qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Emanuele Giuseppe Esposito <eesposit@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>, qemu-block@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>, Hanna Reitz <hreitz@redhat.com>,
	Stefan Weil <sw@weilnetz.de>,
	Aarushi Mehta <mehta.aaru20@gmail.com>,
	Julia Suvorova <jusual@redhat.com>,
	Stefan Hajnoczi <stefanha@redhat.com>,
	Stefano Garzarella <sgarzare@redhat.com>,
	Fam Zheng <fam@euphon.net>,
	qemu-devel@nongnu.org
Subject: Re: [PATCH v2 1/3] linux-aio: use LinuxAioState from the running thread
Date: Fri, 28 Oct 2022 14:16:14 +0200	[thread overview]
Message-ID: <5adbe3ec-c02f-e57e-9242-45910c074105@redhat.com> (raw)
In-Reply-To: <7d34077e-ee9d-26ea-ad94-ecc5c7f7376c@redhat.com>



Am 28/10/2022 um 13:51 schrieb Paolo Bonzini:
> On 10/28/22 09:16, Emanuele Giuseppe Esposito wrote:
>> From: Paolo Bonzini <pbonzini@redhat.com>
>>
>> Remove usage of aio_context_acquire by always submitting asynchronous
>> AIO to the current thread's LinuxAioState.
>>
>> In order to prevent mistakes from the caller side, avoid passing
>> LinuxAioState
>> in laio_io_{plug/unplug} and laio_co_submit.
>>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
>> ---
>>   block/file-posix.c      | 10 +++-------
>>   block/linux-aio.c       | 34 ++++++++++++++++++----------------
>>   include/block/aio.h     |  4 ----
>>   include/block/raw-aio.h | 10 ++++------
>>   4 files changed, 25 insertions(+), 33 deletions(-)
>>
>> diff --git a/block/file-posix.c b/block/file-posix.c
>> index 23acffb9a4..23fe98eb3e 100644
>> --- a/block/file-posix.c
>> +++ b/block/file-posix.c
>> @@ -2099,10 +2099,8 @@ static int coroutine_fn
>> raw_co_prw(BlockDriverState *bs, uint64_t offset,
>>   #endif
>>   #ifdef CONFIG_LINUX_AIO
>>       } else if (s->use_linux_aio) {
>> -        LinuxAioState *aio =
>> aio_get_linux_aio(bdrv_get_aio_context(bs));
>>           assert(qiov->size == bytes);
>> -        return laio_co_submit(bs, aio, s->fd, offset, qiov, type,
>> -                              s->aio_max_batch);
>> +        return laio_co_submit(s->fd, offset, qiov, type,
>> s->aio_max_batch);
>>   #endif
>>       }
>>   @@ -2142,8 +2140,7 @@ static void raw_aio_plug(BlockDriverState *bs)
>>       BDRVRawState __attribute__((unused)) *s = bs->opaque;
>>   #ifdef CONFIG_LINUX_AIO
>>       if (s->use_linux_aio) {
>> -        LinuxAioState *aio =
>> aio_get_linux_aio(bdrv_get_aio_context(bs));
>> -        laio_io_plug(bs, aio);
>> +        laio_io_plug();
>>       }
>>   #endif
>>   #ifdef CONFIG_LINUX_IO_URING
>> @@ -2159,8 +2156,7 @@ static void raw_aio_unplug(BlockDriverState *bs)
>>       BDRVRawState __attribute__((unused)) *s = bs->opaque;
>>   #ifdef CONFIG_LINUX_AIO
>>       if (s->use_linux_aio) {
>> -        LinuxAioState *aio =
>> aio_get_linux_aio(bdrv_get_aio_context(bs));
>> -        laio_io_unplug(bs, aio, s->aio_max_batch);
>> +        laio_io_unplug(s->aio_max_batch);
>>       }
>>   #endif
>>   #ifdef CONFIG_LINUX_IO_URING
>> diff --git a/block/linux-aio.c b/block/linux-aio.c
>> index d2cfb7f523..c806d3bb91 100644
>> --- a/block/linux-aio.c
>> +++ b/block/linux-aio.c
>> @@ -16,6 +16,9 @@
>>   #include "qemu/coroutine.h"
>>   #include "qapi/error.h"
>>   +/* Only used for assertions.  */
>> +#include "qemu/coroutine_int.h"
>> +
>>   #include <libaio.h>
>>     /*
>> @@ -56,10 +59,8 @@ struct LinuxAioState {
>>       io_context_t ctx;
>>       EventNotifier e;
>>   -    /* io queue for submit at batch.  Protected by AioContext lock. */
>> +    /* All data is only used in one I/O thread.  */
>>       LaioQueue io_q;
>> -
>> -    /* I/O completion processing.  Only runs in I/O thread.  */
>>       QEMUBH *completion_bh;
>>       int event_idx;
>>       int event_max;
>> @@ -102,9 +103,8 @@ static void qemu_laio_process_completion(struct
>> qemu_laiocb *laiocb)
>>        * later.  Coroutines cannot be entered recursively so avoid doing
>>        * that!
>>        */
>> -    if (!qemu_coroutine_entered(laiocb->co)) {
>> -        aio_co_wake(laiocb->co);
>> -    }
>> +    assert(laiocb->co->ctx == laiocb->ctx->aio_context);
>> +    qemu_coroutine_enter_if_inactive(laiocb->co);
> 
> This is wrong, it misses the aio_context_acquire/aio_context_release
> pair in aio_co_enter() (which is called by aio_co_wake()).
> 
> Likewise in patch 2.

Makes sense, I'll drop this hunk in v3.

Emanuele



  reply	other threads:[~2022-10-28 12:36 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-28  7:16 [PATCH v2 0/3] AioContext removal: LinuxAioState and ThreadPool Emanuele Giuseppe Esposito
2022-10-28  7:16 ` [PATCH v2 1/3] linux-aio: use LinuxAioState from the running thread Emanuele Giuseppe Esposito
2022-10-28 11:51   ` Paolo Bonzini
2022-10-28 12:16     ` Emanuele Giuseppe Esposito [this message]
2022-10-28  7:16 ` [PATCH v2 2/3] io_uring: use LuringState " Emanuele Giuseppe Esposito
2022-10-28  7:16 ` [PATCH v2 3/3] thread-pool: use ThreadPool " Emanuele Giuseppe Esposito
2022-10-28 11:52   ` Paolo Bonzini

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5adbe3ec-c02f-e57e-9242-45910c074105@redhat.com \
    --to=eesposit@redhat.com \
    --cc=fam@euphon.net \
    --cc=hreitz@redhat.com \
    --cc=jusual@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=mehta.aaru20@gmail.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=sgarzare@redhat.com \
    --cc=stefanha@redhat.com \
    --cc=sw@weilnetz.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).