job_cancel_locked() drops the job list lock temporarily and it may call
aio_poll(). We must assume that the list has changed after this call.
Also, with unlucky timing, it can end up freeing the job during
job_completed_txn_abort_locked(), making the job pointer invalid, too.
For both reasons, we can't just continue at block_job_next_locked(job).
Instead, start at the head of the list again after job_cancel_locked()
and skip those jobs that we already cancelled (or that are completing
anyway).
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230503140142.474404-1-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit e2626874a3)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Change the job_{lock/unlock} and macros to use job_mutex.
Now that they are not nop anymore, remove the aiocontext
to avoid deadlocks.
Therefore:
- when possible, remove completely the aiocontext lock/unlock pair
- if it is used by some other function too, reduce the locking
section as much as possible, leaving the job API outside.
- change AIO_WAIT_WHILE in AIO_WAIT_WHILE_UNLOCKED, since we
are not using the aiocontext lock anymore
The only functions that still need the aiocontext lock are:
- the JobDriver callbacks, already documented in job.h
- job_cancel_sync() in replication.c is called with aio_context_lock
taken, but now job is using AIO_WAIT_WHILE_UNLOCKED so we need to
release the lock.
Reduce the locking section to only cover the callback invocation
and document the functions that take the AioContext lock,
to avoid taking it twice.
Also remove real_job_{lock/unlock}, as they are replaced by the
public functions.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20220926093214.506243-19-eesposit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Now that the API offers also _locked() functions, take advantage
of it and give also the caller control to take the lock and call
_locked functions.
This makes sense especially when we have for loops, because it
makes no sense to have:
for(job = job_next(); ...)
where each job_next() takes the lock internally.
Instead we want
JOB_LOCK_GUARD();
for(job = job_next_locked(); ...)
In addition, protect also direct field accesses, by either creating a
new critical section or widening the existing ones.
Note: at this stage, job_{lock/unlock} and job lock guard macros
are *nop*.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-Id: <20220926093214.506243-12-eesposit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Both blockdev.c and job-qmp.c have TOC/TOU conditions, because
they first search for the job and then perform an action on it.
Therefore, we need to do the search + action under the same
job mutex critical section.
Note: at this stage, job_{lock/unlock} and job lock guard macros
are *nop*.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20220926093214.506243-9-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Similarly to the previous patch, split block_int.h
in block_int-io.h and block_int-global-state.h
block_int-common.h contains the structures shared between
the two headers, and the functions that can't be categorized as
I/O or global state.
Assertions are added in the next patch.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20220303151616.325444-12-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
All the global state (GS) API functions will check that
qemu_in_main_thread() returns true. If not, it means
that the safety of BQL cannot be guaranteed, and
they need to be moved to I/O.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20220303151616.325444-5-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
drive_def is only a particular use case of
qemu_opts_parse_noisily, so it can be inlined.
Also remove drive_mark_claimed_by_board, as it is only defined
but not implemented (nor used) anywhere.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20211215121140.456939-3-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are going to drop BlockJob.blk. So let's retrieve block job context
from underlying job instead of main node.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Nikita Lapshin <nikita.lapshin@virtuozzo.com>
drive_get_next() is basically a bad idea. It returns the "next" block
backend of a certain interface type. "Next" means bus=0,unit=N, where
subsequent calls count N up from zero, per interface type.
This lets you define unit numbers implicitly by execution order. If the
order changes, or new calls appear "in the middle", unit numbers change.
ABI break. Hard to spot in review.
The previous commits eliminated all uses. Drop the function.
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20211117163409.3587705-14-armbru@redhat.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
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,
specifically the active commit job it runs.
As for job_cancel_sync_all(), all callers want it to force-cancel all
jobs, because that is the point of it: To cancel all remaining jobs as
quickly as possible (generally on process termination). So make it
invoke job_cancel_sync() with force=true.
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.)
Buglink: https://gitlab.com/qemu-project/qemu/-/issues/462
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20211006151940.214590-6-hreitz@redhat.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
As the BlockReopenQueue can contain nodes in multiple AioContexts, only
one of which may be locked when AIO_WAIT_WHILE() can be called, we can't
let the caller lock the right contexts. Instead, individually lock the
AioContext of a single node when iterating the queue.
Reintroduce bdrv_reopen() as a wrapper for reopening a single node that
drains the node and temporarily drops the AioContext lock for
bdrv_reopen_multiple().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20210708114709.206487-4-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
