mirror: Handle errors after READY cancel
v3 cover mirror: Handle errors after READY cancel
Hi,
v1 cover letter:
https://lists.nongnu.org/archive/html/qemu-block/2021-07/msg00705.html
v2 cover letter:
https://lists.nongnu.org/archive/html/qemu-block/2021-07/msg00747.html
Changes in v3:
- Patch 1: After adding patch 11, I got a failed assertion in tests/unit/test-block-iothread (failing qemu_mutex_unlock_impl()). That is because before patch 11, for zero-length source devices, mirror clears .cancelled unconditionally before exiting. So even force-cancelled jobs are considered to be completed normally, which doesn’t seem quite right.
Anyway, test-block-iothread does some iothread switching, and cancelling jobs is not really prepared for that. This patch fixes that (I hope...).
- Patch 4: Split off from patch 5
- Patch 7:
- Added a long section in the commit message detailing every choice for every job_is_cancelled() invocation
- Use job_cancel_requested() in the assertion in job_completed_txn_abort(), because it is not quite clear whether soft-cancelled mirror jobs can end up in this path (it seems like a bug if that happens, but I think that’s something to fix in some other series)
- Patch 8: Added: This is kind of preparation for patch 9, but also just a bug fix in itself, I believe
- Patch 9: Moved the job_is_cancelled() check after the last yield point before the mirror_iteration() call
- Patch 10: Added: If force-cancelled jobs should not generate new I/O requests at all (except for forwarding something to the source device), then we need to stop doing active mirroring once the mirror job is force-cancelled
- Patch 11: Added: Clearing .cancelled seemed like a hack, so getting rid of it seems like a good thing to do
(And only with this patch, I can assert that .force_cancel can only be true when .cancelled is true also; if we tried it before this patch, tests/unit/test-block-iothread would fail.)
The discussion around v2 has shown that there are probably more bugs in the job code, but I think this series is becoming long enough that we should tackle those in a different series.
git-backport-diff against v1:
Key:
[----] : patches are identical
[####] : number of functional differences between upstream/downstream patch [down] : patch is downstream-only
The flags [FC] indicate (F)unctional and (C)ontextual differences, respectively
001/12:[down] 'job: Context changes in job_completed_txn_abort()' 002/12:[----] [--] 'mirror: Keep s->synced on error' 003/12:[----] [--] 'mirror: Drop s->synced'
004/12:[down] 'job: Force-cancel jobs in a failed transaction' 005/12:[0007] [FC] 'job: @force parameter for job_cancel_sync{,_all}()' 006/12:[----] [--] 'jobs: Give Job.force_cancel more meaning' 007/12:[0002] [FC] 'job: Add job_cancel_requested()' 008/12:[down] 'mirror: Use job_is_cancelled()'
009/12:[0007] [FC] 'mirror: Check job_is_cancelled() earlier' 010/12:[down] 'mirror: Stop active mirroring after force-cancel' 011/12:[down] 'mirror: Do not clear .cancelled' 012/12:[----] [--] 'iotests: Add mirror-ready-cancel-error test'
Max Reitz (12):
job: Context changes in job_completed_txn_abort()
mirror: Keep s->synced on error
mirror: Drop s->synced
job: Force-cancel jobs in a failed transaction
job: @force parameter for job_cancel_sync{,_all}()
jobs: Give Job.force_cancel more meaning
job: Add job_cancel_requested()
mirror: Use job_is_cancelled()
mirror: Check job_is_cancelled() earlier
mirror: Stop active mirroring after force-cancel
mirror: Do not clear .cancelled
iotests: Add mirror-ready-cancel-error test
include/qemu/job.h | 29 +++- block/backup.c | 3 +- block/mirror.c | 56 ++++--- block/replication.c | 4 +- blockdev.c | 4 +- job.c | 67 ++++++-- qemu-nbd.c | 2 +- softmmu/runstate.c | 2 +- storage-daemon/qemu-storage-daemon.c | 2 +- tests/unit/test-block-iothread.c | 2 +- tests/unit/test-blockjob.c | 2 +- tests/qemu-iotests/109.out | 60 +++----- .../tests/mirror-ready-cancel-error | 143 ++++++++++++++++++ .../tests/mirror-ready-cancel-error.out | 5 + tests/qemu-iotests/tests/qsd-jobs.out | 2 +- 15 files changed, 292 insertions(+), 91 deletions(-) create mode 100755 tests/qemu-iotests/tests/mirror-ready-cancel-error create mode 100644 tests/qemu-iotests/tests/mirror-ready-cancel-error.out
Signature · 1 line
2.31.1
PATCH v3 1/12 job: Context changes in job_completed_txn_abort()
Finalizing the job may cause its AioContext to change. This is noted by job_exit(), which points at job_txn_apply() to take this fact into account.
However, job_completed() does not necessarily invoke job_txn_apply() (through job_completed_txn_success()), but potentially also job_completed_txn_abort(). The latter stores the context in a local variable, and so always acquires the same context at its end that it has released in the beginning -- which may be a different context from the one that job_exit() releases at its end. If it is different, qemu aborts ("qemu_mutex_unlock_impl: Operation not permitted").
Drop the local @outer_ctx variable from job_completed_txn_abort(), and instead re-acquire the actual job's context at the end of the function, so job_exit() will release the same.
--- job.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-)
job.c +18 -5
@@ -737,7 +737,6 @@ static void job_cancel_async(Job *job, bool force)static void job_completed_txn_abort (Job *job) { -AioContext *outer_ctx = job->aio_context;AioContext *ctx;JobTxn *txn = job->txn;Job *other_job;
@@ -751,10 +750,14 @@ static void job_completed_txn_abort(Job *job) txn->aborting = true;job_txn_ref (txn); -/* We can only hold the single job's AioContext lock while calling +/* +* We can only hold the single job's AioContext lock while calling * job_finalize_single() because the finalization callbacks can involve -* calls of AIO_WAIT_WHILE(), which could deadlock otherwise. */ -aio_context_release (outer_ctx); +* calls of AIO_WAIT_WHILE(), which could deadlock otherwise. +* Note that the job's AioContext may change when it is finalized. +*/ +job_ref (job); +aio_context_release (job->aio_context);/* Other jobs are effectively cancelled by us, set the status for * them; this job, however, may or may not be cancelled, depending
@@ -769,6 +772,10 @@ static void job_completed_txn_abort(Job *job) }while (!QLIST_EMPTY (&txn->jobs)) { other_job =QLIST_FIRST (&txn->jobs); +/* +* The job's AioContext may change, so store it in @ctx so we +* release the same context that we have acquired before. +*/ ctx = other_job->aio_context;aio_context_acquire (ctx);if (!job_is_completed (other_job)) {
@@ -779,7 +786,13 @@ static void job_completed_txn_abort(Job *job)aio_context_release (ctx); } -aio_context_acquire (outer_ctx); +/* +* Use job_ref()/job_unref() so we can read the AioContext here +* even if the job went away during job_finalize_single(). +*/ + ctx = job->aio_context; +job_unref (job); +aio_context_acquire (ctx);job_txn_unref (txn); }
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 01/12] job: Context changes in job_completed_txn_abort()
Quoted 11 lines · Max Reitz
Finalizing the job may cause its AioContext to change. This is noted by
job_exit(), which points atjob_txn_apply()to take this fact into account.However,
job_completed()does not necessarily invokejob_txn_apply()(throughjob_completed_txn_success()), but potentially alsojob_completed_txn_abort(). The latter stores the context in a local variable, and so always acquires the same context at its end that it has released in the beginning -- which may be a different context from the one thatjob_exit()releases at its end. If it is different, qemu aborts ("qemu_mutex_unlock_impl: Operation not permitted").
Is this a bug fix that needs to make it into 6.1?
Quoted 9 lines
Drop the local @outer_ctx variable from
job_completed_txn_abort(), and instead re-acquire the actual job's context at the end of the function, sojob_exit()will release the same.Signed-off-by Max Reitz <[email protected]>--- job.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-)
The commit message makes sense, and does a good job at explaining the change. I'm still a bit fuzzy on how jobs are supposed to play nice with contexts, but since your patch matches the commit message, I'm happy to give:
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 01/12] job: Context changes in job_completed_txn_abort()
Quoted 5 lines · Eric Blake
Quoted 4 lines · Max Reitz
Finalizing the job may cause its AioContext to change. This is noted by
job_exit(), which points atjob_txn_apply()to take this fact into account.
Quoted 8 lines
Quoted 7 lines
However,
job_completed()does not necessarily invokejob_txn_apply()(throughjob_completed_txn_success()), but potentially alsojob_completed_txn_abort(). The latter stores the context in a local variable, and so always acquires the same context at its end that it has released in the beginning -- which may be a different context from the one thatjob_exit()releases at its end. If it is different, qemu aborts ("qemu_mutex_unlock_impl: Operation not permitted").Is this a bug fix that needs to make it into 6.1?
Well, I only encountered it as part of this series (which I really don’t think is 6.2 material at this point), and so I don’t know.
Can’t hurt, I suppose, but if we wanted this to be in 6.1, we’d better have a specific test for it, I think.
Quoted 11 lines
Quoted 8 lines
Drop the local @outer_ctx variable from
job_completed_txn_abort(), and instead re-acquire the actual job's context at the end of the function, sojob_exit()will release the same.Signed-off-by Max Reitz <[email protected]>--- job.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-)The commit message makes sense, and does a good job at explaining the change. I'm still a bit fuzzy on how jobs are supposed to play nice with contexts,
I can relate :)
Quoted 4 lines
but since your patch matches the commit message, I'm happy to give:
Reviewed-by Eric Blake <[email protected]>
Thanks!
Re: [PATCH for-6.2 v3 01/12] job: Context changes in job_completed_txn_abort()
In reply to Max ReitzQuoted 64 lines · Max Reitz
Finalizing the job may cause its AioContext to change. This is noted by
job_exit(), which points atjob_txn_apply()to take this fact into account.However,
job_completed()does not necessarily invokejob_txn_apply()(throughjob_completed_txn_success()), but potentially alsojob_completed_txn_abort(). The latter stores the context in a local variable, and so always acquires the same context at its end that it has released in the beginning -- which may be a different context from the one thatjob_exit()releases at its end. If it is different, qemu aborts ("qemu_mutex_unlock_impl: Operation not permitted").Drop the local @outer_ctx variable from
job_completed_txn_abort(), and instead re-acquire the actual job's context at the end of the function, sojob_exit()will release the same.Signed-off-by Max Reitz <[email protected]>--- job.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-)job.c +11
-4@@ -737,7 +737,6 @@ static void job_cancel_async(Job *job, bool force)static void job_completed_txn_abort (Job *job) { -AioContext *outer_ctx = job->aio_context;AioContext *ctx;JobTxn *txn = job->txn;Job *other_job;@@ -751,10 +750,14 @@ static void job_completed_txn_abort(Job *job) txn->aborting = true;job_txn_ref (txn); -/* We can only hold the single job's AioContext lock while calling +/* +* We can only hold the single job's AioContext lock while calling * job_finalize_single() because the finalization callbacks can involve -* calls of AIO_WAIT_WHILE(), which could deadlock otherwise. */ -aio_context_release (outer_ctx); +* calls of AIO_WAIT_WHILE(), which could deadlock otherwise. +* Note that the job's AioContext may change when it is finalized. +*/ +job_ref (job); +aio_context_release (job->aio_context);/* Other jobs are effectively cancelled by us, set the status for * them; this job, however, may or may not be cancelled, depending @@ -769,6 +772,10 @@ static void job_completed_txn_abort(Job *job) }while (!QLIST_EMPTY (&txn->jobs)) { other_job =QLIST_FIRST (&txn->jobs); +/* +* The job's AioContext may change, so store it in @ctx so we +* release the same context that we have acquired before. +*/ ctx = other_job->aio_context;aio_context_acquire (ctx);if (!job_is_completed (other_job)) {@@ -779,7 +786,13 @@ static void job_completed_txn_abort(Job *job)aio_context_release (ctx);
job.c static void job_completed_txn_abort(Job *job) · line 795
Quoted 10 lines
job.c +7
-1} -aio_context_acquire (outer_ctx); +/* +* Use job_ref()/job_unref() so we can read the AioContext here +* even if the job went away during job_finalize_single(). +*/ + ctx = job->aio_context; +job_unref (job); +aio_context_acquire (ctx);
why to use ctx variable and not do it exactly same as in job_txn_apply() :
aio_context_acquire(job->aio_context); job_unref(job);
?
anyway:
Signature · 2 lines
Best regards, Vladimir
Re: [PATCH for-6.2 v3 01/12] job: Context changes in job_completed_txn_abort()
Quoted 75 lines · Vladimir Sementsov-Ogievskiy
Quoted 60 lines · Max Reitz
Finalizing the job may cause its AioContext to change. This is noted by
job_exit(), which points atjob_txn_apply()to take this fact into account.However,
job_completed()does not necessarily invokejob_txn_apply()(throughjob_completed_txn_success()), but potentially also job_completed_txn_abort(). The latter stores the context in a local variable, and so always acquires the same context at its end that it has released in the beginning -- which may be a different context from the one thatjob_exit()releases at its end. If it is different, qemu aborts ("qemu_mutex_unlock_impl: Operation not permitted").Drop the local @outer_ctx variable from
job_completed_txn_abort(), and instead re-acquire the actual job's context at the end of the function, sojob_exit()will release the same.Signed-off-by Max Reitz <[email protected]>---
job.c| 23 ++++++++++++++++++-----1 file changed, 18 insertions(+), 5 deletions(-)job.c +0
-1@@ -737,7 +737,6 @@ static void job_cancel_async(Job *job, bool force) static void job_completed_txn_abort(Job *job) { - AioContext *outer_ctx = job->aio_context; AioContext *ctx; JobTxn *txn = job->txn; Job *other_job;@@ -751,10 +750,14 @@ static void job_completed_txn_abort(Job *job) txn->aborting = true; job_txn_ref(txn);- /* We can only hold the single job's AioContext lock while calling + /*
+ * We can only hold the single job's AioContext lock while calling
*job_finalize_single()because the finalization callbacks can involve - * calls ofAIO_WAIT_WHILE(), which could deadlock otherwise. */ -aio_context_release(outer_ctx);
+ * calls ofAIO_WAIT_WHILE(), which could deadlock otherwise. + * Note that the job's AioContext may change when it is finalized. + */
+job_ref(job);
+aio_context_release(job->aio_context);
/* Other jobs are effectively cancelled by us, set the status for
* them; this job, however, may or may not be cancelled, dependingjob.c +4
-0@@ -769,6 +772,10 @@ static void job_completed_txn_abort(Job *job) } while (!QLIST_EMPTY(&txn->jobs)) { other_job = QLIST_FIRST(&txn->jobs); + /* + * The job's AioContext may change, so store it in @ctx so we + * release the same context that we have acquired before. + */ ctx = other_job->aio_context; aio_context_acquire(ctx); if (!job_is_completed(other_job)) {job.c static void job_completed_txn_abort(Job *job) · line 795
Quoted 10 lines
job.c +7
-0@@ -779,7 +786,13 @@ static void job_completed_txn_abort(Job *job) aio_context_release(ctx); } - aio_context_acquire(outer_ctx); + /* + * Use job_ref()/job_unref() so we can read the AioContext here + * even if the job went away during job_finalize_single(). + */ + ctx = job->aio_context; + job_unref(job); + aio_context_acquire(ctx);why to use ctx variable and not do it exactly same as in
job_txn_apply():
Quoted 4 lines
aio_context_acquire(job->aio_context);
job_unref(job);?
Oh, I just didn’t think of that. Sounds good, thanks!
Hanna
Quoted 3 lines
anyway:
Reviewed-by Vladimir Sementsov-Ogievskiy <[email protected]>
PATCH v3 2/12 mirror: Keep s->synced on error
In reply to Max ReitzAn error does not take us out of the READY phase, which is what s->synced signifies. It does of course mean that source and target are no longer in sync, but that is what s->actively_sync is for -- s->synced never meant that source and target are in sync, only that they were at some point (and at that point we transitioned into the READY phase).
The tangible problem is that we transition to READY once we are in sync and s->synced is false. By resetting s->synced here, we will transition from READY to READY once the error is resolved (if the job keeps running), and that transition is not allowed.
--- block/mirror.c | 1 - 1 file changed, 1 deletion(-)
block/mirror.c +0 -1
@@ -121,7 +121,6 @@ typedef enum MirrorMethod {static BlockErrorAction mirror_error_action (MirrorBlockJob *s,bool read,int error) { - s->synced = false; s->actively_synced = false;if (read) { return block_job_error_action(&s->common, s->on_source_error,
Signature · 1 line
2.31.1
PATCH v3 3/12 mirror: Drop s->synced
In reply to Max ReitzAs of HEAD^, there is no meaning to s->synced other than whether the job is READY or not. job_is_ready() gives us that information, too.
--- block/mirror.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-)
block/mirror.c +9 -10
@@ -56,7 +56,6 @@ typedef struct MirrorBlockJob {bool zero_target;MirrorCopyMode copy_mode;BlockdevOnError on_source_error, on_target_error; -bool synced;/* Set when the target is synced (dirty bitmap is clean, nothing * in flight) and the job is running in active mode */ bool actively_synced;
@@ -936,7 +935,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)if (s->bdev_length ==0 ) {/* Transition to the READY state and wait for complete. */ job_transition_to_ready (&s->common.job); - s->synced = true; s->actively_synced = true;while (!job_is_cancelled (&s->common.job) && !s->should_complete) {job_yield (&s->common.job);
@@ -1028,7 +1026,7 @@ static int coroutine_fn mirror_run(Job *job, Error **errp) should_complete = false;if (s->in_flight ==0 && cnt ==0 ) {trace_mirror_before_flush (s); -if (!s->synced) { +if (!job_is_ready (&s->common.job)) {if (mirror_flush (s) <0 ) {/* Go check s->ret. */ continue ;
@@ -1039,7 +1037,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)* the target in a consistent state. */ job_transition_to_ready (&s->common.job); - s->synced = true;if (s->copy_mode !=MIRROR_COPY_MODE_BACKGROUND ) { s->actively_synced = true; }
@@ -1083,14 +1080,15 @@ static int coroutine_fn mirror_run(Job *job, Error **errp) ret =0 ; -if (s->synced && !should_complete) { +if (job_is_ready (&s->common.job) && !should_complete) { delay_ns = (s->in_flight ==0 && cnt ==0 ?BLOCK_JOB_SLICE_TIME :0 ); } -trace_mirror_before_sleep (s, cnt, s->synced, delay_ns); +trace_mirror_before_sleep (s, cnt,job_is_ready (&s->common.job), + delay_ns);job_sleep_ns (&s->common.job, delay_ns);if (job_is_cancelled (&s->common.job) && - (!s->synced || s->common.job.force_cancel)) + (!job_is_ready (&s->common.job) || s->common.job.force_cancel)) {break ; }
@@ -1103,8 +1101,9 @@ immediate_exit:* or it was cancelled prematurely so that we do not guarantee that * the target is a copy of the source. */ -assert (ret <0 || ((s->common.job.force_cancel || !s->synced) && -job_is_cancelled (&s->common.job))); +assert (ret <0 || + ((s->common.job.force_cancel || !job_is_ready (&s->common.job)) && +job_is_cancelled (&s->common.job)));assert (need_drain);mirror_wait_for_all_io (s); }
@@ -1127,7 +1126,7 @@ static void mirror_complete(Job *job, Error **errp) {MirrorBlockJob *s =container_of (job, MirrorBlockJob, common.job); -if (!s->synced) { +if (!job_is_ready (job)) {error_setg (errp,"The active block job '%s' cannot be completed" , job->id);return ;
Signature · 1 line
2.31.1
PATCH v3 4/12 job: Force-cancel jobs in a failed transaction
In reply to Max ReitzWhen a transaction is aborted, no result matters, and so all jobs within should be force-cancelled.
--- job.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
job.c +6 -1
@@ -766,7 +766,12 @@ static void job_completed_txn_abort(Job *job)if (other_job != job) { ctx = other_job->aio_context;aio_context_acquire (ctx); -job_cancel_async (other_job, false); +/* +* This is a transaction: If one job failed, no result will matter. +* Therefore, pass force=true to terminate all other jobs as quickly +* as possible. +*/ +job_cancel_async (other_job, true);aio_context_release (ctx); } }
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 04/12] job: Force-cancel jobs in a failed transaction
Quoted 7 lines · Max Reitz
When a transaction is aborted, no result matters, and so all jobs within should be force-cancelled.
Signed-off-by Max Reitz <[email protected]>--- job.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
Quoted 23 lines
job.c +6
-1@@ -766,7 +766,12 @@ static void job_completed_txn_abort(Job *job)if (other_job != job) { ctx = other_job->aio_context;aio_context_acquire (ctx); -job_cancel_async (other_job, false); +/* +* This is a transaction: If one job failed, no result will matter. +* Therefore, pass force=true to terminate all other jobs as quickly +* as possible. +*/ +job_cancel_async (other_job, true);aio_context_release (ctx); } }Signature · 1 line
2.31.1
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 04/12] job: Force-cancel jobs in a failed transaction
In reply to Max ReitzQuoted 17 lines · Max Reitz
When a transaction is aborted, no result matters, and so all jobs within should be force-cancelled.
Signed-off-by Max Reitz <[email protected]>--- job.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)job.c +0
-0@@ -766,7 +766,12 @@ static void job_completed_txn_abort(Job *job)if (other_job != job) { ctx = other_job->aio_context;aio_context_acquire (ctx);
job.c static void job_completed_txn_abort(Job *job) · line 777
Quoted 10 lines
job.c +6
-1-job_cancel_async (other_job, false); +/* +* This is a transaction: If one job failed, no result will matter. +* Therefore, pass force=true to terminate all other jobs as quickly +* as possible. +*/ +job_cancel_async (other_job, true);aio_context_release (ctx); } }
Anyway, only backup jobs may be in a transaction, which doesn't distinguish force and soft cancelling. So, that doesn't change any logic.
Signature · 2 lines
Best regards, Vladimir
PATCH v3 5/12 job: @force parameter for job_cancel_sync{, _all}()
In reply to Max ReitzCallers should be able to specify whether they want job_cancel_sync() to force-cancel the job or not.
In fact, almost all invocations do not care about consistency of the result and just want the job to terminate as soon as possible, so they should pass force=true. The replication block driver is the exception.
This changes some iotest outputs, because quitting qemu while a mirror job is active will now lead to it being cancelled instead of completed, which is what we want. (Cancelling a READY mirror job with force=false may take an indefinite amount of time, which we do not want when quitting. If users want consistent results, they must have all jobs be done before they quit qemu.)
--- include/qemu/job.h | 10 ++--- block/replication.c | 4 +- blockdev.c | 4 +- job.c | 20 +++++++-- qemu-nbd.c | 2 +- softmmu/runstate.c | 2 +- storage-daemon/qemu-storage-daemon.c | 2 +- tests/unit/test-block-iothread.c | 2 +- tests/unit/test-blockjob.c | 2 +- tests/qemu-iotests/109.out | 60 +++++++++++---------------- tests/qemu-iotests/tests/qsd-jobs.out | 2 +- 11 files changed, 55 insertions(+), 55 deletions(-)
include/qemu/job.h +5 -5
@@ -506,19 +506,19 @@ void job_user_cancel(Job *job, bool force, Error **errp);/** * Synchronously cancel the @job. The completion callback is called -* before the function returns. The job may actually complete -* instead of canceling itself; the circumstances under which this -* happens depend on the kind of job that is active. +* before the function returns. If @force is false, the job may +* actually complete instead of canceling itself; the circumstances +* under which this happens depend on the kind of job that is active. * * Returns the return value from the job if the job actually completed * during the call, or -ECANCELED if it was canceled. * * Callers must hold the AioContext lock of job->aio_context. */ -int job_cancel_sync (Job *job); +int job_cancel_sync (Job *job,bool force);/** Synchronously cancels all jobs using job_cancel_sync(). */ -void job_cancel_sync_all (void ); +void job_cancel_sync_all (bool force);/** * @job: The job to be completed.
block/replication.c +2 -2
@@ -149,7 +149,7 @@ static void replication_close(BlockDriverState *bs)if (s->stage ==BLOCK_REPLICATION_FAILOVER ) { commit_job = &s->commit_job->job;assert (commit_job->aio_context ==qemu_get_current_aio_context ()); -job_cancel_sync (commit_job); +job_cancel_sync (commit_job, false); }if (s->mode ==REPLICATION_MODE_SECONDARY ) {
@@ -726,7 +726,7 @@ static void replication_stop(ReplicationState *rs, bool failover, Error **errp)* disk, secondary disk in backup_job_completed(). */ if (s->backup_job) { -job_cancel_sync (&s->backup_job->job); +job_cancel_sync (&s->backup_job->job, false); }if (!failover) {
blockdev.c +2 -2
@@ -1848,7 +1848,7 @@ static void drive_backup_abort(BlkActionState *common) aio_context =bdrv_get_aio_context (state->bs);aio_context_acquire (aio_context); -job_cancel_sync (&state->job->job); +job_cancel_sync (&state->job->job, true);aio_context_release (aio_context); }
@@ -1949,7 +1949,7 @@ static void blockdev_backup_abort(BlkActionState *common) aio_context =bdrv_get_aio_context (state->bs);aio_context_acquire (aio_context); -job_cancel_sync (&state->job->job); +job_cancel_sync (&state->job->job, true);aio_context_release (aio_context); }
job.c +16 -4
@@ -982,12 +982,24 @@ static void job_cancel_err(Job *job, Error **errp)job_cancel (job, false); } -int job_cancel_sync (Job *job) +/** +* Same as job_cancel_err(), but force-cancel. +*/ +static void job_force_cancel_err (Job *job,Error **errp) { -return job_finish_sync (job, &job_cancel_err,NULL ); +job_cancel (job, true); +} + +int job_cancel_sync (Job *job,bool force) +{ +if (force) { +return job_finish_sync (job, &job_force_cancel_err,NULL ); + }else { +return job_finish_sync (job, &job_cancel_err,NULL ); + } } -void job_cancel_sync_all (void ) +void job_cancel_sync_all (bool force) {Job *job;AioContext *aio_context;
@@ -995,7 +1007,7 @@ void job_cancel_sync_all(void)while ((job =job_next (NULL ))) { aio_context = job->aio_context;aio_context_acquire (aio_context); -job_cancel_sync (job); +job_cancel_sync (job, force);aio_context_release (aio_context); } }
qemu-nbd.c +1 -1
@@ -479,7 +479,7 @@ static const char *socket_activation_validate_opts(const char *device,static void qemu_nbd_shutdown (void ) { -job_cancel_sync_all (); +job_cancel_sync_all (true);blk_exp_close_all ();bdrv_close_all (); }
softmmu/runstate.c +1 -1
@@ -812,7 +812,7 @@ void qemu_cleanup(void)vm_shutdown ();replay_finish (); -job_cancel_sync_all (); +job_cancel_sync_all (true);bdrv_close_all ();/* vhost-user must be cleaned up before chardevs. */
storage-daemon/qemu-storage-daemon.c +1 -1
@@ -347,7 +347,7 @@ int main(int argc, char *argv[])blk_exp_close_all ();bdrv_drain_all_begin (); -job_cancel_sync_all (); +job_cancel_sync_all (true);bdrv_close_all ();monitor_cleanup ();
tests/unit/test-block-iothread.c +1 -1
@@ -662,7 +662,7 @@ static void test_propagate_mirror(void)g_assert (bdrv_get_aio_context (target) == ctx);g_assert (bdrv_get_aio_context (filter) == ctx); -job_cancel_sync_all (); +job_cancel_sync_all (true);aio_context_acquire (ctx);blk_set_aio_context (blk, main_ctx, &error_abort);
tests/unit/test-blockjob.c +1 -1
@@ -230,7 +230,7 @@ static void cancel_common(CancelJob *s) ctx = job->job.aio_context;aio_context_acquire (ctx); -job_cancel_sync (&job->job); +job_cancel_sync (&job->job, true);if (sts !=JOB_STATUS_CREATED && sts !=JOB_STATUS_CONCLUDED ) {Job *dummy = &job->job;job_dismiss (&dummy, &error_abort);
tests/qemu-iotests/109.out +24 -36
132 lines not on this page · raw message
tests/qemu-iotests/tests/qsd-jobs.out +1 -1
@@ -8,7 +8,7 @@ QMP_VERSION {"return": {}} {"return": {}} {"return": {}} -{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "BLOCK_JOB_COMPLETED", "data": {"device": "job0", "len": 0, "offset": 0, "speed": 0, "type": "commit"}} +{"timestamp": {"seconds": TIMESTAMP, "microseconds": TIMESTAMP}, "event": "BLOCK_JOB_CANCELLED", "data": {"device": "job0", "len": 0, "offset": 0, "speed": 0, "type": "commit"}} === Streaming can't get permission on base node ===
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{, _all}()
Quoted 7 lines · Max Reitz
Callers should be able to specify whether they want
job_cancel_sync()to force-cancel the job or not.In fact, almost all invocations do not care about consistency of the result and just want the job to terminate as soon as possible, so they should pass force=true. The replication block driver is the exception.
Quoted 6 lines
This changes some iotest outputs, because quitting qemu while a mirror job is active will now lead to it being cancelled instead of completed, which is what we want. (Cancelling a READY mirror job with force=false may take an indefinite amount of time, which we do not want when quitting. If users want consistent results, they must have all jobs be done before they quit qemu.)
Feels somewhat like a bug fix, but I also understand why you'd prefer to delay this to 6.2 (it is not a fresh regression, but a longstanding issue).
Quoted 9 lines
Signed-off-by Max Reitz <[email protected]>---
+++
b/job.cjob.c +0
-0@@ -982,12 +982,24 @@ static void job_cancel_err(Job *job, Error **errp)job_cancel (job, false); }
job.c static void job_cancel_err(Job *job, Error **errp) · line 991
Quoted 10 lines
job.c +6
-2-int job_cancel_sync (Job *job) +/** +* Same as job_cancel_err(), but force-cancel. +*/ +static void job_force_cancel_err (Job *job,Error **errp) { -return job_finish_sync (job, &job_cancel_err,NULL ); +job_cancel (job, true); +}
In isolation, it looks odd that errp is passed but not used. But looking further, it's because this is a callback that must have a given signature, so it's okay.
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{, _all}()
Quoted 14 lines · Eric Blake
Quoted 7 lines · Max Reitz
Callers should be able to specify whether they want
job_cancel_sync()to force-cancel the job or not.In fact, almost all invocations do not care about consistency of the result and just want the job to terminate as soon as possible, so they should pass force=true. The replication block driver is the exception.
Quoted 6 lines
This changes some iotest outputs, because quitting qemu while a mirror job is active will now lead to it being cancelled instead of completed, which is what we want. (Cancelling a READY mirror job with force=false may take an indefinite amount of time, which we do not want when quitting. If users want consistent results, they must have all jobs be done before they quit qemu.)
Quoted 3 lines
Feels somewhat like a bug fix, but I also understand why you'd prefer to delay this to 6.2 (it is not a fresh regression, but a longstanding issue).
It is, hence the “Buglink” tag below. However, only all of this series together really fixes that bug (or at least patches 5+7+9 together), just taking one wouldn’t help much. And together, it’s just too much for 6.2 at this point.
Quoted 23 lines
Quoted 7 lines
Signed-off-by Max Reitz <[email protected]>---
+++b/job.cjob.c +0
-0@@ -982,12 +982,24 @@ static void job_cancel_err(Job *job, Error **errp)job_cancel (job, false); }job.c static void job_cancel_err(Job *job, Error **errp) · line 991
Quoted 10 lines
job.c +6
-2-int job_cancel_sync (Job *job) +/** +* Same as job_cancel_err(), but force-cancel. +*/ +static void job_force_cancel_err (Job *job,Error **errp) { -return job_finish_sync (job, &job_cancel_err,NULL ); +job_cancel (job, true); +}In isolation, it looks odd that errp is passed but not used. But looking further, it's because this is a callback that must have a given signature, so it's okay.
Reviewed-by Eric Blake <[email protected]>
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{,_all}()
In reply to Max ReitzQuoted 46 lines · Max Reitz
Callers should be able to specify whether they want
job_cancel_sync()to force-cancel the job or not.In fact, almost all invocations do not care about consistency of the result and just want the job to terminate as soon as possible, so they should pass force=true. The replication block driver is the exception.
This changes some iotest outputs, because quitting qemu while a mirror job is active will now lead to it being cancelled instead of completed, which is what we want. (Cancelling a READY mirror job with force=false may take an indefinite amount of time, which we do not want when quitting. If users want consistent results, they must have all jobs be done before they quit qemu.)
Signed-off-by Max Reitz<[email protected]>--- include/qemu/job.h | 10 ++--- block/replication.c | 4 +- blockdev.c | 4 +- job.c | 20 +++++++-- qemu-nbd.c | 2 +- softmmu/runstate.c | 2 +- storage-daemon/qemu-storage-daemon.c | 2 +- tests/unit/test-block-iothread.c | 2 +- tests/unit/test-blockjob.c | 2 +- tests/qemu-iotests/109.out | 60 +++++++++++---------------- tests/qemu-iotests/tests/qsd-jobs.out | 2 +- 11 files changed, 55 insertions(+), 55 deletions(-)include/qemu/job.h +3
-3@@ -506,19 +506,19 @@ void job_user_cancel(Job *job, bool force, Error **errp);/** * Synchronously cancel the @job. The completion callback is called -* before the function returns. The job may actually complete -* instead of canceling itself; the circumstances under which this -* happens depend on the kind of job that is active. +* before the function returns. If @force is false, the job may +* actually complete instead of canceling itself; the circumstances +* under which this happens depend on the kind of job that is active. * * Returns the return value from the job if the job actually completed
include/qemu/job.h void job_user_cancel(Job *job, bool force, Error **errp); · line 521
Quoted 10 lines
include/qemu/job.h +2
-2* during the call, or -ECANCELED if it was canceled. * * Callers must hold the AioContext lock of job->aio_context. */ -int job_cancel_sync (Job *job); +int job_cancel_sync (Job *job,bool force);/** Synchronously cancels all jobs using job_cancel_sync(). */ -void job_cancel_sync_all (void ); +void job_cancel_sync_all (bool force);
I think it would be better to keep job_cancel_sync_all(void) prototype and just change its behavior to do force-cancel. Anyway, this patch always pass true to it. And it would be strange to do soft-cancel-all, keeping in mind that soft cancelling only make sense for mirror in ready state.
Anyway:
Signature · 2 lines
Best regards, Vladimir
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{,_all}()
Quoted 50 lines · Vladimir Sementsov-Ogievskiy
Quoted 49 lines · Max Reitz
Callers should be able to specify whether they want
job_cancel_sync()to force-cancel the job or not.In fact, almost all invocations do not care about consistency of the result and just want the job to terminate as soon as possible, so they should pass force=true. The replication block driver is the exception.
This changes some iotest outputs, because quitting qemu while a mirror job is active will now lead to it being cancelled instead of completed, which is what we want. (Cancelling a READY mirror job with force=false may take an indefinite amount of time, which we do not want when quitting. If users want consistent results, they must have all jobs be done before they quit qemu.)
Signed-off-by Max Reitz<[email protected]>---
include/qemu/job.h | 10 ++--- block/replication.c | 4 +-
blockdev.c | 4 +-
job.c | 20 +++++++-- qemu-nbd.c | 2 +- softmmu/runstate.c | 2 +- storage-daemon/qemu-storage-daemon.c | 2 +- tests/unit/test-block-iothread.c | 2 +- tests/unit/test-blockjob.c | 2 +- tests/qemu-iotests/109.out | 60 +++++++++++---------------- tests/qemu-iotests/tests/qsd-jobs.out | 2 +-11 files changed, 55 insertions(+), 55 deletions(-)include/qemu/job.h +3
-3@@ -506,19 +506,19 @@ void job_user_cancel(Job *job, bool force, Error **errp); /*** Synchronously cancel the @job. The completion callback is called -* before the function returns. The job may actually complete -* instead of canceling itself; the circumstances under which this -* happens depend on the kind of job that is active. +* before the function returns. If @force is false, the job may +* actually complete instead of canceling itself; the circumstances +* under which this happens depend on the kind of job that is active. * * Returns the return value from the job if the job actually completed * during the call, or -ECANCELED if it was canceled. * * Callers must hold the AioContext lock of job->aio_context.
Quoted 8 lines
include/qemu/job.h void job_user_cancel(Job *job, bool force, Error **errp); · line 521
Quoted 6 lines
include/qemu/job.h +2
-2*/ -int job_cancel_sync (Job *job); +int job_cancel_sync (Job *job,bool force); /** Synchronously cancels all jobs using job_cancel_sync(). */ -void job_cancel_sync_all (void ); +void job_cancel_sync_all (bool force);I think it would be better to keep
job_cancel_sync_all(void)prototype and just change its behavior to do force-cancel. Anyway, this patch always pass true to it. And it would be strange to do soft-cancel-all, keeping in mind that soft cancelling only make sense for mirror in ready state.
Actually, yes, that’s true. I’ll drop the parameter.
Hanna
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{,_all}()
In reply to Max Reitzblock/replication.c static void replication_stop(ReplicationState *rs, bool failover, Error **errp) · line 729
Quoted 6 lines · Max Reitz
block/replication.c +1
-1@@ -726,7 +726,7 @@ static void replication_stop(ReplicationState *rs, bool failover, Error **errp)* disk, secondary disk in backup_job_completed(). */ if (s->backup_job) { -job_cancel_sync (&s->backup_job->job); +job_cancel_sync (&s->backup_job->job, false);
That's not quite correct, as backup is always force cancelled..
Signature · 2 lines
Best regards, Vladimir
Re: [PATCH for-6.2 v3 05/12] job: @force parameter for job_cancel_sync{,_all}()
Quoted 10 lines · Vladimir Sementsov-Ogievskiy
block/replication.c static void replication_stop(ReplicationState · line 730
Quoted 7 lines · Max Reitz
block/replication.c +1
-1@@ -726,7 +726,7 @@ static void replication_stop(ReplicationState*rs, bool failover, Error **errp) * disk, secondary disk in backup_job_completed(). */ if (s->backup_job) { - job_cancel_sync(&s->backup_job->job); + job_cancel_sync(&s->backup_job->job, false);That's not quite correct, as backup is always force cancelled..
Good point. I think functionally it shouldn’t make a difference, right? – but it’s better to be explicit about it and only use force=false where it actually makes a difference.
Hanna
PATCH v3 6/12 jobs: Give Job.force_cancel more meaning
In reply to Max ReitzWe largely have two cancel modes for jobs:
First, there is actual cancelling. The job is terminated as soon as possible, without trying to reach a consistent result.
Second, we have mirror in the READY state. Technically, the job is not really cancelled, but it just is a different completion mode. The job can still run for an indefinite amount of time while it tries to reach a consistent result.
We want to be able to clearly distinguish which cancel mode a job is in (when it has been cancelled). We can use Job.force_cancel for this, but right now it only reflects cancel requests from the user with force=true, but clearly, jobs that do not even distinguish between force=false and force=true are effectively always force-cancelled.
So this patch has Job.force_cancel signify whether the job will terminate as soon as possible (force_cancel=true) or whether it will effectively remain running despite being "cancelled" (force_cancel=false).
To this end, we let jobs that provide JobDriver.cancel() tell the generic job code whether they will terminate as soon as possible or not, and for jobs that do not provide that method we assume they will.
--- include/qemu/job.h | 11 ++++++++++- block/backup.c | 3 ++- block/mirror.c | 24 ++++++++++++++++++------ job.c | 6 +++++- 4 files changed, 35 insertions(+), 9 deletions(-)
include/qemu/job.h +10 -1
@@ -253,8 +253,17 @@ struct JobDriver {/** * If the callback is not NULL, it will be invoked in job_cancel_async +* +* This function must return true if the job will be cancelled +* immediately without any further I/O (mandatory if @force is +* true), and false otherwise. This lets the generic job layer +* know whether a job has been truly (force-)cancelled, or whether +* it is just in a special completion mode (like mirror after +* READY). +* (If the callback is NULL, the job is assumed to terminate +* without I/O.) */ -void (*cancel)(Job *job,bool force); +bool (*cancel)(Job *job,bool force);/** Called when the job is freed */
block/backup.c +2 -1
@@ -331,11 +331,12 @@ static void coroutine_fn backup_set_speed(BlockJob *job, int64_t speed) } } -static void backup_cancel (Job *job,bool force) +static bool backup_cancel (Job *job,bool force) {BackupBlockJob *s =container_of (job, BackupBlockJob, common.job);bdrv_cancel_in_flight (s->target_bs); +return true; }static const BlockJobDriver backup_job_driver = {
block/mirror.c +18 -6
@@ -1087,9 +1087,7 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)trace_mirror_before_sleep (s, cnt,job_is_ready (&s->common.job), delay_ns);job_sleep_ns (&s->common.job, delay_ns); -if (job_is_cancelled (&s->common.job) && - (!job_is_ready (&s->common.job) || s->common.job.force_cancel)) - { +if (job_is_cancelled (&s->common.job) && s->common.job.force_cancel) {break ; } s->last_pause_ns =qemu_clock_get_ns (QEMU_CLOCK_REALTIME );
@@ -1102,7 +1100,7 @@ immediate_exit:* the target is a copy of the source. */ assert (ret <0 || - ((s->common.job.force_cancel || !job_is_ready (&s->common.job)) && + (s->common.job.force_cancel &&job_is_cancelled (&s->common.job)));assert (need_drain);mirror_wait_for_all_io (s);
@@ -1188,14 +1186,27 @@ static bool mirror_drained_poll(BlockJob *job)return !!s->in_flight; } -static void mirror_cancel (Job *job,bool force) +static bool mirror_cancel (Job *job,bool force) {MirrorBlockJob *s =container_of (job, MirrorBlockJob, common.job);BlockDriverState *target =blk_bs (s->target); -if (force || !job_is_ready (job)) { +/* +* Before the job is READY, we treat any cancellation like a +* force-cancellation. +*/ + force = force || !job_is_ready (job); + +if (force) {bdrv_cancel_in_flight (target); } +return force; +} + +static bool commit_active_cancel (Job *job,bool force) +{ +/* Same as above in mirror_cancel() */ +return force || !job_is_ready (job); }static const BlockJobDriver mirror_job_driver = {
@@ -1225,6 +1236,7 @@ static const BlockJobDriver commit_active_job_driver = { .abort = mirror_abort, .pause = mirror_pause, .complete = mirror_complete, + .cancel = commit_active_cancel, }, .drained_poll = mirror_drained_poll, };
job.c +5 -1
@@ -719,8 +719,12 @@ static int job_finalize_single(Job *job)static void job_cancel_async (Job *job,bool force) {if (job->driver->cancel) { - job->driver->cancel (job, force); + force = job->driver->cancel (job, force); + }else { +/* No .cancel() means the job will behave as if force-cancelled */ + force = true; } +if (job->user_paused) {/* Do not call job_enter here, the caller will handle it. */ if (job->driver->user_resume) {
Signature · 1 line
2.31.1
PATCH v3 7/12 job: Add job_cancel_requested()
In reply to Max ReitzMost callers of job_is_cancelled() actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled.
A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.)
Therefore, redefine job_is_cancelled() to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and add job_cancel_requested() as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase.
Finally, here is a justification for how different job_is_cancelled() invocations are treated by this patch:
- block/mirror.c (mirror_run()):
- The first invocation is a while loop that should loop until the job has been cancelled or scheduled for completion. What kind of cancel does not matter, only the fact that the job is supposed to end.
- The second invocation wants to know whether the job has been soft-cancelled. Calling job_cancel_requested() is a bit too broad, but if the job were force-cancelled, we should leave the main loop as soon as possible anyway, so this should not matter here.
- The last two invocations already check force_cancel, so they should continue to use job_is_cancelled().
- block/backup.c, block/commit.c, block/stream.c, anything in tests/:
These jobs know only force-cancel, so there is no difference between job_is_cancelled() and job_cancel_requested(). We can continue using job_is_cancelled().
- job.c:
- job_pause_point(), job_yield(), job_sleep_ns(): Only force-cancelled jobs should be prevented from being paused. Continue using job_is_cancelled().
- job_update_rc(), job_finalize_single(), job_finish_sync(): These functions are all called after the job has left its main loop. The mirror job (the only job that can be soft-cancelled) will clear .cancelled before leaving the main loop if it has been soft-cancelled. Therefore, these functions will observe .cancelled to be true only if the job has been force-cancelled. We can continue to use job_is_cancelled().
(Furthermore, conceptually, a soft-cancelled mirror job should not report to have been cancelled. It should report completion (see also the block-job-cancel QAPI documentation). Therefore, it makes sense for these functions not to distinguish between a soft-cancelled mirror job and a job that has completed as normal.)
- job_completed_txn_abort(): All jobs other than @job have been force-cancelled. job_is_cancelled() must be true for them. Regarding @job itself: job_completed_txn_abort() is mostly called when the job's return value is not 0. A soft-cancelled mirror has a return value of 0, and so will not end up here then. However, job_cancel() invokes job_completed_txn_abort() if the job has been deferred to the main loop, which is mostly the case for completed jobs (which skip the assertion), but not for sure. To be safe, use job_cancel_requested() in this assertion.
- job_complete(): This is function eventually invoked by the user (through qmp_block_job_complete() or qmp_job_complete(), or job_complete_sync(), which comes from qemu-img). The intention here is to prevent a user from invoking job-complete after the job has been cancelled. This should also apply to soft cancelling: After a mirror job has been soft-cancelled, the user should not be able to decide otherwise and have it complete as normal (i.e. pivoting to the target).
--- include/qemu/job.h | 8 +++++++- block/mirror.c | 10 ++++------ job.c | 9 +++++++-- 3 files changed, 18 insertions(+), 9 deletions(-)
include/qemu/job.h +7 -1
@@ -436,9 +436,15 @@ const char *job_type_str(const Job *job);/** Returns true if the job should not be visible to the management layer. */ bool job_is_internal (Job *job); -/** Returns whether the job is scheduled for cancellation. */ +/** Returns whether the job is being cancelled. */ bool job_is_cancelled (Job *job); +/** +* Returns whether the job is scheduled for cancellation (at an +* indefinite point). +*/ +bool job_cancel_requested (Job *job); +/** Returns whether the job is in a completed state. */ bool job_is_completed (Job *job);
block/mirror.c +4 -6
@@ -936,7 +936,7 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)/* Transition to the READY state and wait for complete. */ job_transition_to_ready (&s->common.job); s->actively_synced = true; -while (!job_is_cancelled (&s->common.job) && !s->should_complete) { +while (!job_cancel_requested (&s->common.job) && !s->should_complete) {job_yield (&s->common.job); } s->common.job.cancelled = false;
@@ -1043,7 +1043,7 @@ static int coroutine_fn mirror_run(Job *job, Error **errp) } should_complete = s->should_complete || -job_is_cancelled (&s->common.job); +job_cancel_requested (&s->common.job); cnt =bdrv_get_dirty_count (s->dirty_bitmap); }
@@ -1087,7 +1087,7 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)trace_mirror_before_sleep (s, cnt,job_is_ready (&s->common.job), delay_ns);job_sleep_ns (&s->common.job, delay_ns); -if (job_is_cancelled (&s->common.job) && s->common.job.force_cancel) { +if (job_is_cancelled (&s->common.job)) {break ; } s->last_pause_ns =qemu_clock_get_ns (QEMU_CLOCK_REALTIME );
@@ -1099,9 +1099,7 @@ immediate_exit:* or it was cancelled prematurely so that we do not guarantee that * the target is a copy of the source. */ -assert (ret <0 || - (s->common.job.force_cancel && -job_is_cancelled (&s->common.job))); +assert (ret <0 ||job_is_cancelled (&s->common.job));assert (need_drain);mirror_wait_for_all_io (s); }
job.c +7 -2
@@ -216,6 +216,11 @@ const char *job_type_str(const Job *job) }bool job_is_cancelled (Job *job) +{ +return job->cancelled && job->force_cancel; +} + +bool job_cancel_requested (Job *job) {return job->cancelled; }
@@ -788,7 +793,7 @@ static void job_completed_txn_abort(Job *job) ctx = other_job->aio_context;aio_context_acquire (ctx);if (!job_is_completed (other_job)) { -assert (job_is_cancelled (other_job)); +assert (job_cancel_requested (other_job));job_finish_sync (other_job,NULL ,NULL ); }job_finalize_single (other_job);
@@ -1028,7 +1033,7 @@ void job_complete(Job *job, Error **errp)if (job_apply_verb (job,JOB_VERB_COMPLETE , errp)) {return ; } -if (job_is_cancelled (job) || !job->driver->complete) { +if (job_cancel_requested (job) || !job->driver->complete) {error_setg (errp,"The active block job '%s' cannot be completed" , job->id);return ;
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 07/12] job: Add job_cancel_requested()
Quoted 20 lines · Max Reitz
Most callers of
job_is_cancelled()actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled.A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.)
Therefore, redefine
job_is_cancelled()to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and addjob_cancel_requested()as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase.
Quoted 2 lines
Finally, here is a justification for how different
job_is_cancelled()invocations are treated by this patch:
Thanks for this list; it's really thorough and helpful.
Quoted 3 lines
Signed-off-by Max Reitz <[email protected]>
Although it is fixing a bug, the bug has been long-standing, so I agree with your claim that this is 6.2 material.
Quoted 10 lines
--- include/qemu/job.h | 8 +++++++- block/mirror.c | 10 ++++------ job.c | 9 +++++++-- 3 files changed, 18 insertions(+), 9 deletions(-)+++
b/job.cjob.c +0
-0@@ -216,6 +216,11 @@ const char *job_type_str(const Job *job) }
job.c const char *job_type_str(const Job *job) · line 226
Quoted 10 lines
job.c +5
-0bool job_is_cancelled (Job *job) +{ +return job->cancelled && job->force_cancel; +} + +bool job_cancel_requested (Job *job) {return job->cancelled; }
Works out rather nicely.
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 07/12] job: Add job_cancel_requested()
In reply to Max ReitzQuoted 77 lines · Max Reitz
Most callers of
job_is_cancelled()actually want to know whether the job is on its way to immediate termination. For example, we refuse to pause jobs that are cancelled; but this only makes sense for jobs that are really actually cancelled.A mirror job that is cancelled during READY with force=false should absolutely be allowed to pause. This "cancellation" (which is actually a kind of completion) may take an indefinite amount of time, and so should behave like any job during normal operation. For example, with on-target-error=stop, the job should stop on write errors. (In contrast, force-cancelled jobs should not get write errors, as they should just terminate and not do further I/O.)
Therefore, redefine
job_is_cancelled()to only return true for jobs that are force-cancelled (which as of HEAD^ means any job that interprets the cancellation request as a request for immediate termination), and addjob_cancel_requested()as the general variant, which returns true for any jobs which have been requested to be cancelled, whether it be immediately or after an arbitrarily long completion phase.Finally, here is a justification for how different
job_is_cancelled()invocations are treated by this patch:-
block/mirror.c(mirror_run()):
- The first invocation is a while loop that should loop until the job has been cancelled or scheduled for completion. What kind of cancel does not matter, only the fact that the job is supposed to end.- The second invocation wants to know whether the job has been soft-cancelled. Calling
job_cancel_requested()is a bit too broad, but if the job were force-cancelled, we should leave the main loop as soon as possible anyway, so this should not matter here.- The last two invocations already check
force_cancel, so they should continue to usejob_is_cancelled().-
block/backup.c,block/commit.c,block/stream.c, anything in tests/:
These jobs know only force-cancel, so there is no difference betweenjob_is_cancelled()andjob_cancel_requested(). We can continue usingjob_is_cancelled().-
job.c:
-job_pause_point(),job_yield(),job_sleep_ns(): Only force-cancelled jobs should be prevented from being paused. Continue usingjob_is_cancelled().-
job_update_rc(),job_finalize_single(),job_finish_sync(): These functions are all called after the job has left its main loop. The mirror job (the only job that can be soft-cancelled) will clear .cancelled before leaving the main loop if it has been soft-cancelled. Therefore, these functions will observe .cancelled to be true only if the job has been force-cancelled. We can continue to usejob_is_cancelled().
(Furthermore, conceptually, a soft-cancelled mirror job should not report to have been cancelled. It should report completion (see also the block-job-cancel QAPI documentation). Therefore, it makes sense for these functions not to distinguish between a soft-cancelled mirror job and a job that has completed as normal.)-
job_completed_txn_abort(): All jobs other than @job have been force-cancelled.job_is_cancelled()must be true for them. Regarding @job itself:job_completed_txn_abort()is mostly called when the job's return value is not 0. A soft-cancelled mirror has a return value of 0, and so will not end up here then. However,job_cancel()invokesjob_completed_txn_abort()if the job has been deferred to the main loop, which is mostly the case for completed jobs (which skip the assertion), but not for sure. To be safe, usejob_cancel_requested()in this assertion.-
job_complete(): This is function eventually invoked by the user (throughqmp_block_job_complete()orqmp_job_complete(), orjob_complete_sync(), which comes from qemu-img). The intention here is to prevent a user from invoking job-complete after the job has been cancelled. This should also apply to soft cancelling: After a mirror job has been soft-cancelled, the user should not be able to decide otherwise and have it complete as normal (i.e. pivoting to the target).
Quoted 2 lines
Signed-off-by Max Reitz<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v3 8/12 mirror: Use job_is_cancelled()
In reply to Max Reitzmirror_drained_poll() returns true whenever the job is cancelled, because "we [can] be sure that it won't issue more requests". However, this is only true for force-cancelled jobs, so use job_is_cancelled().
--- block/mirror.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
block/mirror.c +1 -1
@@ -1177,7 +1177,7 @@ static bool mirror_drained_poll(BlockJob *job)* from one of our own drain sections, to avoid a deadlock waiting for * ourselves. */ -if (!s->common.job.paused && !s->common.job.cancelled && !s->in_drain) { +if (!s->common.job.paused && !job_is_cancelled (&job->job) && !s->in_drain) {return true; }
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 08/12] mirror: Use job_is_cancelled()
Quoted 14 lines · Max Reitz
mirror_drained_poll()returns true whenever the job is cancelled, because "we [can] be sure that it won't issue more requests". However, this is only true for force-cancelled jobs, so usejob_is_cancelled().Signed-off-by Max Reitz <[email protected]>--- block/mirror.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)block/mirror.c +0
-0
block/mirror.c static bool mirror_drained_poll(BlockJob *job) · line 1182
Quoted 7 lines
block/mirror.c +1
-1@@ -1177,7 +1177,7 @@ static bool mirror_drained_poll(BlockJob *job)* from one of our own drain sections, to avoid a deadlock waiting for * ourselves. */ -if (!s->common.job.paused && !s->common.job.cancelled && !s->in_drain) { +if (!s->common.job.paused && !job_is_cancelled (&job->job) && !s->in_drain) {return true; }
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 08/12] mirror: Use job_is_cancelled()
In reply to Max ReitzQuoted 5 lines · Max Reitz
mirror_drained_poll()returns true whenever the job is cancelled, because "we [can] be sure that it won't issue more requests". However, this is only true for force-cancelled jobs, so usejob_is_cancelled().Signed-off-by Max Reitz<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v3 9/12 mirror: Check job_is_cancelled() earlier
In reply to Max ReitzWe must check whether the job is force-cancelled early in our main loop, most importantly before any continue statement. For example, we used to have continues before our current checking location that are triggered by mirror_flush() failing. So, if mirror_flush() kept failing, force-cancelling the job would not terminate it.
Jobs can be cancelled while they yield, and once they are (force-cancelled), they should not generate new I/O requests. Therefore, we should put the check after the last yield before mirror_iteration() is invoked.
--- block/mirror.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
block/mirror.c +5 -5
@@ -1000,6 +1000,11 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)job_pause_point (&s->common.job); +if (job_is_cancelled (&s->common.job)) { + ret =0 ; + goto immediate_exit; + } + cnt =bdrv_get_dirty_count (s->dirty_bitmap);/* cnt is the number of dirty bytes remaining and s->bytes_in_flight is * the number of bytes currently being processed; together those are
@@ -1078,8 +1083,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)break ; } - ret =0 ; -if (job_is_ready (&s->common.job) && !should_complete) { delay_ns = (s->in_flight ==0 && cnt ==0 ?BLOCK_JOB_SLICE_TIME :0 );
@@ -1087,9 +1090,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)trace_mirror_before_sleep (s, cnt,job_is_ready (&s->common.job), delay_ns);job_sleep_ns (&s->common.job, delay_ns); -if (job_is_cancelled (&s->common.job)) { -break ; - } s->last_pause_ns =qemu_clock_get_ns (QEMU_CLOCK_REALTIME ); }
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 09/12] mirror: Check job_is_cancelled() earlier
Quoted 11 lines · Max Reitz
We must check whether the job is force-cancelled early in our main loop, most importantly before any
continuestatement. For example, we used to havecontinues before our current checking location that are triggered bymirror_flush()failing. So, ifmirror_flush()kept failing, force-cancelling the job would not terminate it.Jobs can be cancelled while they yield, and once they are (force-cancelled), they should not generate new I/O requests. Therefore, we should put the check after the last yield before
mirror_iteration()is invoked.
Quoted 5 lines
Signed-off-by Max Reitz <[email protected]>--- block/mirror.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 09/12] mirror: Check job_is_cancelled() earlier
In reply to Max ReitzQuoted 6 lines · Max Reitz
We must check whether the job is force-cancelled early in our main loop, most importantly before any
continuestatement. For example, we used to havecontinues before our current checking location that are triggered bymirror_flush()failing. So, ifmirror_flush()kept failing, force-cancelling the job would not terminate it.
Quoted 7 lines
Jobs can be cancelled while they yield, and once they are (force-cancelled), they should not generate new I/O requests. Therefore, we should put the check after the last yield before
mirror_iteration()is invoked.Signed-off-by Max Reitz<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v3 10/12 mirror: Stop active mirroring after force-cancel
In reply to Max ReitzOnce the mirror job is force-cancelled (job_is_cancelled() is true), we should not generate new I/O requests. This applies to active mirroring, too, so stop it once the job is cancelled.
(We must still forward all I/O requests to the source, though, of course, but those are not really I/O requests generated by the job, so this is fine.)
--- block/mirror.c | 2 ++ 1 file changed, 2 insertions(+)
block/mirror.c +2 -0
@@ -1418,6 +1418,7 @@ static int coroutine_fn bdrv_mirror_top_do_write(BlockDriverState *bs,bool copy_to_target; copy_to_target = s->job->ret >=0 && + !job_is_cancelled (&s->job->common.job) && s->job->copy_mode ==MIRROR_COPY_MODE_WRITE_BLOCKING ;if (copy_to_target) {
@@ -1466,6 +1467,7 @@ static int coroutine_fn bdrv_mirror_top_pwritev(BlockDriverState *bs,bool copy_to_target; copy_to_target = s->job->ret >=0 && + !job_is_cancelled (&s->job->common.job) && s->job->copy_mode ==MIRROR_COPY_MODE_WRITE_BLOCKING ;if (copy_to_target) {
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 10/12] mirror: Stop active mirroring after force-cancel
Quoted 4 lines · Max Reitz
Once the mirror job is force-cancelled (
job_is_cancelled()is true), we should not generate new I/O requests. This applies to active mirroring, too, so stop it once the job is cancelled.
Quoted 8 lines
(We must still forward all I/O requests to the source, though, of course, but those are not really I/O requests generated by the job, so this is fine.)
Signed-off-by Max Reitz <[email protected]>--- block/mirror.c | 2 ++ 1 file changed, 2 insertions(+)
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 10/12] mirror: Stop active mirroring after force-cancel
In reply to Max ReitzQuoted 9 lines · Max Reitz
Once the mirror job is force-cancelled (
job_is_cancelled()is true), we should not generate new I/O requests. This applies to active mirroring, too, so stop it once the job is cancelled.(We must still forward all I/O requests to the source, though, of course, but those are not really I/O requests generated by the job, so this is fine.)
Signed-off-by Max Reitz<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v3 11/12 mirror: Do not clear .cancelled
In reply to Max ReitzClearing .cancelled before leaving the main loop when the job has been soft-cancelled is no longer necessary since job_is_cancelled() only returns true for jobs that have been force-cancelled.
Therefore, this only makes a differences in places that call job_cancel_requested(). In block/mirror.c, this is done only before .cancelled was cleared.
In job.c, there are two callers:
- job_completed_txn_abort() asserts that .cancelled is true, so keeping it true will not affect this place.
- job_complete() refuses to let a job complete that has .cancelled set. It is correct to refuse to let the user invoke job-complete on mirror jobs that have already been soft-cancelled.
With this change, there are no places that reset .cancelled to false and so we can be sure that .force_cancel can only be true of .cancelled is true as well. Assert this in job_is_cancelled().
--- block/mirror.c | 2 -- job.c | 4 +++- 2 files changed, 3 insertions(+), 3 deletions(-)
block/mirror.c +0 -2
@@ -939,7 +939,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)while (!job_cancel_requested (&s->common.job) && !s->should_complete) {job_yield (&s->common.job); } - s->common.job.cancelled = false; goto immediate_exit; }
@@ -1078,7 +1077,6 @@ static int coroutine_fn mirror_run(Job *job, Error **errp)* completion. */ assert (QLIST_EMPTY (&bs->tracked_requests)); - s->common.job.cancelled = false; need_drain = false;break ; }
job.c +3 -1
@@ -217,7 +217,9 @@ const char *job_type_str(const Job *job)bool job_is_cancelled (Job *job) { -return job->cancelled && job->force_cancel; +/* force_cancel may be true only if cancelled is true, too */ +assert (job->cancelled || !job->force_cancel); +return job->force_cancel; }bool job_cancel_requested (Job *job)
Signature · 1 line
2.31.1
Re: [PATCH for-6.2 v3 11/12] mirror: Do not clear .cancelled
Quoted 12 lines · Max Reitz
Clearing .cancelled before leaving the main loop when the job has been soft-cancelled is no longer necessary since
job_is_cancelled()only returns true for jobs that have been force-cancelled.Therefore, this only makes a differences in places that call
job_cancel_requested(). Inblock/mirror.c, this is done only before .cancelled was cleared.In
job.c, there are two callers:
-job_completed_txn_abort()asserts that .cancelled is true, so keeping it true will not affect this place.
Quoted 6 lines
-
job_complete()refuses to let a job complete that has .cancelled set. It is correct to refuse to let the user invoke job-complete on mirror jobs that have already been soft-cancelled.With this change, there are no places that reset .cancelled to false and so we can be sure that .force_cancel can only be true of .cancelled is
s/of/if/
Quoted 7 lines
true as well. Assert this in
job_is_cancelled().Signed-off-by Max Reitz <[email protected]>--- block/mirror.c | 2 -- job.c | 4 +++- 2 files changed, 3 insertions(+), 3 deletions(-)
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Re: [PATCH for-6.2 v3 11/12] mirror: Do not clear .cancelled
In reply to Max ReitzQuoted 16 lines · Max Reitz
Clearing .cancelled before leaving the main loop when the job has been soft-cancelled is no longer necessary since
job_is_cancelled()only returns true for jobs that have been force-cancelled.Therefore, this only makes a differences in places that call
job_cancel_requested(). Inblock/mirror.c, this is done only before .cancelled was cleared.In
job.c, there are two callers:
-job_completed_txn_abort()asserts that .cancelled is true, so keeping it true will not affect this place.-
job_complete()refuses to let a job complete that has .cancelled set. It is correct to refuse to let the user invoke job-complete on mirror jobs that have already been soft-cancelled.
Quoted 5 lines
With this change, there are no places that reset .cancelled to false and so we can be sure that .force_cancel can only be true of .cancelled is true as well. Assert this in
job_is_cancelled().Signed-off-by Max Reitz<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v3 12/12 iotests: Add mirror-ready-cancel-error test
In reply to Max ReitzTest what happens when there is an I/O error after a mirror job in the READY phase has been cancelled.
--- .../tests/mirror-ready-cancel-error | 143 ++++++++++++++++++ .../tests/mirror-ready-cancel-error.out | 5 + 2 files changed, 148 insertions(+) create mode 100755 tests/qemu-iotests/tests/mirror-ready-cancel-error create mode 100644 tests/qemu-iotests/tests/mirror-ready-cancel-error.out
tests/qemu-iotests/tests/mirror-ready-cancel-error new file +143 -0
@@ -0,0 +1,143 @@ +#!/usr/bin/env python3 +# group: rw quick +# +# Test what happens when errors occur to a mirror job after it has +# been cancelled in the READY phase +# +# Copyright (C) 2021 Red Hat, Inc. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +import os +import iotests + + +image_size = 1 * 1024 * 1024 +source = os.path.join(iotests.test_dir, 'source.img') +target = os.path.join(iotests.test_dir, 'target.img') + + +class TestMirrorReadyCancelError(iotests.QMPTestCase): + def setUp(self) -> None: + assert iotests.qemu_img_create('-f', iotests.imgfmt, source, + str(image_size)) == 0 + assert iotests.qemu_img_create('-f', iotests.imgfmt, target, + str(image_size)) == 0 + + self.vm = iotests.VM() + self.vm.launch() + + def tearDown(self) -> None: + self.vm.shutdown() + os.remove(source) + os.remove(target) + + def add_blockdevs(self, once: bool) -> None: + res = self.vm.qmp('blockdev-add', + **{'node-name': 'source', + 'driver': iotests.imgfmt, + 'file': { + 'driver': 'file', + 'filename': source + }}) + self.assert_qmp(res, 'return', {}) + + # blkdebug notes: + # Enter state 2 on the first flush, which happens before the + # job enters the READY state. The second flush will happen + # when the job is about to complete, and we want that one to + # fail. + res = self.vm.qmp('blockdev-add', + **{'node-name': 'target', + 'driver': iotests.imgfmt, + 'file': { + 'driver': 'blkdebug', + 'image': { + 'driver': 'file', + 'filename': target + }, + 'set-state': [{ + 'event': 'flush_to_disk', + 'state': 1, + 'new_state': 2 + }], + 'inject-error': [{ + 'event': 'flush_to_disk', + 'once': once, + 'immediately': True, + 'state': 2 + }]}}) + self.assert_qmp(res, 'return', {}) + + def start_mirror(self) -> None: + res = self.vm.qmp('blockdev-mirror', + job_id='mirror', + device='source', + target='target', + filter_node_name='mirror-top', + sync='full', + on_target_error='stop') + self.assert_qmp(res, 'return', {}) + + def cancel_mirror_with_error(self) -> None: + self.vm.event_wait('BLOCK_JOB_READY') + + # Write something so will not leave the job immediately, but + # flush first (which will fail, thanks to blkdebug) + res = self.vm.qmp('human-monitor-command', + command_line='qemu-io mirror-top "write 0 64k"') + self.assert_qmp(res, 'return', '') + + # Drain status change events + while self.vm.event_wait('JOB_STATUS_CHANGE', timeout=0.0) is not None: + pass + + res = self.vm.qmp('block-job-cancel', device='mirror') + self.assert_qmp(res, 'return', {}) + + self.vm.event_wait('BLOCK_JOB_ERROR') + + def test_transient_error(self) -> None: + self.add_blockdevs(True) + self.start_mirror() + self.cancel_mirror_with_error() + + while True: + e = self.vm.event_wait('JOB_STATUS_CHANGE') + if e['data']['status'] == 'standby': + # Transient error, try again + self.vm.qmp('block-job-resume', device='mirror') + elif e['data']['status'] == 'null': + break + + def test_persistent_error(self) -> None: + self.add_blockdevs(False) + self.start_mirror() + self.cancel_mirror_with_error() + + while True: + e = self.vm.event_wait('JOB_STATUS_CHANGE') + if e['data']['status'] == 'standby': + # Persistent error, no point in continuing + self.vm.qmp('block-job-cancel', device='mirror', force=True) + elif e['data']['status'] == 'null': + break + + +if __name__ == '__main__': + # LUKS would require special key-secret handling in add_blockdevs() + iotests.main(supported_fmts=['generic'], + unsupported_fmts=['luks'], + supported_protocols=['file'])
tests/qemu-iotests/tests/mirror-ready-cancel-error.out new file +5 -0
@@ -0,0 +1,5 @@ +.. +---------------------------------------------------------------------- +Ran 2 tests + +OK
Signature · 1 line
2.31.1