[Qemu-devel] [PULL 0/9] Block patches
[libvirt] [PULL 0/3] Block patches
1
The following changes since commit e47f81b617684c4546af286d307b69014a83538a:
1
The following changes since commit aceeaa69d28e6f08a24395d0aa6915b687d0a681:
2
2
3
Merge remote-tracking branch 'remotes/thibault/tags/samuel-thibault' into staging (2019-02-07 18:53:25 +0000)
3
Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2019-12-17' into staging (2019-12-17 15:55:20 +0000)
4
4
5
are available in the Git repository at:
5
are available in the Git repository at:
6
6
7
git://github.com/stefanha/qemu.git tags/block-pull-request
7
https://github.com/stefanha/qemu.git tags/block-pull-request
8
8
9
for you to fetch changes up to 55140166667bb555c5d05165b93b25557d2e6397:
9
for you to fetch changes up to 725fe5d10dbd4259b1853b7d253cef83a3c0d22a:
10
10
11
tests/virtio-blk: add test for WRITE_ZEROES command (2019-02-11 11:58:17 +0800)
11
virtio-blk: fix out-of-bounds access to bitmap in notify_guest_bh (2019-12-19 16:20:25 +0000)
12
12
13
----------------------------------------------------------------
13
----------------------------------------------------------------
14
Pull request
14
Pull request
15
15
16
User-visible changes:
17
18
* virtio-blk: DISCARD and WRITE_ZEROES support
19
20
----------------------------------------------------------------
16
----------------------------------------------------------------
21
17
22
Peter Xu (1):
18
Li Hangjing (1):
23
iothread: fix iothread hang when stop too soon
19
virtio-blk: fix out-of-bounds access to bitmap in notify_guest_bh
24
20
25
Stefano Garzarella (7):
21
Stefan Hajnoczi (2):
26
virtio-blk: cleanup using VirtIOBlock *s and VirtIODevice *vdev
22
virtio-blk: deprecate SCSI passthrough
27
virtio-blk: add acct_failed param to virtio_blk_handle_rw_error()
23
docs: fix rst syntax errors in unbuilt docs
28
virtio-blk: add host_features field in VirtIOBlock
29
virtio-blk: add "discard" and "write-zeroes" properties
30
virtio-blk: add DISCARD and WRITE_ZEROES features
31
tests/virtio-blk: change assert on data_size in virtio_blk_request()
32
tests/virtio-blk: add test for WRITE_ZEROES command
33
24
34
Vladimir Sementsov-Ogievskiy (1):
25
docs/arm-cpu-features.rst | 6 +++---
35
qemugdb/coroutine: fix arch_prctl has unknown return type
26
docs/virtio-net-failover.rst | 4 ++--
36
27
docs/virtio-pmem.rst | 19 ++++++++++---------
37
include/hw/virtio/virtio-blk.h | 5 +-
28
hw/block/dataplane/virtio-blk.c | 2 +-
38
hw/block/virtio-blk.c | 236 +++++++++++++++++++++++++++++----
29
qemu-deprecated.texi | 11 +++++++++++
39
hw/core/machine.c | 2 +
30
5 files changed, 27 insertions(+), 15 deletions(-)
40
iothread.c | 6 +-
41
tests/virtio-blk-test.c | 75 ++++++++++-
42
scripts/qemugdb/coroutine.py | 2 +-
43
6 files changed, 297 insertions(+), 29 deletions(-)
44
31
45
--
32
--
46
2.20.1
33
2.23.0
47
34
35
--
36
libvir-list mailing list
37
libvir-list@redhat.com
38
https://www.redhat.com/mailman/listinfo/libvir-list
48
39
diff view generated by jsdifflib
[Qemu-devel] [PULL 1/9] iothread: fix iothread hang when stop too soon
Deleted patch
1
From: Peter Xu <peterx@redhat.com>
2
1
3
Lukas reported an hard to reproduce QMP iothread hang on s390 that
4
QEMU might hang at pthread_join() of the QMP monitor iothread before
5
quitting:
6
7
Thread 1
8
#0 0x000003ffad10932c in pthread_join
9
#1 0x0000000109e95750 in qemu_thread_join
10
at /home/thuth/devel/qemu/util/qemu-thread-posix.c:570
11
#2 0x0000000109c95a1c in iothread_stop
12
#3 0x0000000109bb0874 in monitor_cleanup
13
#4 0x0000000109b55042 in main
14
15
While the iothread is still in the main loop:
16
17
Thread 4
18
#0 0x000003ffad0010e4 in ??
19
#1 0x000003ffad553958 in g_main_context_iterate.isra.19
20
#2 0x000003ffad553d90 in g_main_loop_run
21
#3 0x0000000109c9585a in iothread_run
22
at /home/thuth/devel/qemu/iothread.c:74
23
#4 0x0000000109e94752 in qemu_thread_start
24
at /home/thuth/devel/qemu/util/qemu-thread-posix.c:502
25
#5 0x000003ffad10825a in start_thread
26
#6 0x000003ffad00dcf2 in thread_start
27
28
IMHO it's because there's a race between the main thread and iothread
29
when stopping the thread in following sequence:
30
31
main thread iothread
32
=========== ==============
33
aio_poll()
34
iothread_get_g_main_context
35
set iothread->worker_context
36
iothread_stop
37
schedule iothread_stop_bh
38
execute iothread_stop_bh [1]
39
set iothread->running=false
40
(since main_loop==NULL so
41
skip to quit main loop.
42
Note: although main_loop is
43
NULL but worker_context is
44
not!)
45
atomic_read(&iothread->worker_context) [2]
46
create main_loop object
47
g_main_loop_run() [3]
48
pthread_join() [4]
49
50
We can see that when execute iothread_stop_bh() at [1] it's possible
51
that main_loop is still NULL because it's only created until the first
52
check of the worker_context later at [2]. Then the iothread will hang
53
in the main loop [3] and it'll starve the main thread too [4].
54
55
Here the simple solution should be that we check again the "running"
56
variable before check against worker_context.
57
58
CC: Thomas Huth <thuth@redhat.com>
59
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
60
CC: Stefan Hajnoczi <stefanha@redhat.com>
61
CC: Lukáš Doktor <ldoktor@redhat.com>
62
CC: Markus Armbruster <armbru@redhat.com>
63
CC: Eric Blake <eblake@redhat.com>
64
CC: Paolo Bonzini <pbonzini@redhat.com>
65
Reported-by: Lukáš Doktor <ldoktor@redhat.com>
66
Signed-off-by: Peter Xu <peterx@redhat.com>
67
Tested-by: Thomas Huth <thuth@redhat.com>
68
Message-id: 20190129051432.22023-1-peterx@redhat.com
69
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
70
---
71
iothread.c | 6 +++++-
72
1 file changed, 5 insertions(+), 1 deletion(-)
73
74
diff --git a/iothread.c b/iothread.c
75
index XXXXXXX..XXXXXXX 100644
76
--- a/iothread.c
77
+++ b/iothread.c
78
@@ -XXX,XX +XXX,XX @@ static void *iothread_run(void *opaque)
79
while (iothread->running) {
80
aio_poll(iothread->ctx, true);
81
82
- if (atomic_read(&iothread->worker_context)) {
83
+ /*
84
+ * We must check the running state again in case it was
85
+ * changed in previous aio_poll()
86
+ */
87
+ if (iothread->running && atomic_read(&iothread->worker_context)) {
88
GMainLoop *loop;
89
90
g_main_context_push_thread_default(iothread->worker_context);
91
--
92
2.20.1
93
94
diff view generated by jsdifflib
[Qemu-devel] [PULL 9/9] tests/virtio-blk: add test for WRITE_ZEROES command
[libvirt] [PULL 1/3] virtio-blk: deprecate SCSI passthrough
1
From: Stefano Garzarella <sgarzare@redhat.com>
1
The Linux virtio_blk.ko guest driver is removing legacy SCSI passthrough
2
support. Deprecate this feature in QEMU too.
2
3
3
If the WRITE_ZEROES feature is enabled, we check this command
4
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4
in the test_basic().
5
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
5
6
Reviewed-by: Christoph Hellwig <hch@lst.de>
6
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
7
Reviewed-by: Thomas Huth <thuth@redhat.com>
7
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
8
Message-id: 20191213144626.1208237-1-stefanha@redhat.com
8
Acked-by: Thomas Huth <thuth@redhat.com>
9
Message-Id: <20191213144626.1208237-1-stefanha@redhat.com>
9
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
10
Acked-by: Pankaj Gupta <pagupta@redhat.com>
11
Message-id: 20190208134950.187665-7-sgarzare@redhat.com
12
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
13
---
11
---
14
tests/virtio-blk-test.c | 60 +++++++++++++++++++++++++++++++++++++++++
12
qemu-deprecated.texi | 11 +++++++++++
15
1 file changed, 60 insertions(+)
13
1 file changed, 11 insertions(+)
16
14
17
diff --git a/tests/virtio-blk-test.c b/tests/virtio-blk-test.c
15
diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi
18
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
19
--- a/tests/virtio-blk-test.c
17
--- a/qemu-deprecated.texi
20
+++ b/tests/virtio-blk-test.c
18
+++ b/qemu-deprecated.texi
21
@@ -XXX,XX +XXX,XX @@ static void test_basic(QVirtioDevice *dev, QGuestAllocator *alloc,
19
@@ -XXX,XX +XXX,XX @@ spec you can use the ``-cpu rv64gcsu,priv_spec=v1.9.1`` command line argument.
22
20
23
guest_free(alloc, req_addr);
21
@section Device options
24
22
25
+ if (features & (1u << VIRTIO_BLK_F_WRITE_ZEROES)) {
23
+@subsection Emulated device options
26
+ struct virtio_blk_discard_write_zeroes dwz_hdr;
27
+ void *expected;
28
+
24
+
29
+ /*
25
+@subsubsection -device virtio-blk,scsi=on|off (since 5.0.0)
30
+ * WRITE_ZEROES request on the same sector of previous test where
31
+ * we wrote "TEST".
32
+ */
33
+ req.type = VIRTIO_BLK_T_WRITE_ZEROES;
34
+ req.data = (char *) &dwz_hdr;
35
+ dwz_hdr.sector = 0;
36
+ dwz_hdr.num_sectors = 1;
37
+ dwz_hdr.flags = 0;
38
+
26
+
39
+ req_addr = virtio_blk_request(alloc, dev, &req, sizeof(dwz_hdr));
27
+The virtio-blk SCSI passthrough feature is a legacy VIRTIO feature. VIRTIO 1.0
28
+and later do not support it because the virtio-scsi device was introduced for
29
+full SCSI support. Use virtio-scsi instead when SCSI passthrough is required.
40
+
30
+
41
+ free_head = qvirtqueue_add(vq, req_addr, 16, false, true);
31
+Note this also applies to ``-device virtio-blk-pci,scsi=on|off'', which is an
42
+ qvirtqueue_add(vq, req_addr + 16, sizeof(dwz_hdr), false, true);
32
+alias.
43
+ qvirtqueue_add(vq, req_addr + 16 + sizeof(dwz_hdr), 1, true, false);
44
+
33
+
45
+ qvirtqueue_kick(dev, vq, free_head);
34
@subsection Block device options
46
+
35
47
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL,
36
@subsubsection "backing": "" (since 2.12.0)
48
+ QVIRTIO_BLK_TIMEOUT_US);
49
+ status = readb(req_addr + 16 + sizeof(dwz_hdr));
50
+ g_assert_cmpint(status, ==, 0);
51
+
52
+ guest_free(alloc, req_addr);
53
+
54
+ /* Read request to check if the sector contains all zeroes */
55
+ req.type = VIRTIO_BLK_T_IN;
56
+ req.ioprio = 1;
57
+ req.sector = 0;
58
+ req.data = g_malloc0(512);
59
+
60
+ req_addr = virtio_blk_request(alloc, dev, &req, 512);
61
+
62
+ g_free(req.data);
63
+
64
+ free_head = qvirtqueue_add(vq, req_addr, 16, false, true);
65
+ qvirtqueue_add(vq, req_addr + 16, 512, true, true);
66
+ qvirtqueue_add(vq, req_addr + 528, 1, true, false);
67
+
68
+ qvirtqueue_kick(dev, vq, free_head);
69
+
70
+ qvirtio_wait_used_elem(dev, vq, free_head, NULL,
71
+ QVIRTIO_BLK_TIMEOUT_US);
72
+ status = readb(req_addr + 528);
73
+ g_assert_cmpint(status, ==, 0);
74
+
75
+ data = g_malloc(512);
76
+ expected = g_malloc0(512);
77
+ memread(req_addr + 16, data, 512);
78
+ g_assert_cmpmem(data, 512, expected, 512);
79
+ g_free(expected);
80
+ g_free(data);
81
+
82
+ guest_free(alloc, req_addr);
83
+ }
84
+
85
if (features & (1u << VIRTIO_F_ANY_LAYOUT)) {
86
/* Write and read with 2 descriptor layout */
87
/* Write request */
88
--
37
--
89
2.20.1
38
2.23.0
90
39
40
--
41
libvir-list mailing list
42
libvir-list@redhat.com
43
https://www.redhat.com/mailman/listinfo/libvir-list
91
44
diff view generated by jsdifflib
[Qemu-devel] [PULL 8/9] tests/virtio-blk: change assert on data_size in virtio_blk_request()
[libvirt] [PULL 2/3] docs: fix rst syntax errors in unbuilt docs
1
From: Stefano Garzarella <sgarzare@redhat.com>
1
The .rst files outside docs/{devel,interop,specs} aren't built yet and
2
therefore a few syntax errors have slipped through. Fix them.
2
3
3
The size of data in the virtio_blk_request must be a multiple
4
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4
of 512 bytes for IN and OUT requests, or a multiple of the size
5
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
5
of struct virtio_blk_discard_write_zeroes for DISCARD and
6
Message-Id: <20191111094411.427174-1-stefanha@redhat.com>
6
WRITE_ZEROES requests.
7
8
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
9
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
10
Reviewed-by: Thomas Huth <thuth@redhat.com>
11
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
12
Acked-by: Pankaj Gupta <pagupta@redhat.com>
13
Message-id: 20190208134950.187665-6-sgarzare@redhat.com
14
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
7
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
15
---
8
---
16
tests/virtio-blk-test.c | 15 ++++++++++++++-
9
docs/arm-cpu-features.rst | 6 +++---
17
1 file changed, 14 insertions(+), 1 deletion(-)
10
docs/virtio-net-failover.rst | 4 ++--
11
docs/virtio-pmem.rst | 19 ++++++++++---------
12
3 files changed, 15 insertions(+), 14 deletions(-)
18
13
19
diff --git a/tests/virtio-blk-test.c b/tests/virtio-blk-test.c
14
diff --git a/docs/arm-cpu-features.rst b/docs/arm-cpu-features.rst
20
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
21
--- a/tests/virtio-blk-test.c
16
--- a/docs/arm-cpu-features.rst
22
+++ b/tests/virtio-blk-test.c
17
+++ b/docs/arm-cpu-features.rst
23
@@ -XXX,XX +XXX,XX @@ static uint64_t virtio_blk_request(QGuestAllocator *alloc, QVirtioDevice *d,
18
@@ -XXX,XX +XXX,XX @@ CPU type is possible with the `query-cpu-model-expansion` QMP command.
24
uint64_t addr;
19
Below are some examples where `scripts/qmp/qmp-shell` (see the top comment
25
uint8_t status = 0xFF;
20
block in the script for usage) is used to issue the QMP commands.
26
21
27
- g_assert_cmpuint(data_size % 512, ==, 0);
22
-(1) Determine which CPU features are available for the `max` CPU type
28
+ switch (req->type) {
23
- (Note, we started QEMU with qemu-system-aarch64, so `max` is
29
+ case VIRTIO_BLK_T_IN:
24
- implementing the ARMv8-A reference manual in this case)::
30
+ case VIRTIO_BLK_T_OUT:
25
+1. Determine which CPU features are available for the `max` CPU type
31
+ g_assert_cmpuint(data_size % 512, ==, 0);
26
+ (Note, we started QEMU with qemu-system-aarch64, so `max` is
32
+ break;
27
+ implementing the ARMv8-A reference manual in this case)::
33
+ case VIRTIO_BLK_T_DISCARD:
28
34
+ case VIRTIO_BLK_T_WRITE_ZEROES:
29
(QEMU) query-cpu-model-expansion type=full model={"name":"max"}
35
+ g_assert_cmpuint(data_size %
30
{ "return": {
36
+ sizeof(struct virtio_blk_discard_write_zeroes), ==, 0);
31
diff --git a/docs/virtio-net-failover.rst b/docs/virtio-net-failover.rst
37
+ break;
32
index XXXXXXX..XXXXXXX 100644
38
+ default:
33
--- a/docs/virtio-net-failover.rst
39
+ g_assert_cmpuint(data_size, ==, 0);
34
+++ b/docs/virtio-net-failover.rst
40
+ }
35
@@ -XXX,XX +XXX,XX @@
36
-========================
37
+======================================
38
QEMU virtio-net standby (net_failover)
39
-========================
40
+======================================
41
42
This document explains the setup and usage of virtio-net standby feature which
43
is used to create a net_failover pair of devices.
44
diff --git a/docs/virtio-pmem.rst b/docs/virtio-pmem.rst
45
index XXXXXXX..XXXXXXX 100644
46
--- a/docs/virtio-pmem.rst
47
+++ b/docs/virtio-pmem.rst
48
@@ -XXX,XX +XXX,XX @@ virtio pmem usage
49
-----------------
50
51
A virtio pmem device backed by a memory-backend-file can be created on
52
- the QEMU command line as in the following example:
53
+ the QEMU command line as in the following example::
54
55
- -object memory-backend-file,id=mem1,share,mem-path=./virtio_pmem.img,size=4G
56
- -device virtio-pmem-pci,memdev=mem1,id=nv1
57
+ -object memory-backend-file,id=mem1,share,mem-path=./virtio_pmem.img,size=4G
58
+ -device virtio-pmem-pci,memdev=mem1,id=nv1
59
60
- where:
61
- - "object memory-backend-file,id=mem1,share,mem-path=<image>, size=<image size>"
62
- creates a backend file with the specified size.
63
+ where:
64
65
- - "device virtio-pmem-pci,id=nvdimm1,memdev=mem1" creates a virtio pmem
66
- pci device whose storage is provided by above memory backend device.
67
+ - "object memory-backend-file,id=mem1,share,mem-path=<image>, size=<image size>"
68
+ creates a backend file with the specified size.
41
+
69
+
42
addr = guest_alloc(alloc, sizeof(*req) + data_size);
70
+ - "device virtio-pmem-pci,id=nvdimm1,memdev=mem1" creates a virtio pmem
43
71
+ pci device whose storage is provided by above memory backend device.
44
virtio_blk_fix_request(d, req);
72
73
Multiple virtio pmem devices can be created if multiple pairs of "-object"
74
and "-device" are provided.
75
@@ -XXX,XX +XXX,XX @@ memory backing has to be added via 'object_add'; afterwards, the virtio
76
pmem device can be added via 'device_add'.
77
78
For example, the following commands add another 4GB virtio pmem device to
79
-the guest:
80
+the guest::
81
82
(qemu) object_add memory-backend-file,id=mem2,share=on,mem-path=virtio_pmem2.img,size=4G
83
(qemu) device_add virtio-pmem-pci,id=virtio_pmem2,memdev=mem2
45
--
84
--
46
2.20.1
85
2.23.0
47
86
48
87
--
88
libvir-list mailing list
89
libvir-list@redhat.com
90
https://www.redhat.com/mailman/listinfo/libvir-list
diff view generated by jsdifflib
[Qemu-devel] [PULL 2/9] qemugdb/coroutine: fix arch_prctl has unknown return type
[libvirt] [PULL 3/3] virtio-blk: fix out-of-bounds access to bitmap in notify_guest_bh
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
1
From: Li Hangjing <lihangjing@baidu.com>
2
2
3
qemu coroutine command results in following error output:
3
When the number of a virtio-blk device's virtqueues is larger than
4
BITS_PER_LONG, the out-of-bounds access to bitmap[ ] will occur.
4
5
5
Python Exception <class 'gdb.error'> 'arch_prctl' has unknown return
6
Fixes: e21737ab15 ("virtio-blk: multiqueue batch notify")
6
type; cast the call to its declared return type: Error occurred in
7
Cc: qemu-stable@nongnu.org
7
Python command: 'arch_prctl' has unknown return type; cast the call to
8
Cc: Stefan Hajnoczi <stefanha@redhat.com>
8
its declared return type
9
Signed-off-by: Li Hangjing <lihangjing@baidu.com>
9
10
Reviewed-by: Xie Yongji <xieyongji@baidu.com>
10
Fix it by giving it what it wants: arch_prctl return type.
11
Reviewed-by: Chai Wen <chaiwen@baidu.com>
11
12
Message-id: 20191216023050.48620-1-lihangjing@baidu.com
12
Information on the topic:
13
Message-Id: <20191216023050.48620-1-lihangjing@baidu.com>
13
https://sourceware.org/gdb/onlinedocs/gdb/Calling.html
14
15
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
16
Message-id: 20190206151425.105871-1-vsementsov@virtuozzo.com
17
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
14
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
18
---
15
---
19
scripts/qemugdb/coroutine.py | 2 +-
16
hw/block/dataplane/virtio-blk.c | 2 +-
20
1 file changed, 1 insertion(+), 1 deletion(-)
17
1 file changed, 1 insertion(+), 1 deletion(-)
21
18
22
diff --git a/scripts/qemugdb/coroutine.py b/scripts/qemugdb/coroutine.py
19
diff --git a/hw/block/dataplane/virtio-blk.c b/hw/block/dataplane/virtio-blk.c
23
index XXXXXXX..XXXXXXX 100644
20
index XXXXXXX..XXXXXXX 100644
24
--- a/scripts/qemugdb/coroutine.py
21
--- a/hw/block/dataplane/virtio-blk.c
25
+++ b/scripts/qemugdb/coroutine.py
22
+++ b/hw/block/dataplane/virtio-blk.c
26
@@ -XXX,XX +XXX,XX @@ def get_fs_base():
23
@@ -XXX,XX +XXX,XX @@ static void notify_guest_bh(void *opaque)
27
pthread_self().'''
24
memset(s->batch_notify_vqs, 0, sizeof(bitmap));
28
# %rsp - 120 is scratch space according to the SystemV ABI
25
29
old = gdb.parse_and_eval('*(uint64_t*)($rsp - 120)')
26
for (j = 0; j < nvqs; j += BITS_PER_LONG) {
30
- gdb.execute('call arch_prctl(0x1003, $rsp - 120)', False, True)
27
- unsigned long bits = bitmap[j];
31
+ gdb.execute('call (int)arch_prctl(0x1003, $rsp - 120)', False, True)
28
+ unsigned long bits = bitmap[j / BITS_PER_LONG];
32
fs_base = gdb.parse_and_eval('*(uint64_t*)($rsp - 120)')
29
33
gdb.execute('set *(uint64_t*)($rsp - 120) = %s' % old, False, True)
30
while (bits != 0) {
34
return fs_base
31
unsigned i = j + ctzl(bits);
35
--
32
--
36
2.20.1
33
2.23.0
37
34
35
--
36
libvir-list mailing list
37
libvir-list@redhat.com
38
https://www.redhat.com/mailman/listinfo/libvir-list
38
39
diff view generated by jsdifflib
[Qemu-devel] [PULL 3/9] virtio-blk: cleanup using VirtIOBlock *s and VirtIODevice *vdev
Deleted patch
1
From: Stefano Garzarella <sgarzare@redhat.com>
2
1
3
In several part we still using req->dev or VIRTIO_DEVICE(req->dev)
4
when we have already defined s and vdev pointers:
5
VirtIOBlock *s = req->dev;
6
VirtIODevice *vdev = VIRTIO_DEVICE(s);
7
8
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
9
Reviewed-by: Liam Merwick <liam.merwick@oracle.com>
10
Message-id: 20190208142347.214815-1-sgarzare@redhat.com
11
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
12
---
13
hw/block/virtio-blk.c | 22 +++++++++-------------
14
1 file changed, 9 insertions(+), 13 deletions(-)
15
16
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
17
index XXXXXXX..XXXXXXX 100644
18
--- a/hw/block/virtio-blk.c
19
+++ b/hw/block/virtio-blk.c
20
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_req_complete(VirtIOBlockReq *req, unsigned char status)
21
static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
22
bool is_read)
23
{
24
- BlockErrorAction action = blk_get_error_action(req->dev->blk,
25
- is_read, error);
26
VirtIOBlock *s = req->dev;
27
+ BlockErrorAction action = blk_get_error_action(s->blk, is_read, error);
28
29
if (action == BLOCK_ERROR_ACTION_STOP) {
30
/* Break the link as the next request is going to be parsed from the
31
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_flush_complete(void *opaque, int ret)
32
}
33
34
virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
35
- block_acct_done(blk_get_stats(req->dev->blk), &req->acct);
36
+ block_acct_done(blk_get_stats(s->blk), &req->acct);
37
virtio_blk_free_request(req);
38
39
out:
40
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
41
- sizeof(struct virtio_blk_inhdr);
42
iov_discard_back(in_iov, &in_num, sizeof(struct virtio_blk_inhdr));
43
44
- type = virtio_ldl_p(VIRTIO_DEVICE(req->dev), &req->out.type);
45
+ type = virtio_ldl_p(vdev, &req->out.type);
46
47
/* VIRTIO_BLK_T_OUT defines the command direction. VIRTIO_BLK_T_BARRIER
48
* is an optional flag. Although a guest should not send this flag if
49
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
50
case VIRTIO_BLK_T_IN:
51
{
52
bool is_write = type & VIRTIO_BLK_T_OUT;
53
- req->sector_num = virtio_ldq_p(VIRTIO_DEVICE(req->dev),
54
- &req->out.sector);
55
+ req->sector_num = virtio_ldq_p(vdev, &req->out.sector);
56
57
if (is_write) {
58
qemu_iovec_init_external(&req->qiov, out_iov, out_num);
59
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
60
req->qiov.size / BDRV_SECTOR_SIZE);
61
}
62
63
- if (!virtio_blk_sect_range_ok(req->dev, req->sector_num,
64
- req->qiov.size)) {
65
+ if (!virtio_blk_sect_range_ok(s, req->sector_num, req->qiov.size)) {
66
virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
67
- block_acct_invalid(blk_get_stats(req->dev->blk),
68
+ block_acct_invalid(blk_get_stats(s->blk),
69
is_write ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ);
70
virtio_blk_free_request(req);
71
return 0;
72
}
73
74
- block_acct_start(blk_get_stats(req->dev->blk),
75
- &req->acct, req->qiov.size,
76
+ block_acct_start(blk_get_stats(s->blk), &req->acct, req->qiov.size,
77
is_write ? BLOCK_ACCT_WRITE : BLOCK_ACCT_READ);
78
79
/* merge would exceed maximum number of requests or IO direction
80
* changes */
81
if (mrb->num_reqs > 0 && (mrb->num_reqs == VIRTIO_BLK_MAX_MERGE_REQS ||
82
is_write != mrb->is_write ||
83
- !req->dev->conf.request_merging)) {
84
- virtio_blk_submit_multireq(req->dev->blk, mrb);
85
+ !s->conf.request_merging)) {
86
+ virtio_blk_submit_multireq(s->blk, mrb);
87
}
88
89
assert(mrb->num_reqs < VIRTIO_BLK_MAX_MERGE_REQS);
90
--
91
2.20.1
92
93
diff view generated by jsdifflib
[Qemu-devel] [PULL 4/9] virtio-blk: add acct_failed param to virtio_blk_handle_rw_error()
Deleted patch
1
From: Stefano Garzarella <sgarzare@redhat.com>
2
1
3
We add acct_failed param in order to use virtio_blk_handle_rw_error()
4
also when is not required to call block_acct_failed(). (eg. a discard
5
operation is failed)
6
7
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
8
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
9
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
10
Acked-by: Pankaj Gupta <pagupta@redhat.com>
11
Message-id: 20190208134950.187665-2-sgarzare@redhat.com
12
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
13
---
14
hw/block/virtio-blk.c | 10 ++++++----
15
1 file changed, 6 insertions(+), 4 deletions(-)
16
17
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
18
index XXXXXXX..XXXXXXX 100644
19
--- a/hw/block/virtio-blk.c
20
+++ b/hw/block/virtio-blk.c
21
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_req_complete(VirtIOBlockReq *req, unsigned char status)
22
}
23
24
static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
25
- bool is_read)
26
+ bool is_read, bool acct_failed)
27
{
28
VirtIOBlock *s = req->dev;
29
BlockErrorAction action = blk_get_error_action(s->blk, is_read, error);
30
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
31
s->rq = req;
32
} else if (action == BLOCK_ERROR_ACTION_REPORT) {
33
virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
34
- block_acct_failed(blk_get_stats(s->blk), &req->acct);
35
+ if (acct_failed) {
36
+ block_acct_failed(blk_get_stats(s->blk), &req->acct);
37
+ }
38
virtio_blk_free_request(req);
39
}
40
41
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_rw_complete(void *opaque, int ret)
42
* the memory until the request is completed (which will
43
* happen on the other side of the migration).
44
*/
45
- if (virtio_blk_handle_rw_error(req, -ret, is_read)) {
46
+ if (virtio_blk_handle_rw_error(req, -ret, is_read, true)) {
47
continue;
48
}
49
}
50
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_flush_complete(void *opaque, int ret)
51
52
aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
53
if (ret) {
54
- if (virtio_blk_handle_rw_error(req, -ret, 0)) {
55
+ if (virtio_blk_handle_rw_error(req, -ret, 0, true)) {
56
goto out;
57
}
58
}
59
--
60
2.20.1
61
62
diff view generated by jsdifflib
[Qemu-devel] [PULL 5/9] virtio-blk: add host_features field in VirtIOBlock
Deleted patch
1
From: Stefano Garzarella <sgarzare@redhat.com>
2
1
3
Since configurable features for virtio-blk are growing, this patch
4
adds host_features field in the struct VirtIOBlock. (as in virtio-net)
5
In this way, we can avoid to add new fields for new properties and
6
we can directly set VIRTIO_BLK_F* flags in the host_features.
7
8
We update "config-wce" and "scsi" property definition to use the new
9
host_features field without change the behaviour.
10
11
Suggested-by: Michael S. Tsirkin <mst@redhat.com>
12
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
13
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
14
Acked-by: Pankaj Gupta <pagupta@redhat.com>
15
Message-id: 20190208134950.187665-3-sgarzare@redhat.com
16
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
17
---
18
include/hw/virtio/virtio-blk.h | 3 +--
19
hw/block/virtio-blk.c | 16 +++++++++-------
20
2 files changed, 10 insertions(+), 9 deletions(-)
21
22
diff --git a/include/hw/virtio/virtio-blk.h b/include/hw/virtio/virtio-blk.h
23
index XXXXXXX..XXXXXXX 100644
24
--- a/include/hw/virtio/virtio-blk.h
25
+++ b/include/hw/virtio/virtio-blk.h
26
@@ -XXX,XX +XXX,XX @@ struct VirtIOBlkConf
27
BlockConf conf;
28
IOThread *iothread;
29
char *serial;
30
- uint32_t scsi;
31
- uint32_t config_wce;
32
uint32_t request_merging;
33
uint16_t num_queues;
34
uint16_t queue_size;
35
@@ -XXX,XX +XXX,XX @@ typedef struct VirtIOBlock {
36
bool dataplane_disabled;
37
bool dataplane_started;
38
struct VirtIOBlockDataPlane *dataplane;
39
+ uint64_t host_features;
40
} VirtIOBlock;
41
42
typedef struct VirtIOBlockReq {
43
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
44
index XXXXXXX..XXXXXXX 100644
45
--- a/hw/block/virtio-blk.c
46
+++ b/hw/block/virtio-blk.c
47
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_scsi_req(VirtIOBlockReq *req)
48
*/
49
scsi = (void *)elem->in_sg[elem->in_num - 2].iov_base;
50
51
- if (!blk->conf.scsi) {
52
+ if (!virtio_has_feature(blk->host_features, VIRTIO_BLK_F_SCSI)) {
53
status = VIRTIO_BLK_S_UNSUPP;
54
goto fail;
55
}
56
@@ -XXX,XX +XXX,XX @@ static uint64_t virtio_blk_get_features(VirtIODevice *vdev, uint64_t features,
57
{
58
VirtIOBlock *s = VIRTIO_BLK(vdev);
59
60
+ /* Firstly sync all virtio-blk possible supported features */
61
+ features |= s->host_features;
62
+
63
virtio_add_feature(&features, VIRTIO_BLK_F_SEG_MAX);
64
virtio_add_feature(&features, VIRTIO_BLK_F_GEOMETRY);
65
virtio_add_feature(&features, VIRTIO_BLK_F_TOPOLOGY);
66
virtio_add_feature(&features, VIRTIO_BLK_F_BLK_SIZE);
67
if (virtio_has_feature(features, VIRTIO_F_VERSION_1)) {
68
- if (s->conf.scsi) {
69
+ if (virtio_has_feature(s->host_features, VIRTIO_BLK_F_SCSI)) {
70
error_setg(errp, "Please set scsi=off for virtio-blk devices in order to use virtio 1.0");
71
return 0;
72
}
73
@@ -XXX,XX +XXX,XX @@ static uint64_t virtio_blk_get_features(VirtIODevice *vdev, uint64_t features,
74
virtio_add_feature(&features, VIRTIO_BLK_F_SCSI);
75
}
76
77
- if (s->conf.config_wce) {
78
- virtio_add_feature(&features, VIRTIO_BLK_F_CONFIG_WCE);
79
- }
80
if (blk_enable_write_cache(s->blk)) {
81
virtio_add_feature(&features, VIRTIO_BLK_F_WCE);
82
}
83
@@ -XXX,XX +XXX,XX @@ static Property virtio_blk_properties[] = {
84
DEFINE_BLOCK_ERROR_PROPERTIES(VirtIOBlock, conf.conf),
85
DEFINE_BLOCK_CHS_PROPERTIES(VirtIOBlock, conf.conf),
86
DEFINE_PROP_STRING("serial", VirtIOBlock, conf.serial),
87
- DEFINE_PROP_BIT("config-wce", VirtIOBlock, conf.config_wce, 0, true),
88
+ DEFINE_PROP_BIT64("config-wce", VirtIOBlock, host_features,
89
+ VIRTIO_BLK_F_CONFIG_WCE, true),
90
#ifdef __linux__
91
- DEFINE_PROP_BIT("scsi", VirtIOBlock, conf.scsi, 0, false),
92
+ DEFINE_PROP_BIT64("scsi", VirtIOBlock, host_features,
93
+ VIRTIO_BLK_F_SCSI, false),
94
#endif
95
DEFINE_PROP_BIT("request-merging", VirtIOBlock, conf.request_merging, 0,
96
true),
97
--
98
2.20.1
99
100
diff view generated by jsdifflib
[Qemu-devel] [PULL 6/9] virtio-blk: add "discard" and "write-zeroes" properties
Deleted patch
1
From: Stefano Garzarella <sgarzare@redhat.com>
2
1
3
In order to avoid migration issues, we enable DISCARD and
4
WRITE_ZEROES features only for machine type >= 4.0
5
6
As discussed with Michael S. Tsirkin and Stefan Hajnoczi on the
7
list [1], DISCARD operation should not have security implications
8
(eg. page cache attacks), so we can enable it by default.
9
10
[1] https://lists.gnu.org/archive/html/qemu-devel/2019-02/msg00504.html
11
12
Suggested-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
13
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
14
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
15
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
16
Acked-by: Pankaj Gupta <pagupta@redhat.com>
17
Message-id: 20190208134950.187665-4-sgarzare@redhat.com
18
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
19
---
20
hw/block/virtio-blk.c | 4 ++++
21
hw/core/machine.c | 2 ++
22
2 files changed, 6 insertions(+)
23
24
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
25
index XXXXXXX..XXXXXXX 100644
26
--- a/hw/block/virtio-blk.c
27
+++ b/hw/block/virtio-blk.c
28
@@ -XXX,XX +XXX,XX @@ static Property virtio_blk_properties[] = {
29
DEFINE_PROP_UINT16("queue-size", VirtIOBlock, conf.queue_size, 128),
30
DEFINE_PROP_LINK("iothread", VirtIOBlock, conf.iothread, TYPE_IOTHREAD,
31
IOThread *),
32
+ DEFINE_PROP_BIT64("discard", VirtIOBlock, host_features,
33
+ VIRTIO_BLK_F_DISCARD, true),
34
+ DEFINE_PROP_BIT64("write-zeroes", VirtIOBlock, host_features,
35
+ VIRTIO_BLK_F_WRITE_ZEROES, true),
36
DEFINE_PROP_END_OF_LIST(),
37
};
38
39
diff --git a/hw/core/machine.c b/hw/core/machine.c
40
index XXXXXXX..XXXXXXX 100644
41
--- a/hw/core/machine.c
42
+++ b/hw/core/machine.c
43
@@ -XXX,XX +XXX,XX @@ GlobalProperty hw_compat_3_1[] = {
44
{ "usb-kbd", "serial", "42" },
45
{ "usb-mouse", "serial", "42" },
46
{ "usb-kbd", "serial", "42" },
47
+ { "virtio-blk-device", "discard", "false" },
48
+ { "virtio-blk-device", "write-zeroes", "false" },
49
};
50
const size_t hw_compat_3_1_len = G_N_ELEMENTS(hw_compat_3_1);
51
52
--
53
2.20.1
54
55
diff view generated by jsdifflib
[Qemu-devel] [PULL 7/9] virtio-blk: add DISCARD and WRITE_ZEROES features
Deleted patch
1
From: Stefano Garzarella <sgarzare@redhat.com>
2
1
3
This patch adds the support of DISCARD and WRITE_ZEROES commands,
4
that have been introduced in the virtio-blk protocol to have
5
better performance when using SSD backend.
6
7
We support only one segment per request since multiple segments
8
are not widely used and there are no userspace APIs that allow
9
applications to submit multiple segments in a single call.
10
11
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
12
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
13
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
14
Acked-by: Pankaj Gupta <pagupta@redhat.com>
15
Message-id: 20190208134950.187665-5-sgarzare@redhat.com
16
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
17
---
18
include/hw/virtio/virtio-blk.h | 2 +
19
hw/block/virtio-blk.c | 184 +++++++++++++++++++++++++++++++++
20
2 files changed, 186 insertions(+)
21
22
diff --git a/include/hw/virtio/virtio-blk.h b/include/hw/virtio/virtio-blk.h
23
index XXXXXXX..XXXXXXX 100644
24
--- a/include/hw/virtio/virtio-blk.h
25
+++ b/include/hw/virtio/virtio-blk.h
26
@@ -XXX,XX +XXX,XX @@ struct VirtIOBlkConf
27
uint32_t request_merging;
28
uint16_t num_queues;
29
uint16_t queue_size;
30
+ uint32_t max_discard_sectors;
31
+ uint32_t max_write_zeroes_sectors;
32
};
33
34
struct VirtIOBlockDataPlane;
35
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
36
index XXXXXXX..XXXXXXX 100644
37
--- a/hw/block/virtio-blk.c
38
+++ b/hw/block/virtio-blk.c
39
@@ -XXX,XX +XXX,XX @@ out:
40
aio_context_release(blk_get_aio_context(s->conf.conf.blk));
41
}
42
43
+static void virtio_blk_discard_write_zeroes_complete(void *opaque, int ret)
44
+{
45
+ VirtIOBlockReq *req = opaque;
46
+ VirtIOBlock *s = req->dev;
47
+ bool is_write_zeroes = (virtio_ldl_p(VIRTIO_DEVICE(s), &req->out.type) &
48
+ ~VIRTIO_BLK_T_BARRIER) == VIRTIO_BLK_T_WRITE_ZEROES;
49
+
50
+ aio_context_acquire(blk_get_aio_context(s->conf.conf.blk));
51
+ if (ret) {
52
+ if (virtio_blk_handle_rw_error(req, -ret, false, is_write_zeroes)) {
53
+ goto out;
54
+ }
55
+ }
56
+
57
+ virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
58
+ if (is_write_zeroes) {
59
+ block_acct_done(blk_get_stats(s->blk), &req->acct);
60
+ }
61
+ virtio_blk_free_request(req);
62
+
63
+out:
64
+ aio_context_release(blk_get_aio_context(s->conf.conf.blk));
65
+}
66
+
67
#ifdef __linux__
68
69
typedef struct {
70
@@ -XXX,XX +XXX,XX @@ static bool virtio_blk_sect_range_ok(VirtIOBlock *dev,
71
return true;
72
}
73
74
+static uint8_t virtio_blk_handle_discard_write_zeroes(VirtIOBlockReq *req,
75
+ struct virtio_blk_discard_write_zeroes *dwz_hdr, bool is_write_zeroes)
76
+{
77
+ VirtIOBlock *s = req->dev;
78
+ VirtIODevice *vdev = VIRTIO_DEVICE(s);
79
+ uint64_t sector;
80
+ uint32_t num_sectors, flags, max_sectors;
81
+ uint8_t err_status;
82
+ int bytes;
83
+
84
+ sector = virtio_ldq_p(vdev, &dwz_hdr->sector);
85
+ num_sectors = virtio_ldl_p(vdev, &dwz_hdr->num_sectors);
86
+ flags = virtio_ldl_p(vdev, &dwz_hdr->flags);
87
+ max_sectors = is_write_zeroes ? s->conf.max_write_zeroes_sectors :
88
+ s->conf.max_discard_sectors;
89
+
90
+ /*
91
+ * max_sectors is at most BDRV_REQUEST_MAX_SECTORS, this check
92
+ * make us sure that "num_sectors << BDRV_SECTOR_BITS" can fit in
93
+ * the integer variable.
94
+ */
95
+ if (unlikely(num_sectors > max_sectors)) {
96
+ err_status = VIRTIO_BLK_S_IOERR;
97
+ goto err;
98
+ }
99
+
100
+ bytes = num_sectors << BDRV_SECTOR_BITS;
101
+
102
+ if (unlikely(!virtio_blk_sect_range_ok(s, sector, bytes))) {
103
+ err_status = VIRTIO_BLK_S_IOERR;
104
+ goto err;
105
+ }
106
+
107
+ /*
108
+ * The device MUST set the status byte to VIRTIO_BLK_S_UNSUPP for discard
109
+ * and write zeroes commands if any unknown flag is set.
110
+ */
111
+ if (unlikely(flags & ~VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP)) {
112
+ err_status = VIRTIO_BLK_S_UNSUPP;
113
+ goto err;
114
+ }
115
+
116
+ if (is_write_zeroes) { /* VIRTIO_BLK_T_WRITE_ZEROES */
117
+ int blk_aio_flags = 0;
118
+
119
+ if (flags & VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP) {
120
+ blk_aio_flags |= BDRV_REQ_MAY_UNMAP;
121
+ }
122
+
123
+ block_acct_start(blk_get_stats(s->blk), &req->acct, bytes,
124
+ BLOCK_ACCT_WRITE);
125
+
126
+ blk_aio_pwrite_zeroes(s->blk, sector << BDRV_SECTOR_BITS,
127
+ bytes, blk_aio_flags,
128
+ virtio_blk_discard_write_zeroes_complete, req);
129
+ } else { /* VIRTIO_BLK_T_DISCARD */
130
+ /*
131
+ * The device MUST set the status byte to VIRTIO_BLK_S_UNSUPP for
132
+ * discard commands if the unmap flag is set.
133
+ */
134
+ if (unlikely(flags & VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP)) {
135
+ err_status = VIRTIO_BLK_S_UNSUPP;
136
+ goto err;
137
+ }
138
+
139
+ blk_aio_pdiscard(s->blk, sector << BDRV_SECTOR_BITS, bytes,
140
+ virtio_blk_discard_write_zeroes_complete, req);
141
+ }
142
+
143
+ return VIRTIO_BLK_S_OK;
144
+
145
+err:
146
+ if (is_write_zeroes) {
147
+ block_acct_invalid(blk_get_stats(s->blk), BLOCK_ACCT_WRITE);
148
+ }
149
+ return err_status;
150
+}
151
+
152
static int virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
153
{
154
uint32_t type;
155
@@ -XXX,XX +XXX,XX @@ static int virtio_blk_handle_request(VirtIOBlockReq *req, MultiReqBuffer *mrb)
156
virtio_blk_free_request(req);
157
break;
158
}
159
+ /*
160
+ * VIRTIO_BLK_T_DISCARD and VIRTIO_BLK_T_WRITE_ZEROES are defined with
161
+ * VIRTIO_BLK_T_OUT flag set. We masked this flag in the switch statement,
162
+ * so we must mask it for these requests, then we will check if it is set.
163
+ */
164
+ case VIRTIO_BLK_T_DISCARD & ~VIRTIO_BLK_T_OUT:
165
+ case VIRTIO_BLK_T_WRITE_ZEROES & ~VIRTIO_BLK_T_OUT:
166
+ {
167
+ struct virtio_blk_discard_write_zeroes dwz_hdr;
168
+ size_t out_len = iov_size(out_iov, out_num);
169
+ bool is_write_zeroes = (type & ~VIRTIO_BLK_T_BARRIER) ==
170
+ VIRTIO_BLK_T_WRITE_ZEROES;
171
+ uint8_t err_status;
172
+
173
+ /*
174
+ * Unsupported if VIRTIO_BLK_T_OUT is not set or the request contains
175
+ * more than one segment.
176
+ */
177
+ if (unlikely(!(type & VIRTIO_BLK_T_OUT) ||
178
+ out_len > sizeof(dwz_hdr))) {
179
+ virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
180
+ virtio_blk_free_request(req);
181
+ return 0;
182
+ }
183
+
184
+ if (unlikely(iov_to_buf(out_iov, out_num, 0, &dwz_hdr,
185
+ sizeof(dwz_hdr)) != sizeof(dwz_hdr))) {
186
+ virtio_error(vdev, "virtio-blk discard/write_zeroes header"
187
+ " too short");
188
+ return -1;
189
+ }
190
+
191
+ err_status = virtio_blk_handle_discard_write_zeroes(req, &dwz_hdr,
192
+ is_write_zeroes);
193
+ if (err_status != VIRTIO_BLK_S_OK) {
194
+ virtio_blk_req_complete(req, err_status);
195
+ virtio_blk_free_request(req);
196
+ }
197
+
198
+ break;
199
+ }
200
default:
201
virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
202
virtio_blk_free_request(req);
203
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_update_config(VirtIODevice *vdev, uint8_t *config)
204
blkcfg.alignment_offset = 0;
205
blkcfg.wce = blk_enable_write_cache(s->blk);
206
virtio_stw_p(vdev, &blkcfg.num_queues, s->conf.num_queues);
207
+ if (virtio_has_feature(s->host_features, VIRTIO_BLK_F_DISCARD)) {
208
+ virtio_stl_p(vdev, &blkcfg.max_discard_sectors,
209
+ s->conf.max_discard_sectors);
210
+ virtio_stl_p(vdev, &blkcfg.discard_sector_alignment,
211
+ blk_size >> BDRV_SECTOR_BITS);
212
+ /*
213
+ * We support only one segment per request since multiple segments
214
+ * are not widely used and there are no userspace APIs that allow
215
+ * applications to submit multiple segments in a single call.
216
+ */
217
+ virtio_stl_p(vdev, &blkcfg.max_discard_seg, 1);
218
+ }
219
+ if (virtio_has_feature(s->host_features, VIRTIO_BLK_F_WRITE_ZEROES)) {
220
+ virtio_stl_p(vdev, &blkcfg.max_write_zeroes_sectors,
221
+ s->conf.max_write_zeroes_sectors);
222
+ blkcfg.write_zeroes_may_unmap = 1;
223
+ virtio_stl_p(vdev, &blkcfg.max_write_zeroes_seg, 1);
224
+ }
225
memcpy(config, &blkcfg, sizeof(struct virtio_blk_config));
226
}
227
228
@@ -XXX,XX +XXX,XX @@ static void virtio_blk_device_realize(DeviceState *dev, Error **errp)
229
return;
230
}
231
232
+ if (virtio_has_feature(s->host_features, VIRTIO_BLK_F_DISCARD) &&
233
+ (!conf->max_discard_sectors ||
234
+ conf->max_discard_sectors > BDRV_REQUEST_MAX_SECTORS)) {
235
+ error_setg(errp, "invalid max-discard-sectors property (%" PRIu32 ")"
236
+ ", must be between 1 and %d",
237
+ conf->max_discard_sectors, (int)BDRV_REQUEST_MAX_SECTORS);
238
+ return;
239
+ }
240
+
241
+ if (virtio_has_feature(s->host_features, VIRTIO_BLK_F_WRITE_ZEROES) &&
242
+ (!conf->max_write_zeroes_sectors ||
243
+ conf->max_write_zeroes_sectors > BDRV_REQUEST_MAX_SECTORS)) {
244
+ error_setg(errp, "invalid max-write-zeroes-sectors property (%" PRIu32
245
+ "), must be between 1 and %d",
246
+ conf->max_write_zeroes_sectors,
247
+ (int)BDRV_REQUEST_MAX_SECTORS);
248
+ return;
249
+ }
250
+
251
virtio_init(vdev, "virtio-blk", VIRTIO_ID_BLOCK,
252
sizeof(struct virtio_blk_config));
253
254
@@ -XXX,XX +XXX,XX @@ static Property virtio_blk_properties[] = {
255
VIRTIO_BLK_F_DISCARD, true),
256
DEFINE_PROP_BIT64("write-zeroes", VirtIOBlock, host_features,
257
VIRTIO_BLK_F_WRITE_ZEROES, true),
258
+ DEFINE_PROP_UINT32("max-discard-sectors", VirtIOBlock,
259
+ conf.max_discard_sectors, BDRV_REQUEST_MAX_SECTORS),
260
+ DEFINE_PROP_UINT32("max-write-zeroes-sectors", VirtIOBlock,
261
+ conf.max_write_zeroes_sectors, BDRV_REQUEST_MAX_SECTORS),
262
DEFINE_PROP_END_OF_LIST(),
263
};
264
265
--
266
2.20.1
267
268
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.