The following changes since commit 91f5f7a5df1fda8c34677a7c49ee8a4bb5b56a36:

Merge remote-tracking branch 'remotes/lvivier-gitlab/tags/linux-user-for-7.0-pull-request' into staging (2022-01-12 11:51:47 +0000)

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

for you to fetch changes up to db608fb78444c58896db69495729e4458eeaace1:

virtio: unify dataplane and non-dataplane ->handle_output() (2022-01-12 17:09:39 +0000)

Stefan Hajnoczi (6):

aio-posix: split poll check from ready handler

virtio: get rid of VirtIOHandleAIOOutput

virtio-blk: drop unused virtio_blk_handle_vq() return value

virtio-scsi: prepare virtio_scsi_handle_cmd for dataplane

virtio: use ->handle_output() instead of ->handle_aio_output()

virtio: unify dataplane and non-dataplane ->handle_output()

include/block/aio.h | 4 +-

include/hw/virtio/virtio-blk.h | 2 +-

include/hw/virtio/virtio.h | 5 +-

util/aio-posix.h | 1 +

block/curl.c | 11 ++--

block/export/fuse.c | 4 +-

block/io_uring.c | 19 ++++---

block/iscsi.c | 4 +-

block/linux-aio.c | 16 +++---

block/nfs.c | 6 +--

block/nvme.c | 51 ++++++++++++-------

block/ssh.c | 4 +-

block/win32-aio.c | 4 +-

hw/block/dataplane/virtio-blk.c | 16 +-----

hw/block/virtio-blk.c | 14 ++----

hw/scsi/virtio-scsi-dataplane.c | 60 +++-------------------

hw/scsi/virtio-scsi.c | 2 +-

hw/virtio/virtio.c | 73 +++++++++------------------

hw/xen/xen-bus.c | 6 +--

io/channel-command.c | 6 ++-

io/channel-file.c | 3 +-

io/channel-socket.c | 3 +-

migration/rdma.c | 8 +--

tests/unit/test-aio.c | 4 +-

tests/unit/test-fdmon-epoll.c | 4 +-

util/aio-posix.c | 89 +++++++++++++++++++++++++--------

util/aio-win32.c | 4 +-

util/async.c | 10 +++-

util/main-loop.c | 4 +-

util/qemu-coroutine-io.c | 5 +-

util/vhost-user-server.c | 11 ++--

31 files changed, 221 insertions(+), 232 deletions(-)

2.34.1

Adaptive polling measures the execution time of the polling check plus

handlers called when a polled event becomes ready. Handlers can take a

significant amount of time, making it look like polling was running for

a long time when in fact the event handler was running for a long time.

For example, on Linux the io_submit(2) syscall invoked when a virtio-blk

device's virtqueue becomes ready can take 10s of microseconds. This

can exceed the default polling interval (32 microseconds) and cause

adaptive polling to stop polling.

By excluding the handler's execution time from the polling check we make

the adaptive polling calculation more accurate. As a result, the event

loop now stays in polling mode where previously it would have fallen

back to file descriptor monitoring.

The following data was collected with virtio-blk num-queues=2

event_idx=off using an IOThread. Before:

168k IOPS, IOThread syscalls:

9837.115 ( 0.020 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 16, iocbpp: 0x7fcb9f937db0) = 16

9837.158 ( 0.002 ms): IO iothread1/620155 write(fd: 103, buf: 0x556a2ef71b88, count: 8) = 8

9837.161 ( 0.001 ms): IO iothread1/620155 write(fd: 104, buf: 0x556a2ef71b88, count: 8) = 8

9837.163 ( 0.001 ms): IO iothread1/620155 ppoll(ufds: 0x7fcb90002800, nfds: 4, tsp: 0x7fcb9f1342d0, sigsetsize: 8) = 3

9837.164 ( 0.001 ms): IO iothread1/620155 read(fd: 107, buf: 0x7fcb9f939cc0, count: 512) = 8

9837.174 ( 0.001 ms): IO iothread1/620155 read(fd: 105, buf: 0x7fcb9f939cc0, count: 512) = 8

9837.176 ( 0.001 ms): IO iothread1/620155 read(fd: 106, buf: 0x7fcb9f939cc0, count: 512) = 8

9837.209 ( 0.035 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 32, iocbpp: 0x7fca7d0cebe0) = 32

174k IOPS (+3.6%), IOThread syscalls:

9809.566 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0cdd62be0) = 32

9809.625 ( 0.001 ms): IO iothread1/623061 write(fd: 103, buf: 0x5647cfba5f58, count: 8) = 8

9809.627 ( 0.002 ms): IO iothread1/623061 write(fd: 104, buf: 0x5647cfba5f58, count: 8) = 8

9809.663 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0d0388b50) = 32

Notice that ppoll(2) and eventfd read(2) syscalls are eliminated because

the IOThread stays in polling mode instead of falling back to file

descriptor monitoring.

As usual, polling is not implemented on Windows so this patch ignores

the new io_poll_read() callback in aio-win32.c.

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>

Message-id: 20211207132336.36627-2-stefanha@redhat.com

[Fixed up aio_set_event_notifier() calls in

tests/unit/test-fdmon-epoll.c added after this series was queued.

--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

include/block/aio.h | 4 +-

util/aio-posix.h | 1 +

block/curl.c | 11 +++--

block/export/fuse.c | 4 +-

block/io_uring.c | 19 ++++----

block/iscsi.c | 4 +-

block/linux-aio.c | 16 ++++---

block/nfs.c | 6 +--

block/nvme.c | 51 +++++++++++++-------

block/ssh.c | 4 +-

block/win32-aio.c | 4 +-

hw/virtio/virtio.c | 16 ++++---

hw/xen/xen-bus.c | 6 +--

io/channel-command.c | 6 ++-

io/channel-file.c | 3 +-

io/channel-socket.c | 3 +-

migration/rdma.c | 8 ++--

tests/unit/test-aio.c | 4 +-

tests/unit/test-fdmon-epoll.c | 4 +-

util/aio-posix.c | 89 ++++++++++++++++++++++++++---------

util/aio-win32.c | 4 +-

util/async.c | 10 +++-

util/main-loop.c | 4 +-

util/qemu-coroutine-io.c | 5 +-

util/vhost-user-server.c | 11 +++--

25 files changed, 193 insertions(+), 104 deletions(-)

@@ -XXX,XX +XXX,XX @@ void aio_set_fd_handler(AioContext *ctx,

IOHandler *io_read,

IOHandler *io_write,

AioPollFn *io_poll,

+ IOHandler *io_poll_ready,

void *opaque);

/* Set polling begin/end callbacks for a file descriptor that has already been

@@ -XXX,XX +XXX,XX @@ void aio_set_event_notifier(AioContext *ctx,

EventNotifier *notifier,

bool is_external,

EventNotifierHandler *io_read,

- AioPollFn *io_poll);

+ AioPollFn *io_poll,

+ EventNotifierHandler *io_poll_ready);

/* Set polling begin/end callbacks for an event notifier that has already been

* registered with aio_set_event_notifier. Do nothing if the event notifier is

diff --git a/util/aio-posix.h b/util/aio-posix.h

index XXXXXXX..XXXXXXX 100644

--- a/util/aio-posix.h

+++ b/util/aio-posix.h

@@ -XXX,XX +XXX,XX @@ struct AioHandler {

IOHandler *io_read;

IOHandler *io_write;

AioPollFn *io_poll;

+ IOHandler *io_poll_ready;

IOHandler *io_poll_begin;

IOHandler *io_poll_end;

void *opaque;

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 gboolean curl_drop_socket(void *key, void *value, void *opaque)

BDRVCURLState *s = socket->s;

aio_set_fd_handler(s->aio_context, socket->fd, false,

- NULL, NULL, NULL, NULL);

+ NULL, NULL, NULL, NULL, NULL);

return true;

}

@@ -XXX,XX +XXX,XX @@ static int curl_sock_cb(CURL *curl, curl_socket_t fd, int action,

switch (action) {

case CURL_POLL_IN:

aio_set_fd_handler(s->aio_context, fd, false,

- curl_multi_do, NULL, NULL, socket);

+ curl_multi_do, NULL, NULL, NULL, socket);

break;

case CURL_POLL_OUT:

aio_set_fd_handler(s->aio_context, fd, false,

- NULL, curl_multi_do, NULL, socket);

+ NULL, curl_multi_do, NULL, NULL, socket);

break;

case CURL_POLL_INOUT:

aio_set_fd_handler(s->aio_context, fd, false,

- curl_multi_do, curl_multi_do, NULL, socket);

+ curl_multi_do, curl_multi_do,

+ NULL, NULL, socket);

break;

case CURL_POLL_REMOVE:

aio_set_fd_handler(s->aio_context, fd, false,

- NULL, NULL, NULL, NULL);

+ NULL, NULL, NULL, NULL, NULL);

break;

}

diff --git a/block/export/fuse.c b/block/export/fuse.c

index XXXXXXX..XXXXXXX 100644

--- a/block/export/fuse.c

+++ b/block/export/fuse.c

@@ -XXX,XX +XXX,XX @@ static int setup_fuse_export(FuseExport *exp, const char *mountpoint,

aio_set_fd_handler(exp->common.ctx,

fuse_session_fd(exp->fuse_session), true,

- read_from_fuse_export, NULL, NULL, exp);

+ read_from_fuse_export, NULL, NULL, NULL, exp);

exp->fd_handler_set_up = true;

return 0;

@@ -XXX,XX +XXX,XX @@ static void fuse_export_shutdown(BlockExport *blk_exp)

if (exp->fd_handler_set_up) {

aio_set_fd_handler(exp->common.ctx,

fuse_session_fd(exp->fuse_session), true,

- NULL, NULL, NULL, NULL);

+ NULL, NULL, NULL, NULL, NULL);

exp->fd_handler_set_up = false;

}

diff --git a/block/io_uring.c b/block/io_uring.c

index XXXXXXX..XXXXXXX 100644

--- a/block/io_uring.c

+++ b/block/io_uring.c

@@ -XXX,XX +XXX,XX @@ static bool qemu_luring_poll_cb(void *opaque)

{

LuringState *s = opaque;

- if (io_uring_cq_ready(&s->ring)) {

- luring_process_completions_and_submit(s);

- return true;

- }

+ return io_uring_cq_ready(&s->ring);

+}

- return false;

+static void qemu_luring_poll_ready(void *opaque)

+{

+ LuringState *s = opaque;

+

+ luring_process_completions_and_submit(s);

}

static void ioq_init(LuringQueue *io_q)

@@ -XXX,XX +XXX,XX @@ int coroutine_fn luring_co_submit(BlockDriverState *bs, LuringState *s, int fd,

void luring_detach_aio_context(LuringState *s, AioContext *old_context)

{

- aio_set_fd_handler(old_context, s->ring.ring_fd, false, NULL, NULL, NULL,

- s);

+ aio_set_fd_handler(old_context, s->ring.ring_fd, false,

+ NULL, NULL, NULL, NULL, s);

qemu_bh_delete(s->completion_bh);

s->aio_context = NULL;

}

@@ -XXX,XX +XXX,XX @@ void luring_attach_aio_context(LuringState *s, AioContext *new_context)

s->aio_context = new_context;

s->completion_bh = aio_bh_new(new_context, qemu_luring_completion_bh, s);

aio_set_fd_handler(s->aio_context, s->ring.ring_fd, false,

- qemu_luring_completion_cb, NULL, qemu_luring_poll_cb, s);

+ qemu_luring_completion_cb, NULL,

+ qemu_luring_poll_cb, qemu_luring_poll_ready, s);

}

LuringState *luring_init(Error **errp)

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_set_events(IscsiLun *iscsilun)

false,

(ev & POLLIN) ? iscsi_process_read : NULL,

(ev & POLLOUT) ? iscsi_process_write : NULL,

- NULL,