Re: [Qemu-block] [PATCH] linux-aio: Cancel BH if not needed

[Stefan Hajnoczi]

On Fri, Jun 17, 2016 at 12:24:11PM +0200, Kevin Wolf wrote:
> Am 17.06.2016 um 12:13 hat Stefan Hajnoczi geschrieben:
> > On Wed, Jun 15, 2016 at 01:16:42PM +0200, Kevin Wolf wrote:
> > > linux-aio uses a BH in order to make sure that the remaining completions
> > > are processed even in nested event loops of completion callbacks in
> > > order to avoid deadlocks.
> > > 
> > > There is no need, however, to have the BH overhead for the first call
> > > into qemu_laio_completion_bh() or after all pending completions have
> > > already been processed. Therefore, this patch calls directly into
> > > qemu_laio_completion_bh() in qemu_laio_completion_cb() and cancels
> > > the BH after qemu_laio_completion_bh() has processed all pending
> > > completions.
> > > 
> > > Signed-off-by: Kevin Wolf <address@hidden>
> > > ---
> > >  block/linux-aio.c | 4 +++-
> > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > 
> > I ran 4 x iodepth=16 random 4KB read I/O benchmarks.  There might be an
> > improvement but it's within the error margin.  My benchmarking setup can
> > be noisy...
> > 
> > Anyway, this patch doesn't hurt performance.  Guest and host are RHEL 7.2.
> 
> Thanks for confirming!
> 
> > $ ./analyze.py runs/
> > Name                                          IOPS   Error
> > linux-aio-bh-optimizations-ccb9dc1      12942616.0 ± 16.83%
> > linux-aio-bh-optimizations-ccb9dc1-2    13833110.4 ± 4.74%
> > linux-aio-bh-optimizations-off-23b0d9f  13303981.4 ± 2.21%
> 
> What are these three commits? Is the first one with your virtio-blk
> changes and this patch, the second only this patch and the third one
> the baseline?

The git trees are from qemu.git/master with no out-of-tree patches (my
stuff isn't there).

1. linux-aio-bh-optimizations-ccb9dc1      12942616.0 ± 16.83%

This is your "linux-aio: Cancel BH if not needed" in qemu.git/master.

2. linux-aio-bh-optimizations-ccb9dc1-2    13833110.4 ± 4.74%

I wanted to try again because the 16.83% error margin is very high and
probably due to noise.  This is just a re-run of #1.

3. linux-aio-bh-optimizations-off-23b0d9f  13303981.4 ± 2.21%

This is the commit before your "linux-aio: Cancel BH if not needed"
(ccb9dc1^ == 23b0d9f).

signature.asc
Description: PGP signature