Re: [Qemu-devel] [RFC PATCH 3/3] linux-aio: Don't reenter request coroutine recursively

[Kevin Wolf <kwolf@redhat.com>]

Am 04.12.2014 um 16:22 hat Ming Lei geschrieben:
> On Thu, Dec 4, 2014 at 10:37 PM, Kevin Wolf <kwolf@redhat.com> wrote:
> > Am 26.11.2014 um 15:46 hat Kevin Wolf geschrieben:
> >> When getting an error while submitting requests, we must be careful to
> >> wake up only inactive coroutines. Therefore we must special-case the
> >> currently active coroutine and communicate an error for that request
> >> using the ordinary return value of ioq_submit().
> >>
> >> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> >> ---
> >>  block/linux-aio.c | 23 ++++++++++++++++-------
> >>  1 file changed, 16 insertions(+), 7 deletions(-)
> >
> > Ming, did you have a look at this patch specifically? Does it fix the
> > issue that you tried to address with a much more complex callback-based
> > patch?
> 
> I think your patch can't fix my issue.
> 
> As I explained, we have to handle -EAGAIN and partial submission,
> which can be triggered quite easily in case of multi-queue, and other
> case like NVME.

Yes, this is an alternative only to the first patch of your series. If
we go that route, your second patch would still be needed as well, of
course. It should be relatively obvious how to change it to apply on top
of this one.

Kevin