[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 16/20] qemu-img: call aio_context_acquire/releas
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-devel] [PATCH 16/20] qemu-img: call aio_context_acquire/release around block job |
Date: |
Wed, 19 Oct 2016 13:05:34 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.3.0 |
On 19/10/2016 02:54, Fam Zheng wrote:
> On Mon, 10/17 15:54, Paolo Bonzini wrote:
>> This will be needed by bdrv_reopen_multiple, which calls
>> bdrv_drain_all and thus will *release* the AioContext.
>
> Looks okay, but I wonder how bdrv_drain_all releasing AioContext break
> anything?
If you don't acquire it first, you get an assertion failure for
releasing a mutex you haven't acquired yet.
Initially I made aio_context_acquire a nop for the default QEMU context,
but then decided it was a hack.
Paolo
> Fam
>
>>
>> Signed-off-by: Paolo Bonzini <address@hidden>
>> ---
>> v1->v2: new [qemu-iotests]
>>
>> qemu-img.c | 6 ++++++
>> 1 file changed, 6 insertions(+)
>>
>> diff --git a/qemu-img.c b/qemu-img.c
>> index 02c07b9..ad7c964 100644
>> --- a/qemu-img.c
>> +++ b/qemu-img.c
>> @@ -795,6 +795,7 @@ static void run_block_job(BlockJob *job, Error **errp)
>> {
>> AioContext *aio_context = blk_get_aio_context(job->blk);
>>
>> + aio_context_acquire(aio_context);
>> do {
>> aio_poll(aio_context, true);
>> qemu_progress_print(job->len ?
>> @@ -802,6 +803,7 @@ static void run_block_job(BlockJob *job, Error **errp)
>> } while (!job->ready);
>>
>> block_job_complete_sync(job, errp);
>> + aio_context_release(aio_context);
>>
>> /* A block job may finish instantaneously without publishing any
>> progress,
>> * so just signal completion here */
>> @@ -819,6 +821,7 @@ static int img_commit(int argc, char **argv)
>> Error *local_err = NULL;
>> CommonBlockJobCBInfo cbi;
>> bool image_opts = false;
>> + AioContext *aio_context;
>>
>> fmt = NULL;
>> cache = BDRV_DEFAULT_CACHE;
>> @@ -928,8 +931,11 @@ static int img_commit(int argc, char **argv)
>> .bs = bs,
>> };
>>
>> + aio_context = bdrv_get_aio_context(bs);
>> + aio_context_acquire(aio_context);
>> commit_active_start("commit", bs, base_bs, 0, BLOCKDEV_ON_ERROR_REPORT,
>> common_block_job_cb, &cbi, &local_err, false);
>> + aio_context_release(aio_context);
>> if (local_err) {
>> goto done;
>> }
>> --
>> 2.7.4
>>
>>
- [Qemu-devel] [PATCH 09/20] nfs: use BDRV_POLL_WHILE, (continued)
- [Qemu-devel] [PATCH 09/20] nfs: use BDRV_POLL_WHILE, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 08/20] nfs: move nfs_set_events out of the while loops, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 11/20] aio: introduce qemu_get_current_aio_context, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 10/20] sheepdog: use BDRV_POLL_WHILE, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 12/20] iothread: detach all block devices before stopping them, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 13/20] replication: pass BlockDriverState to reopen_backing_file, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 14/20] block: prepare bdrv_reopen_multiple to release AioContext, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 15/20] qemu-io: acquire AioContext, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 16/20] qemu-img: call aio_context_acquire/release around block job, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 17/20] block: only call aio_poll on the current thread's AioContext, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 19/20] qemu-thread: introduce QemuRecMutex, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 18/20] iothread: release AioContext around aio_poll, Paolo Bonzini, 2016/10/17
- [Qemu-devel] [PATCH 20/20] aio: convert from RFifoLock to QemuRecMutex, Paolo Bonzini, 2016/10/17
- Re: [Qemu-devel] [PATCH v2 00/20] dataplane: remove RFifoLock, Fam Zheng, 2016/10/18