All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Philippe Mathieu-Daudé" <philmd@redhat.com>
To: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>, Fam Zheng <fam@euphon.net>,
	qemu-block@nongnu.org, qemu-devel@nongnu.org,
	Max Reitz <mreitz@redhat.com>,
	Maxim Levitsky <mlevitsk@redhat.com>
Subject: Re: [PATCH v2 12/12] block/nvme: Use per-queue AIO context
Date: Sat, 4 Jul 2020 23:27:57 +0200	[thread overview]
Message-ID: <feb25dc2-95e0-cfdf-412d-df2e89044339@redhat.com> (raw)
In-Reply-To: <20200701160328.GS126613@stefanha-x1.localdomain>

On 7/1/20 6:03 PM, Stefan Hajnoczi wrote:
> On Tue, Jun 30, 2020 at 09:13:18PM +0200, Philippe Mathieu-Daudé wrote:
>> To be able to use multiple queues on the same hardware,
>> we need to have each queue able to receive IRQ notifications
>> in the correct AIO context.
>> The AIO context and the notification handler have to be proper
>> to each queue, not to the block driver. Move aio_context and
>> irq_notifier from BDRVNVMeState to NVMeQueuePair.
> 
> If I understand correctly, after this patch:
> 
> 1. Each queue pair has an EventNotifier irq_notifier but only the admin
>    queuepair's irq_notifier is hooked up to VFIO. This means all
>    interrupts come via the admin queuepair.

Yes, but this was the behavior before this patch too.

> 
>    (This also means all queuepairs still need to be polled for
>    completions when the interrupt fires because we don't know which
>    queuepair had a completion event.)

Yes.

> 
> 2. AioContexts must be identical across all queuepairs and
>    BlockDriverStates. Although they all have their own AioContext
>    pointer there is no true support for different AioContexts yet.

I'm not sure. Maybe v3 will sort that out.

> 
>    (For example, nvme_cmd_sync() is called with a bs argument but
>    AIO_WAIT_WHILE(q->aio_context, ...) uses the queuepair aio_context so
>    the assumption is that they match.)
> 
> Please confirm and add something equivalent into the commit description
> so the assumptions/limitations are clear.

I'll do my best!

Thanks for your reviews,

Phil.



      reply	other threads:[~2020-07-04 21:29 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-30 19:13 [PATCH v2 00/12] block/nvme: Various cleanups required to use multiple queues Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 01/12] block/nvme: Replace magic value by SCALE_MS definition Philippe Mathieu-Daudé
2020-07-01 15:48   ` Stefan Hajnoczi
2020-06-30 19:13 ` [PATCH v2 02/12] block/nvme: Avoid further processing if trace event not enabled Philippe Mathieu-Daudé
2020-07-01 15:48   ` Stefan Hajnoczi
2020-06-30 19:13 ` [PATCH v2 03/12] block/nvme: Let nvme_create_queue_pair() fail gracefully Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 04/12] block/nvme: Define QUEUE_INDEX macros to ease code review Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 05/12] block/nvme: Rename local variable Philippe Mathieu-Daudé
2020-07-01 15:49   ` Stefan Hajnoczi
2020-06-30 19:13 ` [PATCH v2 06/12] block/nvme: Use union of NvmeIdCtrl / NvmeIdNs structures Philippe Mathieu-Daudé
2020-07-01 15:51   ` Stefan Hajnoczi
2020-06-30 19:13 ` [PATCH v2 07/12] block/nvme: Replace qemu_try_blockalign0 by qemu_try_blockalign/memset Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 08/12] block/nvme: Replace qemu_try_blockalign(bs) by qemu_try_memalign(pg_sz) Philippe Mathieu-Daudé
2020-07-01 15:51   ` Stefan Hajnoczi
2020-06-30 19:13 ` [PATCH v2 09/12] block/nvme: Simplify nvme_init_queue() arguments Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 10/12] block/nvme: Replace BDRV_POLL_WHILE by AIO_WAIT_WHILE Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 11/12] block/nvme: Simplify nvme_create_queue_pair() arguments Philippe Mathieu-Daudé
2020-06-30 19:13 ` [PATCH v2 12/12] block/nvme: Use per-queue AIO context Philippe Mathieu-Daudé
2020-07-01 16:03   ` Stefan Hajnoczi
2020-07-04 21:27     ` Philippe Mathieu-Daudé [this message]

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=feb25dc2-95e0-cfdf-412d-df2e89044339@redhat.com \
    --to=philmd@redhat.com \
    --cc=fam@euphon.net \
    --cc=kwolf@redhat.com \
    --cc=mlevitsk@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.