| Message ID | 20230423110254.35672-1-wangliangzz@126.com |
|---|---|
| State | New |
| Headers | show |
| Series | block/monitor/block-hmp-cmds.c: Fix crash when execute hmp_commit | expand |
Am 23/04/2023 um 13:02 schrieb wangliangzz@126.com: > From: Wang Liang <wangliangzz@inspur.com> > > We need to get the aio_context before calling the blk_is_available. > > Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1615 > Signed-off-by: Wang Liang <wangliangzz@inspur.com> > Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Am 24/04/2023 um 10:15 schrieb Emanuele Giuseppe Esposito: > > > Am 23/04/2023 um 13:02 schrieb wangliangzz@126.com: >> From: Wang Liang <wangliangzz@inspur.com> >> >> We need to get the aio_context before calling the blk_is_available. >> >> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1615 >> Signed-off-by: Wang Liang <wangliangzz@inspur.com> >> > > > Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> > Sorry I forgot, if you need to resend can you add the cause of this issue in the commit message? Something along the lines of: "hmp_commit() calls blk_is_available() from a non-coroutine context (and in the main loop). Since this is a co_wrapper_mixed_bdrv_rdlock function, in this case it calls AIO_WAIT_WHILE(), which crashes if the aio_context lock is not taken before" Thank you, Emanuele
On Mon, 2023-04-24 at 10:36 +0200, Emanuele Giuseppe Esposito wrote: > > Am 24/04/2023 um 10:15 schrieb Emanuele Giuseppe Esposito: > > > > Am 23/04/2023 um 13:02 schrieb wangliangzz@126.com: > > > From: Wang Liang <wangliangzz@inspur.com> > > > > > > We need to get the aio_context before calling the > > > blk_is_available. > > > > > > Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1615 > > > Signed-off-by: Wang Liang <wangliangzz@inspur.com> > > > > > > > Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com> > > > > Sorry I forgot, if you need to resend can you add the cause of this > issue in the commit message? > Something along the lines of: > "hmp_commit() calls blk_is_available() from a non-coroutine context > (and > in the main loop). Since this is a co_wrapper_mixed_bdrv_rdlock > function, in this case it calls AIO_WAIT_WHILE(), which crashes if > the > aio_context lock is not taken before" > > Thank you, > Emanuele Thanks for the detailed explanation. I'll resend the patch. Wang Liang
diff --git a/block/monitor/block-hmp-cmds.c b/block/monitor/block-hmp-cmds.c index 2846083546..ca2599de44 100644 --- a/block/monitor/block-hmp-cmds.c +++ b/block/monitor/block-hmp-cmds.c @@ -214,15 +214,17 @@ void hmp_commit(Monitor *mon, const QDict *qdict) error_report("Device '%s' not found", device); return; } - if (!blk_is_available(blk)) { - error_report("Device '%s' has no medium", device); - return; - } bs = bdrv_skip_implicit_filters(blk_bs(blk)); aio_context = bdrv_get_aio_context(bs); aio_context_acquire(aio_context); + if (!blk_is_available(blk)) { + error_report("Device '%s' has no medium", device); + aio_context_release(aio_context); + return; + } + ret = bdrv_commit(bs); aio_context_release(aio_context);