Re: [PATCH v11 13/21] jobs: protect job.aio_context with BQL and job_mutex

[Emanuele Giuseppe Esposito <eesposit@redhat.com>]

Am 14/09/2022 um 15:25 schrieb Vladimir Sementsov-Ogievskiy:
> On 8/26/22 16:20, Emanuele Giuseppe Esposito wrote:
>> In order to make it thread safe, implement a "fake rwlock",
>> where we allow reads under BQL *or* job_mutex held, but
>> writes only under BQL *and* job_mutex.
>>
>> The only write we have is in child_job_set_aio_ctx, which always
>> happens under drain (so the job is paused).
>> For this reason, introduce job_set_aio_context and make sure that
>> the context is set under BQL, job_mutex and drain.
>> Also make sure all other places where the aiocontext is read
>> are protected.
>>
>> The reads in commit.c and mirror.c are actually safe, because always
>> done under BQL.
>>
>> Note: at this stage, job_{lock/unlock} and job lock guard macros
>> are *nop*.
>>
>> Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
>> Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
>> ---
>>   block/replication.c |  7 +++++--
>>   blockjob.c          |  3 ++-
>>   include/qemu/job.h  | 23 ++++++++++++++++++++---
>>   job.c               | 12 ++++++++++++
>>   4 files changed, 39 insertions(+), 6 deletions(-)
>>
>> diff --git a/block/replication.c b/block/replication.c
>> index 55c8f894aa..6e02d98126 100644
>> --- a/block/replication.c
>> +++ b/block/replication.c
>> @@ -142,14 +142,17 @@ static void replication_close(BlockDriverState *bs)
>>   {
>>       BDRVReplicationState *s = bs->opaque;
>>       Job *commit_job;
>> +    GLOBAL_STATE_CODE();
>>         if (s->stage == BLOCK_REPLICATION_RUNNING) {
>>           replication_stop(s->rs, false, NULL);
>>       }
>>       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, false);
>> +        WITH_JOB_LOCK_GUARD() {
>> +            assert(commit_job->aio_context ==
>> qemu_get_current_aio_context());
>> +            job_cancel_sync_locked(commit_job, false);
>> +        }
> 
> As Kevin said, this hunk seems not needed.. Why to add locking for
> reading aio_context, when we have GLOBAL_STATE_CODE()?

Ok, getting rid of it.
> 
>>       }
>>         if (s->mode == REPLICATION_MODE_SECONDARY) {
>> diff --git a/blockjob.c b/blockjob.c
>> index 96fb9d9f73..c8919cef9b 100644
>> --- a/blockjob.c
>> +++ b/blockjob.c
>> @@ -162,12 +162,13 @@ static void child_job_set_aio_ctx(BdrvChild *c,
>> AioContext *ctx,
>>           bdrv_set_aio_context_ignore(sibling->bs, ctx, ignore);
>>       }
>>   -    job->job.aio_context = ctx;
>> +    job_set_aio_context(&job->job, ctx);
>>   }
>>     static AioContext *child_job_get_parent_aio_context(BdrvChild *c)
>>   {
>>       BlockJob *job = c->opaque;
>> +    GLOBAL_STATE_CODE();
>>         return job->job.aio_context;
>>   }
>> diff --git a/include/qemu/job.h b/include/qemu/job.h
>> index 5709e8d4a8..cede227e67 100644
>> --- a/include/qemu/job.h
>> +++ b/include/qemu/job.h
>> @@ -74,11 +74,17 @@ typedef struct Job {
>>       /* ProgressMeter API is thread-safe */
>>       ProgressMeter progress;
>>   +    /**
>> +     * AioContext to run the job coroutine in.
>> +     * The job Aiocontext can be read when holding *either*
>> +     * the BQL (so we are in the main loop) or the job_mutex.
>> +     * It can only be written when we hold *both* BQL
>> +     * and the job_mutex.
>> +     */
>> +    AioContext *aio_context;
>>   -    /** Protected by AioContext lock */
>>   -    /** AioContext to run the job coroutine in */
>> -    AioContext *aio_context;
>> +    /** Protected by AioContext lock */
>>         /** Reference count of the block job */
>>       int refcnt;
>> @@ -741,4 +747,15 @@ int job_finish_sync(Job *job, void (*finish)(Job
>> *, Error **errp),
>>   int job_finish_sync_locked(Job *job, void (*finish)(Job *, Error
>> **errp),
>>                              Error **errp);
>>   +/**
>> + * Sets the @job->aio_context.
>> + * Called with job_mutex *not* held.
>> + *
>> + * This function must run in the main thread to protect against
>> + * concurrent read in job_finish_sync_locked(), takes the job_mutex
>> + * lock to protect against the read in job_do_yield_locked(), and must
>> + * be called when the coroutine is quiescent.
> 
> May be "job is quiscent" or "job is doing nothing", "no in-flight io
> operations in job".
> 
> For example, backup has several running coroutines in contest of
> block_copy process, and main coroutine of the job
> is almost always "quescent"..

"job is quiescent" seems ok

> 
>> + */
>> +void job_set_aio_context(Job *job, AioContext *ctx);
>> +
>>   #endif
>> diff --git a/job.c b/job.c
>> index 85ae843f03..9f2fb2e73b 100644
>> --- a/job.c
>> +++ b/job.c
>> @@ -396,6 +396,17 @@ Job *job_get(const char *id)
>>       return job_get_locked(id);
>>   }
>>   +void job_set_aio_context(Job *job, AioContext *ctx)
>> +{
>> +    /* protect against read in job_finish_sync_locked and job_start */
>> +    GLOBAL_STATE_CODE();
>> +    /* protect against read in job_do_yield_locked */
>> +    JOB_LOCK_GUARD();
>> +    /* ensure the coroutine is quiescent while the AioContext is
>> changed */
> 
> same not here.

Ok

Thank you,
Emanuele
> 
>> +    assert(job->paused || job_is_completed_locked(job));
>> +    job->aio_context = ctx;
>> +}
>> +
>>   /* Called with job_mutex *not* held. */
>>   static void job_sleep_timer_cb(void *opaque)
>>   {
>> @@ -1379,6 +1390,7 @@ int job_finish_sync_locked(Job *job,
>>   {
>>       Error *local_err = NULL;
>>       int ret;
>> +    GLOBAL_STATE_CODE();
>>         job_ref_locked(job);
>>   
> 
>