[Qemu-devel] [PULL 0/1] Block patches
[PULL 0/1] Block patches
1
The following changes since commit d47a851caeda96d5979bf48d4bae6a87784ad91d:
1
The following changes since commit 887cba855bb6ff4775256f7968409281350b568c:
2
2
3
Merge remote-tracking branch 'remotes/juanquintela/tags/migration/20170601' into staging (2017-06-02 14:07:53 +0100)
3
configure: Fix cross-building for RISCV host (v5) (2023-07-11 17:56:09 +0100)
4
4
5
are available in the git repository at:
5
are available in the Git repository at:
6
6
7
git://github.com/codyprime/qemu-kvm-jtc.git tags/block-pull-request
7
https://gitlab.com/stefanha/qemu.git tags/block-pull-request
8
8
9
for you to fetch changes up to df3a429ae82c0f45becdfab105617701d75e0f05:
9
for you to fetch changes up to 75dcb4d790bbe5327169fd72b185960ca58e2fa6:
10
10
11
gluster: add support for PREALLOC_MODE_FALLOC (2017-06-02 10:51:47 -0400)
11
virtio-blk: fix host notifier issues during dataplane start/stop (2023-07-12 15:20:32 -0400)
12
12
13
----------------------------------------------------------------
13
----------------------------------------------------------------
14
Gluster patch(es)
14
Pull request
15
15
----------------------------------------------------------------
16
----------------------------------------------------------------
16
17
17
Niels de Vos (1):
18
Stefan Hajnoczi (1):
18
gluster: add support for PREALLOC_MODE_FALLOC
19
virtio-blk: fix host notifier issues during dataplane start/stop
19
20
20
block/gluster.c | 78 ++++++++++++++++++++++++++++++---------------------------
21
hw/block/dataplane/virtio-blk.c | 67 +++++++++++++++++++--------------
21
configure | 6 +++++
22
1 file changed, 38 insertions(+), 29 deletions(-)
22
2 files changed, 47 insertions(+), 37 deletions(-)
23
23
24
--
24
--
25
2.9.3
25
2.40.1
26
27
diff view generated by jsdifflib
[Qemu-devel] [PULL 1/1] gluster: add support for PREALLOC_MODE_FALLOC
[PULL 1/1] virtio-blk: fix host notifier issues during dataplane start/stop
1
From: Niels de Vos <ndevos@redhat.com>
1
The main loop thread can consume 100% CPU when using --device
2
virtio-blk-pci,iothread=<iothread>. ppoll() constantly returns but
3
reading virtqueue host notifiers fails with EAGAIN. The file descriptors
4
are stale and remain registered with the AioContext because of bugs in
5
the virtio-blk dataplane start/stop code.
2
6
3
Add missing support for "preallocation=falloc" to the Gluster block
7
The problem is that the dataplane start/stop code involves drain
4
driver. This change bases its logic on that of block/file-posix.c and
8
operations, which call virtio_blk_drained_begin() and
5
removed the gluster_supports_zerofill() and qemu_gluster_zerofill()
9
virtio_blk_drained_end() at points where the host notifier is not
6
functions in favour of #ifdef checks in an easy to read
10
operational:
7
switch-statement.
11
- In virtio_blk_data_plane_start(), blk_set_aio_context() drains after
12
vblk->dataplane_started has been set to true but the host notifier has
13
not been attached yet.
14
- In virtio_blk_data_plane_stop(), blk_drain() and blk_set_aio_context()
15
drain after the host notifier has already been detached but with
16
vblk->dataplane_started still set to true.
8
17
9
Both glfs_zerofill() and glfs_fallocate() have been introduced with
18
I would like to simplify ->ioeventfd_start/stop() to avoid interactions
10
GlusterFS 3.5.0 (pkg-config glusterfs-api = 6). A #define for the
19
with drain entirely, but couldn't find a way to do that. Instead, this
11
availability of glfs_fallocate() has been added to ./configure.
20
patch accepts the fragile nature of the code and reorders it so that
21
vblk->dataplane_started is false during drain operations. This way the
22
virtio_blk_drained_begin() and virtio_blk_drained_end() calls don't
23
touch the host notifier. The result is that
24
virtio_blk_data_plane_start() and virtio_blk_data_plane_stop() have
25
complete control over the host notifier and stale file descriptors are
26
no longer left in the AioContext.
12
27
13
Reported-by: Satheesaran Sundaramoorthi <sasundar@redhat.com>
28
This patch fixes the 100% CPU consumption in the main loop thread and
14
Signed-off-by: Niels de Vos <ndevos@redhat.com>
29
correctly moves host notifier processing to the IOThread.
15
Message-id: 20170528063114.28691-1-ndevos@redhat.com
30
16
URL: https://bugzilla.redhat.com/1450759
31
Fixes: 1665d9326fd2 ("virtio-blk: implement BlockDevOps->drained_begin()")
17
Signed-off-by: Niels de Vos <ndevos@redhat.com>
32
Reported-by: Lukáš Doktor <ldoktor@redhat.com>
18
Signed-off-by: Jeff Cody <jcody@redhat.com>
33
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
34
Tested-by: Lukas Doktor <ldoktor@redhat.com>
35
Message-id: 20230704151527.193586-1-stefanha@redhat.com
36
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
19
---
37
---
20
block/gluster.c | 78 ++++++++++++++++++++++++++++++---------------------------
38
hw/block/dataplane/virtio-blk.c | 67 +++++++++++++++++++--------------
21
configure | 6 +++++
39
1 file changed, 38 insertions(+), 29 deletions(-)
22
2 files changed, 47 insertions(+), 37 deletions(-)
23
40
24
diff --git a/block/gluster.c b/block/gluster.c
41
diff --git a/hw/block/dataplane/virtio-blk.c b/hw/block/dataplane/virtio-blk.c
25
index XXXXXXX..XXXXXXX 100644
42
index XXXXXXX..XXXXXXX 100644
26
--- a/block/gluster.c
43
--- a/hw/block/dataplane/virtio-blk.c
27
+++ b/block/gluster.c
44
+++ b/hw/block/dataplane/virtio-blk.c
28
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qemu_gluster_co_pwrite_zeroes(BlockDriverState *bs,
45
@@ -XXX,XX +XXX,XX @@ int virtio_blk_data_plane_start(VirtIODevice *vdev)
29
qemu_coroutine_yield();
46
30
return acb.ret;
47
memory_region_transaction_commit();
48
49
- /*
50
- * These fields are visible to the IOThread so we rely on implicit barriers
51
- * in aio_context_acquire() on the write side and aio_notify_accept() on
52
- * the read side.
53
- */
54
- s->starting = false;
55
- vblk->dataplane_started = true;
56
trace_virtio_blk_data_plane_start(s);
57
58
old_context = blk_get_aio_context(s->conf->conf.blk);
59
@@ -XXX,XX +XXX,XX @@ int virtio_blk_data_plane_start(VirtIODevice *vdev)
60
event_notifier_set(virtio_queue_get_host_notifier(vq));
61
}
62
63
+ /*
64
+ * These fields must be visible to the IOThread when it processes the
65
+ * virtqueue, otherwise it will think dataplane has not started yet.
66
+ *
67
+ * Make sure ->dataplane_started is false when blk_set_aio_context() is
68
+ * called above so that draining does not cause the host notifier to be
69
+ * detached/attached prematurely.
70
+ */
71
+ s->starting = false;
72
+ vblk->dataplane_started = true;
73
+ smp_wmb(); /* paired with aio_notify_accept() on the read side */
74
+
75
/* Get this show started by hooking up our callbacks */
76
if (!blk_in_drain(s->conf->conf.blk)) {
77
aio_context_acquire(s->ctx);
78
@@ -XXX,XX +XXX,XX @@ int virtio_blk_data_plane_start(VirtIODevice *vdev)
79
fail_guest_notifiers:
80
vblk->dataplane_disabled = true;
81
s->starting = false;
82
- vblk->dataplane_started = true;
83
return -ENOSYS;
31
}
84
}
32
-
85
33
-static inline bool gluster_supports_zerofill(void)
86
@@ -XXX,XX +XXX,XX @@ void virtio_blk_data_plane_stop(VirtIODevice *vdev)
34
-{
87
aio_wait_bh_oneshot(s->ctx, virtio_blk_data_plane_stop_bh, s);
35
- return 1;
36
-}
37
-
38
-static inline int qemu_gluster_zerofill(struct glfs_fd *fd, int64_t offset,
39
- int64_t size)
40
-{
41
- return glfs_zerofill(fd, offset, size);
42
-}
43
-
44
-#else
45
-static inline bool gluster_supports_zerofill(void)
46
-{
47
- return 0;
48
-}
49
-
50
-static inline int qemu_gluster_zerofill(struct glfs_fd *fd, int64_t offset,
51
- int64_t size)
52
-{
53
- return 0;
54
-}
55
#endif
56
57
static int qemu_gluster_create(const char *filename,
58
@@ -XXX,XX +XXX,XX @@ static int qemu_gluster_create(const char *filename,
59
struct glfs *glfs;
60
struct glfs_fd *fd;
61
int ret = 0;
62
- int prealloc = 0;
63
+ PreallocMode prealloc;
64
int64_t total_size = 0;
65
char *tmp = NULL;
66
+ Error *local_err = NULL;
67
68
gconf = g_new0(BlockdevOptionsGluster, 1);
69
gconf->debug = qemu_opt_get_number_del(opts, GLUSTER_OPT_DEBUG,
70
@@ -XXX,XX +XXX,XX @@ static int qemu_gluster_create(const char *filename,
71
BDRV_SECTOR_SIZE);
72
73
tmp = qemu_opt_get_del(opts, BLOCK_OPT_PREALLOC);
74
- if (!tmp || !strcmp(tmp, "off")) {
75
- prealloc = 0;
76
- } else if (!strcmp(tmp, "full") && gluster_supports_zerofill()) {
77
- prealloc = 1;
78
- } else {
79
- error_setg(errp, "Invalid preallocation mode: '%s'"
80
- " or GlusterFS doesn't support zerofill API", tmp);
81
+ prealloc = qapi_enum_parse(PreallocMode_lookup, tmp,
82
+ PREALLOC_MODE__MAX, PREALLOC_MODE_OFF,
83
+ &local_err);
84
+ g_free(tmp);
85
+ if (local_err) {
86
+ error_propagate(errp, local_err);
87
ret = -EINVAL;
88
goto out;
89
}
88
}
90
@@ -XXX,XX +XXX,XX @@ static int qemu_gluster_create(const char *filename,
89
91
O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, S_IRUSR | S_IWUSR);
90
+ /*
92
if (!fd) {
91
+ * Batch all the host notifiers in a single transaction to avoid
93
ret = -errno;
92
+ * quadratic time complexity in address_space_update_ioeventfds().
94
- } else {
93
+ */
95
+ goto out;
94
+ memory_region_transaction_begin();
95
+
96
+ for (i = 0; i < nvqs; i++) {
97
+ virtio_bus_set_host_notifier(VIRTIO_BUS(qbus), i, false);
96
+ }
98
+ }
97
+
99
+
98
+ switch (prealloc) {
100
+ /*
99
+#ifdef CONFIG_GLUSTERFS_FALLOCATE
101
+ * The transaction expects the ioeventfds to be open when it
100
+ case PREALLOC_MODE_FALLOC:
102
+ * commits. Do it now, before the cleanup loop.
101
+ if (glfs_fallocate(fd, 0, 0, total_size)) {
103
+ */
102
+ error_setg(errp, "Could not preallocate data for the new file");
104
+ memory_region_transaction_commit();
103
+ ret = -errno;
105
+
104
+ }
106
+ for (i = 0; i < nvqs; i++) {
105
+ break;
107
+ virtio_bus_cleanup_host_notifier(VIRTIO_BUS(qbus), i);
106
+#endif /* CONFIG_GLUSTERFS_FALLOCATE */
107
+#ifdef CONFIG_GLUSTERFS_ZEROFILL
108
+ case PREALLOC_MODE_FULL:
109
if (!glfs_ftruncate(fd, total_size)) {
110
- if (prealloc && qemu_gluster_zerofill(fd, 0, total_size)) {
111
+ if (glfs_zerofill(fd, 0, total_size)) {
112
+ error_setg(errp, "Could not zerofill the new file");
113
ret = -errno;
114
}
115
} else {
116
+ error_setg(errp, "Could not resize file");
117
ret = -errno;
118
}
119
+ break;
120
+#endif /* CONFIG_GLUSTERFS_ZEROFILL */
121
+ case PREALLOC_MODE_OFF:
122
+ if (glfs_ftruncate(fd, total_size) != 0) {
123
+ ret = -errno;
124
+ error_setg(errp, "Could not resize file");
125
+ }
126
+ break;
127
+ default:
128
+ ret = -EINVAL;
129
+ error_setg(errp, "Unsupported preallocation mode: %s",
130
+ PreallocMode_lookup[prealloc]);
131
+ break;
132
+ }
108
+ }
133
134
- if (glfs_close(fd) != 0) {
135
- ret = -errno;
136
- }
137
+ if (glfs_close(fd) != 0) {
138
+ ret = -errno;
139
}
140
out:
141
- g_free(tmp);
142
qapi_free_BlockdevOptionsGluster(gconf);
143
glfs_clear_preopened(glfs);
144
return ret;
145
diff --git a/configure b/configure
146
index XXXXXXX..XXXXXXX 100755
147
--- a/configure
148
+++ b/configure
149
@@ -XXX,XX +XXX,XX @@ seccomp=""
150
glusterfs=""
151
glusterfs_xlator_opt="no"
152
glusterfs_discard="no"
153
+glusterfs_fallocate="no"
154
glusterfs_zerofill="no"
155
gtk=""
156
gtkabi=""
157
@@ -XXX,XX +XXX,XX @@ if test "$glusterfs" != "no" ; then
158
glusterfs_discard="yes"
159
fi
160
if $pkg_config --atleast-version=6 glusterfs-api; then
161
+ glusterfs_fallocate="yes"
162
glusterfs_zerofill="yes"
163
fi
164
else
165
@@ -XXX,XX +XXX,XX @@ if test "$glusterfs_discard" = "yes" ; then
166
echo "CONFIG_GLUSTERFS_DISCARD=y" >> $config_host_mak
167
fi
168
169
+if test "$glusterfs_fallocate" = "yes" ; then
170
+ echo "CONFIG_GLUSTERFS_FALLOCATE=y" >> $config_host_mak
171
+fi
172
+
109
+
173
if test "$glusterfs_zerofill" = "yes" ; then
110
+ /*
174
echo "CONFIG_GLUSTERFS_ZEROFILL=y" >> $config_host_mak
111
+ * Set ->dataplane_started to false before draining so that host notifiers
175
fi
112
+ * are not detached/attached anymore.
113
+ */
114
+ vblk->dataplane_started = false;
115
+
116
aio_context_acquire(s->ctx);
117
118
/* Wait for virtio_blk_dma_restart_bh() and in flight I/O to complete */
119
@@ -XXX,XX +XXX,XX @@ void virtio_blk_data_plane_stop(VirtIODevice *vdev)
120
121
aio_context_release(s->ctx);
122
123
- /*
124
- * Batch all the host notifiers in a single transaction to avoid
125
- * quadratic time complexity in address_space_update_ioeventfds().
126
- */
127
- memory_region_transaction_begin();
128
-
129
- for (i = 0; i < nvqs; i++) {
130
- virtio_bus_set_host_notifier(VIRTIO_BUS(qbus), i, false);
131
- }
132
-
133
- /*
134
- * The transaction expects the ioeventfds to be open when it
135
- * commits. Do it now, before the cleanup loop.
136
- */
137
- memory_region_transaction_commit();
138
-
139
- for (i = 0; i < nvqs; i++) {
140
- virtio_bus_cleanup_host_notifier(VIRTIO_BUS(qbus), i);
141
- }
142
-
143
qemu_bh_cancel(s->bh);
144
notify_guest_bh(s); /* final chance to notify guest */
145
146
/* Clean up guest notifier (irq) */
147
k->set_guest_notifiers(qbus->parent, nvqs, false);
148
149
- vblk->dataplane_started = false;
150
s->stopping = false;
151
}
176
--
152
--
177
2.9.3
153
2.40.1
178
154
179
155
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.