qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] another locking issue in current dataplane code?


From: Paolo Bonzini
Subject: Re: [Qemu-devel] another locking issue in current dataplane code?
Date: Tue, 08 Jul 2014 21:50:29 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.6.0

Il 08/07/2014 21:07, Christian Borntraeger ha scritto:
On 08/07/14 19:08, Paolo Bonzini wrote:
Il 08/07/2014 17:59, Stefan Hajnoczi ha scritto:
I sent Christian an initial patch to fix this but now both threads are
stuck in rfifolock_lock() inside cond wait.  That's very strange and
should never happen.

I had this patch pending for 2.2:

commit 6c81e31615c3cda5ea981a998ba8b1b8ed17de6f
Author: Paolo Bonzini <address@hidden>
Date:   Mon Jul 7 10:39:49 2014 +0200

    iothread: do not rely on aio_poll(ctx, true) result to end a loop

    Currently, whenever aio_poll(ctx, true) has completed all pending
    work it returns true *and* the next call to aio_poll(ctx, true)
    will not block.

    This invariant has its roots in qemu_aio_flush()'s implementation
    as "while (qemu_aio_wait()) {}".  However, qemu_aio_flush() does
    not exist anymore and bdrv_drain_all() is implemented differently;
    and this invariant is complicated to maintain and subtly different
    from the return value of GMainLoop's g_main_context_iteration.

    All calls to aio_poll(ctx, true) except one are guarded by a
    while() loop checking for a request to be incomplete, or a
    BlockDriverState to be idle.  Modify that one exception in
    iothread.c.

    Signed-off-by: Paolo Bonzini <address@hidden>

The hangs are gone. Looks like 2.1 material now...

Acked-by: Christian Borntraeger <address@hidden>
Tested-by: Christian Borntraeger <address@hidden>

Great, I'll send it out tomorrow morning.

Paolo




reply via email to

[Prev in Thread] Current Thread [Next in Thread]