_img, backing_img)
+        if not self.image_len:
+            return
         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     def test_commit_node(self):
         self.run_commit_test("mid", "base", node_names=True)
+        if not self.image_len:
+            return
         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(ImageCommitTestCase):
 
     def test_top_is_active(self):
         self.run_commit_test(test_img, backing_img, need_ready=True)
+        if not self.image_len:
+            return
         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
     def test_top_is_default_active(self):
         self.run_default_commit_test()
+        if not self.image_len:
+            return
         qemu_io('-f', 'raw', '-c', 'read -P 0xab 0 524288', backing_img)
         qemu_io('-f', 'raw', '-c', 'read -P 0xef 524288 524288', backing_img)
 
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

Without this change, asserting that qemu_io always returns 0 causes this
test to fail in a way we happened not to be catching previously:

qemu.utils.VerboseProcessError: Command
  '('/home/jsnow/src/qemu/bin/git/tests/qemu-iotests/../../qemu-io',
  '--cache', 'writeback', '--aio', 'threads', '-f', 'qcow2', '-c',
  'read -P 4 3M 1M',
  '/home/jsnow/src/qemu/bin/git/tests/qemu-iotests/scratch/3.img')'
  returned non-zero exit status 1.
  ┏━ output ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  ┃ qemu-io: can't open device
  ┃ /home/jsnow/src/qemu/bin/git/tests/qemu-iotests/scratch/3.img:
  ┃ Could not open backing file: Could not open backing file: Throttle
  ┃ group 'tg' does not exist
  ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

The commit jobs changes the backing file string stored in the image file
header belonging to the node above the commit’s top node to point to the
commit target (the base node).  QEMU tries to be as accurate as
possible, and so in these test cases will include the filter that is
part of the block graph in that backing file string (by virtue of making
it a json:{} description of the post-commit subgraph).  This makes
little sense outside of QEMU, though: Specifically, the throttle node in
that subgraph will dearly miss its supposedly associated throttle group
object.

When starting the commit job, we can specify a custom backing file
string to write into said image file, so let’s use that feature to write
the plain filename of the backing chain’s next actual image file there.

Explicitly provide the backing file so that opening the file outside of
QEMU (Where we will not have throttle groups) will succeed.

Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20220418211504.943969-6-jsnow@redhat.com>
---
 tests/qemu-iotests/040 | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/040
+++ b/tests/qemu-iotests/040
@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
                              job_id='commit',
                              device='top-filter',
                              top_node='cow-2',
-                             base_node='cow-1')
+                             base_node='cow-1',
+                             backing_file=self.img1)
         self.assert_qmp(result, 'return', {})
         self.wait_until_completed(drive='commit')
 
@@ -XXX,XX +XXX,XX @@ class TestCommitWithFilters(iotests.QMPTestCase):
                              job_id='commit',
                              device='top-filter',
                              top_node='cow-1',
-                             base_node='cow-0')
+                             base_node='cow-0',
+                             backing_file=self.img0)
         self.assert_qmp(result, 'return', {})
         self.wait_until_completed(drive='commit')
 
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

reimplement qemu_img() in terms of qemu_tool() in preparation for doing
the same with qemu_io().

diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_img_create_prepare_args(args: List[str]) -> List[str]:
 
     return result
 
-def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
-             ) -> 'subprocess.CompletedProcess[str]':
-    """
-    Run qemu_img and return the status code and console output.
 
-    This function always prepends QEMU_IMG_OPTIONS and may further alter
-    the args for 'create' commands.
+def qemu_tool(*args: str, check: bool = True, combine_stdio: bool = True
+              ) -> 'subprocess.CompletedProcess[str]':
+    """
+    Run a qemu tool and return its status code and console output.
 
-    :param args: command-line arguments to qemu-img.
+    :param args: full command line to run.
     :param check: Enforce a return code of zero.
     :param combine_stdio: set to False to keep stdout/stderr separated.
 
@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
         properties. If streams are not combined, it will also have a
         stderr property.
     """
-    full_args = qemu_img_args + qemu_img_create_prepare_args(list(args))
-
     subp = subprocess.run(
-        full_args,
+        args,
         stdout=subprocess.PIPE,
         stderr=subprocess.STDOUT if combine_stdio else subprocess.PIPE,
         universal_newlines=True,
@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
 
     if check and subp.returncode or (subp.returncode < 0):
         raise VerboseProcessError(
-            subp.returncode, full_args,
+            subp.returncode, args,
             output=subp.stdout,
             stderr=subp.stderr,
         )
@@ -XXX,XX +XXX,XX @@ def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
     return subp
 
 
+def qemu_img(*args: str, check: bool = True, combine_stdio: bool = True
+             ) -> 'subprocess.CompletedProcess[str]':
+    """
+    Run QEMU_IMG_PROG and return its status code and console output.
+
+    This function always prepends QEMU_IMG_OPTIONS and may further alter
+    the args for 'create' commands.
+
+    See `qemu_tool()` for greater detail.
+    """
+    full_args = qemu_img_args + qemu_img_create_prepare_args(list(args))
+    return qemu_tool(*full_args, check=check, combine_stdio=combine_stdio)
+
+
 def ordered_qmp(qmsg, conv_keys=True):
     # Dictionaries are not ordered prior to 3.6, therefore:
     if isinstance(qmsg, list):
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

Rework qemu_io() to be analogous to qemu_img(); a function that requires
a return code of zero by default unless disabled explicitly.

Tests that use qemu_io():
030 040 041 044 055 056 093 124 129 132 136 148 149 151 152 163 165 205
209 219 236 245 248 254 255 257 260 264 280 298 300 302 304
image-fleecing migrate-bitmaps-postcopy-test migrate-bitmaps-test
migrate-during-backup migration-permissions

Test that use qemu_io_log():
242 245 255 274 303 307 nbd-reconnect-on-open

Copy-pastables for testing/verification:

./check -qcow2 030 040 041 044 055 056 124 129 132 151 152 163 165 209 \
               219 236 242 245 248 254 255 257 260 264 274 \
               280 298 300 302 303 304 307 image-fleecing \
               migrate-bitmaps-postcopy-test migrate-bitmaps-test \
               migrate-during-backup nbd-reconnect-on-open
./check -raw 093 136 148 migration-permissions
./check -nbd 205

# ./configure configure --disable-gnutls --enable-gcrypt
# this ALSO requires passwordless sudo.
./check -luks 149

# Just the tests that were edited in this commit:
./check -qcow2 030 040 242 245
./check -raw migration-permissions
./check -nbd 205
./check -luks 149

Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20220418211504.943969-8-jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
 tests/qemu-iotests/030                        | 85 +++++++++++--------
 tests/qemu-iotests/149                        |  6 +-
 tests/qemu-iotests/205                        |  4 +-
 tests/qemu-iotests/245                        | 17 ++--
 tests/qemu-iotests/iotests.py                 | 19 +++--
 .../qemu-iotests/tests/migration-permissions  |  4 +-
 6 files changed, 81 insertions(+), 54 deletions(-)

diff --git a/tests/qemu-iotests/030 b/tests/qemu-iotests/030
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/030
+++ b/tests/qemu-iotests/030
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
-                         'image file map does not match backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
+            'image file map does not match backing file after streaming')
 
     def test_stream_intermediate(self):
         self.assert_no_active_block_jobs()
 
-        self.assertNotEqual(qemu_io('-f', 'raw', '-rU', '-c', 'map', backing_img),
-                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', mid_img),
-                            'image file map matches backing file before streaming')
+        self.assertNotEqual(
+            qemu_io('-f', 'raw', '-rU', '-c', 'map', backing_img).stdout,
+            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', mid_img).stdout,
+            'image file map matches backing file before streaming')
 
         result = self.vm.qmp('block-stream', device='mid', job_id='stream-mid')
         self.assert_qmp(result, 'return', {})
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img),
-                         'image file map does not match backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img).stdout,
+            'image file map does not match backing file after streaming')
 
     def test_stream_pause(self):
         self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', 'raw', '-c', 'map', backing_img),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
-                         'image file map does not match backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', 'raw', '-c', 'map', backing_img).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
+            'image file map does not match backing file after streaming')
 
     def test_stream_no_op(self):
         self.assert_no_active_block_jobs()
 
         # The image map is empty before the operation
-        empty_map = qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', test_img)
+        empty_map = qemu_io(
+            '-f', iotests.imgfmt, '-rU', '-c', 'map', test_img).stdout
 
         # This is a no-op: no data should ever be copied from the base image
         result = self.vm.qmp('block-stream', device='drive0', base=mid_img)
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
-                         empty_map, 'image file map changed after a no-op')
+        self.assertEqual(
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
+            empty_map, 'image file map changed after a no-op')
 
     def test_stream_partial(self):
         self.assert_no_active_block_jobs()
@@ -XXX,XX +XXX,XX @@ class TestSingleDrive(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img),
-                         'image file map does not match backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', mid_img).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', test_img).stdout,
+            'image file map does not match backing file after streaming')
 
     def test_device_not_found(self):
         result = self.vm.qmp('block-stream', device='nonexistent')
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
 
         # Check that the maps don't match before the streaming operations
         for i in range(2, self.num_imgs, 2):
-            self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i]),
-                                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i-1]),
-                                'image file map matches backing file before streaming')
+            self.assertNotEqual(
+                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i]).stdout,
+                qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[i-1]).stdout,
+                'image file map matches backing file before streaming')
 
         # Create all streaming jobs
         pending_jobs = []
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
 
         # Check that all maps match now
         for i in range(2, self.num_imgs, 2):
-            self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i]),
-                             qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i-1]),
-                             'image file map does not match backing file after streaming')
+            self.assertEqual(
+                qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i]).stdout,
+                qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[i-1]).stdout,
+                'image file map does not match backing file after streaming')
 
     # Test that it's not possible to perform two block-stream
     # operations if there are nodes involved in both.
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
     def test_stream_base_node_name(self):
         self.assert_no_active_block_jobs()
 
-        self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[4]),
-                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[3]),
-                            'image file map matches backing file before streaming')
+        self.assertNotEqual(
+            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[4]).stdout,
+            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.imgs[3]).stdout,
+            'image file map matches backing file before streaming')
 
         # Error: the base node does not exist
         result = self.vm.qmp('block-stream', device='node4', base_node='none', job_id='stream')
@@ -XXX,XX +XXX,XX @@ class TestParallelOps(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[4]),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[3]),
-                         'image file map matches backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[4]).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.imgs[3]).stdout,
+            'image file map matches backing file after streaming')
 
 class TestQuorum(iotests.QMPTestCase):
     num_children = 3
@@ -XXX,XX +XXX,XX @@ class TestQuorum(iotests.QMPTestCase):
             os.remove(img)
 
     def test_stream_quorum(self):
-        self.assertNotEqual(qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.children[0]),
-                            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.backing[0]),
-                            'image file map matches backing file before streaming')
+        self.assertNotEqual(
+            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.children[0]).stdout,
+            qemu_io('-f', iotests.imgfmt, '-rU', '-c', 'map', self.backing[0]).stdout,
+            'image file map matches backing file before streaming')
 
         self.assert_no_active_block_jobs()
 
@@ -XXX,XX +XXX,XX @@ class TestQuorum(iotests.QMPTestCase):
         self.assert_no_active_block_jobs()
         self.vm.shutdown()
 
-        self.assertEqual(qemu_io('-f', iotests.imgfmt, '-c', 'map', self.children[0]),
-                         qemu_io('-f', iotests.imgfmt, '-c', 'map', self.backing[0]),
-                         'image file map does not match backing file after streaming')
+        self.assertEqual(
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.children[0]).stdout,
+            qemu_io('-f', iotests.imgfmt, '-c', 'map', self.backing[0]).stdout,
+            'image file map does not match backing file after streaming')
 
 class TestSmallerBackingFile(iotests.QMPTestCase):
     backing_len = 1 * 1024 * 1024 # MB
diff --git a/tests/qemu-iotests/149 b/tests/qemu-iotests/149
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/149
+++ b/tests/qemu-iotests/149
@@ -XXX,XX +XXX,XX @@ def qemu_io_write_pattern(config, pattern, offset_mb, size_mb, dev=False):
     args = ["-c", "write -P 0x%x %dM %dM" % (pattern, offset_mb, size_mb)]
     args.extend(qemu_io_image_args(config, dev))
     iotests.log("qemu-io " + " ".join(args), filters=[iotests.filter_test_dir])
-    iotests.log(check_cipher_support(config, iotests.qemu_io(*args)),
+    output = iotests.qemu_io(*args, check=False).stdout
+    iotests.log(check_cipher_support(config, output),
                 filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
 
 
@@ -XXX,XX +XXX,XX @@ def qemu_io_read_pattern(config, pattern, offset_mb, size_mb, dev=False):
     args = ["-c", "read -P 0x%x %dM %dM" % (pattern, offset_mb, size_mb)]
     args.extend(qemu_io_image_args(config, dev))
     iotests.log("qemu-io " + " ".join(args), filters=[iotests.filter_test_dir])
-    iotests.log(check_cipher_support(config, iotests.qemu_io(*args)),
+    output = iotests.qemu_io(*args, check=False).stdout
+    iotests.log(check_cipher_support(config, output),
                 filters=[iotests.filter_test_dir, iotests.filter_qemu_io])
 
 
diff --git a/tests/qemu-iotests/205 b/tests/qemu-iotests/205
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/205
+++ b/tests/qemu-iotests/205
@@ -XXX,XX +XXX,XX @@ class TestNbdServerRemove(iotests.QMPTestCase):
 
     def do_test_connect_after_remove(self, mode=None):
         args = ('-r', '-f', 'raw', '-c', 'read 0 512', nbd_uri)
-        self.assertReadOk(qemu_io(*args))
+        self.assertReadOk(qemu_io(*args).stdout)
 
         result = self.remove_export('exp', mode)
         self.assert_qmp(result, 'return', {})
 
         self.assertExportNotFound('exp')
-        self.assertConnectFailed(qemu_io(*args))
+        self.assertConnectFailed(qemu_io(*args, check=False).stdout)
 
     def test_connect_after_remove_default(self):
         self.do_test_connect_after_remove()
diff --git a/tests/qemu-iotests/245 b/tests/qemu-iotests/245
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/245
+++ b/tests/qemu-iotests/245
@@ -XXX,XX +XXX,XX @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 
+import copy
+import json
 import os
 import re
+from subprocess import CalledProcessError
+
 import iotests
-import copy
-import json
 from iotests import qemu_img, qemu_io
 
 hd_path = [
@@ -XXX,XX +XXX,XX @@ class TestBlockdevReopen(iotests.QMPTestCase):
 
     # Reopen an image several times changing some of its options
     def test_reopen(self):
-        # Check whether the filesystem supports O_DIRECT
-        if 'O_DIRECT' in qemu_io('-f', 'raw', '-t', 'none', '-c', 'quit', hd_path[0]):
-            supports_direct = False
-        else:
+        try:
+            qemu_io('-f', 'raw', '-t', 'none', '-c', 'quit', hd_path[0])
             supports_direct = True
+        except CalledProcessError as exc:
+            if 'O_DIRECT' in exc.stdout:
+                supports_direct = False
+            else:
+                raise
 
         # Open the hd1 image passing all backing options
         opts = hd_opts(1)
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_io_wrap_args(args: Sequence[str]) -> List[str]:
 def qemu_io_popen(*args):
     return qemu_tool_popen(qemu_io_wrap_args(args))
 
-def qemu_io(*args):
-    '''Run qemu-io and return the stdout data'''
-    return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))[0]
+def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
+            ) -> 'subprocess.CompletedProcess[str]':
+    """
+    Run QEMU_IO_PROG and return the status code and console output.
+
+    This function always prepends either QEMU_IO_OPTIONS or
+    QEMU_IO_OPTIONS_NO_FMT.
+    """
+    return qemu_tool(*qemu_io_wrap_args(args),
+                     check=check, combine_stdio=combine_stdio)
 
 def qemu_io_pipe_and_status(*args):
     return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))
 
-def qemu_io_log(*args):
-    result = qemu_io(*args)
-    log(result, filters=[filter_testfiles, filter_qemu_io])
+def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
+    result = qemu_io(*args, check=False)
+    log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
     return result
 
 def qemu_io_silent(*args):
diff --git a/tests/qemu-iotests/tests/migration-permissions b/tests/qemu-iotests/tests/migration-permissions
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/migration-permissions
+++ b/tests/qemu-iotests/tests/migration-permissions
@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
     def test_post_migration_permissions(self):
         # Try to access the image R/W, which should fail because virtio-blk
         # has not been configured with share-rw=on
-        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img)
+        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
         if not log.strip():
             print('ERROR (pre-migration): qemu-io should not be able to '
                   'access this image, but it reported no error')
@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
 
         # Try the same qemu-io access again, verifying that the WRITE
         # permission remains unshared
-        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img)
+        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
         if not log.strip():
             print('ERROR (post-migration): qemu-io should not be able to '
                   'access this image, but it reported no error')
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

Modify this test to use assertRaises for its negative testing of
qemu_io. If the exception raised does not match the one we tell it to
expect, we get *that* exception unhandled. If we get no exception, we
get a unittest assertion failure and the provided emsg printed to
screen.

If we get the CalledProcessError exception but the output is not what we
expect, we re-raise the original CalledProcessError.

Tidy.

(Note: Yes, you can reference "with" objects after that block ends; it
just means that ctx.__exit__(...) will have been called on it. It does
not *actually* go out of scope. unittests expects you to want to inspect
the Exception object, so they leave it defined post-exit.)

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220418211504.943969-9-jsnow@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
 .../qemu-iotests/tests/migration-permissions  | 28 +++++++++----------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/tests/qemu-iotests/tests/migration-permissions b/tests/qemu-iotests/tests/migration-permissions
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/migration-permissions
+++ b/tests/qemu-iotests/tests/migration-permissions
@@ -XXX,XX +XXX,XX @@
 #
 
 import os
+from subprocess import CalledProcessError
+
 import iotests
 from iotests import imgfmt, qemu_img_create, qemu_io
 
@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
     def test_post_migration_permissions(self):
         # Try to access the image R/W, which should fail because virtio-blk
         # has not been configured with share-rw=on
-        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
-        if not log.strip():
-            print('ERROR (pre-migration): qemu-io should not be able to '
-                  'access this image, but it reported no error')
-        else:
-            # This is the expected output
-            assert 'Is another process using the image' in log
+        emsg = ('ERROR (pre-migration): qemu-io should not be able to '
+                'access this image, but it reported no error')
+        with self.assertRaises(CalledProcessError, msg=emsg) as ctx:
+            qemu_io('-f', imgfmt, '-c', 'quit', test_img)
+        if 'Is another process using the image' not in ctx.exception.stdout:
+            raise ctx.exception
 
         # Now migrate the VM
         self.vm_s.qmp('migrate', uri=f'unix:{mig_sock}')
@@ -XXX,XX +XXX,XX @@ class TestMigrationPermissions(iotests.QMPTestCase):
 
         # Try the same qemu-io access again, verifying that the WRITE
         # permission remains unshared
-        log = qemu_io('-f', imgfmt, '-c', 'quit', test_img, check=False).stdout
-        if not log.strip():
-            print('ERROR (post-migration): qemu-io should not be able to '
-                  'access this image, but it reported no error')
-        else:
-            # This is the expected output
-            assert 'Is another process using the image' in log
+        emsg = ('ERROR (post-migration): qemu-io should not be able to '
+                'access this image, but it reported no error')
+        with self.assertRaises(CalledProcessError, msg=emsg) as ctx:
+            qemu_io('-f', imgfmt, '-c', 'quit', test_img)
+        if 'Is another process using the image' not in ctx.exception.stdout:
+            raise ctx.exception
 
 
 if __name__ == '__main__':
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

This test expects failure ... but only sometimes. When? Why?

It's for reads of a region not defined by a bitmap. Adjust the test to
be more explicit about what it expects to fail and why.

diff --git a/tests/qemu-iotests/tests/image-fleecing b/tests/qemu-iotests/tests/image-fleecing
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/image-fleecing
+++ b/tests/qemu-iotests/tests/image-fleecing
@@ -XXX,XX +XXX,XX @@
 #
 # Creator/Owner: John Snow <jsnow@redhat.com>
 
+from subprocess import CalledProcessError
+
 import iotests
-from iotests import log, qemu_img, qemu_io, qemu_io_silent, \
-    qemu_io_pipe_and_status
+from iotests import log, qemu_img, qemu_io, qemu_io_silent
 
 iotests.script_initialize(
     supported_fmts=['qcow2'],
@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
         for p in patterns + zeroes:
             cmd = 'read -P%s %s %s' % p
             log(cmd)
-            out, ret = qemu_io_pipe_and_status('-r', '-f', 'raw', '-c', cmd,
-                                               nbd_uri)
-            if ret != 0:
-                print(out)
+
+            try:
+                qemu_io('-r', '-f', 'raw', '-c', cmd, nbd_uri)
+            except CalledProcessError as exc:
+                if bitmap and p in zeroes:
+                    log(exc.stdout)
+                else:
+                    raise
 
     log('')
     log('--- Testing COW ---')
@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
             args += [target_img_path]
         else:
             args += ['-f', 'raw', nbd_uri]
-        out, ret = qemu_io_pipe_and_status(*args)
-        if ret != 0:
-            print(out)
+
+        try:
+            qemu_io(*args)
+        except CalledProcessError as exc:
+            if bitmap and p in zeroes:
+                log(exc.stdout)
+            else:
+                raise
 
     log('')
     log('--- Cleanup ---')
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

I know we just added it, sorry. This is done in favor of qemu_io() which
*also* returns the console output and status, but with more robust error
handling on failure.

diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
     return qemu_tool(*qemu_io_wrap_args(args),
                      check=check, combine_stdio=combine_stdio)
 
-def qemu_io_pipe_and_status(*args):
-    return qemu_tool_pipe_and_status('qemu-io', qemu_io_wrap_args(args))
-
 def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
     result = qemu_io(*args, check=False)
     log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

Like qemu-img, qemu-io returning 0 should be the norm and not the
exception. Remove all calls to qemu_io_silent that just assert the
return code is zero (That's every last call, as it turns out), and
replace them with a normal qemu_io() call.

qemu_io_silent_check() appeared to have been unused already.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220418211504.943969-12-jsnow@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
 tests/qemu-iotests/216                        | 12 +++++-----
 tests/qemu-iotests/218                        |  5 ++---
 tests/qemu-iotests/224                        |  4 ++--
 tests/qemu-iotests/258                        | 11 +++++-----
 tests/qemu-iotests/298                        | 17 ++++++--------
 tests/qemu-iotests/310                        | 22 +++++++++----------
 tests/qemu-iotests/iotests.py                 | 16 --------------
 tests/qemu-iotests/tests/image-fleecing       |  4 ++--
 .../tests/mirror-ready-cancel-error           |  2 +-
 .../qemu-iotests/tests/stream-error-on-reset  |  4 ++--
 10 files changed, 38 insertions(+), 59 deletions(-)

diff --git a/tests/qemu-iotests/216 b/tests/qemu-iotests/216
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/216
+++ b/tests/qemu-iotests/216
@@ -XXX,XX +XXX,XX @@
 # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
 
 import iotests
-from iotests import log, qemu_img, qemu_io_silent
+from iotests import log, qemu_img, qemu_io
 
 # Need backing file support
 iotests.script_initialize(supported_fmts=['qcow2', 'qcow', 'qed', 'vmdk'],
@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
     log('')
 
     qemu_img('create', '-f', iotests.imgfmt, base_img_path, '64M')
-    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 0M 1M') == 0
+    qemu_io(base_img_path, '-c', 'write -P 1 0M 1M')
     qemu_img('create', '-f', iotests.imgfmt, '-b', base_img_path,
              '-F', iotests.imgfmt, top_img_path)
-    assert qemu_io_silent(top_img_path,  '-c', 'write -P 2 1M 1M') == 0
+    qemu_io(top_img_path,  '-c', 'write -P 2 1M 1M')
 
     log('Done')
 
@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
     log('--- Checking COR result ---')
     log('')
 
-    assert qemu_io_silent(base_img_path, '-c', 'discard 0 64M') == 0
-    assert qemu_io_silent(top_img_path,  '-c', 'read -P 1 0M 1M') == 0
-    assert qemu_io_silent(top_img_path,  '-c', 'read -P 2 1M 1M') == 0
+    qemu_io(base_img_path, '-c', 'discard 0 64M')
+    qemu_io(top_img_path,  '-c', 'read -P 1 0M 1M')
+    qemu_io(top_img_path,  '-c', 'read -P 2 1M 1M')
 
     log('Done')
diff --git a/tests/qemu-iotests/218 b/tests/qemu-iotests/218
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/218
+++ b/tests/qemu-iotests/218
@@ -XXX,XX +XXX,XX @@
 # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
 
 import iotests
-from iotests import log, qemu_img, qemu_io_silent
+from iotests import log, qemu_img, qemu_io
 
 iotests.script_initialize(supported_fmts=['qcow2', 'raw'])
 
@@ -XXX,XX +XXX,XX @@ with iotests.VM() as vm, \
      iotests.FilePath('src.img') as src_img_path:
 
     qemu_img('create', '-f', iotests.imgfmt, src_img_path, '64M')
-    assert qemu_io_silent('-f', iotests.imgfmt, src_img_path,
-                          '-c', 'write -P 42 0M 64M') == 0
+    qemu_io('-f', iotests.imgfmt, src_img_path, '-c', 'write -P 42 0M 64M')
 
     vm.launch()
 
diff --git a/tests/qemu-iotests/224 b/tests/qemu-iotests/224
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/224
+++ b/tests/qemu-iotests/224
@@ -XXX,XX +XXX,XX @@
 # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
 
 import iotests
-from iotests import log, qemu_img, qemu_io_silent, filter_qmp_testfiles, \
+from iotests import log, qemu_img, qemu_io, filter_qmp_testfiles, \
                     filter_qmp_imgfmt
 import json
 
@@ -XXX,XX +XXX,XX @@ for filter_node_name in False, True:
                  '-F', iotests.imgfmt, top_img_path)
 
         # Something to commit
-        assert qemu_io_silent(mid_img_path, '-c', 'write -P 1 0 1M') == 0
+        qemu_io(mid_img_path, '-c', 'write -P 1 0 1M')
 
         vm.launch()
 
diff --git a/tests/qemu-iotests/258 b/tests/qemu-iotests/258
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/258
+++ b/tests/qemu-iotests/258
@@ -XXX,XX +XXX,XX @@
 # Creator/Owner: Hanna Reitz <hreitz@redhat.com>
 
 import iotests
-from iotests import log, qemu_img, qemu_io_silent, \
+from iotests import log, qemu_img, qemu_io, \
         filter_qmp_testfiles, filter_qmp_imgfmt
 
 # Returns a node for blockdev-add
@@ -XXX,XX +XXX,XX @@ def test_concurrent_finish(write_to_stream_node):
         if write_to_stream_node:
             # This is what (most of the time) makes commit finish
             # earlier and then pull in stream
-            assert qemu_io_silent(node2_path,
-                                  '-c', 'write %iK 64K' % (65536 - 192),
-                                  '-c', 'write %iK 64K' % (65536 -  64)) == 0
+            qemu_io(node2_path,
+                    '-c', 'write %iK 64K' % (65536 - 192),
+                    '-c', 'write %iK 64K' % (65536 -  64))
 
             stream_throttle='tg'
         else:
             # And this makes stream finish earlier
-            assert qemu_io_silent(node1_path,
-                                  '-c', 'write %iK 64K' % (65536 - 64)) == 0
+            qemu_io(node1_path, '-c', 'write %iK 64K' % (65536 - 64))
 
             commit_throttle='tg'
 
diff --git a/tests/qemu-iotests/298 b/tests/qemu-iotests/298
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/298
+++ b/tests/qemu-iotests/298
@@ -XXX,XX +XXX,XX @@ class TestTruncate(iotests.QMPTestCase):
         os.remove(refdisk)
 
     def do_test(self, prealloc_mode, new_size):
-        ret = iotests.qemu_io_silent('--image-opts', '-c', 'write 0 10M', '-c',
-                                     f'truncate -m {prealloc_mode} {new_size}',
-                                     drive_opts)
-        self.assertEqual(ret, 0)
-
-        ret = iotests.qemu_io_silent('-f', iotests.imgfmt, '-c', 'write 0 10M',
-                                     '-c',
-                                     f'truncate -m {prealloc_mode} {new_size}',
-                                     refdisk)
-        self.assertEqual(ret, 0)
+        iotests.qemu_io('--image-opts', '-c', 'write 0 10M', '-c',
+                        f'truncate -m {prealloc_mode} {new_size}',
+                        drive_opts)
+
+        iotests.qemu_io('-f', iotests.imgfmt, '-c', 'write 0 10M',
+                        '-c', f'truncate -m {prealloc_mode} {new_size}',
+                        refdisk)
 
         stat = os.stat(disk)
         refstat = os.stat(refdisk)
diff --git a/tests/qemu-iotests/310 b/tests/qemu-iotests/310
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/310
+++ b/tests/qemu-iotests/310
@@ -XXX,XX +XXX,XX @@
 #
 
 import iotests
-from iotests import log, qemu_img, qemu_io_silent
+from iotests import log, qemu_img, qemu_io
 
 # Need backing file support
 iotests.script_initialize(supported_fmts=['qcow2'],
@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
     log('')
 
     qemu_img('create', '-f', iotests.imgfmt, base_img_path, '64M')
-    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 0M 1M') == 0
-    assert qemu_io_silent(base_img_path, '-c', 'write -P 1 3M 1M') == 0
+    qemu_io(base_img_path, '-c', 'write -P 1 0M 1M')
+    qemu_io(base_img_path, '-c', 'write -P 1 3M 1M')
     qemu_img('create', '-f', iotests.imgfmt, '-b', base_img_path,
              '-F', iotests.imgfmt, mid_img_path)
-    assert qemu_io_silent(mid_img_path, '-c', 'write -P 3 2M 1M') == 0
-    assert qemu_io_silent(mid_img_path, '-c', 'write -P 3 4M 1M') == 0
+    qemu_io(mid_img_path, '-c', 'write -P 3 2M 1M')
+    qemu_io(mid_img_path, '-c', 'write -P 3 4M 1M')
     qemu_img('create', '-f', iotests.imgfmt, '-b', mid_img_path,
              '-F', iotests.imgfmt, top_img_path)
-    assert qemu_io_silent(top_img_path, '-c', 'write -P 2 1M 1M') == 0
+    qemu_io(top_img_path, '-c', 'write -P 2 1M 1M')
 
 #      0 1 2 3 4
 # top    2
@@ -XXX,XX +XXX,XX @@ with iotests.FilePath('base.img') as base_img_path, \
     # Detach backing to check that we can read the data from the top level now
     qemu_img('rebase', '-u', '-b', '', '-f', iotests.imgfmt, top_img_path)
 
-    assert qemu_io_silent(top_img_path, '-c', 'read -P 0 0 1M') == 0
-    assert qemu_io_silent(top_img_path, '-c', 'read -P 2 1M 1M') == 0
-    assert qemu_io_silent(top_img_path, '-c', 'read -P 3 2M 1M') == 0
-    assert qemu_io_silent(top_img_path, '-c', 'read -P 0 3M 1M') == 0
-    assert qemu_io_silent(top_img_path, '-c', 'read -P 3 4M 1M') == 0
+    qemu_io(top_img_path, '-c', 'read -P 0 0 1M')
+    qemu_io(top_img_path, '-c', 'read -P 2 1M 1M')
+    qemu_io(top_img_path, '-c', 'read -P 3 2M 1M')
+    qemu_io(top_img_path, '-c', 'read -P 0 3M 1M')
+    qemu_io(top_img_path, '-c', 'read -P 3 4M 1M')
 
     log('Done')
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
     log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
     return result
 
-def qemu_io_silent(*args):
-    '''Run qemu-io and return the exit code, suppressing stdout'''
-    args = qemu_io_wrap_args(args)
-    result = subprocess.run(args, stdout=subprocess.DEVNULL, check=False)
-    if result.returncode < 0:
-        sys.stderr.write('qemu-io received signal %i: %s\n' %
-                         (-result.returncode, ' '.join(args)))
-    return result.returncode
-
-def qemu_io_silent_check(*args):
-    '''Run qemu-io and return the true if subprocess returned 0'''
-    args = qemu_io_wrap_args(args)
-    result = subprocess.run(args, stdout=subprocess.DEVNULL,
-                            stderr=subprocess.STDOUT, check=False)
-    return result.returncode == 0
-
 class QemuIoInteractive:
     def __init__(self, *args):
         self.args = qemu_io_wrap_args(args)
diff --git a/tests/qemu-iotests/tests/image-fleecing b/tests/qemu-iotests/tests/image-fleecing
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/image-fleecing
+++ b/tests/qemu-iotests/tests/image-fleecing
@@ -XXX,XX +XXX,XX @@
 from subprocess import CalledProcessError
 
 import iotests
-from iotests import log, qemu_img, qemu_io, qemu_io_silent
+from iotests import log, qemu_img, qemu_io
 
 iotests.script_initialize(
     supported_fmts=['qcow2'],
@@ -XXX,XX +XXX,XX @@ def do_test(vm, use_cbw, use_snapshot_access_filter, base_img_path,
     for p in overwrite + remainder:
         cmd = 'read -P%s %s %s' % p
         log(cmd)
-        assert qemu_io_silent(base_img_path, '-c', cmd) == 0
+        qemu_io(base_img_path, '-c', cmd)
 
     log('')
     log('Done')
diff --git a/tests/qemu-iotests/tests/mirror-ready-cancel-error b/tests/qemu-iotests/tests/mirror-ready-cancel-error
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/mirror-ready-cancel-error
+++ b/tests/qemu-iotests/tests/mirror-ready-cancel-error
@@ -XXX,XX +XXX,XX @@ class TestMirrorReadyCancelError(iotests.QMPTestCase):
         # Ensure that mirror will copy something before READY so the
         # target format layer will forward the pre-READY flush to its
         # file child
-        assert iotests.qemu_io_silent('-c', 'write -P 1 0 64k', source) == 0
+        iotests.qemu_io('-c', 'write -P 1 0 64k', source)
 
         self.vm = iotests.VM()
         self.vm.launch()
diff --git a/tests/qemu-iotests/tests/stream-error-on-reset b/tests/qemu-iotests/tests/stream-error-on-reset
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/stream-error-on-reset
+++ b/tests/qemu-iotests/tests/stream-error-on-reset
@@ -XXX,XX +XXX,XX @@
 
 import os
 import iotests
-from iotests import imgfmt, qemu_img_create, qemu_io_silent, QMPTestCase
+from iotests import imgfmt, qemu_img_create, qemu_io, QMPTestCase
 
 
 image_size = 1 * 1024 * 1024
@@ -XXX,XX +XXX,XX @@ class TestStreamErrorOnReset(QMPTestCase):
         - top image is attached to a virtio-scsi device
         """
         qemu_img_create('-f', imgfmt, base, str(image_size))
-        assert qemu_io_silent('-c', f'write 0 {data_size}', base) == 0
+        qemu_io('-c', f'write 0 {data_size}', base)
         qemu_img_create('-f', imgfmt, top, str(image_size))
 
         self.vm = iotests.VM()
-- 
2.35.1

From: John Snow <jsnow@redhat.com>

Just like qemu_img_log(), upgrade qemu_io_log() to enforce a return code
of zero by default.

Tests that use qemu_io_log(): 242 245 255 274 303 307 nbd-reconnect-on-open

diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -XXX,XX +XXX,XX @@ def qemu_io(*args: str, check: bool = True, combine_stdio: bool = True
     return qemu_tool(*qemu_io_wrap_args(args),
                      check=check, combine_stdio=combine_stdio)
 
-def qemu_io_log(*args: str) -> 'subprocess.CompletedProcess[str]':
-    result = qemu_io(*args, check=False)
+def qemu_io_log(*args: str, check: bool = True
+                ) -> 'subprocess.CompletedProcess[str]':
+    result = qemu_io(*args, check=check)
     log(result.stdout, filters=[filter_testfiles, filter_qemu_io])
     return result
 
diff --git a/tests/qemu-iotests/tests/nbd-reconnect-on-open b/tests/qemu-iotests/tests/nbd-reconnect-on-open
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/tests/nbd-reconnect-on-open
+++ b/tests/qemu-iotests/tests/nbd-reconnect-on-open
@@ -XXX,XX +XXX,XX @@ def check_fail_to_connect(open_timeout):
     log(f'Check fail to connect with {open_timeout} seconds of timeout')
 
     start_t = time.time()
-    qemu_io_log(*create_args(open_timeout))
+    qemu_io_log(*create_args(open_timeout), check=False)
     delta_t = time.time() - start_t
 
     max_delta = open_timeout + 0.2
-- 
2.35.1

FUSE exports' allow-other option defaults to "auto", which means that it
will try passing allow_other as a mount option, and fall back to not
using it when an error occurs.  We make no effort to hide fusermount's
error message (because it would be difficult, and because users might
want to know about the fallback occurring), and so when allow_other does
not work (primarily when /etc/fuse.conf does not contain
user_allow_other), this error message will appear and break the
reference output.

We do not need allow_other here, though, so we can just pass
allow-other=off to fix that.

Reported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220421142435.569600-1-hreitz@redhat.com>
Tested-by: Markus Armbruster <armbru@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
---
 tests/qemu-iotests/108 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/qemu-iotests/108 b/tests/qemu-iotests/108
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/108
+++ b/tests/qemu-iotests/108
@@ -XXX,XX +XXX,XX @@ else
 
     $QSD \
         --blockdev file,node-name=export-node,filename="$TEST_IMG" \
-        --export fuse,id=fuse-export,node-name=export-node,mountpoint="$export_mp",writable=on,growable=off \
+        --export fuse,id=fuse-export,node-name=export-node,mountpoint="$export_mp",writable=on,growable=off,allow-other=off \
         --pidfile "$TEST_DIR/qsd.pid" \
         --daemonize
 fi
-- 
2.35.1

The following changes since commit 56f9e46b841c7be478ca038d8d4085d776ab4b0d:

Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2017-02-20' into staging (2017-02-20 17:42:47 +0000)

are available in the git repository at:

git://github.com/stefanha/qemu.git tags/block-pull-request

for you to fetch changes up to a7b91d35bab97a2d3e779d0c64c9b837b52a6cf7:

coroutine-lock: make CoRwlock thread-safe and fair (2017-02-21 11:39:40 +0000)

----------------------------------------------------------------
Pull request

v2:
 * Rebased to resolve scsi conflicts

----------------------------------------------------------------

Paolo Bonzini (24):
  block: move AioContext, QEMUTimer, main-loop to libqemuutil
  aio: introduce aio_co_schedule and aio_co_wake
  block-backend: allow blk_prw from coroutine context
  test-thread-pool: use generic AioContext infrastructure
  io: add methods to set I/O handlers on AioContext
  io: make qio_channel_yield aware of AioContexts
  nbd: convert to use qio_channel_yield
  coroutine-lock: reschedule coroutine on the AioContext it was running
    on
  blkdebug: reschedule coroutine on the AioContext it is running on
  qed: introduce qed_aio_start_io and qed_aio_next_io_cb
  aio: push aio_context_acquire/release down to dispatching
  block: explicitly acquire aiocontext in timers that need it
  block: explicitly acquire aiocontext in callbacks that need it
  block: explicitly acquire aiocontext in bottom halves that need it
  block: explicitly acquire aiocontext in aio callbacks that need it
  aio-posix: partially inline aio_dispatch into aio_poll
  async: remove unnecessary inc/dec pairs
  block: document fields protected by AioContext lock
  coroutine-lock: make CoMutex thread-safe
  coroutine-lock: add limited spinning to CoMutex
  test-aio-multithread: add performance comparison with thread-based
    mutexes
  coroutine-lock: place CoMutex before CoQueue in header
  coroutine-lock: add mutex argument to CoQueue APIs
  coroutine-lock: make CoRwlock thread-safe and fair

-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

AioContext is fairly self contained, the only dependency is QEMUTimer but
that in turn doesn't need anything else.  So move them out of block-obj-y
to avoid introducing a dependency from io/ to block-obj-y.

main-loop and its dependency iohandler also need to be moved, because
later in this series io/ will call iohandler_get_aio_context.

[Changed copyright "the QEMU team" to "other QEMU contributors" as
suggested by Daniel Berrange and agreed by Paolo.
--Stefan]

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-2-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 Makefile.objs                       |  4 ---
 stubs/Makefile.objs                 |  1 +
 tests/Makefile.include              | 11 ++++----
 util/Makefile.objs                  |  6 +++-
 block/io.c                          | 29 -------------------
 stubs/linux-aio.c                   | 32 +++++++++++++++++++++
 stubs/set-fd-handler.c              | 11 --------
 aio-posix.c => util/aio-posix.c     |  2 +-
 aio-win32.c => util/aio-win32.c     |  0
 util/aiocb.c                        | 55 +++++++++++++++++++++++++++++++++++++
 async.c => util/async.c             |  3 +-
 iohandler.c => util/iohandler.c     |  0
 main-loop.c => util/main-loop.c     |  0
 qemu-timer.c => util/qemu-timer.c   |  0
 thread-pool.c => util/thread-pool.c |  2 +-
 trace-events                        | 11 --------
 util/trace-events                   | 11 ++++++++
 17 files changed, 114 insertions(+), 64 deletions(-)
 create mode 100644 stubs/linux-aio.c
 rename aio-posix.c => util/aio-posix.c (99%)
 rename aio-win32.c => util/aio-win32.c (100%)
 create mode 100644 util/aiocb.c
 rename async.c => util/async.c (99%)
 rename iohandler.c => util/iohandler.c (100%)
 rename main-loop.c => util/main-loop.c (100%)
 rename qemu-timer.c => util/qemu-timer.c (100%)
 rename thread-pool.c => util/thread-pool.c (99%)

diff --git a/Makefile.objs b/Makefile.objs
index XXXXXXX..XXXXXXX 100644
--- a/Makefile.objs
+++ b/Makefile.objs
@@ -XXX,XX +XXX,XX @@ chardev-obj-y = chardev/
 #######################################################################
 # block-obj-y is code used by both qemu system emulation and qemu-img
 
-block-obj-y = async.o thread-pool.o
 block-obj-y += nbd/
 block-obj-y += block.o blockjob.o
-block-obj-y += main-loop.o iohandler.o qemu-timer.o
-block-obj-$(CONFIG_POSIX) += aio-posix.o
-block-obj-$(CONFIG_WIN32) += aio-win32.o
 block-obj-y += block/
 block-obj-y += qemu-io-cmds.o
 block-obj-$(CONFIG_REPLICATION) += replication.o
diff --git a/stubs/Makefile.objs b/stubs/Makefile.objs
index XXXXXXX..XXXXXXX 100644
--- a/stubs/Makefile.objs
+++ b/stubs/Makefile.objs
@@ -XXX,XX +XXX,XX @@ stub-obj-y += get-vm-name.o
 stub-obj-y += iothread.o
 stub-obj-y += iothread-lock.o
 stub-obj-y += is-daemonized.o
+stub-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
 stub-obj-y += machine-init-done.o
 stub-obj-y += migr-blocker.o
 stub-obj-y += monitor.o
diff --git a/tests/Makefile.include b/tests/Makefile.include
index XXXXXXX..XXXXXXX 100644
--- a/tests/Makefile.include
+++ b/tests/Makefile.include
@@ -XXX,XX +XXX,XX @@ check-unit-y += tests/test-visitor-serialization$(EXESUF)
 check-unit-y += tests/test-iov$(EXESUF)
 gcov-files-test-iov-y = util/iov.c
 check-unit-y += tests/test-aio$(EXESUF)
+gcov-files-test-aio-y = util/async.c util/qemu-timer.o
+gcov-files-test-aio-$(CONFIG_WIN32) += util/aio-win32.c
+gcov-files-test-aio-$(CONFIG_POSIX) += util/aio-posix.c
 check-unit-y += tests/test-throttle$(EXESUF)
 gcov-files-test-aio-$(CONFIG_WIN32) = aio-win32.c
 gcov-files-test-aio-$(CONFIG_POSIX) = aio-posix.c
@@ -XXX,XX +XXX,XX @@ tests/check-qjson$(EXESUF): tests/check-qjson.o $(test-util-obj-y)
 tests/check-qom-interface$(EXESUF): tests/check-qom-interface.o $(test-qom-obj-y)
 tests/check-qom-proplist$(EXESUF): tests/check-qom-proplist.o $(test-qom-obj-y)
 
-tests/test-char$(EXESUF): tests/test-char.o qemu-timer.o \
-	$(test-util-obj-y) $(qtest-obj-y) $(test-block-obj-y) $(chardev-obj-y)
+tests/test-char$(EXESUF): tests/test-char.o $(test-util-obj-y) $(qtest-obj-y) $(test-io-obj-y) $(chardev-obj-y)
 tests/test-coroutine$(EXESUF): tests/test-coroutine.o $(test-block-obj-y)
 tests/test-aio$(EXESUF): tests/test-aio.o $(test-block-obj-y)
 tests/test-throttle$(EXESUF): tests/test-throttle.o $(test-block-obj-y)
@@ -XXX,XX +XXX,XX @@ tests/test-vmstate$(EXESUF): tests/test-vmstate.o \
 	migration/vmstate.o migration/qemu-file.o \
         migration/qemu-file-channel.o migration/qjson.o \
 	$(test-io-obj-y)
-tests/test-timed-average$(EXESUF): tests/test-timed-average.o qemu-timer.o \
-	$(test-util-obj-y)
+tests/test-timed-average$(EXESUF): tests/test-timed-average.o $(test-util-obj-y)
 tests/test-base64$(EXESUF): tests/test-base64.o \
 	libqemuutil.a libqemustub.a
 tests/ptimer-test$(EXESUF): tests/ptimer-test.o tests/ptimer-test-stubs.o hw/core/ptimer.o libqemustub.a
@@ -XXX,XX +XXX,XX @@ tests/usb-hcd-ehci-test$(EXESUF): tests/usb-hcd-ehci-test.o $(libqos-usb-obj-y)
 tests/usb-hcd-xhci-test$(EXESUF): tests/usb-hcd-xhci-test.o $(libqos-usb-obj-y)
 tests/pc-cpu-test$(EXESUF): tests/pc-cpu-test.o
 tests/postcopy-test$(EXESUF): tests/postcopy-test.o
-tests/vhost-user-test$(EXESUF): tests/vhost-user-test.o qemu-timer.o \
+tests/vhost-user-test$(EXESUF): tests/vhost-user-test.o $(test-util-obj-y) \
 	$(qtest-obj-y) $(test-io-obj-y) $(libqos-virtio-obj-y) $(libqos-pc-obj-y) \
 	$(chardev-obj-y)
 tests/qemu-iotests/socket_scm_helper$(EXESUF): tests/qemu-iotests/socket_scm_helper.o
diff --git a/util/Makefile.objs b/util/Makefile.objs
index XXXXXXX..XXXXXXX 100644
--- a/util/Makefile.objs
+++ b/util/Makefile.objs
@@ -XXX,XX +XXX,XX @@
 util-obj-y = osdep.o cutils.o unicode.o qemu-timer-common.o
 util-obj-y += bufferiszero.o
 util-obj-y += lockcnt.o
+util-obj-y += aiocb.o async.o thread-pool.o qemu-timer.o
+util-obj-y += main-loop.o iohandler.o
+util-obj-$(CONFIG_POSIX) += aio-posix.o
 util-obj-$(CONFIG_POSIX) += compatfd.o
 util-obj-$(CONFIG_POSIX) += event_notifier-posix.o
 util-obj-$(CONFIG_POSIX) += mmap-alloc.o
 util-obj-$(CONFIG_POSIX) += oslib-posix.o
 util-obj-$(CONFIG_POSIX) += qemu-openpty.o
 util-obj-$(CONFIG_POSIX) += qemu-thread-posix.o
-util-obj-$(CONFIG_WIN32) += event_notifier-win32.o
 util-obj-$(CONFIG_POSIX) += memfd.o
+util-obj-$(CONFIG_WIN32) += aio-win32.o
+util-obj-$(CONFIG_WIN32) += event_notifier-win32.o
 util-obj-$(CONFIG_WIN32) += oslib-win32.o
 util-obj-$(CONFIG_WIN32) += qemu-thread-win32.o
 util-obj-y += envlist.o path.o module.o
diff --git a/block/io.c b/block/io.c
index XXXXXXX..XXXXXXX 100644
--- a/block/io.c
+++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ BlockAIOCB *bdrv_aio_flush(BlockDriverState *bs,
     return &acb->common;
 }
 
-void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
-                   BlockCompletionFunc *cb, void *opaque)
-{
-    BlockAIOCB *acb;
-
-    acb = g_malloc(aiocb_info->aiocb_size);
-    acb->aiocb_info = aiocb_info;
-    acb->bs = bs;
-    acb->cb = cb;
-    acb->opaque = opaque;
-    acb->refcnt = 1;
-    return acb;
-}
-
-void qemu_aio_ref(void *p)
-{
-    BlockAIOCB *acb = p;
-    acb->refcnt++;
-}
-
-void qemu_aio_unref(void *p)
-{
-    BlockAIOCB *acb = p;
-    assert(acb->refcnt > 0);
-    if (--acb->refcnt == 0) {
-        g_free(acb);
-    }
-}
-
 /**************************************************************/
 /* Coroutine block device emulation */
 
diff --git a/stubs/linux-aio.c b/stubs/linux-aio.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/stubs/linux-aio.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Linux native AIO support.
+ *
+ * Copyright (C) 2009 IBM, Corp.
+ * Copyright (C) 2009 Red Hat, Inc.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+#include "qemu/osdep.h"
+#include "block/aio.h"
+#include "block/raw-aio.h"
+
+void laio_detach_aio_context(LinuxAioState *s, AioContext *old_context)
+{
+    abort();
+}
+
+void laio_attach_aio_context(LinuxAioState *s, AioContext *new_context)
+{
+    abort();
+}
+
+LinuxAioState *laio_init(void)
+{
+    abort();
+}
+
+void laio_cleanup(LinuxAioState *s)
+{
+    abort();
+}
diff --git a/stubs/set-fd-handler.c b/stubs/set-fd-handler.c
index XXXXXXX..XXXXXXX 100644
--- a/stubs/set-fd-handler.c
+++ b/stubs/set-fd-handler.c
@@ -XXX,XX +XXX,XX @@ void qemu_set_fd_handler(int fd,
 {
     abort();
 }
-
-void aio_set_fd_handler(AioContext *ctx,
-                        int fd,
-                        bool is_external,
-                        IOHandler *io_read,
-                        IOHandler *io_write,
-                        AioPollFn *io_poll,
-                        void *opaque)
-{
-    abort();
-}
diff --git a/aio-posix.c b/util/aio-posix.c
similarity index 99%
rename from aio-posix.c
rename to util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/rcu_queue.h"
 #include "qemu/sockets.h"
 #include "qemu/cutils.h"
-#include "trace-root.h"
+#include "trace.h"
 #ifdef CONFIG_EPOLL_CREATE1
 #include <sys/epoll.h>
 #endif
diff --git a/aio-win32.c b/util/aio-win32.c
similarity index 100%
rename from aio-win32.c
rename to util/aio-win32.c
diff --git a/util/aiocb.c b/util/aiocb.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/util/aiocb.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * BlockAIOCB allocation
+ *
+ * Copyright (c) 2003-2017 Fabrice Bellard and other QEMU contributors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu/osdep.h"
+#include "block/aio.h"
+
+void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
+                   BlockCompletionFunc *cb, void *opaque)
+{
+    BlockAIOCB *acb;
+
+    acb = g_malloc(aiocb_info->aiocb_size);
+    acb->aiocb_info = aiocb_info;
+    acb->bs = bs;
+    acb->cb = cb;
+    acb->opaque = opaque;
+    acb->refcnt = 1;
+    return acb;
+}
+
+void qemu_aio_ref(void *p)
+{
+    BlockAIOCB *acb = p;
+    acb->refcnt++;
+}
+
+void qemu_aio_unref(void *p)
+{
+    BlockAIOCB *acb = p;
+    assert(acb->refcnt > 0);
+    if (--acb->refcnt == 0) {
+        g_free(acb);
+    }
+}
diff --git a/async.c b/util/async.c
similarity index 99%
rename from async.c
rename to util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@
 /*
- * QEMU System Emulator
+ * Data plane event loop
  *
  * Copyright (c) 2003-2008 Fabrice Bellard
+ * Copyright (c) 2009-2017 QEMU contributors
  *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
diff --git a/iohandler.c b/util/iohandler.c
similarity index 100%
rename from iohandler.c
rename to util/iohandler.c
diff --git a/main-loop.c b/util/main-loop.c
similarity index 100%
rename from main-loop.c
rename to util/main-loop.c
diff --git a/qemu-timer.c b/util/qemu-timer.c
similarity index 100%
rename from qemu-timer.c
rename to util/qemu-timer.c
diff --git a/thread-pool.c b/util/thread-pool.c
similarity index 99%
rename from thread-pool.c
rename to util/thread-pool.c
index XXXXXXX..XXXXXXX 100644
--- a/thread-pool.c
+++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/queue.h"
 #include "qemu/thread.h"
 #include "qemu/coroutine.h"
-#include "trace-root.h"
+#include "trace.h"
 #include "block/thread-pool.h"
 #include "qemu/main-loop.h"
 
diff --git a/trace-events b/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/trace-events
+++ b/trace-events
@@ -XXX,XX +XXX,XX @@
 #
 # The <format-string> should be a sprintf()-compatible format string.
 
-# aio-posix.c
-run_poll_handlers_begin(void *ctx, int64_t max_ns) "ctx %p max_ns %"PRId64
-run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
-poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
-poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
-
-# thread-pool.c
-thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
-thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
-thread_pool_cancel(void *req, void *opaque) "req %p opaque %p"
-
 # ioport.c
 cpu_in(unsigned int addr, char size, unsigned int val) "addr %#x(%c) value %u"
 cpu_out(unsigned int addr, char size, unsigned int val) "addr %#x(%c) value %u"
diff --git a/util/trace-events b/util/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/util/trace-events
+++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@
 # See docs/tracing.txt for syntax documentation.
 
+# util/aio-posix.c
+run_poll_handlers_begin(void *ctx, int64_t max_ns) "ctx %p max_ns %"PRId64
+run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
+poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
+poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
+
+# util/thread-pool.c
+thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
+thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
+thread_pool_cancel(void *req, void *opaque) "req %p opaque %p"
+
 # util/buffer.c
 buffer_resize(const char *buf, size_t olen, size_t len) "%s: old %zd, new %zd"
 buffer_move_empty(const char *buf, size_t len, const char *from) "%s: %zd bytes from %s"
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

aio_co_wake provides the infrastructure to start a coroutine on a "home"
AioContext.  It will be used by CoMutex and CoQueue, so that coroutines
don't jump from one context to another when they go to sleep on a
mutex or waitqueue.  However, it can also be used as a more efficient
alternative to one-shot bottom halves, and saves the effort of tracking
which AioContext a coroutine is running on.

aio_co_schedule is the part of aio_co_wake that starts a coroutine
on a remove AioContext, but it is also useful to implement e.g.
bdrv_set_aio_context callbacks.

The implementation of aio_co_schedule is based on a lock-free
multiple-producer, single-consumer queue.  The multiple producers use
cmpxchg to add to a LIFO stack.  The consumer (a per-AioContext bottom
half) grabs all items added so far, inverts the list to make it FIFO,
and goes through it one item at a time until it's empty.  The data
structure was inspired by OSv, which uses it in the very code we'll
"port" to QEMU for the thread-safe CoMutex.

Most of the new code is really tests.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-3-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/Makefile.include       |   8 +-
 include/block/aio.h          |  32 +++++++
 include/qemu/coroutine_int.h |  11 ++-
 tests/iothread.h             |  25 +++++
 tests/iothread.c             |  91 ++++++++++++++++++
 tests/test-aio-multithread.c | 213 +++++++++++++++++++++++++++++++++++++++++++
 util/async.c                 |  65 +++++++++++++
 util/qemu-coroutine.c        |   8 ++
 util/trace-events            |   4 +
 9 files changed, 453 insertions(+), 4 deletions(-)
 create mode 100644 tests/iothread.h
 create mode 100644 tests/iothread.c
 create mode 100644 tests/test-aio-multithread.c

diff --git a/tests/Makefile.include b/tests/Makefile.include
index XXXXXXX..XXXXXXX 100644
--- a/tests/Makefile.include
+++ b/tests/Makefile.include
@@ -XXX,XX +XXX,XX @@ check-unit-y += tests/test-aio$(EXESUF)
 gcov-files-test-aio-y = util/async.c util/qemu-timer.o
 gcov-files-test-aio-$(CONFIG_WIN32) += util/aio-win32.c
 gcov-files-test-aio-$(CONFIG_POSIX) += util/aio-posix.c
+check-unit-y += tests/test-aio-multithread$(EXESUF)
+gcov-files-test-aio-multithread-y = $(gcov-files-test-aio-y)
+gcov-files-test-aio-multithread-y += util/qemu-coroutine.c tests/iothread.c
 check-unit-y += tests/test-throttle$(EXESUF)
-gcov-files-test-aio-$(CONFIG_WIN32) = aio-win32.c
-gcov-files-test-aio-$(CONFIG_POSIX) = aio-posix.c
 check-unit-y += tests/test-thread-pool$(EXESUF)
 gcov-files-test-thread-pool-y = thread-pool.c
 gcov-files-test-hbitmap-y = util/hbitmap.c
@@ -XXX,XX +XXX,XX @@ test-qapi-obj-y = tests/test-qapi-visit.o tests/test-qapi-types.o \
 	$(test-qom-obj-y)
 test-crypto-obj-y = $(crypto-obj-y) $(test-qom-obj-y)
 test-io-obj-y = $(io-obj-y) $(test-crypto-obj-y)
-test-block-obj-y = $(block-obj-y) $(test-io-obj-y)
+test-block-obj-y = $(block-obj-y) $(test-io-obj-y) tests/iothread.o
 
 tests/check-qint$(EXESUF): tests/check-qint.o $(test-util-obj-y)
 tests/check-qstring$(EXESUF): tests/check-qstring.o $(test-util-obj-y)
@@ -XXX,XX +XXX,XX @@ tests/check-qom-proplist$(EXESUF): tests/check-qom-proplist.o $(test-qom-obj-y)
 tests/test-char$(EXESUF): tests/test-char.o $(test-util-obj-y) $(qtest-obj-y) $(test-io-obj-y) $(chardev-obj-y)
 tests/test-coroutine$(EXESUF): tests/test-coroutine.o $(test-block-obj-y)
 tests/test-aio$(EXESUF): tests/test-aio.o $(test-block-obj-y)
+tests/test-aio-multithread$(EXESUF): tests/test-aio-multithread.o $(test-block-obj-y)
 tests/test-throttle$(EXESUF): tests/test-throttle.o $(test-block-obj-y)
 tests/test-blockjob$(EXESUF): tests/test-blockjob.o $(test-block-obj-y) $(test-util-obj-y)
 tests/test-blockjob-txn$(EXESUF): tests/test-blockjob-txn.o $(test-block-obj-y) $(test-util-obj-y)
diff --git a/include/block/aio.h b/include/block/aio.h
index XXXXXXX..XXXXXXX 100644
--- a/include/block/aio.h
+++ b/include/block/aio.h
@@ -XXX,XX +XXX,XX @@ typedef void QEMUBHFunc(void *opaque);
 typedef bool AioPollFn(void *opaque);
 typedef void IOHandler(void *opaque);
 
+struct Coroutine;
 struct ThreadPool;
 struct LinuxAioState;
 
@@ -XXX,XX +XXX,XX @@ struct AioContext {
     bool notified;
     EventNotifier notifier;
 
+    QSLIST_HEAD(, Coroutine) scheduled_coroutines;
+    QEMUBH *co_schedule_bh;
+
     /* Thread pool for performing work and receiving completion callbacks.
      * Has its own locking.
      */
@@ -XXX,XX +XXX,XX @@ static inline bool aio_node_check(AioContext *ctx, bool is_external)
 }
 
 /**
+ * aio_co_schedule:
+ * @ctx: the aio context
+ * @co: the coroutine
+ *
+ * Start a coroutine on a remote AioContext.
+ *
+ * The coroutine must not be entered by anyone else while aio_co_schedule()
+ * is active.  In addition the coroutine must have yielded unless ctx
+ * is the context in which the coroutine is running (i.e. the value of
+ * qemu_get_current_aio_context() from the coroutine itself).
+ */
+void aio_co_schedule(AioContext *ctx, struct Coroutine *co);
+
+/**
+ * aio_co_wake:
+ * @co: the coroutine
+ *
+ * Restart a coroutine on the AioContext where it was running last, thus
+ * preventing coroutines from jumping from one context to another when they
+ * go to sleep.
+ *
+ * aio_co_wake may be executed either in coroutine or non-coroutine
+ * context.  The coroutine must not be entered by anyone else while
+ * aio_co_wake() is active.
+ */
+void aio_co_wake(struct Coroutine *co);
+
+/**
  * Return the AioContext whose event loop runs in the current thread.
  *
  * If called from an IOThread this will be the IOThread's AioContext.  If
diff --git a/include/qemu/coroutine_int.h b/include/qemu/coroutine_int.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine_int.h
+++ b/include/qemu/coroutine_int.h
@@ -XXX,XX +XXX,XX @@ struct Coroutine {
     CoroutineEntry *entry;
     void *entry_arg;
     Coroutine *caller;
+
+    /* Only used when the coroutine has terminated.  */
     QSLIST_ENTRY(Coroutine) pool_next;
+
     size_t locks_held;
 
-    /* Coroutines that should be woken up when we yield or terminate */
+    /* Coroutines that should be woken up when we yield or terminate.
+     * Only used when the coroutine is running.
+     */
     QSIMPLEQ_HEAD(, Coroutine) co_queue_wakeup;
+
+    /* Only used when the coroutine has yielded.  */
+    AioContext *ctx;
     QSIMPLEQ_ENTRY(Coroutine) co_queue_next;
+    QSLIST_ENTRY(Coroutine) co_scheduled_next;
 };
 
 Coroutine *qemu_coroutine_new(void);
diff --git a/tests/iothread.h b/tests/iothread.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/tests/iothread.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Event loop thread implementation for unit tests
+ *
+ * Copyright Red Hat Inc., 2013, 2016
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@redhat.com>
+ *  Paolo Bonzini     <pbonzini@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+#ifndef TEST_IOTHREAD_H
+#define TEST_IOTHREAD_H
+
+#include "block/aio.h"
+#include "qemu/thread.h"
+
+typedef struct IOThread IOThread;
+
+IOThread *iothread_new(void);
+void iothread_join(IOThread *iothread);
+AioContext *iothread_get_aio_context(IOThread *iothread);
+
+#endif
diff --git a/tests/iothread.c b/tests/iothread.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/tests/iothread.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Event loop thread implementation for unit tests
+ *
+ * Copyright Red Hat Inc., 2013, 2016
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@redhat.com>
+ *  Paolo Bonzini     <pbonzini@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "block/aio.h"
+#include "qemu/main-loop.h"
+#include "qemu/rcu.h"
+#include "iothread.h"
+
+struct IOThread {
+    AioContext *ctx;
+
+    QemuThread thread;
+    QemuMutex init_done_lock;
+    QemuCond init_done_cond;    /* is thread initialization done? */
+    bool stopping;
+};
+
+static __thread IOThread *my_iothread;
+
+AioContext *qemu_get_current_aio_context(void)
+{
+    return my_iothread ? my_iothread->ctx : qemu_get_aio_context();
+}
+
+static void *iothread_run(void *opaque)
+{
+    IOThread *iothread = opaque;
+
+    rcu_register_thread();
+
+    my_iothread = iothread;
+    qemu_mutex_lock(&iothread->init_done_lock);
+    iothread->ctx = aio_context_new(&error_abort);
+    qemu_cond_signal(&iothread->init_done_cond);
+    qemu_mutex_unlock(&iothread->init_done_lock);
+
+    while (!atomic_read(&iothread->stopping)) {
+        aio_poll(iothread->ctx, true);
+    }
+
+    rcu_unregister_thread();
+    return NULL;
+}
+
+void iothread_join(IOThread *iothread)
+{
+    iothread->stopping = true;
+    aio_notify(iothread->ctx);
+    qemu_thread_join(&iothread->thread);
+    qemu_cond_destroy(&iothread->init_done_cond);
+    qemu_mutex_destroy(&iothread->init_done_lock);
+    aio_context_unref(iothread->ctx);
+    g_free(iothread);
+}
+
+IOThread *iothread_new(void)
+{
+    IOThread *iothread = g_new0(IOThread, 1);
+
+    qemu_mutex_init(&iothread->init_done_lock);
+    qemu_cond_init(&iothread->init_done_cond);
+    qemu_thread_create(&iothread->thread, NULL, iothread_run,
+                       iothread, QEMU_THREAD_JOINABLE);
+
+    /* Wait for initialization to complete */
+    qemu_mutex_lock(&iothread->init_done_lock);
+    while (iothread->ctx == NULL) {
+        qemu_cond_wait(&iothread->init_done_cond,
+                       &iothread->init_done_lock);
+    }
+    qemu_mutex_unlock(&iothread->init_done_lock);
+    return iothread;
+}
+
+AioContext *iothread_get_aio_context(IOThread *iothread)
+{
+    return iothread->ctx;
+}
diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/tests/test-aio-multithread.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * AioContext multithreading tests
+ *
+ * Copyright Red Hat, Inc. 2016
+ *
+ * Authors:
+ *  Paolo Bonzini    <pbonzini@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include <glib.h>
+#include "block/aio.h"
+#include "qapi/error.h"
+#include "qemu/coroutine.h"
+#include "qemu/thread.h"
+#include "qemu/error-report.h"
+#include "iothread.h"
+
+/* AioContext management */
+
+#define NUM_CONTEXTS 5
+
+static IOThread *threads[NUM_CONTEXTS];
+static AioContext *ctx[NUM_CONTEXTS];
+static __thread int id = -1;
+
+static QemuEvent done_event;
+
+/* Run a function synchronously on a remote iothread. */
+
+typedef struct CtxRunData {
+    QEMUBHFunc *cb;
+    void *arg;
+} CtxRunData;
+
+static void ctx_run_bh_cb(void *opaque)
+{
+    CtxRunData *data = opaque;
+
+    data->cb(data->arg);
+    qemu_event_set(&done_event);
+}
+
+static void ctx_run(int i, QEMUBHFunc *cb, void *opaque)
+{
+    CtxRunData data = {
+        .cb = cb,
+        .arg = opaque
+    };
+
+    qemu_event_reset(&done_event);
+    aio_bh_schedule_oneshot(ctx[i], ctx_run_bh_cb, &data);
+    qemu_event_wait(&done_event);
+}
+
+/* Starting the iothreads. */
+
+static void set_id_cb(void *opaque)
+{
+    int *i = opaque;
+
+    id = *i;
+}
+
+static void create_aio_contexts(void)
+{
+    int i;
+
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        threads[i] = iothread_new();
+        ctx[i] = iothread_get_aio_context(threads[i]);
+    }
+
+    qemu_event_init(&done_event, false);
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        ctx_run(i, set_id_cb, &i);
+    }
+}
+
+/* Stopping the iothreads. */
+
+static void join_aio_contexts(void)
+{
+    int i;
+
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        aio_context_ref(ctx[i]);
+    }
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        iothread_join(threads[i]);
+    }
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        aio_context_unref(ctx[i]);
+    }
+    qemu_event_destroy(&done_event);
+}
+
+/* Basic test for the stuff above. */
+
+static void test_lifecycle(void)
+{
+    create_aio_contexts();
+    join_aio_contexts();
+}
+
+/* aio_co_schedule test.  */
+
+static Coroutine *to_schedule[NUM_CONTEXTS];
+
+static bool now_stopping;
+
+static int count_retry;
+static int count_here;
+static int count_other;
+
+static bool schedule_next(int n)
+{
+    Coroutine *co;
+
+    co = atomic_xchg(&to_schedule[n], NULL);
+    if (!co) {
+        atomic_inc(&count_retry);
+        return false;
+    }
+
+    if (n == id) {
+        atomic_inc(&count_here);
+    } else {
+        atomic_inc(&count_other);
+    }
+
+    aio_co_schedule(ctx[n], co);
+    return true;
+}
+
+static void finish_cb(void *opaque)
+{
+    schedule_next(id);
+}
+
+static coroutine_fn void test_multi_co_schedule_entry(void *opaque)
+{
+    g_assert(to_schedule[id] == NULL);
+    atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
+
+    while (!atomic_mb_read(&now_stopping)) {
+        int n;
+
+        n = g_test_rand_int_range(0, NUM_CONTEXTS);
+        schedule_next(n);
+        qemu_coroutine_yield();
+
+        g_assert(to_schedule[id] == NULL);
+        atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
+    }
+}
+
+
+static void test_multi_co_schedule(int seconds)
+{
+    int i;
+
+    count_here = count_other = count_retry = 0;
+    now_stopping = false;
+
+    create_aio_contexts();
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        Coroutine *co1 = qemu_coroutine_create(test_multi_co_schedule_entry, NULL);
+        aio_co_schedule(ctx[i], co1);
+    }
+
+    g_usleep(seconds * 1000000);
+
+    atomic_mb_set(&now_stopping, true);
+    for (i = 0; i < NUM_CONTEXTS; i++) {
+        ctx_run(i, finish_cb, NULL);
+        to_schedule[i] = NULL;
+    }
+
+    join_aio_contexts();
+    g_test_message("scheduled %d, queued %d, retry %d, total %d\n",
+                  count_other, count_here, count_retry,
+                  count_here + count_other + count_retry);
+}
+
+static void test_multi_co_schedule_1(void)
+{
+    test_multi_co_schedule(1);
+}
+
+static void test_multi_co_schedule_10(void)
+{
+    test_multi_co_schedule(10);
+}
+
+/* End of tests.  */
+
+int main(int argc, char **argv)
+{
+    init_clocks();
+
+    g_test_init(&argc, &argv, NULL);
+    g_test_add_func("/aio/multi/lifecycle", test_lifecycle);
+    if (g_test_quick()) {
+        g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
+    } else {
+        g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
+    }
+    return g_test_run();
+}
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/main-loop.h"
 #include "qemu/atomic.h"
 #include "block/raw-aio.h"
+#include "qemu/coroutine_int.h"
+#include "trace.h"
 
 /***********************************************************/
 /* bottom halves (can be seen as timers which expire ASAP) */
@@ -XXX,XX +XXX,XX @@ aio_ctx_finalize(GSource     *source)
     }
 #endif
 
+    assert(QSLIST_EMPTY(&ctx->scheduled_coroutines));
+    qemu_bh_delete(ctx->co_schedule_bh);
+
     qemu_lockcnt_lock(&ctx->list_lock);
     assert(!qemu_lockcnt_count(&ctx->list_lock));
     while (ctx->first_bh) {
@@ -XXX,XX +XXX,XX @@ static bool event_notifier_poll(void *opaque)
     return atomic_read(&ctx->notified);
 }
 
+static void co_schedule_bh_cb(void *opaque)
+{
+    AioContext *ctx = opaque;
+    QSLIST_HEAD(, Coroutine) straight, reversed;
+
+    QSLIST_MOVE_ATOMIC(&reversed, &ctx->scheduled_coroutines);
+    QSLIST_INIT(&straight);
+
+    while (!QSLIST_EMPTY(&reversed)) {
+        Coroutine *co = QSLIST_FIRST(&reversed);
+        QSLIST_REMOVE_HEAD(&reversed, co_scheduled_next);
+        QSLIST_INSERT_HEAD(&straight, co, co_scheduled_next);
+    }
+
+    while (!QSLIST_EMPTY(&straight)) {
+        Coroutine *co = QSLIST_FIRST(&straight);
+        QSLIST_REMOVE_HEAD(&straight, co_scheduled_next);
+        trace_aio_co_schedule_bh_cb(ctx, co);
+        qemu_coroutine_enter(co);
+    }
+}
+
 AioContext *aio_context_new(Error **errp)
 {
     int ret;
@@ -XXX,XX +XXX,XX @@ AioContext *aio_context_new(Error **errp)
     }
     g_source_set_can_recurse(&ctx->source, true);
     qemu_lockcnt_init(&ctx->list_lock);
+
+    ctx->co_schedule_bh = aio_bh_new(ctx, co_schedule_bh_cb, ctx);
+    QSLIST_INIT(&ctx->scheduled_coroutines);
+
     aio_set_event_notifier(ctx, &ctx->notifier,
                            false,
                            (EventNotifierHandler *)
@@ -XXX,XX +XXX,XX @@ fail:
     return NULL;
 }
 
+void aio_co_schedule(AioContext *ctx, Coroutine *co)
+{
+    trace_aio_co_schedule(ctx, co);
+    QSLIST_INSERT_HEAD_ATOMIC(&ctx->scheduled_coroutines,
+                              co, co_scheduled_next);
+    qemu_bh_schedule(ctx->co_schedule_bh);
+}
+
+void aio_co_wake(struct Coroutine *co)
+{
+    AioContext *ctx;
+
+    /* Read coroutine before co->ctx.  Matches smp_wmb in
+     * qemu_coroutine_enter.
+     */
+    smp_read_barrier_depends();
+    ctx = atomic_read(&co->ctx);
+
+    if (ctx != qemu_get_current_aio_context()) {
+        aio_co_schedule(ctx, co);
+        return;
+    }
+
+    if (qemu_in_coroutine()) {
+        Coroutine *self = qemu_coroutine_self();
+        assert(self != co);
+        QSIMPLEQ_INSERT_TAIL(&self->co_queue_wakeup, co, co_queue_next);
+    } else {
+        aio_context_acquire(ctx);
+        qemu_coroutine_enter(co);
+        aio_context_release(ctx);
+    }
+}
+
 void aio_context_ref(AioContext *ctx)
 {
     g_source_ref(&ctx->source);
diff --git a/util/qemu-coroutine.c b/util/qemu-coroutine.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine.c
+++ b/util/qemu-coroutine.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/atomic.h"
 #include "qemu/coroutine.h"
 #include "qemu/coroutine_int.h"
+#include "block/aio.h"
 
 enum {
     POOL_BATCH_SIZE = 64,
@@ -XXX,XX +XXX,XX @@ void qemu_coroutine_enter(Coroutine *co)
     }
 
     co->caller = self;
+    co->ctx = qemu_get_current_aio_context();
+
+    /* Store co->ctx before anything that stores co.  Matches
+     * barrier in aio_co_wake.
+     */
+    smp_wmb();
+
     ret = qemu_coroutine_switch(self, co, COROUTINE_ENTER);
 
     qemu_co_queue_run_restart(co);
diff --git a/util/trace-events b/util/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/util/trace-events
+++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@ run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
 poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
 poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
 
+# util/async.c
+aio_co_schedule(void *ctx, void *co) "ctx %p co %p"
+aio_co_schedule_bh_cb(void *ctx, void *co) "ctx %p co %p"
+
 # util/thread-pool.c
 thread_pool_submit(void *pool, void *req, void *opaque) "pool %p req %p opaque %p"
 thread_pool_complete(void *pool, void *req, void *opaque, int ret) "pool %p req %p opaque %p ret %d"
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

qcow2_create2 calls this.  Do not run a nested event loop, as that
breaks when aio_co_wake tries to queue the coroutine on the co_queue_wakeup
list of the currently running one.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-4-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/block-backend.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/block/block-backend.c b/block/block-backend.c
index XXXXXXX..XXXXXXX 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -XXX,XX +XXX,XX @@ static int blk_prw(BlockBackend *blk, int64_t offset, uint8_t *buf,
 {
     QEMUIOVector qiov;
     struct iovec iov;
-    Coroutine *co;
     BlkRwCo rwco;
 
     iov = (struct iovec) {
@@ -XXX,XX +XXX,XX @@ static int blk_prw(BlockBackend *blk, int64_t offset, uint8_t *buf,
         .ret    = NOT_DONE,
     };
 
-    co = qemu_coroutine_create(co_entry, &rwco);
-    qemu_coroutine_enter(co);
-    BDRV_POLL_WHILE(blk_bs(blk), rwco.ret == NOT_DONE);
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        co_entry(&rwco);
+    } else {
+        Coroutine *co = qemu_coroutine_create(co_entry, &rwco);
+        qemu_coroutine_enter(co);
+        BDRV_POLL_WHILE(blk_bs(blk), rwco.ret == NOT_DONE);
+    }
 
     return rwco.ret;
 }
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Once the thread pool starts using aio_co_wake, it will also need
qemu_get_current_aio_context().  Make test-thread-pool create
an AioContext with qemu_init_main_loop, so that stubs/iothread.c
and tests/iothread.c can provide the rest.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-5-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/test-thread-pool.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)

diff --git a/tests/test-thread-pool.c b/tests/test-thread-pool.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/test-thread-pool.c
+++ b/tests/test-thread-pool.c
@@ -XXX,XX +XXX,XX @@
 #include "qapi/error.h"
 #include "qemu/timer.h"
 #include "qemu/error-report.h"
+#include "qemu/main-loop.h"
 
 static AioContext *ctx;
 static ThreadPool *pool;
@@ -XXX,XX +XXX,XX @@ static void test_cancel_async(void)
 int main(int argc, char **argv)
 {
     int ret;
-    Error *local_error = NULL;
 
-    init_clocks();
-
-    ctx = aio_context_new(&local_error);
-    if (!ctx) {
-        error_reportf_err(local_error, "Failed to create AIO Context: ");
-        exit(1);
-    }
+    qemu_init_main_loop(&error_abort);
+    ctx = qemu_get_current_aio_context();
     pool = aio_get_thread_pool(ctx);
 
     g_test_init(&argc, &argv, NULL);
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
 
     ret = g_test_run();
 
-    aio_context_unref(ctx);
     return ret;
 }
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This is in preparation for making qio_channel_yield work on
AioContexts other than the main one.

Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-6-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/io/channel.h | 25 +++++++++++++++++++++++++
 io/channel-command.c | 13 +++++++++++++
 io/channel-file.c    | 11 +++++++++++
 io/channel-socket.c  | 16 +++++++++++-----
 io/channel-tls.c     | 12 ++++++++++++
 io/channel-watch.c   |  6 ++++++
 io/channel.c         | 11 +++++++++++
 7 files changed, 89 insertions(+), 5 deletions(-)

diff --git a/include/io/channel.h b/include/io/channel.h
index XXXXXXX..XXXXXXX 100644
--- a/include/io/channel.h
+++ b/include/io/channel.h
@@ -XXX,XX +XXX,XX @@
 
 #include "qemu-common.h"
 #include "qom/object.h"
+#include "block/aio.h"
 
 #define TYPE_QIO_CHANNEL "qio-channel"
 #define QIO_CHANNEL(obj)                                    \
@@ -XXX,XX +XXX,XX @@ struct QIOChannelClass {
                      off_t offset,
                      int whence,
                      Error **errp);
+    void (*io_set_aio_fd_handler)(QIOChannel *ioc,
+                                  AioContext *ctx,
+                                  IOHandler *io_read,
+                                  IOHandler *io_write,
+                                  void *opaque);
 };
 
 /* General I/O handling functions */
@@ -XXX,XX +XXX,XX @@ void qio_channel_yield(QIOChannel *ioc,
 void qio_channel_wait(QIOChannel *ioc,
                       GIOCondition condition);
 
+/**
+ * qio_channel_set_aio_fd_handler:
+ * @ioc: the channel object
+ * @ctx: the AioContext to set the handlers on
+ * @io_read: the read handler
+ * @io_write: the write handler
+ * @opaque: the opaque value passed to the handler
+ *
+ * This is used internally by qio_channel_yield().  It can
+ * be used by channel implementations to forward the handlers
+ * to another channel (e.g. from #QIOChannelTLS to the
+ * underlying socket).
+ */
+void qio_channel_set_aio_fd_handler(QIOChannel *ioc,
+                                    AioContext *ctx,
+                                    IOHandler *io_read,
+                                    IOHandler *io_write,
+                                    void *opaque);
+
 #endif /* QIO_CHANNEL_H */
diff --git a/io/channel-command.c b/io/channel-command.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel-command.c
+++ b/io/channel-command.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_command_close(QIOChannel *ioc,
 }
 
 
+static void qio_channel_command_set_aio_fd_handler(QIOChannel *ioc,
+                                                   AioContext *ctx,
+                                                   IOHandler *io_read,
+                                                   IOHandler *io_write,
+                                                   void *opaque)
+{
+    QIOChannelCommand *cioc = QIO_CHANNEL_COMMAND(ioc);
+    aio_set_fd_handler(ctx, cioc->readfd, false, io_read, NULL, NULL, opaque);
+    aio_set_fd_handler(ctx, cioc->writefd, false, NULL, io_write, NULL, opaque);
+}
+
+
 static GSource *qio_channel_command_create_watch(QIOChannel *ioc,
                                                  GIOCondition condition)
 {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_command_class_init(ObjectClass *klass,
     ioc_klass->io_set_blocking = qio_channel_command_set_blocking;
     ioc_klass->io_close = qio_channel_command_close;
     ioc_klass->io_create_watch = qio_channel_command_create_watch;
+    ioc_klass->io_set_aio_fd_handler = qio_channel_command_set_aio_fd_handler;
 }
 
 static const TypeInfo qio_channel_command_info = {
diff --git a/io/channel-file.c b/io/channel-file.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel-file.c
+++ b/io/channel-file.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_file_close(QIOChannel *ioc,
 }
 
 
+static void qio_channel_file_set_aio_fd_handler(QIOChannel *ioc,
+                                                AioContext *ctx,
+                                                IOHandler *io_read,
+                                                IOHandler *io_write,
+                                                void *opaque)
+{
+    QIOChannelFile *fioc = QIO_CHANNEL_FILE(ioc);
+    aio_set_fd_handler(ctx, fioc->fd, false, io_read, io_write, NULL, opaque);
+}
+
 static GSource *qio_channel_file_create_watch(QIOChannel *ioc,
                                               GIOCondition condition)
 {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_file_class_init(ObjectClass *klass,
     ioc_klass->io_seek = qio_channel_file_seek;
     ioc_klass->io_close = qio_channel_file_close;
     ioc_klass->io_create_watch = qio_channel_file_create_watch;
+    ioc_klass->io_set_aio_fd_handler = qio_channel_file_set_aio_fd_handler;
 }
 
 static const TypeInfo qio_channel_file_info = {
diff --git a/io/channel-socket.c b/io/channel-socket.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel-socket.c
+++ b/io/channel-socket.c
@@ -XXX,XX +XXX,XX @@ qio_channel_socket_set_blocking(QIOChannel *ioc,
         qemu_set_block(sioc->fd);
     } else {
         qemu_set_nonblock(sioc->fd);
-#ifdef WIN32
-        WSAEventSelect(sioc->fd, ioc->event,
-                       FD_READ | FD_ACCEPT | FD_CLOSE |
-                       FD_CONNECT | FD_WRITE | FD_OOB);
-#endif
     }
     return 0;
 }
@@ -XXX,XX +XXX,XX @@ qio_channel_socket_shutdown(QIOChannel *ioc,
     return 0;
 }
 
+static void qio_channel_socket_set_aio_fd_handler(QIOChannel *ioc,
+                                                  AioContext *ctx,
+                                                  IOHandler *io_read,
+                                                  IOHandler *io_write,
+                                                  void *opaque)
+{
+    QIOChannelSocket *sioc = QIO_CHANNEL_SOCKET(ioc);
+    aio_set_fd_handler(ctx, sioc->fd, false, io_read, io_write, NULL, opaque);
+}
+
 static GSource *qio_channel_socket_create_watch(QIOChannel *ioc,
                                                 GIOCondition condition)
 {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_socket_class_init(ObjectClass *klass,
     ioc_klass->io_set_cork = qio_channel_socket_set_cork;
     ioc_klass->io_set_delay = qio_channel_socket_set_delay;
     ioc_klass->io_create_watch = qio_channel_socket_create_watch;
+    ioc_klass->io_set_aio_fd_handler = qio_channel_socket_set_aio_fd_handler;
 }
 
 static const TypeInfo qio_channel_socket_info = {
diff --git a/io/channel-tls.c b/io/channel-tls.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel-tls.c
+++ b/io/channel-tls.c
@@ -XXX,XX +XXX,XX @@ static int qio_channel_tls_close(QIOChannel *ioc,
     return qio_channel_close(tioc->master, errp);
 }
 
+static void qio_channel_tls_set_aio_fd_handler(QIOChannel *ioc,
+                                               AioContext *ctx,
+                                               IOHandler *io_read,
+                                               IOHandler *io_write,
+                                               void *opaque)
+{
+    QIOChannelTLS *tioc = QIO_CHANNEL_TLS(ioc);
+
+    qio_channel_set_aio_fd_handler(tioc->master, ctx, io_read, io_write, opaque);
+}
+
 static GSource *qio_channel_tls_create_watch(QIOChannel *ioc,
                                              GIOCondition condition)
 {
@@ -XXX,XX +XXX,XX @@ static void qio_channel_tls_class_init(ObjectClass *klass,
     ioc_klass->io_close = qio_channel_tls_close;
     ioc_klass->io_shutdown = qio_channel_tls_shutdown;
     ioc_klass->io_create_watch = qio_channel_tls_create_watch;
+    ioc_klass->io_set_aio_fd_handler = qio_channel_tls_set_aio_fd_handler;
 }
 
 static const TypeInfo qio_channel_tls_info = {
diff --git a/io/channel-watch.c b/io/channel-watch.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel-watch.c
+++ b/io/channel-watch.c
@@ -XXX,XX +XXX,XX @@ GSource *qio_channel_create_socket_watch(QIOChannel *ioc,
     GSource *source;
     QIOChannelSocketSource *ssource;
 
+#ifdef WIN32
+    WSAEventSelect(socket, ioc->event,
+                   FD_READ | FD_ACCEPT | FD_CLOSE |
+                   FD_CONNECT | FD_WRITE | FD_OOB);
+#endif
+
     source = g_source_new(&qio_channel_socket_source_funcs,
                           sizeof(QIOChannelSocketSource));
     ssource = (QIOChannelSocketSource *)source;
diff --git a/io/channel.c b/io/channel.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel.c
+++ b/io/channel.c
@@ -XXX,XX +XXX,XX @@ GSource *qio_channel_create_watch(QIOChannel *ioc,
 }
 
 
+void qio_channel_set_aio_fd_handler(QIOChannel *ioc,
+                                    AioContext *ctx,
+                                    IOHandler *io_read,
+                                    IOHandler *io_write,
+                                    void *opaque)
+{
+    QIOChannelClass *klass = QIO_CHANNEL_GET_CLASS(ioc);
+
+    klass->io_set_aio_fd_handler(ioc, ctx, io_read, io_write, opaque);
+}
+
 guint qio_channel_add_watch(QIOChannel *ioc,
                             GIOCondition condition,
                             QIOChannelFunc func,
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Support separate coroutines for reading and writing, and place the
read/write handlers on the AioContext that the QIOChannel is registered
with.

Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-7-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/io/channel.h | 47 ++++++++++++++++++++++++++--
 io/channel.c         | 86 +++++++++++++++++++++++++++++++++++++++-------------
 2 files changed, 109 insertions(+), 24 deletions(-)

diff --git a/include/io/channel.h b/include/io/channel.h
index XXXXXXX..XXXXXXX 100644
--- a/include/io/channel.h
+++ b/include/io/channel.h
@@ -XXX,XX +XXX,XX @@
 
 #include "qemu-common.h"
 #include "qom/object.h"
+#include "qemu/coroutine.h"
 #include "block/aio.h"
 
 #define TYPE_QIO_CHANNEL "qio-channel"
@@ -XXX,XX +XXX,XX @@ struct QIOChannel {
     Object parent;
     unsigned int features; /* bitmask of QIOChannelFeatures */
     char *name;
+    AioContext *ctx;
+    Coroutine *read_coroutine;
+    Coroutine *write_coroutine;
 #ifdef _WIN32
     HANDLE event; /* For use with GSource on Win32 */
 #endif
@@ -XXX,XX +XXX,XX @@ guint qio_channel_add_watch(QIOChannel *ioc,
 
 
 /**
+ * qio_channel_attach_aio_context:
+ * @ioc: the channel object
+ * @ctx: the #AioContext to set the handlers on
+ *
+ * Request that qio_channel_yield() sets I/O handlers on
+ * the given #AioContext.  If @ctx is %NULL, qio_channel_yield()
+ * uses QEMU's main thread event loop.
+ *
+ * You can move a #QIOChannel from one #AioContext to another even if
+ * I/O handlers are set for a coroutine.  However, #QIOChannel provides
+ * no synchronization between the calls to qio_channel_yield() and
+ * qio_channel_attach_aio_context().
+ *
+ * Therefore you should first call qio_channel_detach_aio_context()
+ * to ensure that the coroutine is not entered concurrently.  Then,
+ * while the coroutine has yielded, call qio_channel_attach_aio_context(),
+ * and then aio_co_schedule() to place the coroutine on the new
+ * #AioContext.  The calls to qio_channel_detach_aio_context()
+ * and qio_channel_attach_aio_context() should be protected with
+ * aio_context_acquire() and aio_context_release().
+ */
+void qio_channel_attach_aio_context(QIOChannel *ioc,
+                                    AioContext *ctx);
+
+/**
+ * qio_channel_detach_aio_context:
+ * @ioc: the channel object
+ *
+ * Disable any I/O handlers set by qio_channel_yield().  With the
+ * help of aio_co_schedule(), this allows moving a coroutine that was
+ * paused by qio_channel_yield() to another context.
+ */
+void qio_channel_detach_aio_context(QIOChannel *ioc);
+
+/**
  * qio_channel_yield:
  * @ioc: the channel object
  * @condition: the I/O condition to wait for
  *
- * Yields execution from the current coroutine until
- * the condition indicated by @condition becomes
- * available.
+ * Yields execution from the current coroutine until the condition
+ * indicated by @condition becomes available.  @condition must
+ * be either %G_IO_IN or %G_IO_OUT; it cannot contain both.  In
+ * addition, no two coroutine can be waiting on the same condition
+ * and channel at the same time.
  *
  * This must only be called from coroutine context
  */
diff --git a/io/channel.c b/io/channel.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel.c
+++ b/io/channel.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/osdep.h"
 #include "io/channel.h"
 #include "qapi/error.h"
-#include "qemu/coroutine.h"
+#include "qemu/main-loop.h"
 
 bool qio_channel_has_feature(QIOChannel *ioc,
                              QIOChannelFeature feature)
@@ -XXX,XX +XXX,XX @@ off_t qio_channel_io_seek(QIOChannel *ioc,
 }
 
 
-typedef struct QIOChannelYieldData QIOChannelYieldData;
-struct QIOChannelYieldData {
-    QIOChannel *ioc;
-    Coroutine *co;
-};
+static void qio_channel_set_aio_fd_handlers(QIOChannel *ioc);
 
+static void qio_channel_restart_read(void *opaque)
+{
+    QIOChannel *ioc = opaque;
+    Coroutine *co = ioc->read_coroutine;
+
+    ioc->read_coroutine = NULL;
+    qio_channel_set_aio_fd_handlers(ioc);
+    aio_co_wake(co);
+}
 
-static gboolean qio_channel_yield_enter(QIOChannel *ioc,
-                                        GIOCondition condition,
-                                        gpointer opaque)
+static void qio_channel_restart_write(void *opaque)
 {
-    QIOChannelYieldData *data = opaque;
-    qemu_coroutine_enter(data->co);
-    return FALSE;
+    QIOChannel *ioc = opaque;
+    Coroutine *co = ioc->write_coroutine;
+
+    ioc->write_coroutine = NULL;
+    qio_channel_set_aio_fd_handlers(ioc);
+    aio_co_wake(co);
 }
 
+static void qio_channel_set_aio_fd_handlers(QIOChannel *ioc)
+{
+    IOHandler *rd_handler = NULL, *wr_handler = NULL;
+    AioContext *ctx;
+
+    if (ioc->read_coroutine) {
+        rd_handler = qio_channel_restart_read;
+    }
+    if (ioc->write_coroutine) {
+        wr_handler = qio_channel_restart_write;
+    }
+
+    ctx = ioc->ctx ? ioc->ctx : iohandler_get_aio_context();
+    qio_channel_set_aio_fd_handler(ioc, ctx, rd_handler, wr_handler, ioc);
+}
+
+void qio_channel_attach_aio_context(QIOChannel *ioc,
+                                    AioContext *ctx)
+{
+    AioContext *old_ctx;
+    if (ioc->ctx == ctx) {
+        return;
+    }
+
+    old_ctx = ioc->ctx ? ioc->ctx : iohandler_get_aio_context();
+    qio_channel_set_aio_fd_handler(ioc, old_ctx, NULL, NULL, NULL);
+    ioc->ctx = ctx;
+    qio_channel_set_aio_fd_handlers(ioc);
+}
+
+void qio_channel_detach_aio_context(QIOChannel *ioc)
+{
+    ioc->read_coroutine = NULL;
+    ioc->write_coroutine = NULL;
+    qio_channel_set_aio_fd_handlers(ioc);
+    ioc->ctx = NULL;
+}
 
 void coroutine_fn qio_channel_yield(QIOChannel *ioc,
                                     GIOCondition condition)
 {
-    QIOChannelYieldData data;
-
     assert(qemu_in_coroutine());
-    data.ioc = ioc;
-    data.co = qemu_coroutine_self();
-    qio_channel_add_watch(ioc,
-                          condition,
-                          qio_channel_yield_enter,
-                          &data,
-                          NULL);
+    if (condition == G_IO_IN) {
+        assert(!ioc->read_coroutine);
+        ioc->read_coroutine = qemu_coroutine_self();
+    } else if (condition == G_IO_OUT) {
+        assert(!ioc->write_coroutine);
+        ioc->write_coroutine = qemu_coroutine_self();
+    } else {
+        abort();
+    }
+    qio_channel_set_aio_fd_handlers(ioc);
     qemu_coroutine_yield();
 }
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

In the client, read the reply headers from a coroutine, switching the
read side between the "read header" coroutine and the I/O coroutine that
reads the body of the reply.

In the server, if the server can read more requests it will create a new
"read request" coroutine as soon as a request has been read.  Otherwise,
the new coroutine is created in nbd_request_put.

diff --git a/block/nbd-client.h b/block/nbd-client.h
index XXXXXXX..XXXXXXX 100644
--- a/block/nbd-client.h
+++ b/block/nbd-client.h
@@ -XXX,XX +XXX,XX @@ typedef struct NBDClientSession {
 
     CoMutex send_mutex;
     CoQueue free_sema;
-    Coroutine *send_coroutine;
+    Coroutine *read_reply_co;
     int in_flight;
 
     Coroutine *recv_coroutine[MAX_NBD_REQUESTS];
diff --git a/block/nbd-client.c b/block/nbd-client.c
index XXXXXXX..XXXXXXX 100644
--- a/block/nbd-client.c
+++ b/block/nbd-client.c
@@ -XXX,XX +XXX,XX @@
 #define HANDLE_TO_INDEX(bs, handle) ((handle) ^ ((uint64_t)(intptr_t)bs))
 #define INDEX_TO_HANDLE(bs, index)  ((index)  ^ ((uint64_t)(intptr_t)bs))
 
-static void nbd_recv_coroutines_enter_all(NBDClientSession *s)
+static void nbd_recv_coroutines_enter_all(BlockDriverState *bs)
 {
+    NBDClientSession *s = nbd_get_client_session(bs);
     int i;
 
     for (i = 0; i < MAX_NBD_REQUESTS; i++) {
@@ -XXX,XX +XXX,XX @@ static void nbd_recv_coroutines_enter_all(NBDClientSession *s)
             qemu_coroutine_enter(s->recv_coroutine[i]);
         }
     }
+    BDRV_POLL_WHILE(bs, s->read_reply_co);
 }
 
 static void nbd_teardown_connection(BlockDriverState *bs)
@@ -XXX,XX +XXX,XX @@ static void nbd_teardown_connection(BlockDriverState *bs)
     qio_channel_shutdown(client->ioc,
                          QIO_CHANNEL_SHUTDOWN_BOTH,
                          NULL);
-    nbd_recv_coroutines_enter_all(client);
+    nbd_recv_coroutines_enter_all(bs);
 
     nbd_client_detach_aio_context(bs);
     object_unref(OBJECT(client->sioc));
@@ -XXX,XX +XXX,XX @@ static void nbd_teardown_connection(BlockDriverState *bs)
     client->ioc = NULL;
 }
 
-static void nbd_reply_ready(void *opaque)
+static coroutine_fn void nbd_read_reply_entry(void *opaque)
 {
-    BlockDriverState *bs = opaque;
-    NBDClientSession *s = nbd_get_client_session(bs);
+    NBDClientSession *s = opaque;
     uint64_t i;
     int ret;
 
-    if (!s->ioc) { /* Already closed */
-        return;
-    }
-
-    if (s->reply.handle == 0) {
-        /* No reply already in flight.  Fetch a header.  It is possible
-         * that another thread has done the same thing in parallel, so
-         * the socket is not readable anymore.
-         */
+    for (;;) {
+        assert(s->reply.handle == 0);
         ret = nbd_receive_reply(s->ioc, &s->reply);
-        if (ret == -EAGAIN) {
-            return;
-        }
         if (ret < 0) {
-            s->reply.handle = 0;
-            goto fail;
+            break;
         }
-    }
 
-    /* There's no need for a mutex on the receive side, because the
-     * handler acts as a synchronization point and ensures that only
-     * one coroutine is called until the reply finishes.  */
-    i = HANDLE_TO_INDEX(s, s->reply.handle);
-    if (i >= MAX_NBD_REQUESTS) {
-        goto fail;
-    }
+        /* There's no need for a mutex on the receive side, because the
+         * handler acts as a synchronization point and ensures that only
+         * one coroutine is called until the reply finishes.
+         */
+        i = HANDLE_TO_INDEX(s, s->reply.handle);
+        if (i >= MAX_NBD_REQUESTS || !s->recv_coroutine[i]) {
+            break;
+        }
 
-    if (s->recv_coroutine[i]) {
-        qemu_coroutine_enter(s->recv_coroutine[i]);
-        return;
+        /* We're woken up by the recv_coroutine itself.  Note that there
+         * is no race between yielding and reentering read_reply_co.  This
+         * is because:
+         *
+         * - if recv_coroutine[i] runs on the same AioContext, it is only
+         *   entered after we yield
+         *
+         * - if recv_coroutine[i] runs on a different AioContext, reentering
+         *   read_reply_co happens through a bottom half, which can only
+         *   run after we yield.
+         */
+        aio_co_wake(s->recv_coroutine[i]);
+        qemu_coroutine_yield();
     }
-
-fail:
-    nbd_teardown_connection(bs);
-}
-
-static void nbd_restart_write(void *opaque)
-{
-    BlockDriverState *bs = opaque;
-
-    qemu_coroutine_enter(nbd_get_client_session(bs)->send_coroutine);
+    s->read_reply_co = NULL;
 }
 
 static int nbd_co_send_request(BlockDriverState *bs,
@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
                                QEMUIOVector *qiov)
 {
     NBDClientSession *s = nbd_get_client_session(bs);
-    AioContext *aio_context;
     int rc, ret, i;
 
     qemu_co_mutex_lock(&s->send_mutex);
@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
         return -EPIPE;
     }
 
-    s->send_coroutine = qemu_coroutine_self();
-    aio_context = bdrv_get_aio_context(bs);
-
-    aio_set_fd_handler(aio_context, s->sioc->fd, false,
-                       nbd_reply_ready, nbd_restart_write, NULL, bs);
     if (qiov) {
         qio_channel_set_cork(s->ioc, true);
         rc = nbd_send_request(s->ioc, request);
@@ -XXX,XX +XXX,XX @@ static int nbd_co_send_request(BlockDriverState *bs,
     } else {
         rc = nbd_send_request(s->ioc, request);
     }
-    aio_set_fd_handler(aio_context, s->sioc->fd, false,
-                       nbd_reply_ready, NULL, NULL, bs);
-    s->send_coroutine = NULL;
     qemu_co_mutex_unlock(&s->send_mutex);
     return rc;
 }
@@ -XXX,XX +XXX,XX @@ static void nbd_co_receive_reply(NBDClientSession *s,
 {
     int ret;
 
-    /* Wait until we're woken up by the read handler.  TODO: perhaps
-     * peek at the next reply and avoid yielding if it's ours?  */
+    /* Wait until we're woken up by nbd_read_reply_entry.  */
     qemu_coroutine_yield();
     *reply = s->reply;
     if (reply->handle != request->handle ||
@@ -XXX,XX +XXX,XX @@ static void nbd_coroutine_start(NBDClientSession *s,
     /* s->recv_coroutine[i] is set as soon as we get the send_lock.  */
 }
 
-static void nbd_coroutine_end(NBDClientSession *s,
+static void nbd_coroutine_end(BlockDriverState *bs,
                               NBDRequest *request)
 {
+    NBDClientSession *s = nbd_get_client_session(bs);
     int i = HANDLE_TO_INDEX(s, request->handle);
+
     s->recv_coroutine[i] = NULL;
-    if (s->in_flight-- == MAX_NBD_REQUESTS) {
-        qemu_co_queue_next(&s->free_sema);
+    s->in_flight--;
+    qemu_co_queue_next(&s->free_sema);
+
+    /* Kick the read_reply_co to get the next reply.  */
+    if (s->read_reply_co) {
+        aio_co_wake(s->read_reply_co);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ int nbd_client_co_preadv(BlockDriverState *bs, uint64_t offset,
     } else {
         nbd_co_receive_reply(client, &request, &reply, qiov);
     }
-    nbd_coroutine_end(client, &request);
+    nbd_coroutine_end(bs, &request);
     return -reply.error;
 }
 
@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pwritev(BlockDriverState *bs, uint64_t offset,
     } else {
         nbd_co_receive_reply(client, &request, &reply, NULL);
     }
-    nbd_coroutine_end(client, &request);
+    nbd_coroutine_end(bs, &request);
     return -reply.error;
 }
 
@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset,
     } else {
         nbd_co_receive_reply(client, &request, &reply, NULL);
     }
-    nbd_coroutine_end(client, &request);
+    nbd_coroutine_end(bs, &request);
     return -reply.error;
 }
 
@@ -XXX,XX +XXX,XX @@ int nbd_client_co_flush(BlockDriverState *bs)
     } else {
         nbd_co_receive_reply(client, &request, &reply, NULL);
     }
-    nbd_coroutine_end(client, &request);
+    nbd_coroutine_end(bs, &request);
     return -reply.error;
 }
 
@@ -XXX,XX +XXX,XX @@ int nbd_client_co_pdiscard(BlockDriverState *bs, int64_t offset, int count)
     } else {
         nbd_co_receive_reply(client, &request, &reply, NULL);
     }
-    nbd_coroutine_end(client, &request);
+    nbd_coroutine_end(bs, &request);
     return -reply.error;
 
 }
 
 void nbd_client_detach_aio_context(BlockDriverState *bs)
 {
-    aio_set_fd_handler(bdrv_get_aio_context(bs),
-                       nbd_get_client_session(bs)->sioc->fd,
-                       false, NULL, NULL, NULL, NULL);
+    NBDClientSession *client = nbd_get_client_session(bs);
+    qio_channel_detach_aio_context(QIO_CHANNEL(client->sioc));
 }
 
 void nbd_client_attach_aio_context(BlockDriverState *bs,
                                    AioContext *new_context)
 {
-    aio_set_fd_handler(new_context, nbd_get_client_session(bs)->sioc->fd,
-                       false, nbd_reply_ready, NULL, NULL, bs);
+    NBDClientSession *client = nbd_get_client_session(bs);
+    qio_channel_attach_aio_context(QIO_CHANNEL(client->sioc), new_context);
+    aio_co_schedule(new_context, client->read_reply_co);
 }
 
 void nbd_client_close(BlockDriverState *bs)
@@ -XXX,XX +XXX,XX @@ int nbd_client_init(BlockDriverState *bs,
     /* Now that we're connected, set the socket to be non-blocking and
      * kick the reply mechanism.  */
     qio_channel_set_blocking(QIO_CHANNEL(sioc), false, NULL);
-
+    client->read_reply_co = qemu_coroutine_create(nbd_read_reply_entry, client);
     nbd_client_attach_aio_context(bs, bdrv_get_aio_context(bs));
 
     logout("Established connection with NBD server\n");
diff --git a/nbd/client.c b/nbd/client.c
index XXXXXXX..XXXXXXX 100644
--- a/nbd/client.c
+++ b/nbd/client.c
@@ -XXX,XX +XXX,XX @@ ssize_t nbd_receive_reply(QIOChannel *ioc, NBDReply *reply)
     ssize_t ret;
 
     ret = read_sync(ioc, buf, sizeof(buf));
-    if (ret < 0) {
+    if (ret <= 0) {
         return ret;
     }
 
diff --git a/nbd/common.c b/nbd/common.c
index XXXXXXX..XXXXXXX 100644
--- a/nbd/common.c
+++ b/nbd/common.c
@@ -XXX,XX +XXX,XX @@ ssize_t nbd_wr_syncv(QIOChannel *ioc,
         }
         if (len == QIO_CHANNEL_ERR_BLOCK) {
             if (qemu_in_coroutine()) {
-                /* XXX figure out if we can create a variant on
-                 * qio_channel_yield() that works with AIO contexts
-                 * and consider using that in this branch */
-                qemu_coroutine_yield();
-            } else if (done) {
-                /* XXX this is needed by nbd_reply_ready.  */
-                qio_channel_wait(ioc,
-                                 do_read ? G_IO_IN : G_IO_OUT);
+                qio_channel_yield(ioc, do_read ? G_IO_IN : G_IO_OUT);
             } else {
                 return -EAGAIN;
             }
diff --git a/nbd/server.c b/nbd/server.c
index XXXXXXX..XXXXXXX 100644
--- a/nbd/server.c
+++ b/nbd/server.c
@@ -XXX,XX +XXX,XX @@ struct NBDClient {
     CoMutex send_lock;
     Coroutine *send_coroutine;
 
-    bool can_read;
-
     QTAILQ_ENTRY(NBDClient) next;
     int nb_requests;
     bool closing;
@@ -XXX,XX +XXX,XX @@ struct NBDClient {
 
 /* That's all folks */
 
-static void nbd_set_handlers(NBDClient *client);
-static void nbd_unset_handlers(NBDClient *client);
-static void nbd_update_can_read(NBDClient *client);
+static void nbd_client_receive_next_request(NBDClient *client);
 
 static gboolean nbd_negotiate_continue(QIOChannel *ioc,
                                        GIOCondition condition,
@@ -XXX,XX +XXX,XX @@ void nbd_client_put(NBDClient *client)
          */
         assert(client->closing);
 
-        nbd_unset_handlers(client);
+        qio_channel_detach_aio_context(client->ioc);
         object_unref(OBJECT(client->sioc));
         object_unref(OBJECT(client->ioc));
         if (client->tlscreds) {
@@ -XXX,XX +XXX,XX @@ static NBDRequestData *nbd_request_get(NBDClient *client)
 
     assert(client->nb_requests <= MAX_NBD_REQUESTS - 1);
     client->nb_requests++;
-    nbd_update_can_read(client);
 
     req = g_new0(NBDRequestData, 1);
     nbd_client_get(client);
@@ -XXX,XX +XXX,XX @@ static void nbd_request_put(NBDRequestData *req)
     g_free(req);
 
     client->nb_requests--;
-    nbd_update_can_read(client);
+    nbd_client_receive_next_request(client);
+
     nbd_client_put(client);
 }
 
@@ -XXX,XX +XXX,XX @@ static void blk_aio_attached(AioContext *ctx, void *opaque)
     exp->ctx = ctx;
 
     QTAILQ_FOREACH(client, &exp->clients, next) {
-        nbd_set_handlers(client);
+        qio_channel_attach_aio_context(client->ioc, ctx);
+        if (client->recv_coroutine) {
+            aio_co_schedule(ctx, client->recv_coroutine);
+        }
+        if (client->send_coroutine) {
+            aio_co_schedule(ctx, client->send_coroutine);
+        }
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static void blk_aio_detach(void *opaque)
     TRACE("Export %s: Detaching clients from AIO context %p\n", exp->name, exp->ctx);
 
     QTAILQ_FOREACH(client, &exp->clients, next) {
-        nbd_unset_handlers(client);
+        qio_channel_detach_aio_context(client->ioc);
     }
 
     exp->ctx = NULL;
@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_send_reply(NBDRequestData *req, NBDReply *reply,
     g_assert(qemu_in_coroutine());
     qemu_co_mutex_lock(&client->send_lock);
     client->send_coroutine = qemu_coroutine_self();
-    nbd_set_handlers(client);
 
     if (!len) {
         rc = nbd_send_reply(client->ioc, reply);
@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_send_reply(NBDRequestData *req, NBDReply *reply,
     }
 
     client->send_coroutine = NULL;
-    nbd_set_handlers(client);
     qemu_co_mutex_unlock(&client->send_lock);
     return rc;
 }
@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_receive_request(NBDRequestData *req,
     ssize_t rc;
 
     g_assert(qemu_in_coroutine());
-    client->recv_coroutine = qemu_coroutine_self();
-    nbd_update_can_read(client);
-
+    assert(client->recv_coroutine == qemu_coroutine_self());
     rc = nbd_receive_request(client->ioc, request);
     if (rc < 0) {
         if (rc != -EAGAIN) {
@@ -XXX,XX +XXX,XX @@ static ssize_t nbd_co_receive_request(NBDRequestData *req,
 
 out:
     client->recv_coroutine = NULL;
-    nbd_update_can_read(client);
+    nbd_client_receive_next_request(client);
 
     return rc;
 }
 
-static void nbd_trip(void *opaque)
+/* Owns a reference to the NBDClient passed as opaque.  */
+static coroutine_fn void nbd_trip(void *opaque)
 {
     NBDClient *client = opaque;
     NBDExport *exp = client->exp;
     NBDRequestData *req;
-    NBDRequest request;
+    NBDRequest request = { 0 };    /* GCC thinks it can be used uninitialized */
     NBDReply reply;
     ssize_t ret;
     int flags;
 
     TRACE("Reading request.");
     if (client->closing) {
+        nbd_client_put(client);
         return;
     }
 
@@ -XXX,XX +XXX,XX @@ static void nbd_trip(void *opaque)
 
 done:
     nbd_request_put(req);
+    nbd_client_put(client);
     return;
 
 out:
     nbd_request_put(req);
     client_close(client);
+    nbd_client_put(client);
 }
 
-static void nbd_read(void *opaque)
+static void nbd_client_receive_next_request(NBDClient *client)
 {
-    NBDClient *client = opaque;
-
-    if (client->recv_coroutine) {
-        qemu_coroutine_enter(client->recv_coroutine);
-    } else {
-        qemu_coroutine_enter(qemu_coroutine_create(nbd_trip, client));
-    }
-}
-
-static void nbd_restart_write(void *opaque)
-{
-    NBDClient *client = opaque;
-
-    qemu_coroutine_enter(client->send_coroutine);
-}
-
-static void nbd_set_handlers(NBDClient *client)
-{
-    if (client->exp && client->exp->ctx) {
-        aio_set_fd_handler(client->exp->ctx, client->sioc->fd, true,
-                           client->can_read ? nbd_read : NULL,
-                           client->send_coroutine ? nbd_restart_write : NULL,
-                           NULL, client);
-    }
-}
-
-static void nbd_unset_handlers(NBDClient *client)
-{
-    if (client->exp && client->exp->ctx) {
-        aio_set_fd_handler(client->exp->ctx, client->sioc->fd, true, NULL,
-                           NULL, NULL, NULL);
-    }
-}
-
-static void nbd_update_can_read(NBDClient *client)
-{
-    bool can_read = client->recv_coroutine ||
-                    client->nb_requests < MAX_NBD_REQUESTS;
-
-    if (can_read != client->can_read) {
-        client->can_read = can_read;
-        nbd_set_handlers(client);
-
-        /* There is no need to invoke aio_notify(), since aio_set_fd_handler()
-         * in nbd_set_handlers() will have taken care of that */
+    if (!client->recv_coroutine && client->nb_requests < MAX_NBD_REQUESTS) {
+        nbd_client_get(client);
+        client->recv_coroutine = qemu_coroutine_create(nbd_trip, client);
+        aio_co_schedule(client->exp->ctx, client->recv_coroutine);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void nbd_co_client_start(void *opaque)
         goto out;
     }
     qemu_co_mutex_init(&client->send_lock);
-    nbd_set_handlers(client);
 
     if (exp) {
         QTAILQ_INSERT_TAIL(&exp->clients, client, next);
     }
+
+    nbd_client_receive_next_request(client);
+
 out:
     g_free(data);
 }
@@ -XXX,XX +XXX,XX @@ void nbd_client_new(NBDExport *exp,
     object_ref(OBJECT(client->sioc));
     client->ioc = QIO_CHANNEL(sioc);
     object_ref(OBJECT(client->ioc));
-    client->can_read = true;
     client->close = close_fn;
 
     data->client = client;
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

As a small step towards the introduction of multiqueue, we want
coroutines to remain on the same AioContext that started them,
unless they are moved explicitly with e.g. aio_co_schedule.  This patch
avoids that coroutines switch AioContext when they use a CoMutex.
For now it does not make much of a difference, because the CoMutex
is not thread-safe and the AioContext itself is used to protect the
CoMutex from concurrent access.  However, this is going to change.

diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-lock.c
+++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/coroutine.h"
 #include "qemu/coroutine_int.h"
 #include "qemu/queue.h"
+#include "block/aio.h"
 #include "trace.h"
 
 void qemu_co_queue_init(CoQueue *queue)
@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_run_restart(Coroutine *co)
 
 static bool qemu_co_queue_do_restart(CoQueue *queue, bool single)
 {
-    Coroutine *self = qemu_coroutine_self();
     Coroutine *next;
 
     if (QSIMPLEQ_EMPTY(&queue->entries)) {
@@ -XXX,XX +XXX,XX @@ static bool qemu_co_queue_do_restart(CoQueue *queue, bool single)
 
     while ((next = QSIMPLEQ_FIRST(&queue->entries)) != NULL) {
         QSIMPLEQ_REMOVE_HEAD(&queue->entries, co_queue_next);
-        QSIMPLEQ_INSERT_TAIL(&self->co_queue_wakeup, next, co_queue_next);
-        trace_qemu_co_queue_next(next);
+        aio_co_wake(next);
         if (single) {
             break;
         }
diff --git a/util/trace-events b/util/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/util/trace-events
+++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@ qemu_coroutine_terminate(void *co) "self %p"
 
 # util/qemu-coroutine-lock.c
 qemu_co_queue_run_restart(void *co) "co %p"
-qemu_co_queue_next(void *nxt) "next %p"
 qemu_co_mutex_lock_entry(void *mutex, void *self) "mutex %p self %p"
 qemu_co_mutex_lock_return(void *mutex, void *self) "mutex %p self %p"
 qemu_co_mutex_unlock_entry(void *mutex, void *self) "mutex %p self %p"
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Keep the coroutine on the same AioContext.  Without this change,
there would be a race between yielding the coroutine and reentering it.
While the race cannot happen now, because the code only runs from a single
AioContext, this will change with multiqueue support in the block layer.

While doing the change, replace custom bottom half with aio_co_schedule.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-10-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/blkdebug.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)

diff --git a/block/blkdebug.c b/block/blkdebug.c
index XXXXXXX..XXXXXXX 100644
--- a/block/blkdebug.c
+++ b/block/blkdebug.c
@@ -XXX,XX +XXX,XX @@ out:
     return ret;
 }
 
-static void error_callback_bh(void *opaque)
-{
-    Coroutine *co = opaque;
-    qemu_coroutine_enter(co);
-}
-
 static int inject_error(BlockDriverState *bs, BlkdebugRule *rule)
 {
     BDRVBlkdebugState *s = bs->opaque;
@@ -XXX,XX +XXX,XX @@ static int inject_error(BlockDriverState *bs, BlkdebugRule *rule)
     }
 
     if (!immediately) {
-        aio_bh_schedule_oneshot(bdrv_get_aio_context(bs), error_callback_bh,
-                                qemu_coroutine_self());
+        aio_co_schedule(qemu_get_current_aio_context(), qemu_coroutine_self());
         qemu_coroutine_yield();
     }
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

qed_aio_start_io and qed_aio_next_io will not have to acquire/release
the AioContext, while qed_aio_next_io_cb will.  Split the functionality
and gain a little type-safety in the process.

diff --git a/block/qed.c b/block/qed.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static CachedL2Table *qed_new_l2_table(BDRVQEDState *s)
     return l2_table;
 }
 
-static void qed_aio_next_io(void *opaque, int ret);
+static void qed_aio_next_io(QEDAIOCB *acb, int ret);
+
+static void qed_aio_start_io(QEDAIOCB *acb)
+{
+    qed_aio_next_io(acb, 0);
+}
+
+static void qed_aio_next_io_cb(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+
+    qed_aio_next_io(acb, ret);
+}
 
 static void qed_plug_allocating_write_reqs(BDRVQEDState *s)
 {
@@ -XXX,XX +XXX,XX @@ static void qed_unplug_allocating_write_reqs(BDRVQEDState *s)
 
     acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
     if (acb) {
-        qed_aio_next_io(acb, 0);
+        qed_aio_start_io(acb);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static void qed_aio_complete(QEDAIOCB *acb, int ret)
         QSIMPLEQ_REMOVE_HEAD(&s->allocating_write_reqs, next);
         acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
         if (acb) {
-            qed_aio_next_io(acb, 0);
+            qed_aio_start_io(acb);
         } else if (s->header.features & QED_F_NEED_CHECK) {
             qed_start_need_check_timer(s);
         }
@@ -XXX,XX +XXX,XX @@ static void qed_commit_l2_update(void *opaque, int ret)
     acb->request.l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
     assert(acb->request.l2_table != NULL);
 
-    qed_aio_next_io(opaque, ret);
+    qed_aio_next_io(acb, ret);
 }
 
 /**
@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_l2_update(QEDAIOCB *acb, int ret, uint64_t offset)
     if (need_alloc) {
         /* Write out the whole new L2 table */
         qed_write_l2_table(s, &acb->request, 0, s->table_nelems, true,
-                            qed_aio_write_l1_update, acb);
+                           qed_aio_write_l1_update, acb);
     } else {
         /* Write out only the updated part of the L2 table */
         qed_write_l2_table(s, &acb->request, index, acb->cur_nclusters, false,
-                            qed_aio_next_io, acb);
+                           qed_aio_next_io_cb, acb);
     }
     return;
 
@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_main(void *opaque, int ret)
     }
 
     if (acb->find_cluster_ret == QED_CLUSTER_FOUND) {
-        next_fn = qed_aio_next_io;
+        next_fn = qed_aio_next_io_cb;
     } else {
         if (s->bs->backing) {
             next_fn = qed_aio_write_flush_before_l2_update;
@@ -XXX,XX +XXX,XX @@ static void qed_aio_write_alloc(QEDAIOCB *acb, size_t len)
     if (acb->flags & QED_AIOCB_ZERO) {
         /* Skip ahead if the clusters are already zero */
         if (acb->find_cluster_ret == QED_CLUSTER_ZERO) {
-            qed_aio_next_io(acb, 0);
+            qed_aio_start_io(acb);
             return;
         }
 
@@ -XXX,XX +XXX,XX @@ static void qed_aio_read_data(void *opaque, int ret,
     /* Handle zero cluster and backing file reads */
     if (ret == QED_CLUSTER_ZERO) {
         qemu_iovec_memset(&acb->cur_qiov, 0, 0, acb->cur_qiov.size);
-        qed_aio_next_io(acb, 0);
+        qed_aio_start_io(acb);
         return;
     } else if (ret != QED_CLUSTER_FOUND) {
         qed_read_backing_file(s, acb->cur_pos, &acb->cur_qiov,
-                              &acb->backing_qiov, qed_aio_next_io, acb);
+                              &acb->backing_qiov, qed_aio_next_io_cb, acb);
         return;
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
     bdrv_aio_readv(bs->file, offset / BDRV_SECTOR_SIZE,
                    &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
-                   qed_aio_next_io, acb);
+                   qed_aio_next_io_cb, acb);
     return;
 
 err:
@@ -XXX,XX +XXX,XX @@ err:
 /**
  * Begin next I/O or complete the request
  */
-static void qed_aio_next_io(void *opaque, int ret)
+static void qed_aio_next_io(QEDAIOCB *acb, int ret)
 {
-    QEDAIOCB *acb = opaque;
     BDRVQEDState *s = acb_to_s(acb);
     QEDFindClusterFunc *io_fn = (acb->flags & QED_AIOCB_WRITE) ?
                                 qed_aio_write_data : qed_aio_read_data;
@@ -XXX,XX +XXX,XX @@ static BlockAIOCB *qed_aio_setup(BlockDriverState *bs,
     qemu_iovec_init(&acb->cur_qiov, qiov->niov);
 
     /* Start request */
-    qed_aio_next_io(acb, 0);
+    qed_aio_start_io(acb);
     return &acb->common;
 }
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

The AioContext data structures are now protected by list_lock and/or
they are walked with FOREACH_RCU primitives.  There is no need anymore
to acquire the AioContext for the entire duration of aio_dispatch.
Instead, just acquire it before and after invoking the callbacks.
The next step is then to push it further down.

diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
             (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
             aio_node_check(ctx, node->is_external) &&
             node->io_read) {
+            aio_context_acquire(ctx);
             node->io_read(node->opaque);
+            aio_context_release(ctx);
 
             /* aio_notify() does not count as progress */
             if (node->opaque != &ctx->notifier) {
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
             (revents & (G_IO_OUT | G_IO_ERR)) &&
             aio_node_check(ctx, node->is_external) &&
             node->io_write) {
+            aio_context_acquire(ctx);
             node->io_write(node->opaque);
+            aio_context_release(ctx);
             progress = true;
         }
 
@@ -XXX,XX +XXX,XX @@ bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
     }
 
     /* Run our timers */
+    aio_context_acquire(ctx);
     progress |= timerlistgroup_run_timers(&ctx->tlg);
+    aio_context_release(ctx);
 
     return progress;
 }
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
     int64_t timeout;
     int64_t start = 0;
 
-    aio_context_acquire(ctx);
-    progress = false;
-
     /* aio_notify can avoid the expensive event_notifier_set if
      * everything (file descriptors, bottom halves, timers) will
      * be re-evaluated before the next blocking poll().  This is
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         start = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
     }
 
-    if (try_poll_mode(ctx, blocking)) {
-        progress = true;
-    } else {
+    aio_context_acquire(ctx);
+    progress = try_poll_mode(ctx, blocking);
+    aio_context_release(ctx);
+
+    if (!progress) {
         assert(npfd == 0);
 
         /* fill pollfds */
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         timeout = blocking ? aio_compute_timeout(ctx) : 0;
 
         /* wait until next event */
-        if (timeout) {
-            aio_context_release(ctx);
-        }
         if (aio_epoll_check_poll(ctx, pollfds, npfd, timeout)) {
             AioHandler epoll_handler;
 
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         } else  {
             ret = qemu_poll_ns(pollfds, npfd, timeout);
         }
-        if (timeout) {
-            aio_context_acquire(ctx);
-        }
     }
 
     if (blocking) {
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         progress = true;
     }
 
-    aio_context_release(ctx);
-
     return progress;
 }
 
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
             (revents || event_notifier_get_handle(node->e) == event) &&
             node->io_notify) {
             node->pfd.revents = 0;
+            aio_context_acquire(ctx);
             node->io_notify(node->e);
+            aio_context_release(ctx);
 
             /* aio_notify() does not count as progress */
             if (node->e != &ctx->notifier) {
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
             (node->io_read || node->io_write)) {
             node->pfd.revents = 0;
             if ((revents & G_IO_IN) && node->io_read) {
+                aio_context_acquire(ctx);
                 node->io_read(node->opaque);
+                aio_context_release(ctx);
                 progress = true;
             }
             if ((revents & G_IO_OUT) && node->io_write) {
+                aio_context_acquire(ctx);
                 node->io_write(node->opaque);
+                aio_context_release(ctx);
                 progress = true;
             }
 
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
     int count;
     int timeout;
 
-    aio_context_acquire(ctx);
     progress = false;
 
     /* aio_notify can avoid the expensive event_notifier_set if
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
 
         timeout = blocking && !have_select_revents
             ? qemu_timeout_ns_to_ms(aio_compute_timeout(ctx)) : 0;
-        if (timeout) {
-            aio_context_release(ctx);
-        }
         ret = WaitForMultipleObjects(count, events, FALSE, timeout);
         if (blocking) {
             assert(first);
             atomic_sub(&ctx->notify_me, 2);
         }
-        if (timeout) {
-            aio_context_acquire(ctx);
-        }
 
         if (first) {
             aio_notify_accept(ctx);
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         progress |= aio_dispatch_handlers(ctx, event);
     } while (count > 0);
 
+    aio_context_acquire(ctx);
     progress |= timerlistgroup_run_timers(&ctx->tlg);
-
     aio_context_release(ctx);
     return progress;
 }
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
                 ret = 1;
             }
             bh->idle = 0;
+            aio_context_acquire(ctx);
             aio_bh_call(bh);
+            aio_context_release(ctx);
         }
         if (bh->deleted) {
             deleted = true;
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-13-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/qed.h                 |  3 +++
 block/curl.c                |  2 ++
 block/io.c                  |  5 +++++
 block/iscsi.c               |  8 ++++++--
 block/null.c                |  4 ++++
 block/qed.c                 | 12 ++++++++++++
 block/throttle-groups.c     |  2 ++
 util/aio-posix.c            |  2 --
 util/aio-win32.c            |  2 --
 util/qemu-coroutine-sleep.c |  2 +-
 10 files changed, 35 insertions(+), 7 deletions(-)

diff --git a/block/qed.h b/block/qed.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qed.h
+++ b/block/qed.h
@@ -XXX,XX +XXX,XX @@ enum {
  */
 typedef void QEDFindClusterFunc(void *opaque, int ret, uint64_t offset, size_t len);
 
+void qed_acquire(BDRVQEDState *s);
+void qed_release(BDRVQEDState *s);
+
 /**
  * Generic callback for chaining async callbacks
  */
diff --git a/block/curl.c b/block/curl.c
index XXXXXXX..XXXXXXX 100644
--- a/block/curl.c
+++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_multi_timeout_do(void *arg)
         return;
     }
 
+    aio_context_acquire(s->aio_context);
     curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
 
     curl_multi_check_completion(s);
+    aio_context_release(s->aio_context);
 #else
     abort();
 #endif
diff --git a/block/io.c b/block/io.c
index XXXXXXX..XXXXXXX 100644
--- a/block/io.c
+++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ void bdrv_aio_cancel(BlockAIOCB *acb)
         if (acb->aiocb_info->get_aio_context) {
             aio_poll(acb->aiocb_info->get_aio_context(acb), true);
         } else if (acb->bs) {
+            /* qemu_aio_ref and qemu_aio_unref are not thread-safe, so
+             * assert that we're not using an I/O thread.  Thread-safe
+             * code should use bdrv_aio_cancel_async exclusively.
+             */
+            assert(bdrv_get_aio_context(acb->bs) == qemu_get_aio_context());
             aio_poll(bdrv_get_aio_context(acb->bs), true);
         } else {
             abort();
diff --git a/block/iscsi.c b/block/iscsi.c
index XXXXXXX..XXXXXXX 100644
--- a/block/iscsi.c
+++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void iscsi_retry_timer_expired(void *opaque)
     struct IscsiTask *iTask = opaque;
     iTask->complete = 1;
     if (iTask->co) {
-        qemu_coroutine_enter(iTask->co);
+        aio_co_wake(iTask->co);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static void iscsi_nop_timed_event(void *opaque)
 {
     IscsiLun *iscsilun = opaque;
 
+    aio_context_acquire(iscsilun->aio_context);
     if (iscsi_get_nops_in_flight(iscsilun->iscsi) >= MAX_NOP_FAILURES) {
         error_report("iSCSI: NOP timeout. Reconnecting...");
         iscsilun->request_timed_out = true;
     } else if (iscsi_nop_out_async(iscsilun->iscsi, NULL, NULL, 0, NULL) != 0) {
         error_report("iSCSI: failed to sent NOP-Out. Disabling NOP messages.");
-        return;
+        goto out;
     }
 
     timer_mod(iscsilun->nop_timer, qemu_clock_get_ms(QEMU_CLOCK_REALTIME) + NOP_INTERVAL);
     iscsi_set_events(iscsilun);
+
+out:
+    aio_context_release(iscsilun->aio_context);
 }
 
 static void iscsi_readcapacity_sync(IscsiLun *iscsilun, Error **errp)
diff --git a/block/null.c b/block/null.c
index XXXXXXX..XXXXXXX 100644
--- a/block/null.c
+++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static void null_bh_cb(void *opaque)
 static void null_timer_cb(void *opaque)
 {
     NullAIOCB *acb = opaque;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
+
+    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, 0);
+    aio_context_release(ctx);
     timer_deinit(&acb->timer);
     qemu_aio_unref(acb);
 }
diff --git a/block/qed.c b/block/qed.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_need_check_timer_cb(void *opaque)
 
     trace_qed_need_check_timer_cb(s);
 
+    qed_acquire(s);
     qed_plug_allocating_write_reqs(s);
 
     /* Ensure writes are on disk before clearing flag */
     bdrv_aio_flush(s->bs->file->bs, qed_clear_need_check, s);
+    qed_release(s);
+}
+
+void qed_acquire(BDRVQEDState *s)
+{
+    aio_context_acquire(bdrv_get_aio_context(s->bs));
+}
+
+void qed_release(BDRVQEDState *s)
+{
+    aio_context_release(bdrv_get_aio_context(s->bs));
 }
 
 static void qed_start_need_check_timer(BDRVQEDState *s)
diff --git a/block/throttle-groups.c b/block/throttle-groups.c
index XXXXXXX..XXXXXXX 100644
--- a/block/throttle-groups.c
+++ b/block/throttle-groups.c
@@ -XXX,XX +XXX,XX @@ static void timer_cb(BlockBackend *blk, bool is_write)
     qemu_mutex_unlock(&tg->lock);
 
     /* Run the request that was waiting for this timer */
+    aio_context_acquire(blk_get_aio_context(blk));
     empty_queue = !qemu_co_enter_next(&blkp->throttled_reqs[is_write]);
+    aio_context_release(blk_get_aio_context(blk));
 
     /* If the request queue was empty then we have to take care of
      * scheduling the next one */
diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
     }
 
     /* Run our timers */
-    aio_context_acquire(ctx);
     progress |= timerlistgroup_run_timers(&ctx->tlg);
-    aio_context_release(ctx);
 
     return progress;
 }
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         progress |= aio_dispatch_handlers(ctx, event);
     } while (count > 0);
 
-    aio_context_acquire(ctx);
     progress |= timerlistgroup_run_timers(&ctx->tlg);
-    aio_context_release(ctx);
     return progress;
 }
 
diff --git a/util/qemu-coroutine-sleep.c b/util/qemu-coroutine-sleep.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-sleep.c
+++ b/util/qemu-coroutine-sleep.c
@@ -XXX,XX +XXX,XX @@ static void co_sleep_cb(void *opaque)
 {
     CoSleepCB *sleep_cb = opaque;
 
-    qemu_coroutine_enter(sleep_cb->co);
+    aio_co_wake(sleep_cb->co);
 }
 
 void coroutine_fn co_aio_sleep_ns(AioContext *ctx, QEMUClockType type,
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This covers both file descriptor callbacks and polling callbacks,
since they execute related code.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-14-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/curl.c          | 16 +++++++++++++---
 block/iscsi.c         |  4 ++++
 block/linux-aio.c     |  4 ++++
 block/nfs.c           |  6 ++++++
 block/sheepdog.c      | 29 +++++++++++++++--------------
 block/ssh.c           | 29 +++++++++--------------------
 block/win32-aio.c     | 10 ++++++----
 hw/block/virtio-blk.c |  5 ++++-
 hw/scsi/virtio-scsi.c |  7 +++++++
 util/aio-posix.c      |  7 -------
 util/aio-win32.c      |  6 ------
 11 files changed, 68 insertions(+), 55 deletions(-)

diff --git a/block/curl.c b/block/curl.c
index XXXXXXX..XXXXXXX 100644
--- a/block/curl.c
+++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_multi_check_completion(BDRVCURLState *s)
     }
 }
 
-static void curl_multi_do(void *arg)
+static void curl_multi_do_locked(CURLState *s)
 {
-    CURLState *s = (CURLState *)arg;
     CURLSocket *socket, *next_socket;
     int running;
     int r;
@@ -XXX,XX +XXX,XX @@ static void curl_multi_do(void *arg)
     }
 }
 
+static void curl_multi_do(void *arg)
+{
+    CURLState *s = (CURLState *)arg;
+
+    aio_context_acquire(s->s->aio_context);
+    curl_multi_do_locked(s);
+    aio_context_release(s->s->aio_context);
+}
+
 static void curl_multi_read(void *arg)
 {
     CURLState *s = (CURLState *)arg;
 
-    curl_multi_do(arg);
+    aio_context_acquire(s->s->aio_context);
+    curl_multi_do_locked(s);
     curl_multi_check_completion(s->s);
+    aio_context_release(s->s->aio_context);
 }
 
 static void curl_multi_timeout_do(void *arg)
diff --git a/block/iscsi.c b/block/iscsi.c
index XXXXXXX..XXXXXXX 100644
--- a/block/iscsi.c
+++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ iscsi_process_read(void *arg)
     IscsiLun *iscsilun = arg;
     struct iscsi_context *iscsi = iscsilun->iscsi;
 
+    aio_context_acquire(iscsilun->aio_context);
     iscsi_service(iscsi, POLLIN);
     iscsi_set_events(iscsilun);
+    aio_context_release(iscsilun->aio_context);
 }
 
 static void
@@ -XXX,XX +XXX,XX @@ iscsi_process_write(void *arg)
     IscsiLun *iscsilun = arg;
     struct iscsi_context *iscsi = iscsilun->iscsi;
 
+    aio_context_acquire(iscsilun->aio_context);
     iscsi_service(iscsi, POLLOUT);
     iscsi_set_events(iscsilun);
+    aio_context_release(iscsilun->aio_context);
 }
 
 static int64_t sector_lun2qemu(int64_t sector, IscsiLun *iscsilun)
diff --git a/block/linux-aio.c b/block/linux-aio.c
index XXXXXXX..XXXXXXX 100644
--- a/block/linux-aio.c
+++ b/block/linux-aio.c
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_completion_cb(EventNotifier *e)
     LinuxAioState *s = container_of(e, LinuxAioState, e);
 
     if (event_notifier_test_and_clear(&s->e)) {
+        aio_context_acquire(s->aio_context);
         qemu_laio_process_completions_and_submit(s);
+        aio_context_release(s->aio_context);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static bool qemu_laio_poll_cb(void *opaque)
         return false;
     }
 
+    aio_context_acquire(s->aio_context);
     qemu_laio_process_completions_and_submit(s);
+    aio_context_release(s->aio_context);
     return true;
 }
 
diff --git a/block/nfs.c b/block/nfs.c
index XXXXXXX..XXXXXXX 100644
--- a/block/nfs.c
+++ b/block/nfs.c
@@ -XXX,XX +XXX,XX @@ static void nfs_set_events(NFSClient *client)
 static void nfs_process_read(void *arg)
 {
     NFSClient *client = arg;
+
+    aio_context_acquire(client->aio_context);
     nfs_service(client->context, POLLIN);
     nfs_set_events(client);
+    aio_context_release(client->aio_context);
 }
 
 static void nfs_process_write(void *arg)
 {
     NFSClient *client = arg;
+
+    aio_context_acquire(client->aio_context);
     nfs_service(client->context, POLLOUT);
     nfs_set_events(client);
+    aio_context_release(client->aio_context);
 }
 
 static void nfs_co_init_task(BlockDriverState *bs, NFSRPC *task)
diff --git a/block/sheepdog.c b/block/sheepdog.c
index XXXXXXX..XXXXXXX 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int send_co_req(int sockfd, SheepdogReq *hdr, void *data,
     return ret;
 }
 
-static void restart_co_req(void *opaque)
-{
-    Coroutine *co = opaque;
-
-    qemu_coroutine_enter(co);
-}
-
 typedef struct SheepdogReqCo {
     int sockfd;
     BlockDriverState *bs;
@@ -XXX,XX +XXX,XX @@ typedef struct SheepdogReqCo {
     unsigned int *rlen;
     int ret;
     bool finished;
+    Coroutine *co;
 } SheepdogReqCo;
 
+static void restart_co_req(void *opaque)
+{
+    SheepdogReqCo *srco = opaque;
+
+    aio_co_wake(srco->co);
+}
+
 static coroutine_fn void do_co_req(void *opaque)
 {
     int ret;
-    Coroutine *co;
     SheepdogReqCo *srco = opaque;
     int sockfd = srco->sockfd;
     SheepdogReq *hdr = srco->hdr;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void do_co_req(void *opaque)
     unsigned int *wlen = srco->wlen;
     unsigned int *rlen = srco->rlen;
 
-    co = qemu_coroutine_self();
+    srco->co = qemu_coroutine_self();
     aio_set_fd_handler(srco->aio_context, sockfd, false,
-                       NULL, restart_co_req, NULL, co);
+                       NULL, restart_co_req, NULL, srco);
 
     ret = send_co_req(sockfd, hdr, data, wlen);
     if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void do_co_req(void *opaque)
     }
 
     aio_set_fd_handler(srco->aio_context, sockfd, false,
-                       restart_co_req, NULL, NULL, co);
+                       restart_co_req, NULL, NULL, srco);
 
     ret = qemu_co_recv(sockfd, hdr, sizeof(*hdr));
     if (ret != sizeof(*hdr)) {
@@ -XXX,XX +XXX,XX @@ out:
     aio_set_fd_handler(srco->aio_context, sockfd, false,
                        NULL, NULL, NULL, NULL);
 
+    srco->co = NULL;
     srco->ret = ret;
     srco->finished = true;
     if (srco->bs) {
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn aio_read_response(void *opaque)
          * We've finished all requests which belong to the AIOCB, so
          * we can switch back to sd_co_readv/writev now.
          */
-        qemu_coroutine_enter(acb->coroutine);
+        aio_co_wake(acb->coroutine);
     }
 
     return;
@@ -XXX,XX +XXX,XX @@ static void co_read_response(void *opaque)
         s->co_recv = qemu_coroutine_create(aio_read_response, opaque);
     }
 
-    qemu_coroutine_enter(s->co_recv);
+    aio_co_wake(s->co_recv);
 }
 
 static void co_write_request(void *opaque)
 {
     BDRVSheepdogState *s = opaque;
 
-    qemu_coroutine_enter(s->co_send);
+    aio_co_wake(s->co_send);
 }
 
 /*
diff --git a/block/ssh.c b/block/ssh.c
index XXXXXXX..XXXXXXX 100644
--- a/block/ssh.c
+++ b/block/ssh.c
@@ -XXX,XX +XXX,XX @@ static void restart_coroutine(void *opaque)
 
     DPRINTF("co=%p", co);
 
-    qemu_coroutine_enter(co);
+    aio_co_wake(co);
 }
 
-static coroutine_fn void set_fd_handler(BDRVSSHState *s, BlockDriverState *bs)
+/* A non-blocking call returned EAGAIN, so yield, ensuring the
+ * handlers are set up so that we'll be rescheduled when there is an
+ * interesting event on the socket.
+ */
+static coroutine_fn void co_yield(BDRVSSHState *s, BlockDriverState *bs)
 {
     int r;
     IOHandler *rd_handler = NULL, *wr_handler = NULL;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn void set_fd_handler(BDRVSSHState *s, BlockDriverState *bs)
 
     aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
                        false, rd_handler, wr_handler, NULL, co);
-}
-
-static coroutine_fn void clear_fd_handler(BDRVSSHState *s,
-                                          BlockDriverState *bs)
-{
-    DPRINTF("s->sock=%d", s->sock);
-    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
-                       false, NULL, NULL, NULL, NULL);
-}
-
-/* A non-blocking call returned EAGAIN, so yield, ensuring the
- * handlers are set up so that we'll be rescheduled when there is an
- * interesting event on the socket.
- */
-static coroutine_fn void co_yield(BDRVSSHState *s, BlockDriverState *bs)
-{
-    set_fd_handler(s, bs);
     qemu_coroutine_yield();
-    clear_fd_handler(s, bs);
+    DPRINTF("s->sock=%d - back", s->sock);
+    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock, false,
+                       NULL, NULL, NULL, NULL);
 }
 
 /* SFTP has a function `libssh2_sftp_seek64' which seeks to a position
diff --git a/block/win32-aio.c b/block/win32-aio.c
index XXXXXXX..XXXXXXX 100644
--- a/block/win32-aio.c
+++ b/block/win32-aio.c
@@ -XXX,XX +XXX,XX @@ struct QEMUWin32AIOState {
     HANDLE hIOCP;
     EventNotifier e;
     int count;
-    bool is_aio_context_attached;
+    AioContext *aio_ctx;
 };
 
 typedef struct QEMUWin32AIOCB {
@@ -XXX,XX +XXX,XX @@ static void win32_aio_process_completion(QEMUWin32AIOState *s,
     }
 
 
+    aio_context_acquire(s->aio_ctx);
     waiocb->common.cb(waiocb->common.opaque, ret);
+    aio_context_release(s->aio_ctx);
     qemu_aio_unref(waiocb);
 }
 
@@ -XXX,XX +XXX,XX @@ void win32_aio_detach_aio_context(QEMUWin32AIOState *aio,
                                   AioContext *old_context)
 {
     aio_set_event_notifier(old_context, &aio->e, false, NULL, NULL);
-    aio->is_aio_context_attached = false;
+    aio->aio_ctx = NULL;
 }
 
 void win32_aio_attach_aio_context(QEMUWin32AIOState *aio,
                                   AioContext *new_context)
 {
-    aio->is_aio_context_attached = true;
+    aio->aio_ctx = new_context;
     aio_set_event_notifier(new_context, &aio->e, false,
                            win32_aio_completion_cb, NULL);
 }
@@ -XXX,XX +XXX,XX @@ out_free_state:
 
 void win32_aio_cleanup(QEMUWin32AIOState *aio)
 {
-    assert(!aio->is_aio_context_attached);
+    assert(!aio->aio_ctx);
     CloseHandle(aio->hIOCP);
     event_notifier_cleanup(&aio->e);
     g_free(aio);
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_ioctl_complete(void *opaque, int status)
 {
     VirtIOBlockIoctlReq *ioctl_req = opaque;
     VirtIOBlockReq *req = ioctl_req->req;
-    VirtIODevice *vdev = VIRTIO_DEVICE(req->dev);
+    VirtIOBlock *s = req->dev;
+    VirtIODevice *vdev = VIRTIO_DEVICE(s);
     struct virtio_scsi_inhdr *scsi;
     struct sg_io_hdr *hdr;
 
@@ -XXX,XX +XXX,XX @@ bool virtio_blk_handle_vq(VirtIOBlock *s, VirtQueue *vq)
     MultiReqBuffer mrb = {};
     bool progress = false;
 
+    aio_context_acquire(blk_get_aio_context(s->blk));
     blk_io_plug(s->blk);
 
     do {
@@ -XXX,XX +XXX,XX @@ bool virtio_blk_handle_vq(VirtIOBlock *s, VirtQueue *vq)
     }
 
     blk_io_unplug(s->blk);
+    aio_context_release(blk_get_aio_context(s->blk));
     return progress;
 }
 
diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/scsi/virtio-scsi.c
+++ b/hw/scsi/virtio-scsi.c
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_ctrl_vq(VirtIOSCSI *s, VirtQueue *vq)
     VirtIOSCSIReq *req;
     bool progress = false;
 
+    virtio_scsi_acquire(s);
     while ((req = virtio_scsi_pop_req(s, vq))) {
         progress = true;
         virtio_scsi_handle_ctrl_req(s, req);
     }
+    virtio_scsi_release(s);
     return progress;
 }
 
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_cmd_vq(VirtIOSCSI *s, VirtQueue *vq)
 
     QTAILQ_HEAD(, VirtIOSCSIReq) reqs = QTAILQ_HEAD_INITIALIZER(reqs);
 
+    virtio_scsi_acquire(s);
     do {
         virtio_queue_set_notification(vq, 0);
 
@@ -XXX,XX +XXX,XX @@ bool virtio_scsi_handle_cmd_vq(VirtIOSCSI *s, VirtQueue *vq)
     QTAILQ_FOREACH_SAFE(req, &reqs, next, next) {
         virtio_scsi_handle_cmd_req_submit(s, req);
     }
+    virtio_scsi_release(s);
     return progress;
 }
 
@@ -XXX,XX +XXX,XX @@ out:
 
 bool virtio_scsi_handle_event_vq(VirtIOSCSI *s, VirtQueue *vq)
 {
+    virtio_scsi_acquire(s);
     if (s->events_dropped) {
         virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
+        virtio_scsi_release(s);
         return true;
     }
+    virtio_scsi_release(s);
     return false;
 }
 
diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
             (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
             aio_node_check(ctx, node->is_external) &&
             node->io_read) {
-            aio_context_acquire(ctx);
             node->io_read(node->opaque);
-            aio_context_release(ctx);
 
             /* aio_notify() does not count as progress */
             if (node->opaque != &ctx->notifier) {
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
             (revents & (G_IO_OUT | G_IO_ERR)) &&
             aio_node_check(ctx, node->is_external) &&
             node->io_write) {
-            aio_context_acquire(ctx);
             node->io_write(node->opaque);
-            aio_context_release(ctx);
             progress = true;
         }
 
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         start = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
     }
 
-    aio_context_acquire(ctx);
     progress = try_poll_mode(ctx, blocking);
-    aio_context_release(ctx);
-
     if (!progress) {
         assert(npfd == 0);
 
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
             (revents || event_notifier_get_handle(node->e) == event) &&
             node->io_notify) {
             node->pfd.revents = 0;
-            aio_context_acquire(ctx);
             node->io_notify(node->e);
-            aio_context_release(ctx);
 
             /* aio_notify() does not count as progress */
             if (node->e != &ctx->notifier) {
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
             (node->io_read || node->io_write)) {
             node->pfd.revents = 0;
             if ((revents & G_IO_IN) && node->io_read) {
-                aio_context_acquire(ctx);
                 node->io_read(node->opaque);
-                aio_context_release(ctx);
                 progress = true;
             }
             if ((revents & G_IO_OUT) && node->io_write) {
-                aio_context_acquire(ctx);
                 node->io_write(node->opaque);
-                aio_context_release(ctx);
                 progress = true;
             }
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-15-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/archipelago.c   |  3 +++
 block/blkreplay.c     |  2 +-
 block/block-backend.c |  6 ++++++
 block/curl.c          | 26 ++++++++++++++++++--------
 block/gluster.c       |  9 +--------
 block/io.c            |  6 +++++-
 block/iscsi.c         |  6 +++++-
 block/linux-aio.c     | 15 +++++++++------
 block/nfs.c           |  3 ++-
 block/null.c          |  4 ++++
 block/qed.c           |  3 +++
 block/rbd.c           |  4 ++++
 dma-helpers.c         |  2 ++
 hw/block/virtio-blk.c |  2 ++
 hw/scsi/scsi-bus.c    |  2 ++
 util/async.c          |  4 ++--
 util/thread-pool.c    |  2 ++
 17 files changed, 71 insertions(+), 28 deletions(-)

diff --git a/block/archipelago.c b/block/archipelago.c
index XXXXXXX..XXXXXXX 100644
--- a/block/archipelago.c
+++ b/block/archipelago.c
@@ -XXX,XX +XXX,XX @@ static void qemu_archipelago_complete_aio(void *opaque)
 {
     AIORequestData *reqdata = (AIORequestData *) opaque;
     ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
+    AioContext *ctx = bdrv_get_aio_context(aio_cb->common.bs);
 
+    aio_context_acquire(ctx);
     aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
+    aio_context_release(ctx);
     aio_cb->status = 0;
 
     qemu_aio_unref(aio_cb);
diff --git a/block/blkreplay.c b/block/blkreplay.c
index XXXXXXX..XXXXXXX 100755
--- a/block/blkreplay.c
+++ b/block/blkreplay.c
@@ -XXX,XX +XXX,XX @@ static int64_t blkreplay_getlength(BlockDriverState *bs)
 static void blkreplay_bh_cb(void *opaque)
 {
     Request *req = opaque;
-    qemu_coroutine_enter(req->co);
+    aio_co_wake(req->co);
     qemu_bh_delete(req->bh);
     g_free(req);
 }
diff --git a/block/block-backend.c b/block/block-backend.c
index XXXXXXX..XXXXXXX 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -XXX,XX +XXX,XX @@ int blk_make_zero(BlockBackend *blk, BdrvRequestFlags flags)
 static void error_callback_bh(void *opaque)
 {
     struct BlockBackendAIOCB *acb = opaque;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 
     bdrv_dec_in_flight(acb->common.bs);
+    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, acb->ret);
+    aio_context_release(ctx);
     qemu_aio_unref(acb);
 }
 
@@ -XXX,XX +XXX,XX @@ static void blk_aio_complete(BlkAioEmAIOCB *acb)
 static void blk_aio_complete_bh(void *opaque)
 {
     BlkAioEmAIOCB *acb = opaque;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 
     assert(acb->has_returned);
+    aio_context_acquire(ctx);
     blk_aio_complete(acb);
+    aio_context_release(ctx);
 }
 
 static BlockAIOCB *blk_aio_prwv(BlockBackend *blk, int64_t offset, int bytes,
diff --git a/block/curl.c b/block/curl.c
index XXXXXXX..XXXXXXX 100644
--- a/block/curl.c
+++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
 {
     CURLState *state;
     int running;
+    int ret = -EINPROGRESS;
 
     CURLAIOCB *acb = p;
-    BDRVCURLState *s = acb->common.bs->opaque;
+    BlockDriverState *bs = acb->common.bs;
+    BDRVCURLState *s = bs->opaque;
+    AioContext *ctx = bdrv_get_aio_context(bs);
 
     size_t start = acb->sector_num * BDRV_SECTOR_SIZE;
     size_t end;
 
+    aio_context_acquire(ctx);
+
     // In case we have the requested data already (e.g. read-ahead),
     // we can just call the callback and be done.
     switch (curl_find_buf(s, start, acb->nb_sectors * BDRV_SECTOR_SIZE, acb)) {
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
             qemu_aio_unref(acb);
             // fall through
         case FIND_RET_WAIT:
-            return;
+            goto out;
         default:
             break;
     }
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
     // No cache found, so let's start a new request
     state = curl_init_state(acb->common.bs, s);
     if (!state) {
-        acb->common.cb(acb->common.opaque, -EIO);
-        qemu_aio_unref(acb);
-        return;
+        ret = -EIO;
+        goto out;
     }
 
     acb->start = 0;
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
     state->orig_buf = g_try_malloc(state->buf_len);
     if (state->buf_len && state->orig_buf == NULL) {
         curl_clean_state(state);
-        acb->common.cb(acb->common.opaque, -ENOMEM);
-        qemu_aio_unref(acb);
-        return;
+        ret = -ENOMEM;
+        goto out;
     }
     state->acb[0] = acb;
 
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
 
     /* Tell curl it needs to kick things off */
     curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
+
+out:
+    if (ret != -EINPROGRESS) {
+        acb->common.cb(acb->common.opaque, ret);
+        qemu_aio_unref(acb);
+    }
+    aio_context_release(ctx);
 }
 
 static BlockAIOCB *curl_aio_readv(BlockDriverState *bs,
diff --git a/block/gluster.c b/block/gluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/gluster.c
+++ b/block/gluster.c
@@ -XXX,XX +XXX,XX @@ static struct glfs *qemu_gluster_init(BlockdevOptionsGluster *gconf,
     return qemu_gluster_glfs_init(gconf, errp);
 }
 
-static void qemu_gluster_complete_aio(void *opaque)
-{
-    GlusterAIOCB *acb = (GlusterAIOCB *)opaque;
-
-    qemu_coroutine_enter(acb->coroutine);
-}
-
 /*
  * AIO callback routine called from GlusterFS thread.
  */
@@ -XXX,XX +XXX,XX @@ static void gluster_finish_aiocb(struct glfs_fd *fd, ssize_t ret, void *arg)
         acb->ret = -EIO; /* Partial read/write - fail it */
     }
 
-    aio_bh_schedule_oneshot(acb->aio_context, qemu_gluster_complete_aio, acb);
+    aio_co_schedule(acb->aio_context, acb->coroutine);
 }
 
 static void qemu_gluster_parse_flags(int bdrv_flags, int *open_flags)
diff --git a/block/io.c b/block/io.c
index XXXXXXX..XXXXXXX 100644
--- a/block/io.c
+++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_drain_bh_cb(void *opaque)
     bdrv_dec_in_flight(bs);
     bdrv_drained_begin(bs);
     data->done = true;
-    qemu_coroutine_enter(co);
+    aio_co_wake(co);
 }
 
 static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs)
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_complete(BlockAIOCBCoroutine *acb)
 static void bdrv_co_em_bh(void *opaque)
 {
     BlockAIOCBCoroutine *acb = opaque;
+    BlockDriverState *bs = acb->common.bs;
+    AioContext *ctx = bdrv_get_aio_context(bs);
 
     assert(!acb->need_bh);
+    aio_context_acquire(ctx);
     bdrv_co_complete(acb);
+    aio_context_release(ctx);
 }
 
 static void bdrv_co_maybe_schedule_bh(BlockAIOCBCoroutine *acb)
diff --git a/block/iscsi.c b/block/iscsi.c
index XXXXXXX..XXXXXXX 100644
--- a/block/iscsi.c
+++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void
 iscsi_bh_cb(void *p)
 {
     IscsiAIOCB *acb = p;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 
     qemu_bh_delete(acb->bh);
 
     g_free(acb->buf);
     acb->buf = NULL;
 
+    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, acb->status);
+    aio_context_release(ctx);
 
     if (acb->task != NULL) {
         scsi_free_scsi_task(acb->task);
@@ -XXX,XX +XXX,XX @@ iscsi_schedule_bh(IscsiAIOCB *acb)
 static void iscsi_co_generic_bh_cb(void *opaque)
 {
     struct IscsiTask *iTask = opaque;
+
     iTask->complete = 1;
-    qemu_coroutine_enter(iTask->co);
+    aio_co_wake(iTask->co);
 }
 
 static void iscsi_retry_timer_expired(void *opaque)
diff --git a/block/linux-aio.c b/block/linux-aio.c
index XXXXXXX..XXXXXXX 100644
--- a/block/linux-aio.c
+++ b/block/linux-aio.c
@@ -XXX,XX +XXX,XX @@ struct LinuxAioState {
     io_context_t ctx;
     EventNotifier e;
 
-    /* io queue for submit at batch */
+    /* io queue for submit at batch.  Protected by AioContext lock. */
     LaioQueue io_q;
 
-    /* I/O completion processing */
+    /* I/O completion processing.  Only runs in I/O thread.  */
     QEMUBH *completion_bh;
     int event_idx;
     int event_max;
@@ -XXX,XX +XXX,XX @@ static inline ssize_t io_event_ret(struct io_event *ev)
  */
 static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
 {
+    LinuxAioState *s = laiocb->ctx;
     int ret;
 
     ret = laiocb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
     }
 
     laiocb->ret = ret;
+    aio_context_acquire(s->aio_context);
     if (laiocb->co) {
         /* If the coroutine is already entered it must be in ioq_submit() and
          * will notice laio->ret has been filled in when it eventually runs
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
         laiocb->common.cb(laiocb->common.opaque, ret);
         qemu_aio_unref(laiocb);
     }
+    aio_context_release(s->aio_context);
 }
 
 /**
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completions(LinuxAioState *s)
 static void qemu_laio_process_completions_and_submit(LinuxAioState *s)
 {
     qemu_laio_process_completions(s);
+
+    aio_context_acquire(s->aio_context);
     if (!s->io_q.plugged && !QSIMPLEQ_EMPTY(&s->io_q.pending)) {
         ioq_submit(s);
     }
+    aio_context_release(s->aio_context);
 }
 
 static void qemu_laio_completion_bh(void *opaque)
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_completion_cb(EventNotifier *e)
     LinuxAioState *s = container_of(e, LinuxAioState, e);
 
     if (event_notifier_test_and_clear(&s->e)) {
-        aio_context_acquire(s->aio_context);
         qemu_laio_process_completions_and_submit(s);
-        aio_context_release(s->aio_context);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static bool qemu_laio_poll_cb(void *opaque)
         return false;
     }
 
-    aio_context_acquire(s->aio_context);
     qemu_laio_process_completions_and_submit(s);
-    aio_context_release(s->aio_context);
     return true;
 }
 
@@ -XXX,XX +XXX,XX @@ void laio_detach_aio_context(LinuxAioState *s, AioContext *old_context)
 {
     aio_set_event_notifier(old_context, &s->e, false, NULL, NULL);
     qemu_bh_delete(s->completion_bh);
+    s->aio_context = NULL;
 }
 
 void laio_attach_aio_context(LinuxAioState *s, AioContext *new_context)
diff --git a/block/nfs.c b/block/nfs.c
index XXXXXXX..XXXXXXX 100644
--- a/block/nfs.c
+++ b/block/nfs.c
@@ -XXX,XX +XXX,XX @@ static void nfs_co_init_task(BlockDriverState *bs, NFSRPC *task)
 static void nfs_co_generic_bh_cb(void *opaque)
 {
     NFSRPC *task = opaque;
+
     task->complete = 1;
-    qemu_coroutine_enter(task->co);
+    aio_co_wake(task->co);
 }
 
 static void
diff --git a/block/null.c b/block/null.c
index XXXXXXX..XXXXXXX 100644
--- a/block/null.c
+++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static const AIOCBInfo null_aiocb_info = {
 static void null_bh_cb(void *opaque)
 {
     NullAIOCB *acb = opaque;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
+
+    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, 0);
+    aio_context_release(ctx);
     qemu_aio_unref(acb);
 }
 
diff --git a/block/qed.c b/block/qed.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_update_l2_table(BDRVQEDState *s, QEDTable *table, int index,
 static void qed_aio_complete_bh(void *opaque)
 {
     QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
     BlockCompletionFunc *cb = acb->common.cb;
     void *user_opaque = acb->common.opaque;
     int ret = acb->bh_ret;
@@ -XXX,XX +XXX,XX @@ static void qed_aio_complete_bh(void *opaque)
     qemu_aio_unref(acb);
 
     /* Invoke callback */
+    qed_acquire(s);
     cb(user_opaque, ret);
+    qed_release(s);
 }
 
 static void qed_aio_complete(QEDAIOCB *acb, int ret)
diff --git a/block/rbd.c b/block/rbd.c
index XXXXXXX..XXXXXXX 100644
--- a/block/rbd.c
+++ b/block/rbd.c
@@ -XXX,XX +XXX,XX @@ shutdown:
 static void qemu_rbd_complete_aio(RADOSCB *rcb)
 {
     RBDAIOCB *acb = rcb->acb;
+    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
     int64_t r;
 
     r = rcb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
         qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
     }
     qemu_vfree(acb->bounce);
+
+    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
+    aio_context_release(ctx);
 
     qemu_aio_unref(acb);
 }
diff --git a/dma-helpers.c b/dma-helpers.c
index XXXXXXX..XXXXXXX 100644
--- a/dma-helpers.c
+++ b/dma-helpers.c
@@ -XXX,XX +XXX,XX @@ static void dma_blk_cb(void *opaque, int ret)
                                 QEMU_ALIGN_DOWN(dbs->iov.size, dbs->align));
     }
 
+    aio_context_acquire(dbs->ctx);
     dbs->acb = dbs->io_func(dbs->offset, &dbs->iov,
                             dma_blk_cb, dbs, dbs->io_func_opaque);
+    aio_context_release(dbs->ctx);
     assert(dbs->acb);
 }
 
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_dma_restart_bh(void *opaque)
 
     s->rq = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
     while (req) {
         VirtIOBlockReq *next = req->next;
         if (virtio_blk_handle_request(req, &mrb)) {
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_dma_restart_bh(void *opaque)
     if (mrb.num_reqs) {
         virtio_blk_submit_multireq(s->blk, &mrb);
     }
+    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
 }
 
 static void virtio_blk_dma_restart_cb(void *opaque, int running,
diff --git a/hw/scsi/scsi-bus.c b/hw/scsi/scsi-bus.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/scsi/scsi-bus.c
+++ b/hw/scsi/scsi-bus.c
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_restart_bh(void *opaque)
     qemu_bh_delete(s->bh);
     s->bh = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.blk));
     QTAILQ_FOREACH_SAFE(req, &s->requests, next, next) {
         scsi_req_ref(req);
         if (req->retry) {
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_restart_bh(void *opaque)
         }
         scsi_req_unref(req);
     }
+    aio_context_release(blk_get_aio_context(s->conf.blk));
 }
 
 void scsi_req_retry(SCSIRequest *req)
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
                 ret = 1;
             }
             bh->idle = 0;
-            aio_context_acquire(ctx);
             aio_bh_call(bh);
-            aio_context_release(ctx);
         }
         if (bh->deleted) {
             deleted = true;
@@ -XXX,XX +XXX,XX @@ static void co_schedule_bh_cb(void *opaque)
         Coroutine *co = QSLIST_FIRST(&straight);
         QSLIST_REMOVE_HEAD(&straight, co_scheduled_next);
         trace_aio_co_schedule_bh_cb(ctx, co);
+        aio_context_acquire(ctx);
         qemu_coroutine_enter(co);
+        aio_context_release(ctx);
     }
 }
 
diff --git a/util/thread-pool.c b/util/thread-pool.c
index XXXXXXX..XXXXXXX 100644
--- a/util/thread-pool.c
+++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@ static void thread_pool_completion_bh(void *opaque)
     ThreadPool *pool = opaque;
     ThreadPoolElement *elem, *next;
 
+    aio_context_acquire(pool->ctx);
 restart:
     QLIST_FOREACH_SAFE(elem, &pool->head, all, next) {
         if (elem->state != THREAD_DONE) {
@@ -XXX,XX +XXX,XX @@ restart:
             qemu_aio_unref(elem);
         }
     }
+    aio_context_release(pool->ctx);
 }
 
 static void thread_pool_cancel(BlockAIOCB *acb)
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-16-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/archipelago.c    |  3 ---
 block/block-backend.c  |  7 -------
 block/curl.c           |  2 +-
 block/io.c             |  6 +-----
 block/iscsi.c          |  3 ---
 block/linux-aio.c      |  5 +----
 block/mirror.c         | 12 +++++++++---
 block/null.c           |  8 --------
 block/qed-cluster.c    |  2 ++
 block/qed-table.c      | 12 ++++++++++--
 block/qed.c            |  4 ++--
 block/rbd.c            |  4 ----
 block/win32-aio.c      |  3 ---
 hw/block/virtio-blk.c  | 12 +++++++++++-
 hw/scsi/scsi-disk.c    | 15 +++++++++++++++
 hw/scsi/scsi-generic.c | 20 +++++++++++++++++---
 util/thread-pool.c     |  4 +++-
 17 files changed, 72 insertions(+), 50 deletions(-)

diff --git a/block/archipelago.c b/block/archipelago.c
index XXXXXXX..XXXXXXX 100644
--- a/block/archipelago.c
+++ b/block/archipelago.c
@@ -XXX,XX +XXX,XX @@ static void qemu_archipelago_complete_aio(void *opaque)
 {
     AIORequestData *reqdata = (AIORequestData *) opaque;
     ArchipelagoAIOCB *aio_cb = (ArchipelagoAIOCB *) reqdata->aio_cb;
-    AioContext *ctx = bdrv_get_aio_context(aio_cb->common.bs);
 
-    aio_context_acquire(ctx);
     aio_cb->common.cb(aio_cb->common.opaque, aio_cb->ret);
-    aio_context_release(ctx);
     aio_cb->status = 0;
 
     qemu_aio_unref(aio_cb);
diff --git a/block/block-backend.c b/block/block-backend.c
index XXXXXXX..XXXXXXX 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -XXX,XX +XXX,XX @@ int blk_make_zero(BlockBackend *blk, BdrvRequestFlags flags)
 static void error_callback_bh(void *opaque)
 {
     struct BlockBackendAIOCB *acb = opaque;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 
     bdrv_dec_in_flight(acb->common.bs);
-    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, acb->ret);
-    aio_context_release(ctx);
     qemu_aio_unref(acb);
 }
 
@@ -XXX,XX +XXX,XX @@ static void blk_aio_complete(BlkAioEmAIOCB *acb)
 static void blk_aio_complete_bh(void *opaque)
 {
     BlkAioEmAIOCB *acb = opaque;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
-
     assert(acb->has_returned);
-    aio_context_acquire(ctx);
     blk_aio_complete(acb);
-    aio_context_release(ctx);
 }
 
 static BlockAIOCB *blk_aio_prwv(BlockBackend *blk, int64_t offset, int bytes,
diff --git a/block/curl.c b/block/curl.c
index XXXXXXX..XXXXXXX 100644
--- a/block/curl.c
+++ b/block/curl.c
@@ -XXX,XX +XXX,XX @@ static void curl_readv_bh_cb(void *p)
     curl_multi_socket_action(s->multi, CURL_SOCKET_TIMEOUT, 0, &running);
 
 out:
+    aio_context_release(ctx);
     if (ret != -EINPROGRESS) {
         acb->common.cb(acb->common.opaque, ret);
         qemu_aio_unref(acb);
     }
-    aio_context_release(ctx);
 }
 
 static BlockAIOCB *curl_aio_readv(BlockDriverState *bs,
diff --git a/block/io.c b/block/io.c
index XXXXXXX..XXXXXXX 100644
--- a/block/io.c
+++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_io_em_complete(void *opaque, int ret)
     CoroutineIOCompletion *co = opaque;
 
     co->ret = ret;
-    qemu_coroutine_enter(co->coroutine);
+    aio_co_wake(co->coroutine);
 }
 
 static int coroutine_fn bdrv_driver_preadv(BlockDriverState *bs,
@@ -XXX,XX +XXX,XX @@ static void bdrv_co_complete(BlockAIOCBCoroutine *acb)
 static void bdrv_co_em_bh(void *opaque)
 {
     BlockAIOCBCoroutine *acb = opaque;
-    BlockDriverState *bs = acb->common.bs;
-    AioContext *ctx = bdrv_get_aio_context(bs);
 
     assert(!acb->need_bh);
-    aio_context_acquire(ctx);
     bdrv_co_complete(acb);
-    aio_context_release(ctx);
 }
 
 static void bdrv_co_maybe_schedule_bh(BlockAIOCBCoroutine *acb)
diff --git a/block/iscsi.c b/block/iscsi.c
index XXXXXXX..XXXXXXX 100644
--- a/block/iscsi.c
+++ b/block/iscsi.c
@@ -XXX,XX +XXX,XX @@ static void
 iscsi_bh_cb(void *p)
 {
     IscsiAIOCB *acb = p;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
 
     qemu_bh_delete(acb->bh);
 
     g_free(acb->buf);
     acb->buf = NULL;
 
-    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, acb->status);
-    aio_context_release(ctx);
 
     if (acb->task != NULL) {
         scsi_free_scsi_task(acb->task);
diff --git a/block/linux-aio.c b/block/linux-aio.c
index XXXXXXX..XXXXXXX 100644
--- a/block/linux-aio.c
+++ b/block/linux-aio.c
@@ -XXX,XX +XXX,XX @@ static inline ssize_t io_event_ret(struct io_event *ev)
  */
 static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
 {
-    LinuxAioState *s = laiocb->ctx;
     int ret;
 
     ret = laiocb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
     }
 
     laiocb->ret = ret;
-    aio_context_acquire(s->aio_context);
     if (laiocb->co) {
         /* If the coroutine is already entered it must be in ioq_submit() and
          * will notice laio->ret has been filled in when it eventually runs
@@ -XXX,XX +XXX,XX @@ static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
          * that!
          */
         if (!qemu_coroutine_entered(laiocb->co)) {
-            qemu_coroutine_enter(laiocb->co);
+            aio_co_wake(laiocb->co);
         }
     } else {
         laiocb->common.cb(laiocb->common.opaque, ret);
         qemu_aio_unref(laiocb);
     }
-    aio_context_release(s->aio_context);
 }
 
 /**
diff --git a/block/mirror.c b/block/mirror.c
index XXXXXXX..XXXXXXX 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -XXX,XX +XXX,XX @@ static void mirror_write_complete(void *opaque, int ret)
 {
     MirrorOp *op = opaque;
     MirrorBlockJob *s = op->s;
+
+    aio_context_acquire(blk_get_aio_context(s->common.blk));
     if (ret < 0) {
         BlockErrorAction action;
 
@@ -XXX,XX +XXX,XX @@ static void mirror_write_complete(void *opaque, int ret)
         }
     }
     mirror_iteration_done(op, ret);
+    aio_context_release(blk_get_aio_context(s->common.blk));
 }
 
 static void mirror_read_complete(void *opaque, int ret)
 {
     MirrorOp *op = opaque;
     MirrorBlockJob *s = op->s;
+
+    aio_context_acquire(blk_get_aio_context(s->common.blk));
     if (ret < 0) {
         BlockErrorAction action;
 
@@ -XXX,XX +XXX,XX @@ static void mirror_read_complete(void *opaque, int ret)
         }
 
         mirror_iteration_done(op, ret);
-        return;
+    } else {
+        blk_aio_pwritev(s->target, op->sector_num * BDRV_SECTOR_SIZE, &op->qiov,
+                        0, mirror_write_complete, op);
     }
-    blk_aio_pwritev(s->target, op->sector_num * BDRV_SECTOR_SIZE, &op->qiov,
-                    0, mirror_write_complete, op);
+    aio_context_release(blk_get_aio_context(s->common.blk));
 }
 
 static inline void mirror_clip_sectors(MirrorBlockJob *s,
diff --git a/block/null.c b/block/null.c
index XXXXXXX..XXXXXXX 100644
--- a/block/null.c
+++ b/block/null.c
@@ -XXX,XX +XXX,XX @@ static const AIOCBInfo null_aiocb_info = {
 static void null_bh_cb(void *opaque)
 {
     NullAIOCB *acb = opaque;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
-
-    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, 0);
-    aio_context_release(ctx);
     qemu_aio_unref(acb);
 }
 
 static void null_timer_cb(void *opaque)
 {
     NullAIOCB *acb = opaque;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
-
-    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, 0);
-    aio_context_release(ctx);
     timer_deinit(&acb->timer);
     qemu_aio_unref(acb);
 }
diff --git a/block/qed-cluster.c b/block/qed-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed-cluster.c
+++ b/block/qed-cluster.c
@@ -XXX,XX +XXX,XX @@ static void qed_find_cluster_cb(void *opaque, int ret)
     unsigned int index;
     unsigned int n;
 
+    qed_acquire(s);
     if (ret) {
         goto out;
     }
@@ -XXX,XX +XXX,XX @@ static void qed_find_cluster_cb(void *opaque, int ret)
 
 out:
     find_cluster_cb->cb(find_cluster_cb->opaque, ret, offset, len);
+    qed_release(s);
     g_free(find_cluster_cb);
 }
 
diff --git a/block/qed-table.c b/block/qed-table.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed-table.c
+++ b/block/qed-table.c
@@ -XXX,XX +XXX,XX @@ static void qed_read_table_cb(void *opaque, int ret)
 {
     QEDReadTableCB *read_table_cb = opaque;
     QEDTable *table = read_table_cb->table;
+    BDRVQEDState *s = read_table_cb->s;
     int noffsets = read_table_cb->qiov.size / sizeof(uint64_t);
     int i;
 
@@ -XXX,XX +XXX,XX @@ static void qed_read_table_cb(void *opaque, int ret)
     }
 
     /* Byteswap offsets */
+    qed_acquire(s);
     for (i = 0; i < noffsets; i++) {
         table->offsets[i] = le64_to_cpu(table->offsets[i]);
     }
+    qed_release(s);
 
 out:
     /* Completion */
-    trace_qed_read_table_cb(read_table_cb->s, read_table_cb->table, ret);
+    trace_qed_read_table_cb(s, read_table_cb->table, ret);
     gencb_complete(&read_table_cb->gencb, ret);
 }
 
@@ -XXX,XX +XXX,XX @@ typedef struct {
 static void qed_write_table_cb(void *opaque, int ret)
 {
     QEDWriteTableCB *write_table_cb = opaque;
+    BDRVQEDState *s = write_table_cb->s;
 
-    trace_qed_write_table_cb(write_table_cb->s,
+    trace_qed_write_table_cb(s,
                              write_table_cb->orig_table,
                              write_table_cb->flush,
                              ret);
@@ -XXX,XX +XXX,XX @@ static void qed_write_table_cb(void *opaque, int ret)
     if (write_table_cb->flush) {
         /* We still need to flush first */
         write_table_cb->flush = false;
+        qed_acquire(s);
         bdrv_aio_flush(write_table_cb->s->bs, qed_write_table_cb,
                        write_table_cb);
+        qed_release(s);
         return;
     }
 
@@ -XXX,XX +XXX,XX @@ static void qed_read_l2_table_cb(void *opaque, int ret)
     CachedL2Table *l2_table = request->l2_table;
     uint64_t l2_offset = read_l2_table_cb->l2_offset;
 
+    qed_acquire(s);
     if (ret) {
         /* can't trust loaded L2 table anymore */
         qed_unref_l2_cache_entry(l2_table);
@@ -XXX,XX +XXX,XX @@ static void qed_read_l2_table_cb(void *opaque, int ret)
         request->l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
         assert(request->l2_table != NULL);
     }
+    qed_release(s);
 
     gencb_complete(&read_l2_table_cb->gencb, ret);
 }
diff --git a/block/qed.c b/block/qed.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -XXX,XX +XXX,XX @@ static void qed_is_allocated_cb(void *opaque, int ret, uint64_t offset, size_t l
     }
 
     if (cb->co) {
-        qemu_coroutine_enter(cb->co);
+        aio_co_wake(cb->co);
     }
 }
 
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qed_co_pwrite_zeroes_cb(void *opaque, int ret)
     cb->done = true;
     cb->ret = ret;
     if (cb->co) {
-        qemu_coroutine_enter(cb->co);
+        aio_co_wake(cb->co);
     }
 }
 
diff --git a/block/rbd.c b/block/rbd.c
index XXXXXXX..XXXXXXX 100644
--- a/block/rbd.c
+++ b/block/rbd.c
@@ -XXX,XX +XXX,XX @@ shutdown:
 static void qemu_rbd_complete_aio(RADOSCB *rcb)
 {
     RBDAIOCB *acb = rcb->acb;
-    AioContext *ctx = bdrv_get_aio_context(acb->common.bs);
     int64_t r;
 
     r = rcb->ret;
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_complete_aio(RADOSCB *rcb)
         qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
     }
     qemu_vfree(acb->bounce);
-
-    aio_context_acquire(ctx);
     acb->common.cb(acb->common.opaque, (acb->ret > 0 ? 0 : acb->ret));
-    aio_context_release(ctx);
 
     qemu_aio_unref(acb);
 }
diff --git a/block/win32-aio.c b/block/win32-aio.c
index XXXXXXX..XXXXXXX 100644
--- a/block/win32-aio.c
+++ b/block/win32-aio.c
@@ -XXX,XX +XXX,XX @@ static void win32_aio_process_completion(QEMUWin32AIOState *s,
         qemu_vfree(waiocb->buf);
     }
 
-
-    aio_context_acquire(s->aio_ctx);
     waiocb->common.cb(waiocb->common.opaque, ret);
-    aio_context_release(s->aio_ctx);
     qemu_aio_unref(waiocb);
 }
 
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
 static void virtio_blk_rw_complete(void *opaque, int ret)
 {
     VirtIOBlockReq *next = opaque;
+    VirtIOBlock *s = next->dev;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
     while (next) {
         VirtIOBlockReq *req = next;
         next = req->mr_next;
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_rw_complete(void *opaque, int ret)
         block_acct_done(blk_get_stats(req->dev->blk), &req->acct);
         virtio_blk_free_request(req);
     }
+    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
 }
 
 static void virtio_blk_flush_complete(void *opaque, int ret)
 {
     VirtIOBlockReq *req = opaque;
+    VirtIOBlock *s = req->dev;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
     if (ret) {
         if (virtio_blk_handle_rw_error(req, -ret, 0)) {
-            return;
+            goto out;
         }
     }
 
     virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
     block_acct_done(blk_get_stats(req->dev->blk), &req->acct);
     virtio_blk_free_request(req);
+
+out:
+    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
 }
 
 #ifdef __linux__
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_ioctl_complete(void *opaque, int status)
     virtio_stl_p(vdev, &scsi->data_len, hdr->dxfer_len);
 
 out:
+    aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
     virtio_blk_req_complete(req, status);
     virtio_blk_free_request(req);
+    aio_context_release(blk_get_aio_context(s->conf.conf.blk));
     g_free(ioctl_req);
 }
 
diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/scsi/scsi-disk.c
+++ b/hw/scsi/scsi-disk.c
@@ -XXX,XX +XXX,XX @@ static void scsi_aio_complete(void *opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (scsi_disk_req_check_error(r, ret, true)) {
         goto done;
     }
@@ -XXX,XX +XXX,XX @@ static void scsi_aio_complete(void *opaque, int ret)
     scsi_req_complete(&r->req, GOOD);
 
 done:
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
     scsi_req_unref(&r->req);
 }
 
@@ -XXX,XX +XXX,XX @@ static void scsi_dma_complete(void *opaque, int ret)
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (ret < 0) {
         block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
     } else {
         block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     }
     scsi_dma_complete_noio(r, ret);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 static void scsi_read_complete(void * opaque, int ret)
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (scsi_disk_req_check_error(r, ret, true)) {
         goto done;
     }
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
 
 done:
     scsi_req_unref(&r->req);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 /* Actually issue a read to the block device.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_do_read_cb(void *opaque, int ret)
     assert (r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (ret < 0) {
         block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
     } else {
         block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     }
     scsi_do_read(opaque, ret);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 /* Read more data from scsi device into buffer.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
     assert (r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (ret < 0) {
         block_acct_failed(blk_get_stats(s->qdev.conf.blk), &r->acct);
     } else {
         block_acct_done(blk_get_stats(s->qdev.conf.blk), &r->acct);
     }
     scsi_write_complete_noio(r, ret);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 static void scsi_write_data(SCSIRequest *req)
@@ -XXX,XX +XXX,XX @@ static void scsi_unmap_complete(void *opaque, int ret)
 {
     UnmapCBData *data = opaque;
     SCSIDiskReq *r = data->r;
+    SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     scsi_unmap_complete_noio(data, ret);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 static void scsi_disk_emulate_unmap(SCSIDiskReq *r, uint8_t *inbuf)
@@ -XXX,XX +XXX,XX @@ static void scsi_write_same_complete(void *opaque, int ret)
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
+    aio_context_acquire(blk_get_aio_context(s->qdev.conf.blk));
     if (scsi_disk_req_check_error(r, ret, true)) {
         goto done;
     }
@@ -XXX,XX +XXX,XX @@ done:
     scsi_req_unref(&r->req);
     qemu_vfree(data->iov.iov_base);
     g_free(data);
+    aio_context_release(blk_get_aio_context(s->qdev.conf.blk));
 }
 
 static void scsi_disk_emulate_write_same(SCSIDiskReq *r, uint8_t *inbuf)
diff --git a/hw/scsi/scsi-generic.c b/hw/scsi/scsi-generic.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/scsi/scsi-generic.c
+++ b/hw/scsi/scsi-generic.c
@@ -XXX,XX +XXX,XX @@ done:
 static void scsi_command_complete(void *opaque, int ret)
 {
     SCSIGenericReq *r = (SCSIGenericReq *)opaque;
+    SCSIDevice *s = r->req.dev;
 
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
+
+    aio_context_acquire(blk_get_aio_context(s->conf.blk));
     scsi_command_complete_noio(r, ret);
+    aio_context_release(blk_get_aio_context(s->conf.blk));
 }
 
 static int execute_command(BlockBackend *blk,
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.blk));
+
     if (ret || r->req.io_canceled) {
         scsi_command_complete_noio(r, ret);
-        return;
+        goto done;
     }
 
     len = r->io_header.dxfer_len - r->io_header.resid;
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
     r->len = -1;
     if (len == 0) {
         scsi_command_complete_noio(r, 0);
-        return;
+        goto done;
     }
 
     /* Snoop READ CAPACITY output to set the blocksize.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_read_complete(void * opaque, int ret)
     }
     scsi_req_data(&r->req, len);
     scsi_req_unref(&r->req);
+
+done:
+    aio_context_release(blk_get_aio_context(s->conf.blk));
 }
 
 /* Read more data from scsi device into buffer.  */
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
     assert(r->req.aiocb != NULL);
     r->req.aiocb = NULL;
 
+    aio_context_acquire(blk_get_aio_context(s->conf.blk));
+
     if (ret || r->req.io_canceled) {
         scsi_command_complete_noio(r, ret);
-        return;
+        goto done;
     }
 
     if (r->req.cmd.buf[0] == MODE_SELECT && r->req.cmd.buf[4] == 12 &&
@@ -XXX,XX +XXX,XX @@ static void scsi_write_complete(void * opaque, int ret)
     }
 
     scsi_command_complete_noio(r, ret);
+
+done:
+    aio_context_release(blk_get_aio_context(s->conf.blk));
 }
 
 /* Write data to a scsi device.  Returns nonzero on failure.
diff --git a/util/thread-pool.c b/util/thread-pool.c
index XXXXXXX..XXXXXXX 100644
--- a/util/thread-pool.c
+++ b/util/thread-pool.c
@@ -XXX,XX +XXX,XX @@ restart:
              */
             qemu_bh_schedule(pool->completion_bh);
 
+            aio_context_release(pool->ctx);
             elem->common.cb(elem->common.opaque, elem->ret);
+            aio_context_acquire(pool->ctx);
             qemu_aio_unref(elem);
             goto restart;
         } else {
@@ -XXX,XX +XXX,XX @@ static void thread_pool_co_cb(void *opaque, int ret)
     ThreadPoolCo *co = opaque;
 
     co->ret = ret;
-    qemu_coroutine_enter(co->co);
+    aio_co_wake(co->co);
 }
 
 int coroutine_fn thread_pool_submit_co(ThreadPool *pool, ThreadPoolFunc *func,
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This patch prepares for the removal of unnecessary lockcnt inc/dec pairs.
Extract the dispatching loop for file descriptor handlers into a new
function aio_dispatch_handlers, and then inline aio_dispatch into
aio_poll.

aio_dispatch can now become void.

diff --git a/include/block/aio.h b/include/block/aio.h
index XXXXXXX..XXXXXXX 100644
--- a/include/block/aio.h
+++ b/include/block/aio.h
@@ -XXX,XX +XXX,XX @@ bool aio_pending(AioContext *ctx);
 /* Dispatch any pending callbacks from the GSource attached to the AioContext.
  *
  * This is used internally in the implementation of the GSource.
- *
- * @dispatch_fds: true to process fds, false to skip them
- *                (can be used as an optimization by callers that know there
- *                are no fds ready)
  */
-bool aio_dispatch(AioContext *ctx, bool dispatch_fds);
+void aio_dispatch(AioContext *ctx);
 
 /* Progress in completing AIO work to occur.  This can issue new pending
  * aio as a result of executing I/O completion or bh callbacks.
diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
     AioHandler *node, *tmp;
     bool progress = false;
 
-    /*
-     * We have to walk very carefully in case aio_set_fd_handler is
-     * called while we're walking.
-     */
-    qemu_lockcnt_inc(&ctx->list_lock);
-
     QLIST_FOREACH_SAFE_RCU(node, &ctx->aio_handlers, node, tmp) {
         int revents;
 
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
         }
     }
 
-    qemu_lockcnt_dec(&ctx->list_lock);
     return progress;
 }
 
-/*
- * Note that dispatch_fds == false has the side-effect of post-poning the
- * freeing of deleted handlers.
- */
-bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
+void aio_dispatch(AioContext *ctx)
 {
-    bool progress;
+    aio_bh_poll(ctx);
 
-    /*
-     * If there are callbacks left that have been queued, we need to call them.
-     * Do not call select in this case, because it is possible that the caller
-     * does not need a complete flush (as is the case for aio_poll loops).
-     */
-    progress = aio_bh_poll(ctx);
+    qemu_lockcnt_inc(&ctx->list_lock);
+    aio_dispatch_handlers(ctx);
+    qemu_lockcnt_dec(&ctx->list_lock);
 
-    if (dispatch_fds) {
-        progress |= aio_dispatch_handlers(ctx);
-    }
-
-    /* Run our timers */
-    progress |= timerlistgroup_run_timers(&ctx->tlg);
-
-    return progress;
+    timerlistgroup_run_timers(&ctx->tlg);
 }
 
 /* These thread-local variables are used only in a small part of aio_poll
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
     npfd = 0;
     qemu_lockcnt_dec(&ctx->list_lock);
 
-    /* Run dispatch even if there were no readable fds to run timers */
-    if (aio_dispatch(ctx, ret > 0)) {
-        progress = true;
+    progress |= aio_bh_poll(ctx);
+
+    if (ret > 0) {
+        qemu_lockcnt_inc(&ctx->list_lock);
+        progress |= aio_dispatch_handlers(ctx);
+        qemu_lockcnt_dec(&ctx->list_lock);
     }
 
+    progress |= timerlistgroup_run_timers(&ctx->tlg);
+
     return progress;
 }
 
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
     return progress;
 }
 
-bool aio_dispatch(AioContext *ctx, bool dispatch_fds)
+void aio_dispatch(AioContext *ctx)
 {
-    bool progress;
-
-    progress = aio_bh_poll(ctx);
-    if (dispatch_fds) {
-        progress |= aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
-    }
-    progress |= timerlistgroup_run_timers(&ctx->tlg);
-    return progress;
+    aio_bh_poll(ctx);
+    aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
+    timerlistgroup_run_timers(&ctx->tlg);
 }
 
 bool aio_poll(AioContext *ctx, bool blocking)
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ aio_ctx_dispatch(GSource     *source,
     AioContext *ctx = (AioContext *) source;
 
     assert(callback == NULL);
-    aio_dispatch(ctx, true);
+    aio_dispatch(ctx);
     return true;
 }
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Pull the increment/decrement pair out of aio_bh_poll and into the
callers.

diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx)
 
 void aio_dispatch(AioContext *ctx)
 {
+    qemu_lockcnt_inc(&ctx->list_lock);
     aio_bh_poll(ctx);
-
-    qemu_lockcnt_inc(&ctx->list_lock);
     aio_dispatch_handlers(ctx);
     qemu_lockcnt_dec(&ctx->list_lock);
 
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
     }
 
     npfd = 0;
-    qemu_lockcnt_dec(&ctx->list_lock);
 
     progress |= aio_bh_poll(ctx);
 
     if (ret > 0) {
-        qemu_lockcnt_inc(&ctx->list_lock);
         progress |= aio_dispatch_handlers(ctx);
-        qemu_lockcnt_dec(&ctx->list_lock);
     }
 
+    qemu_lockcnt_dec(&ctx->list_lock);
+
     progress |= timerlistgroup_run_timers(&ctx->tlg);
 
     return progress;
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
     bool progress = false;
     AioHandler *tmp;
 
-    qemu_lockcnt_inc(&ctx->list_lock);
-
     /*
      * We have to walk very carefully in case aio_set_fd_handler is
      * called while we're walking.
@@ -XXX,XX +XXX,XX @@ static bool aio_dispatch_handlers(AioContext *ctx, HANDLE event)
         }
     }
 
-    qemu_lockcnt_dec(&ctx->list_lock);
     return progress;
 }
 
 void aio_dispatch(AioContext *ctx)
 {
+    qemu_lockcnt_inc(&ctx->list_lock);
     aio_bh_poll(ctx);
     aio_dispatch_handlers(ctx, INVALID_HANDLE_VALUE);
+    qemu_lockcnt_dec(&ctx->list_lock);
     timerlistgroup_run_timers(&ctx->tlg);
 }
 
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         }
     }
 
-    qemu_lockcnt_dec(&ctx->list_lock);
     first = true;
 
     /* ctx->notifier is always registered.  */
@@ -XXX,XX +XXX,XX @@ bool aio_poll(AioContext *ctx, bool blocking)
         progress |= aio_dispatch_handlers(ctx, event);
     } while (count > 0);
 
+    qemu_lockcnt_dec(&ctx->list_lock);
+
     progress |= timerlistgroup_run_timers(&ctx->tlg);
     return progress;
 }
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ void aio_bh_call(QEMUBH *bh)
     bh->cb(bh->opaque);
 }
 
-/* Multiple occurrences of aio_bh_poll cannot be called concurrently */
+/* Multiple occurrences of aio_bh_poll cannot be called concurrently.
+ * The count in ctx->list_lock is incremented before the call, and is
+ * not affected by the call.
+ */
 int aio_bh_poll(AioContext *ctx)
 {
     QEMUBH *bh, **bhp, *next;
     int ret;
     bool deleted = false;
 
-    qemu_lockcnt_inc(&ctx->list_lock);
-
     ret = 0;
     for (bh = atomic_rcu_read(&ctx->first_bh); bh; bh = next) {
         next = atomic_rcu_read(&bh->next);
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
 
     /* remove deleted bhs */
     if (!deleted) {
-        qemu_lockcnt_dec(&ctx->list_lock);
         return ret;
     }
 
-    if (qemu_lockcnt_dec_and_lock(&ctx->list_lock)) {
+    if (qemu_lockcnt_dec_if_lock(&ctx->list_lock)) {
         bhp = &ctx->first_bh;
         while (*bhp) {
             bh = *bhp;
@@ -XXX,XX +XXX,XX @@ int aio_bh_poll(AioContext *ctx)
                 bhp = &bh->next;
             }
         }
-        qemu_lockcnt_unlock(&ctx->list_lock);
+        qemu_lockcnt_inc_and_unlock(&ctx->list_lock);
     }
     return ret;
 }
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/include/block/block_int.h b/include/block/block_int.h
index XXXXXXX..XXXXXXX 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -XXX,XX +XXX,XX @@ struct BdrvChild {
  * copied as well.
  */
 struct BlockDriverState {
-    int64_t total_sectors; /* if we are reading a disk image, give its
-                              size in sectors */
+    /* Protected by big QEMU lock or read-only after opening.  No special
+     * locking needed during I/O...
+     */
     int open_flags; /* flags used to open the file, re-used for re-open */
     bool read_only; /* if true, the media is read only */
     bool encrypted; /* if true, the media is encrypted */
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
     bool sg;        /* if true, the device is a /dev/sg* */
     bool probed;    /* if true, format was probed rather than specified */
 
-    int copy_on_read; /* if nonzero, copy read backing sectors into image.
-                         note this is a reference count */
-
-    CoQueue flush_queue;            /* Serializing flush queue */
-    bool active_flush_req;          /* Flush request in flight? */
-    unsigned int write_gen;         /* Current data generation */
-    unsigned int flushed_gen;       /* Flushed write generation */
-
     BlockDriver *drv; /* NULL means no media */
     void *opaque;
 
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
     BdrvChild *backing;
     BdrvChild *file;
 
-    /* Callback before write request is processed */
-    NotifierWithReturnList before_write_notifiers;
-
-    /* number of in-flight requests; overall and serialising */
-    unsigned int in_flight;
-    unsigned int serialising_in_flight;
-
-    bool wakeup;
-
-    /* Offset after the highest byte written to */
-    uint64_t wr_highest_offset;
-
     /* I/O Limits */
     BlockLimits bl;
 
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
     QTAILQ_ENTRY(BlockDriverState) bs_list;
     /* element of the list of monitor-owned BDS */
     QTAILQ_ENTRY(BlockDriverState) monitor_list;
-    QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
     int refcnt;
 
-    QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
-
     /* operation blockers */
     QLIST_HEAD(, BdrvOpBlocker) op_blockers[BLOCK_OP_TYPE_MAX];
 
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
     /* The error object in use for blocking operations on backing_hd */
     Error *backing_blocker;
 
+    /* Protected by AioContext lock */
+
+    /* If true, copy read backing sectors into image.  Can be >1 if more
+     * than one client has requested copy-on-read.
+     */
+    int copy_on_read;
+
+    /* If we are reading a disk image, give its size in sectors.
+     * Generally read-only; it is written to by load_vmstate and save_vmstate,
+     * but the block layer is quiescent during those.
+     */
+    int64_t total_sectors;
+
+    /* Callback before write request is processed */
+    NotifierWithReturnList before_write_notifiers;
+
+    /* number of in-flight requests; overall and serialising */
+    unsigned int in_flight;
+    unsigned int serialising_in_flight;
+
+    bool wakeup;
+
+    /* Offset after the highest byte written to */
+    uint64_t wr_highest_offset;
+
     /* threshold limit for writes, in bytes. "High water mark". */
     uint64_t write_threshold_offset;
     NotifierWithReturn write_threshold_notifier;
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
     /* counter for nested bdrv_io_plug */
     unsigned io_plugged;
 
+    QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
+    CoQueue flush_queue;                  /* Serializing flush queue */
+    bool active_flush_req;                /* Flush request in flight? */
+    unsigned int write_gen;               /* Current data generation */
+    unsigned int flushed_gen;             /* Flushed write generation */
+
+    QLIST_HEAD(, BdrvDirtyBitmap) dirty_bitmaps;
+
+    /* do we need to tell the quest if we have a volatile write cache? */
+    int enable_write_cache;
+
     int quiesce_counter;
 };
 
diff --git a/include/sysemu/block-backend.h b/include/sysemu/block-backend.h
index XXXXXXX..XXXXXXX 100644
--- a/include/sysemu/block-backend.h
+++ b/include/sysemu/block-backend.h
@@ -XXX,XX +XXX,XX @@ typedef struct BlockDevOps {
  * fields that must be public. This is in particular for QLIST_ENTRY() and
  * friends so that BlockBackends can be kept in lists outside block-backend.c */
 typedef struct BlockBackendPublic {
-    /* I/O throttling.
-     * throttle_state tells us if this BlockBackend has I/O limits configured.
-     * io_limits_disabled tells us if they are currently being enforced */
+    /* I/O throttling has its own locking, but also some fields are
+     * protected by the AioContext lock.
+     */
+
+    /* Protected by AioContext lock.  */
     CoQueue      throttled_reqs[2];
+
+    /* Nonzero if the I/O limits are currently being ignored; generally
+     * it is zero.  */
     unsigned int io_limits_disabled;
 
     /* The following fields are protected by the ThrottleGroup lock.
-     * See the ThrottleGroup documentation for details. */
+     * See the ThrottleGroup documentation for details.
+     * throttle_state tells us if I/O limits are configured. */
     ThrottleState *throttle_state;
     ThrottleTimers throttle_timers;
     unsigned       pending_reqs[2];
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This uses the lock-free mutex described in the paper '"Blocking without
Locking", or LFTHREADS: A lock-free thread library' by Gidenstam and
Papatriantafilou.  The same technique is used in OSv, and in fact
the code is essentially a conversion to C of OSv's code.

[Added missing coroutine_fn in tests/test-aio-multithread.c.
--Stefan]

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-2-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/qemu/coroutine.h     |  17 ++++-
 tests/test-aio-multithread.c |  86 ++++++++++++++++++++++++
 util/qemu-coroutine-lock.c   | 155 ++++++++++++++++++++++++++++++++++++++++---
 util/trace-events            |   1 +
 4 files changed, 246 insertions(+), 13 deletions(-)

diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine.h
+++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue);
 /**
  * Provides a mutex that can be used to synchronise coroutines
  */
+struct CoWaitRecord;
 typedef struct CoMutex {
-    bool locked;
+    /* Count of pending lockers; 0 for a free mutex, 1 for an
+     * uncontended mutex.
+     */
+    unsigned locked;
+
+    /* A queue of waiters.  Elements are added atomically in front of
+     * from_push.  to_pop is only populated, and popped from, by whoever
+     * is in charge of the next wakeup.  This can be an unlocker or,
+     * through the handoff protocol, a locker that is about to go to sleep.
+     */
+    QSLIST_HEAD(, CoWaitRecord) from_push, to_pop;
+
+    unsigned handoff, sequence;
+
     Coroutine *holder;
-    CoQueue queue;
 } CoMutex;
 
 /**
diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/test-aio-multithread.c
+++ b/tests/test-aio-multithread.c
@@ -XXX,XX +XXX,XX @@ static void test_multi_co_schedule_10(void)
     test_multi_co_schedule(10);
 }
 
+/* CoMutex thread-safety.  */
+
+static uint32_t atomic_counter;
+static uint32_t running;
+static uint32_t counter;
+static CoMutex comutex;
+
+static void coroutine_fn test_multi_co_mutex_entry(void *opaque)
+{
+    while (!atomic_mb_read(&now_stopping)) {
+        qemu_co_mutex_lock(&comutex);
+        counter++;
+        qemu_co_mutex_unlock(&comutex);
+
+        /* Increase atomic_counter *after* releasing the mutex.  Otherwise
+         * there is a chance (it happens about 1 in 3 runs) that the iothread
+         * exits before the coroutine is woken up, causing a spurious
+         * assertion failure.
+         */
+        atomic_inc(&atomic_counter);
+    }
+    atomic_dec(&running);
+}
+
+static void test_multi_co_mutex(int threads, int seconds)
+{
+    int i;
+
+    qemu_co_mutex_init(&comutex);
+    counter = 0;
+    atomic_counter = 0;
+    now_stopping = false;
+
+    create_aio_contexts();
+    assert(threads <= NUM_CONTEXTS);
+    running = threads;
+    for (i = 0; i < threads; i++) {
+        Coroutine *co1 = qemu_coroutine_create(test_multi_co_mutex_entry, NULL);
+        aio_co_schedule(ctx[i], co1);
+    }
+
+    g_usleep(seconds * 1000000);
+
+    atomic_mb_set(&now_stopping, true);
+    while (running > 0) {
+        g_usleep(100000);
+    }
+
+    join_aio_contexts();
+    g_test_message("%d iterations/second\n", counter / seconds);
+    g_assert_cmpint(counter, ==, atomic_counter);
+}
+
+/* Testing with NUM_CONTEXTS threads focuses on the queue.  The mutex however
+ * is too contended (and the threads spend too much time in aio_poll)
+ * to actually stress the handoff protocol.
+ */
+static void test_multi_co_mutex_1(void)
+{
+    test_multi_co_mutex(NUM_CONTEXTS, 1);
+}
+
+static void test_multi_co_mutex_10(void)
+{
+    test_multi_co_mutex(NUM_CONTEXTS, 10);
+}
+
+/* Testing with fewer threads stresses the handoff protocol too.  Still, the
+ * case where the locker _can_ pick up a handoff is very rare, happening
+ * about 10 times in 1 million, so increase the runtime a bit compared to
+ * other "quick" testcases that only run for 1 second.
+ */
+static void test_multi_co_mutex_2_3(void)
+{
+    test_multi_co_mutex(2, 3);
+}
+
+static void test_multi_co_mutex_2_30(void)
+{
+    test_multi_co_mutex(2, 30);
+}
+
 /* End of tests.  */
 
 int main(int argc, char **argv)
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
     g_test_add_func("/aio/multi/lifecycle", test_lifecycle);
     if (g_test_quick()) {
         g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
+        g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_1);
+        g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_3);
     } else {
         g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
+        g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_10);
+        g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_30);
     }
     return g_test_run();
 }
diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-lock.c
+++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@
  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  * THE SOFTWARE.
+ *
+ * The lock-free mutex implementation is based on OSv
+ * (core/lfmutex.cc, include/lockfree/mutex.hh).
+ * Copyright (C) 2013 Cloudius Systems, Ltd.
  */
 
 #include "qemu/osdep.h"
@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue)
     return QSIMPLEQ_FIRST(&queue->entries) == NULL;
 }
 
+/* The wait records are handled with a multiple-producer, single-consumer
+ * lock-free queue.  There cannot be two concurrent pop_waiter() calls
+ * because pop_waiter() can only be called while mutex->handoff is zero.
+ * This can happen in three cases:
+ * - in qemu_co_mutex_unlock, before the hand-off protocol has started.
+ *   In this case, qemu_co_mutex_lock will see mutex->handoff == 0 and
+ *   not take part in the handoff.
+ * - in qemu_co_mutex_lock, if it steals the hand-off responsibility from
+ *   qemu_co_mutex_unlock.  In this case, qemu_co_mutex_unlock will fail
+ *   the cmpxchg (it will see either 0 or the next sequence value) and
+ *   exit.  The next hand-off cannot begin until qemu_co_mutex_lock has
+ *   woken up someone.
+ * - in qemu_co_mutex_unlock, if it takes the hand-off token itself.
+ *   In this case another iteration starts with mutex->handoff == 0;
+ *   a concurrent qemu_co_mutex_lock will fail the cmpxchg, and
+ *   qemu_co_mutex_unlock will go back to case (1).
+ *
+ * The following functions manage this queue.
+ */
+typedef struct CoWaitRecord {
+    Coroutine *co;
+    QSLIST_ENTRY(CoWaitRecord) next;
+} CoWaitRecord;
+
+static void push_waiter(CoMutex *mutex, CoWaitRecord *w)
+{
+    w->co = qemu_coroutine_self();
+    QSLIST_INSERT_HEAD_ATOMIC(&mutex->from_push, w, next);
+}
+
+static void move_waiters(CoMutex *mutex)
+{
+    QSLIST_HEAD(, CoWaitRecord) reversed;
+    QSLIST_MOVE_ATOMIC(&reversed, &mutex->from_push);
+    while (!QSLIST_EMPTY(&reversed)) {
+        CoWaitRecord *w = QSLIST_FIRST(&reversed);
+        QSLIST_REMOVE_HEAD(&reversed, next);
+        QSLIST_INSERT_HEAD(&mutex->to_pop, w, next);
+    }
+}
+
+static CoWaitRecord *pop_waiter(CoMutex *mutex)
+{
+    CoWaitRecord *w;
+
+    if (QSLIST_EMPTY(&mutex->to_pop)) {
+        move_waiters(mutex);
+        if (QSLIST_EMPTY(&mutex->to_pop)) {
+            return NULL;
+        }
+    }
+    w = QSLIST_FIRST(&mutex->to_pop);
+    QSLIST_REMOVE_HEAD(&mutex->to_pop, next);
+    return w;
+}
+
+static bool has_waiters(CoMutex *mutex)
+{
+    return QSLIST_EMPTY(&mutex->to_pop) || QSLIST_EMPTY(&mutex->from_push);
+}
+
 void qemu_co_mutex_init(CoMutex *mutex)
 {
     memset(mutex, 0, sizeof(*mutex));
-    qemu_co_queue_init(&mutex->queue);
 }
 
-void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
+static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
 {
     Coroutine *self = qemu_coroutine_self();
+    CoWaitRecord w;
+    unsigned old_handoff;
 
     trace_qemu_co_mutex_lock_entry(mutex, self);
+    w.co = self;
+    push_waiter(mutex, &w);
 
-    while (mutex->locked) {
-        qemu_co_queue_wait(&mutex->queue);
+    /* This is the "Responsibility Hand-Off" protocol; a lock() picks from
+     * a concurrent unlock() the responsibility of waking somebody up.
+     */
+    old_handoff = atomic_mb_read(&mutex->handoff);
+    if (old_handoff &&
+        has_waiters(mutex) &&
+        atomic_cmpxchg(&mutex->handoff, old_handoff, 0) == old_handoff) {
+        /* There can be no concurrent pops, because there can be only
+         * one active handoff at a time.
+         */
+        CoWaitRecord *to_wake = pop_waiter(mutex);
+        Coroutine *co = to_wake->co;
+        if (co == self) {
+            /* We got the lock ourselves!  */
+            assert(to_wake == &w);
+            return;
+        }
+
+        aio_co_wake(co);
     }
 
-    mutex->locked = true;
-    mutex->holder = self;
-    self->locks_held++;
-
+    qemu_coroutine_yield();
     trace_qemu_co_mutex_lock_return(mutex, self);
 }
 
+void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
+{
+    Coroutine *self = qemu_coroutine_self();
+
+    if (atomic_fetch_inc(&mutex->locked) == 0) {
+        /* Uncontended.  */
+        trace_qemu_co_mutex_lock_uncontended(mutex, self);
+    } else {
+        qemu_co_mutex_lock_slowpath(mutex);
+    }
+    mutex->holder = self;
+    self->locks_held++;
+}
+
 void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
 {
     Coroutine *self = qemu_coroutine_self();
 
     trace_qemu_co_mutex_unlock_entry(mutex, self);
 
-    assert(mutex->locked == true);
+    assert(mutex->locked);
     assert(mutex->holder == self);
     assert(qemu_in_coroutine());
 
-    mutex->locked = false;
     mutex->holder = NULL;
     self->locks_held--;
-    qemu_co_queue_next(&mutex->queue);
+    if (atomic_fetch_dec(&mutex->locked) == 1) {
+        /* No waiting qemu_co_mutex_lock().  Pfew, that was easy!  */
+        return;
+    }
+
+    for (;;) {
+        CoWaitRecord *to_wake = pop_waiter(mutex);
+        unsigned our_handoff;
+
+        if (to_wake) {
+            Coroutine *co = to_wake->co;
+            aio_co_wake(co);
+            break;
+        }
+
+        /* Some concurrent lock() is in progress (we know this because
+         * mutex->locked was >1) but it hasn't yet put itself on the wait
+         * queue.  Pick a sequence number for the handoff protocol (not 0).
+         */
+        if (++mutex->sequence == 0) {
+            mutex->sequence = 1;
+        }
+
+        our_handoff = mutex->sequence;
+        atomic_mb_set(&mutex->handoff, our_handoff);
+        if (!has_waiters(mutex)) {
+            /* The concurrent lock has not added itself yet, so it
+             * will be able to pick our handoff.
+             */
+            break;
+        }
+
+        /* Try to do the handoff protocol ourselves; if somebody else has
+         * already taken it, however, we're done and they're responsible.
+         */
+        if (atomic_cmpxchg(&mutex->handoff, our_handoff, 0) != our_handoff) {
+            break;
+        }
+    }
 
     trace_qemu_co_mutex_unlock_return(mutex, self);
 }
diff --git a/util/trace-events b/util/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/util/trace-events
+++ b/util/trace-events
@@ -XXX,XX +XXX,XX @@ qemu_coroutine_terminate(void *co) "self %p"
 
 # util/qemu-coroutine-lock.c
 qemu_co_queue_run_restart(void *co) "co %p"
+qemu_co_mutex_lock_uncontended(void *mutex, void *self) "mutex %p self %p"
 qemu_co_mutex_lock_entry(void *mutex, void *self) "mutex %p self %p"
 qemu_co_mutex_lock_return(void *mutex, void *self) "mutex %p self %p"
 qemu_co_mutex_unlock_entry(void *mutex, void *self) "mutex %p self %p"
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Running a very small critical section on pthread_mutex_t and CoMutex
shows that pthread_mutex_t is much faster because it doesn't actually
go to sleep.  What happens is that the critical section is shorter
than the latency of entering the kernel and thus FUTEX_WAIT always
fails.  With CoMutex there is no such latency but you still want to
avoid wait and wakeup.  So introduce it artificially.

This only works with one waiters; because CoMutex is fair, it will
always have more waits and wakeups than a pthread_mutex_t.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-3-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/qemu/coroutine.h   |  5 +++++
 util/qemu-coroutine-lock.c | 51 ++++++++++++++++++++++++++++++++++++++++------
 util/qemu-coroutine.c      |  2 +-
 3 files changed, 51 insertions(+), 7 deletions(-)

diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine.h
+++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ typedef struct CoMutex {
      */
     unsigned locked;
 
+    /* Context that is holding the lock.  Useful to avoid spinning
+     * when two coroutines on the same AioContext try to get the lock. :)
+     */
+    AioContext *ctx;
+
     /* A queue of waiters.  Elements are added atomically in front of
      * from_push.  to_pop is only populated, and popped from, by whoever
      * is in charge of the next wakeup.  This can be an unlocker or,
diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-lock.c
+++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu-common.h"
 #include "qemu/coroutine.h"
 #include "qemu/coroutine_int.h"
+#include "qemu/processor.h"
 #include "qemu/queue.h"
 #include "block/aio.h"
 #include "trace.h"
@@ -XXX,XX +XXX,XX @@ void qemu_co_mutex_init(CoMutex *mutex)
     memset(mutex, 0, sizeof(*mutex));
 }
 
-static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
+static void coroutine_fn qemu_co_mutex_wake(CoMutex *mutex, Coroutine *co)
+{
+    /* Read co before co->ctx; pairs with smp_wmb() in
+     * qemu_coroutine_enter().
+     */
+    smp_read_barrier_depends();
+    mutex->ctx = co->ctx;
+    aio_co_wake(co);
+}
+
+static void coroutine_fn qemu_co_mutex_lock_slowpath(AioContext *ctx,
+                                                     CoMutex *mutex)
 {
     Coroutine *self = qemu_coroutine_self();
     CoWaitRecord w;
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
         if (co == self) {
             /* We got the lock ourselves!  */
             assert(to_wake == &w);
+            mutex->ctx = ctx;
             return;
         }
 
-        aio_co_wake(co);
+        qemu_co_mutex_wake(mutex, co);
     }
 
     qemu_coroutine_yield();
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn qemu_co_mutex_lock_slowpath(CoMutex *mutex)
 
 void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex)
 {
+    AioContext *ctx = qemu_get_current_aio_context();
     Coroutine *self = qemu_coroutine_self();
+    int waiters, i;
 
-    if (atomic_fetch_inc(&mutex->locked) == 0) {
+    /* Running a very small critical section on pthread_mutex_t and CoMutex
+     * shows that pthread_mutex_t is much faster because it doesn't actually
+     * go to sleep.  What happens is that the critical section is shorter
+     * than the latency of entering the kernel and thus FUTEX_WAIT always
+     * fails.  With CoMutex there is no such latency but you still want to
+     * avoid wait and wakeup.  So introduce it artificially.
+     */
+    i = 0;
+retry_fast_path:
+    waiters = atomic_cmpxchg(&mutex->locked, 0, 1);
+    if (waiters != 0) {
+        while (waiters == 1 && ++i < 1000) {
+            if (atomic_read(&mutex->ctx) == ctx) {
+                break;
+            }
+            if (atomic_read(&mutex->locked) == 0) {
+                goto retry_fast_path;
+            }
+            cpu_relax();
+        }
+        waiters = atomic_fetch_inc(&mutex->locked);
+    }
+
+    if (waiters == 0) {
         /* Uncontended.  */
         trace_qemu_co_mutex_lock_uncontended(mutex, self);
+        mutex->ctx = ctx;
     } else {
-        qemu_co_mutex_lock_slowpath(mutex);
+        qemu_co_mutex_lock_slowpath(ctx, mutex);
     }
     mutex->holder = self;
     self->locks_held++;
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
     assert(mutex->holder == self);
     assert(qemu_in_coroutine());
 
+    mutex->ctx = NULL;
     mutex->holder = NULL;
     self->locks_held--;
     if (atomic_fetch_dec(&mutex->locked) == 1) {
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex)
         unsigned our_handoff;
 
         if (to_wake) {
-            Coroutine *co = to_wake->co;
-            aio_co_wake(co);
+            qemu_co_mutex_wake(mutex, to_wake->co);
             break;
         }
 
diff --git a/util/qemu-coroutine.c b/util/qemu-coroutine.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine.c
+++ b/util/qemu-coroutine.c
@@ -XXX,XX +XXX,XX @@ void qemu_coroutine_enter(Coroutine *co)
     co->ctx = qemu_get_current_aio_context();
 
     /* Store co->ctx before anything that stores co.  Matches
-     * barrier in aio_co_wake.
+     * barrier in aio_co_wake and qemu_co_mutex_wake.
      */
     smp_wmb();
 
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

Add two implementations of the same benchmark as the previous patch,
but using pthreads.  One uses a normal QemuMutex, the other is Linux
only and implements a fair mutex based on MCS locks and futexes.
This shows that the slower performance of the 5-thread case is due to
the fairness of CoMutex, rather than to coroutines.  If fairness does
not matter, as is the case with two threads, CoMutex can actually be
faster than pthreads.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-4-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/test-aio-multithread.c | 164 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 164 insertions(+)

diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/test-aio-multithread.c
+++ b/tests/test-aio-multithread.c
@@ -XXX,XX +XXX,XX @@ static void test_multi_co_mutex_2_30(void)
     test_multi_co_mutex(2, 30);
 }
 
+/* Same test with fair mutexes, for performance comparison.  */
+
+#ifdef CONFIG_LINUX
+#include "qemu/futex.h"
+
+/* The nodes for the mutex reside in this structure (on which we try to avoid
+ * false sharing).  The head of the mutex is in the "mutex_head" variable.
+ */
+static struct {
+    int next, locked;
+    int padding[14];
+} nodes[NUM_CONTEXTS] __attribute__((__aligned__(64)));
+
+static int mutex_head = -1;
+
+static void mcs_mutex_lock(void)
+{
+    int prev;
+
+    nodes[id].next = -1;
+    nodes[id].locked = 1;
+    prev = atomic_xchg(&mutex_head, id);
+    if (prev != -1) {
+        atomic_set(&nodes[prev].next, id);
+        qemu_futex_wait(&nodes[id].locked, 1);
+    }
+}
+
+static void mcs_mutex_unlock(void)
+{
+    int next;
+    if (nodes[id].next == -1) {
+        if (atomic_read(&mutex_head) == id &&
+            atomic_cmpxchg(&mutex_head, id, -1) == id) {
+            /* Last item in the list, exit.  */
+            return;
+        }
+        while (atomic_read(&nodes[id].next) == -1) {
+            /* mcs_mutex_lock did the xchg, but has not updated
+             * nodes[prev].next yet.
+             */
+        }
+    }
+
+    /* Wake up the next in line.  */
+    next = nodes[id].next;
+    nodes[next].locked = 0;
+    qemu_futex_wake(&nodes[next].locked, 1);
+}
+
+static void test_multi_fair_mutex_entry(void *opaque)
+{
+    while (!atomic_mb_read(&now_stopping)) {
+        mcs_mutex_lock();
+        counter++;
+        mcs_mutex_unlock();
+        atomic_inc(&atomic_counter);
+    }
+    atomic_dec(&running);
+}
+
+static void test_multi_fair_mutex(int threads, int seconds)
+{
+    int i;
+
+    assert(mutex_head == -1);
+    counter = 0;
+    atomic_counter = 0;
+    now_stopping = false;
+
+    create_aio_contexts();
+    assert(threads <= NUM_CONTEXTS);
+    running = threads;
+    for (i = 0; i < threads; i++) {
+        Coroutine *co1 = qemu_coroutine_create(test_multi_fair_mutex_entry, NULL);
+        aio_co_schedule(ctx[i], co1);
+    }
+
+    g_usleep(seconds * 1000000);
+
+    atomic_mb_set(&now_stopping, true);
+    while (running > 0) {
+        g_usleep(100000);
+    }
+
+    join_aio_contexts();
+    g_test_message("%d iterations/second\n", counter / seconds);
+    g_assert_cmpint(counter, ==, atomic_counter);
+}
+
+static void test_multi_fair_mutex_1(void)
+{
+    test_multi_fair_mutex(NUM_CONTEXTS, 1);
+}
+
+static void test_multi_fair_mutex_10(void)
+{
+    test_multi_fair_mutex(NUM_CONTEXTS, 10);
+}
+#endif
+
+/* Same test with pthread mutexes, for performance comparison and
+ * portability.  */
+
+static QemuMutex mutex;
+
+static void test_multi_mutex_entry(void *opaque)
+{
+    while (!atomic_mb_read(&now_stopping)) {
+        qemu_mutex_lock(&mutex);
+        counter++;
+        qemu_mutex_unlock(&mutex);
+        atomic_inc(&atomic_counter);
+    }
+    atomic_dec(&running);
+}
+
+static void test_multi_mutex(int threads, int seconds)
+{
+    int i;
+
+    qemu_mutex_init(&mutex);
+    counter = 0;
+    atomic_counter = 0;
+    now_stopping = false;
+
+    create_aio_contexts();
+    assert(threads <= NUM_CONTEXTS);
+    running = threads;
+    for (i = 0; i < threads; i++) {
+        Coroutine *co1 = qemu_coroutine_create(test_multi_mutex_entry, NULL);
+        aio_co_schedule(ctx[i], co1);
+    }
+
+    g_usleep(seconds * 1000000);
+
+    atomic_mb_set(&now_stopping, true);
+    while (running > 0) {
+        g_usleep(100000);
+    }
+
+    join_aio_contexts();
+    g_test_message("%d iterations/second\n", counter / seconds);
+    g_assert_cmpint(counter, ==, atomic_counter);
+}
+
+static void test_multi_mutex_1(void)
+{
+    test_multi_mutex(NUM_CONTEXTS, 1);
+}
+
+static void test_multi_mutex_10(void)
+{
+    test_multi_mutex(NUM_CONTEXTS, 10);
+}
+
 /* End of tests.  */
 
 int main(int argc, char **argv)
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
         g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_1);
         g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_1);
         g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_3);
+#ifdef CONFIG_LINUX
+        g_test_add_func("/aio/multi/mutex/mcs", test_multi_fair_mutex_1);
+#endif
+        g_test_add_func("/aio/multi/mutex/pthread", test_multi_mutex_1);
     } else {
         g_test_add_func("/aio/multi/schedule", test_multi_co_schedule_10);
         g_test_add_func("/aio/multi/mutex/contended", test_multi_co_mutex_10);
         g_test_add_func("/aio/multi/mutex/handoff", test_multi_co_mutex_2_30);
+#ifdef CONFIG_LINUX
+        g_test_add_func("/aio/multi/mutex/mcs", test_multi_fair_mutex_10);
+#endif
+        g_test_add_func("/aio/multi/mutex/pthread", test_multi_mutex_10);
     }
     return g_test_run();
 }
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This will avoid forward references in the next patch.  It is also
more logical because CoQueue is not anymore the basic primitive.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-5-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/qemu/coroutine.h | 89 ++++++++++++++++++++++++------------------------
 1 file changed, 44 insertions(+), 45 deletions(-)

diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine.h
+++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ bool qemu_in_coroutine(void);
  */
 bool qemu_coroutine_entered(Coroutine *co);
 
-
-/**
- * CoQueues are a mechanism to queue coroutines in order to continue executing
- * them later. They provide the fundamental primitives on which coroutine locks
- * are built.
- */
-typedef struct CoQueue {
-    QSIMPLEQ_HEAD(, Coroutine) entries;
-} CoQueue;
-
-/**
- * Initialise a CoQueue. This must be called before any other operation is used
- * on the CoQueue.
- */
-void qemu_co_queue_init(CoQueue *queue);
-
-/**
- * Adds the current coroutine to the CoQueue and transfers control to the
- * caller of the coroutine.
- */
-void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
-
-/**
- * Restarts the next coroutine in the CoQueue and removes it from the queue.
- *
- * Returns true if a coroutine was restarted, false if the queue is empty.
- */
-bool coroutine_fn qemu_co_queue_next(CoQueue *queue);
-
-/**
- * Restarts all coroutines in the CoQueue and leaves the queue empty.
- */
-void coroutine_fn qemu_co_queue_restart_all(CoQueue *queue);
-
-/**
- * Enter the next coroutine in the queue
- */
-bool qemu_co_enter_next(CoQueue *queue);
-
-/**
- * Checks if the CoQueue is empty.
- */
-bool qemu_co_queue_empty(CoQueue *queue);
-
-
 /**
  * Provides a mutex that can be used to synchronise coroutines
  */
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex);
  */
 void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex);
 
+
+/**
+ * CoQueues are a mechanism to queue coroutines in order to continue executing
+ * them later.
+ */
+typedef struct CoQueue {
+    QSIMPLEQ_HEAD(, Coroutine) entries;
+} CoQueue;
+
+/**
+ * Initialise a CoQueue. This must be called before any other operation is used
+ * on the CoQueue.
+ */
+void qemu_co_queue_init(CoQueue *queue);
+
+/**
+ * Adds the current coroutine to the CoQueue and transfers control to the
+ * caller of the coroutine.
+ */
+void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
+
+/**
+ * Restarts the next coroutine in the CoQueue and removes it from the queue.
+ *
+ * Returns true if a coroutine was restarted, false if the queue is empty.
+ */
+bool coroutine_fn qemu_co_queue_next(CoQueue *queue);
+
+/**
+ * Restarts all coroutines in the CoQueue and leaves the queue empty.
+ */
+void coroutine_fn qemu_co_queue_restart_all(CoQueue *queue);
+
+/**
+ * Enter the next coroutine in the queue
+ */
+bool qemu_co_enter_next(CoQueue *queue);
+
+/**
+ * Checks if the CoQueue is empty.
+ */
+bool qemu_co_queue_empty(CoQueue *queue);
+
+
 typedef struct CoRwlock {
     bool writer;
     int reader;
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

All that CoQueue needs in order to become thread-safe is help
from an external mutex.  Add this to the API.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-6-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/qemu/coroutine.h   |  8 +++++---
 block/backup.c             |  2 +-
 block/io.c                 |  4 ++--
 block/nbd-client.c         |  2 +-
 block/qcow2-cluster.c      |  4 +---
 block/sheepdog.c           |  2 +-
 block/throttle-groups.c    |  2 +-
 hw/9pfs/9p.c               |  2 +-
 util/qemu-coroutine-lock.c | 24 +++++++++++++++++++++---
 9 files changed, 34 insertions(+), 16 deletions(-)

diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine.h
+++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex);
 
 /**
  * CoQueues are a mechanism to queue coroutines in order to continue executing
- * them later.
+ * them later.  They are similar to condition variables, but they need help
+ * from an external mutex in order to maintain thread-safety.
  */
 typedef struct CoQueue {
     QSIMPLEQ_HEAD(, Coroutine) entries;
@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_init(CoQueue *queue);
 
 /**
  * Adds the current coroutine to the CoQueue and transfers control to the
- * caller of the coroutine.
+ * caller of the coroutine.  The mutex is unlocked during the wait and
+ * locked again afterwards.
  */
-void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
+void coroutine_fn qemu_co_queue_wait(CoQueue *queue, CoMutex *mutex);
 
 /**
  * Restarts the next coroutine in the CoQueue and removes it from the queue.
diff --git a/block/backup.c b/block/backup.c
index XXXXXXX..XXXXXXX 100644
--- a/block/backup.c
+++ b/block/backup.c
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn wait_for_overlapping_requests(BackupBlockJob *job,
         retry = false;
         QLIST_FOREACH(req, &job->inflight_reqs, list) {
             if (end > req->start && start < req->end) {
-                qemu_co_queue_wait(&req->wait_queue);
+                qemu_co_queue_wait(&req->wait_queue, NULL);
                 retry = true;
                 break;
             }
diff --git a/block/io.c b/block/io.c
index XXXXXXX..XXXXXXX 100644
--- a/block/io.c
+++ b/block/io.c
@@ -XXX,XX +XXX,XX @@ static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
                  * (instead of producing a deadlock in the former case). */
                 if (!req->waiting_for) {
                     self->waiting_for = req;
-                    qemu_co_queue_wait(&req->wait_queue);
+                    qemu_co_queue_wait(&req->wait_queue, NULL);
                     self->waiting_for = NULL;
                     retry = true;
                     waited = true;
@@ -XXX,XX +XXX,XX @@ int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
 
     /* Wait until any previous flushes are completed */
     while (bs->active_flush_req) {
-        qemu_co_queue_wait(&bs->flush_queue);
+        qemu_co_queue_wait(&bs->flush_queue, NULL);
     }
 
     bs->active_flush_req = true;
diff --git a/block/nbd-client.c b/block/nbd-client.c
index XXXXXXX..XXXXXXX 100644
--- a/block/nbd-client.c
+++ b/block/nbd-client.c
@@ -XXX,XX +XXX,XX @@ static void nbd_coroutine_start(NBDClientSession *s,
     /* Poor man semaphore.  The free_sema is locked when no other request
      * can be accepted, and unlocked after receiving one reply.  */
     if (s->in_flight == MAX_NBD_REQUESTS) {
-        qemu_co_queue_wait(&s->free_sema);
+        qemu_co_queue_wait(&s->free_sema, NULL);
         assert(s->in_flight < MAX_NBD_REQUESTS);
     }
     s->in_flight++;
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
             if (bytes == 0) {
                 /* Wait for the dependency to complete. We need to recheck
                  * the free/allocated clusters when we continue. */
-                qemu_co_mutex_unlock(&s->lock);
-                qemu_co_queue_wait(&old_alloc->dependent_requests);
-                qemu_co_mutex_lock(&s->lock);
+                qemu_co_queue_wait(&old_alloc->dependent_requests, &s->lock);
                 return -EAGAIN;
             }
         }
diff --git a/block/sheepdog.c b/block/sheepdog.c
index XXXXXXX..XXXXXXX 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -XXX,XX +XXX,XX @@ static void wait_for_overlapping_aiocb(BDRVSheepdogState *s, SheepdogAIOCB *acb)
 retry:
     QLIST_FOREACH(cb, &s->inflight_aiocb_head, aiocb_siblings) {
         if (AIOCBOverlapping(acb, cb)) {
-            qemu_co_queue_wait(&s->overlapping_queue);
+            qemu_co_queue_wait(&s->overlapping_queue, NULL);
             goto retry;
         }
     }
diff --git a/block/throttle-groups.c b/block/throttle-groups.c
index XXXXXXX..XXXXXXX 100644
--- a/block/throttle-groups.c
+++ b/block/throttle-groups.c
@@ -XXX,XX +XXX,XX @@ void coroutine_fn throttle_group_co_io_limits_intercept(BlockBackend *blk,
     if (must_wait || blkp->pending_reqs[is_write]) {
         blkp->pending_reqs[is_write]++;
         qemu_mutex_unlock(&tg->lock);
-        qemu_co_queue_wait(&blkp->throttled_reqs[is_write]);
+        qemu_co_queue_wait(&blkp->throttled_reqs[is_write], NULL);
         qemu_mutex_lock(&tg->lock);
         blkp->pending_reqs[is_write]--;
     }
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn v9fs_flush(void *opaque)
         /*
          * Wait for pdu to complete.
          */
-        qemu_co_queue_wait(&cancel_pdu->complete);
+        qemu_co_queue_wait(&cancel_pdu->complete, NULL);
         cancel_pdu->cancelled = 0;
         pdu_free(cancel_pdu);
     }
diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-lock.c
+++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@ void qemu_co_queue_init(CoQueue *queue)
     QSIMPLEQ_INIT(&queue->entries);
 }
 
-void coroutine_fn qemu_co_queue_wait(CoQueue *queue)
+void coroutine_fn qemu_co_queue_wait(CoQueue *queue, CoMutex *mutex)
 {
     Coroutine *self = qemu_coroutine_self();
     QSIMPLEQ_INSERT_TAIL(&queue->entries, self, co_queue_next);
+
+    if (mutex) {
+        qemu_co_mutex_unlock(mutex);
+    }
+
+    /* There is no race condition here.  Other threads will call
+     * aio_co_schedule on our AioContext, which can reenter this
+     * coroutine but only after this yield and after the main loop
+     * has gone through the next iteration.
+     */
     qemu_coroutine_yield();
     assert(qemu_in_coroutine());
+
+    /* TODO: OSv implements wait morphing here, where the wakeup
+     * primitive automatically places the woken coroutine on the
+     * mutex's queue.  This avoids the thundering herd effect.
+     */
+    if (mutex) {
+        qemu_co_mutex_lock(mutex);
+    }
 }
 
 /**
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_rdlock(CoRwlock *lock)
     Coroutine *self = qemu_coroutine_self();
 
     while (lock->writer) {
-        qemu_co_queue_wait(&lock->queue);
+        qemu_co_queue_wait(&lock->queue, NULL);
     }
     lock->reader++;
     self->locks_held++;
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_wrlock(CoRwlock *lock)
     Coroutine *self = qemu_coroutine_self();
 
     while (lock->writer || lock->reader) {
-        qemu_co_queue_wait(&lock->queue);
+        qemu_co_queue_wait(&lock->queue, NULL);
     }
     lock->writer = true;
     self->locks_held++;
-- 
2.9.3

From: Paolo Bonzini <pbonzini@redhat.com>

This adds a CoMutex around the existing CoQueue.  Because the write-side
can just take CoMutex, the old "writer" field is not necessary anymore.
Instead of removing it altogether, count the number of pending writers
during a read-side critical section and forbid further readers from
entering.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-7-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/qemu/coroutine.h   |  3 ++-
 util/qemu-coroutine-lock.c | 35 ++++++++++++++++++++++++-----------
 2 files changed, 26 insertions(+), 12 deletions(-)

diff --git a/include/qemu/coroutine.h b/include/qemu/coroutine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/coroutine.h
+++ b/include/qemu/coroutine.h
@@ -XXX,XX +XXX,XX @@ bool qemu_co_queue_empty(CoQueue *queue);
 
 
 typedef struct CoRwlock {
-    bool writer;
+    int pending_writer;
     int reader;
+    CoMutex mutex;
     CoQueue queue;
 } CoRwlock;
 
diff --git a/util/qemu-coroutine-lock.c b/util/qemu-coroutine-lock.c
index XXXXXXX..XXXXXXX 100644
--- a/util/qemu-coroutine-lock.c
+++ b/util/qemu-coroutine-lock.c
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_init(CoRwlock *lock)
 {
     memset(lock, 0, sizeof(*lock));
     qemu_co_queue_init(&lock->queue);
+    qemu_co_mutex_init(&lock->mutex);
 }
 
 void qemu_co_rwlock_rdlock(CoRwlock *lock)
 {
     Coroutine *self = qemu_coroutine_self();
 
-    while (lock->writer) {
-        qemu_co_queue_wait(&lock->queue, NULL);
+    qemu_co_mutex_lock(&lock->mutex);
+    /* For fairness, wait if a writer is in line.  */
+    while (lock->pending_writer) {
+        qemu_co_queue_wait(&lock->queue, &lock->mutex);
     }
     lock->reader++;
+    qemu_co_mutex_unlock(&lock->mutex);
+
+    /* The rest of the read-side critical section is run without the mutex.  */
     self->locks_held++;
 }
 
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_unlock(CoRwlock *lock)
     Coroutine *self = qemu_coroutine_self();
 
     assert(qemu_in_coroutine());
-    if (lock->writer) {
-        lock->writer = false;
+    if (!lock->reader) {
+        /* The critical section started in qemu_co_rwlock_wrlock.  */
         qemu_co_queue_restart_all(&lock->queue);
     } else {
+        self->locks_held--;
+
+        qemu_co_mutex_lock(&lock->mutex);
         lock->reader--;
         assert(lock->reader >= 0);
         /* Wakeup only one waiting writer */
@@ -XXX,XX +XXX,XX @@ void qemu_co_rwlock_unlock(CoRwlock *lock)
             qemu_co_queue_next(&lock->queue);
         }
     }
-    self->locks_held--;
+    qemu_co_mutex_unlock(&lock->mutex);
 }
 
 void qemu_co_rwlock_wrlock(CoRwlock *lock)
 {
-    Coroutine *self = qemu_coroutine_self();
-
-    while (lock->writer || lock->reader) {
-        qemu_co_queue_wait(&lock->queue, NULL);
+    qemu_co_mutex_lock(&lock->mutex);
+    lock->pending_writer++;
+    while (lock->reader) {
+        qemu_co_queue_wait(&lock->queue, &lock->mutex);
     }
-    lock->writer = true;
-    self->locks_held++;
+    lock->pending_writer--;
+
+    /* The rest of the write-side critical section is run with
+     * the mutex taken, so that lock->reader remains zero.
+     * There is no need to update self->locks_held.
+     */
 }
-- 
2.9.3