[PULL 0/3] Block patches
[Qemu-devel] [PULL for-2.11-rc1 v2 0/2] Block patches
1
The following changes since commit 9c125d17e9402c232c46610802e5931b3639d77b:
1
The following changes since commit b0fbe46ad82982b289a44ee2495b59b0bad8a842:
2
2
3
Merge tag 'pull-tcg-20220420' of https://gitlab.com/rth7680/qemu into staging (2022-04-20 16:43:11 -0700)
3
Update version for v2.11.0-rc0 release (2017-11-07 16:05:28 +0000)
4
4
5
are available in the Git repository at:
5
are available in the git repository at:
6
6
7
https://gitlab.com/stefanha/qemu.git tags/block-pull-request
7
git://github.com/stefanha/qemu.git tags/block-pull-request
8
8
9
for you to fetch changes up to d45c83328feab2e4083991693160f0a417cfd9b0:
9
for you to fetch changes up to ef6dada8b44e1e7c4bec5c1115903af9af415b50:
10
10
11
virtiofsd: Add docs/helper for killpriv_v2/no_killpriv_v2 option (2022-04-21 12:05:15 +0200)
11
util/async: use atomic_mb_set in qemu_bh_cancel (2017-11-08 19:09:15 +0000)
12
12
13
----------------------------------------------------------------
13
----------------------------------------------------------------
14
Pull request
14
Pull request
15
15
16
Small contrib/vhost-user-blk, contrib/vhost-user-scsi, and tools/virtiofsd
16
v2:
17
improvements.
17
* v1 emails 2/3 and 3/3 weren't sent due to an email failure
18
* Included Sergio's updated wording in the commit description
18
19
19
----------------------------------------------------------------
20
----------------------------------------------------------------
20
21
21
Liu Yiding (1):
22
Sergio Lopez (1):
22
virtiofsd: Add docs/helper for killpriv_v2/no_killpriv_v2 option
23
util/async: use atomic_mb_set in qemu_bh_cancel
23
24
Sakshi Kaushik (1):
25
Implements Backend Program conventions for vhost-user-scsi
26
24
27
Stefan Hajnoczi (1):
25
Stefan Hajnoczi (1):
28
contrib/vhost-user-blk: add missing GOptionEntry NULL terminator
26
tests-aio-multithread: fix /aio/multi/schedule race condition
29
27
30
docs/tools/virtiofsd.rst | 5 ++
28
tests/test-aio-multithread.c | 5 ++---
31
contrib/vhost-user-blk/vhost-user-blk.c | 3 +-
29
util/async.c | 2 +-
32
contrib/vhost-user-scsi/vhost-user-scsi.c | 77 +++++++++++++++--------
30
2 files changed, 3 insertions(+), 4 deletions(-)
33
tools/virtiofsd/helper.c | 3 +
34
4 files changed, 62 insertions(+), 26 deletions(-)
35
31
36
--
32
--
37
2.35.1
33
2.13.6
34
35
diff view generated by jsdifflib
[PULL 3/3] virtiofsd: Add docs/helper for killpriv_v2/no_killpriv_v2 option
[Qemu-devel] [PULL for-2.11-rc1 v2 1/2] tests-aio-multithread: fix /aio/multi/schedule race condition
1
From: Liu Yiding <liuyd.fnst@fujitsu.com>
1
test_multi_co_schedule_entry() set to_schedule[id] in the final loop
2
iteration before terminating the coroutine. There is a race condition
3
where the main thread attempts to enter the terminating or terminated
4
coroutine when signalling coroutines to stop:
2
5
3
virtiofsd has introduced killpriv_v2/no_killpriv_v2 for a while. Add
6
atomic_mb_set(&now_stopping, true);
4
description of it to docs/helper.
7
for (i = 0; i < NUM_CONTEXTS; i++) {
8
ctx_run(i, finish_cb, NULL); <--- enters dead coroutine!
9
to_schedule[i] = NULL;
10
}
5
11
6
Signed-off-by: Liu Yiding <liuyd.fnst@fujitsu.com>
12
Make sure only to set to_schedule[id] if this coroutine really needs to
7
Message-Id: <20220421095151.2231099-1-liuyd.fnst@fujitsu.com>
13
be scheduled!
8
14
9
[Small documentation fixes: s/as client supports/as the client supports/
15
Reported-by: "R.Nageswara Sastry" <nasastry@in.ibm.com>
10
and s/. /. /.
16
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
11
--Stefan]
17
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
12
18
Message-id: 20171106190233.1175-1-stefanha@redhat.com
13
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
19
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
14
---
20
---
15
docs/tools/virtiofsd.rst | 5 +++++
21
tests/test-aio-multithread.c | 5 ++---
16
tools/virtiofsd/helper.c | 3 +++
22
1 file changed, 2 insertions(+), 3 deletions(-)
17
2 files changed, 8 insertions(+)
18
23
19
diff --git a/docs/tools/virtiofsd.rst b/docs/tools/virtiofsd.rst
24
diff --git a/tests/test-aio-multithread.c b/tests/test-aio-multithread.c
20
index XXXXXXX..XXXXXXX 100644
25
index XXXXXXX..XXXXXXX 100644
21
--- a/docs/tools/virtiofsd.rst
26
--- a/tests/test-aio-multithread.c
22
+++ b/docs/tools/virtiofsd.rst
27
+++ b/tests/test-aio-multithread.c
23
@@ -XXX,XX +XXX,XX @@ Options
28
@@ -XXX,XX +XXX,XX @@ static void finish_cb(void *opaque)
24
label. Server will try to set that label on newly created file
29
static coroutine_fn void test_multi_co_schedule_entry(void *opaque)
25
atomically wherever possible.
30
{
26
31
g_assert(to_schedule[id] == NULL);
27
+ * killpriv_v2|no_killpriv_v2 -
32
- atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
28
+ Enable/disable ``FUSE_HANDLE_KILLPRIV_V2`` support. KILLPRIV_V2 is enabled
33
29
+ by default as long as the client supports it. Enabling this option helps
34
while (!atomic_mb_read(&now_stopping)) {
30
+ with performance in write path.
35
int n;
36
37
n = g_test_rand_int_range(0, NUM_CONTEXTS);
38
schedule_next(n);
31
+
39
+
32
.. option:: --socket-path=PATH
40
+ atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
33
41
qemu_coroutine_yield();
34
Listen on vhost-user UNIX domain socket at PATH.
42
-
35
diff --git a/tools/virtiofsd/helper.c b/tools/virtiofsd/helper.c
43
g_assert(to_schedule[id] == NULL);
36
index XXXXXXX..XXXXXXX 100644
44
- atomic_mb_set(&to_schedule[id], qemu_coroutine_self());
37
--- a/tools/virtiofsd/helper.c
45
}
38
+++ b/tools/virtiofsd/helper.c
39
@@ -XXX,XX +XXX,XX @@ void fuse_cmdline_help(void)
40
" -o announce_submounts Announce sub-mount points to the guest\n"
41
" -o posix_acl/no_posix_acl Enable/Disable posix_acl. (default: disabled)\n"
42
" -o security_label/no_security_label Enable/Disable security label. (default: disabled)\n"
43
+ " -o killpriv_v2/no_killpriv_v2\n"
44
+ " Enable/Disable FUSE_HANDLE_KILLPRIV_V2.\n"
45
+ " (default: enabled as long as client supports it)\n"
46
);
47
}
46
}
48
47
49
--
48
--
50
2.35.1
49
2.13.6
50
51
diff view generated by jsdifflib
[PULL 1/3] Implements Backend Program conventions for vhost-user-scsi
[Qemu-devel] [PULL for-2.11-rc1 v2 2/2] util/async: use atomic_mb_set in qemu_bh_cancel
1
From: Sakshi Kaushik <sakshikaushik717@gmail.com>
1
From: Sergio Lopez <slp@redhat.com>
2
2
3
Signed-off-by: Sakshi Kaushik <sakshikaushik717@gmail.com>
3
Commit b7a745d added a qemu_bh_cancel call to the completion function
4
Message-id: 20220406162410.8536-1-sakshikaushik717@gmail.com
4
as an optimization to prevent it from unnecessarily rescheduling itself.
5
5
6
[Name the iSCSI URL long option --iscsi-uri instead of --iscsi_uri for
6
This completion function is scheduled from worker_thread, after setting
7
consistency, fix --fd which was rejected due to an outdated
7
the state of a ThreadPoolElement to THREAD_DONE.
8
--socket-path check, and add missing entries[] terminator.
8
9
This was considered to be safe, as the completion function restarts the
10
loop just after the call to qemu_bh_cancel. But, as this loop lacks a HW
11
memory barrier, the read of req->state may actually happen _before_ the
12
call, seeing it still as THREAD_QUEUED, and ending the completion
13
function without having processed a pending TPE linked at pool->head:
14
15
worker thread | I/O thread
16
------------------------------------------------------------------------
17
| speculatively read req->state
18
req->state = THREAD_DONE; |
19
qemu_bh_schedule(p->completion_bh) |
20
bh->scheduled = 1; |
21
| qemu_bh_cancel(p->completion_bh)
22
| bh->scheduled = 0;
23
| if (req->state == THREAD_DONE)
24
| // sees THREAD_QUEUED
25
26
The source of the misunderstanding was that qemu_bh_cancel is now being
27
used by the _consumer_ rather than the producer, and therefore now needs
28
to have acquire semantics just like e.g. aio_bh_poll.
29
30
In some situations, if there are no other independent requests in the
31
same aio context that could eventually trigger the scheduling of the
32
completion function, the omitted TPE and all operations pending on it
33
will get stuck forever.
34
35
[Added Sergio's updated wording about the HW memory barrier.
9
--Stefan]
36
--Stefan]
10
37
38
Signed-off-by: Sergio Lopez <slp@redhat.com>
39
Message-id: 20171108063447.2842-1-slp@redhat.com
11
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
40
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
12
---
41
---
13
contrib/vhost-user-scsi/vhost-user-scsi.c | 77 +++++++++++++++--------
42
util/async.c | 2 +-
14
1 file changed, 52 insertions(+), 25 deletions(-)
43
1 file changed, 1 insertion(+), 1 deletion(-)
15
44
16
diff --git a/contrib/vhost-user-scsi/vhost-user-scsi.c b/contrib/vhost-user-scsi/vhost-user-scsi.c
45
diff --git a/util/async.c b/util/async.c
17
index XXXXXXX..XXXXXXX 100644
46
index XXXXXXX..XXXXXXX 100644
18
--- a/contrib/vhost-user-scsi/vhost-user-scsi.c
47
--- a/util/async.c
19
+++ b/contrib/vhost-user-scsi/vhost-user-scsi.c
48
+++ b/util/async.c
20
@@ -XXX,XX +XXX,XX @@ fail:
49
@@ -XXX,XX +XXX,XX @@ void qemu_bh_schedule(QEMUBH *bh)
21
50
*/
22
/** vhost-user-scsi **/
51
void qemu_bh_cancel(QEMUBH *bh)
23
24
+static int opt_fdnum = -1;
25
+static char *opt_socket_path;
26
+static gboolean opt_print_caps;
27
+static char *iscsi_uri;
28
+
29
+static GOptionEntry entries[] = {
30
+ { "print-capabilities", 'c', 0, G_OPTION_ARG_NONE, &opt_print_caps,
31
+ "Print capabilities", NULL },
32
+ { "fd", 'f', 0, G_OPTION_ARG_INT, &opt_fdnum,
33
+ "Use inherited fd socket", "FDNUM" },
34
+ { "iscsi-uri", 'i', 0, G_OPTION_ARG_FILENAME, &iscsi_uri,
35
+ "iSCSI URI to connect to", "FDNUM" },
36
+ { "socket-path", 's', 0, G_OPTION_ARG_FILENAME, &opt_socket_path,
37
+ "Use UNIX socket path", "PATH" },
38
+ { NULL, }
39
+};
40
+
41
int main(int argc, char **argv)
42
{
52
{
43
VusDev *vdev_scsi = NULL;
53
- bh->scheduled = 0;
44
- char *unix_fn = NULL;
54
+ atomic_mb_set(&bh->scheduled, 0);
45
- char *iscsi_uri = NULL;
55
}
46
- int lsock = -1, csock = -1, opt, err = EXIT_SUCCESS;
56
47
+ int lsock = -1, csock = -1, err = EXIT_SUCCESS;
57
/* This func is async.The bottom half will do the delete action at the finial
48
49
- while ((opt = getopt(argc, argv, "u:i:")) != -1) {
50
- switch (opt) {
51
- case 'h':
52
- goto help;
53
- case 'u':
54
- unix_fn = g_strdup(optarg);
55
- break;
56
- case 'i':
57
- iscsi_uri = g_strdup(optarg);
58
- break;
59
- default:
60
- goto help;
61
- }
62
+ GError *error = NULL;
63
+ GOptionContext *context;
64
+
65
+ context = g_option_context_new(NULL);
66
+ g_option_context_add_main_entries(context, entries, NULL);
67
+ if (!g_option_context_parse(context, &argc, &argv, &error)) {
68
+ g_printerr("Option parsing failed: %s\n", error->message);
69
+ exit(EXIT_FAILURE);
70
+ }
71
+
72
+ if (opt_print_caps) {
73
+ g_print("{\n");
74
+ g_print(" \"type\": \"scsi\"\n");
75
+ g_print("}\n");
76
+ goto out;
77
}
78
- if (!unix_fn || !iscsi_uri) {
79
+
80
+ if (!iscsi_uri) {
81
goto help;
82
}
83
84
- lsock = unix_sock_new(unix_fn);
85
- if (lsock < 0) {
86
- goto err;
87
+ if (opt_socket_path) {
88
+ lsock = unix_sock_new(opt_socket_path);
89
+ if (lsock < 0) {
90
+ exit(EXIT_FAILURE);
91
+ }
92
+ } else if (opt_fdnum < 0) {
93
+ g_print("%s\n", g_option_context_get_help(context, true, NULL));
94
+ exit(EXIT_FAILURE);
95
+ } else {
96
+ lsock = opt_fdnum;
97
}
98
99
csock = accept(lsock, NULL, NULL);
100
@@ -XXX,XX +XXX,XX @@ out:
101
if (vdev_scsi) {
102
g_main_loop_unref(vdev_scsi->loop);
103
g_free(vdev_scsi);
104
- unlink(unix_fn);
105
+ unlink(opt_socket_path);
106
}
107
if (csock >= 0) {
108
close(csock);
109
@@ -XXX,XX +XXX,XX @@ out:
110
if (lsock >= 0) {
111
close(lsock);
112
}
113
- g_free(unix_fn);
114
+ g_free(opt_socket_path);
115
g_free(iscsi_uri);
116
117
return err;
118
@@ -XXX,XX +XXX,XX @@ err:
119
goto out;
120
121
help:
122
- fprintf(stderr, "Usage: %s [ -u unix_sock_path -i iscsi_uri ] | [ -h ]\n",
123
+ fprintf(stderr, "Usage: %s [ -s socket-path -i iscsi-uri -f fd -p print-capabilities ] | [ -h ]\n",
124
argv[0]);
125
- fprintf(stderr, " -u path to unix socket\n");
126
- fprintf(stderr, " -i iscsi uri for lun 0\n");
127
+ fprintf(stderr, " -s, --socket-path=SOCKET_PATH path to unix socket\n");
128
+ fprintf(stderr, " -i, --iscsi-uri=ISCSI_URI iscsi uri for lun 0\n");
129
+ fprintf(stderr, " -f, --fd=FILE_DESCRIPTOR file-descriptor\n");
130
+ fprintf(stderr, " -p, --print-capabilities=PRINT_CAPABILITIES denotes print-capabilities\n");
131
fprintf(stderr, " -h print help and quit\n");
132
133
goto err;
134
--
58
--
135
2.35.1
59
2.13.6
60
61
diff view generated by jsdifflib
[PULL 2/3] contrib/vhost-user-blk: add missing GOptionEntry NULL terminator
Deleted patch
1
The GLib documentation says "a NULL-terminated array of GOptionEntrys"
2
so we'd better make sure there is a terminator that lets
3
g_option_context_add_main_entries() know when the end of the array has
4
been reached.
5
1
6
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
7
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
8
Message-id: 20220411150057.3009667-1-stefanha@redhat.com
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10
---
11
contrib/vhost-user-blk/vhost-user-blk.c | 3 ++-
12
1 file changed, 2 insertions(+), 1 deletion(-)
13
14
diff --git a/contrib/vhost-user-blk/vhost-user-blk.c b/contrib/vhost-user-blk/vhost-user-blk.c
15
index XXXXXXX..XXXXXXX 100644
16
--- a/contrib/vhost-user-blk/vhost-user-blk.c
17
+++ b/contrib/vhost-user-blk/vhost-user-blk.c
18
@@ -XXX,XX +XXX,XX @@ static GOptionEntry entries[] = {
19
{"blk-file", 'b', 0, G_OPTION_ARG_FILENAME, &opt_blk_file,
20
"block device or file path", "PATH"},
21
{ "read-only", 'r', 0, G_OPTION_ARG_NONE, &opt_read_only,
22
- "Enable read-only", NULL }
23
+ "Enable read-only", NULL },
24
+ { NULL, },
25
};
26
27
int main(int argc, char **argv)
28
--
29
2.35.1
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.