Re: [Qemu-devel] [Qemu-block] [PATCH 6/7] qemu-iotests: 141: reduce likelihood of race condition on systems with fast IO

[Kevin Wolf]

Am 06.04.2016 um 18:15 hat Max Reitz geschrieben:
> On 05.04.2016 11:21, Sascha Silbe wrote:
> > On systems with fast IO, qemu-io may write more than 1 MiB before
> > receiving the block-job-cancel command, causing the test case to fail.
> > 
> > 141 is inherently racy, but we can at least reduce the likelihood of the
> > job completing before the cancel command arrives by bumping the size of
> > the data getting written; we'll try 32 MiB for a start.
> 
> Hm, interesting. I tried to prevent this by setting the block jobs'
> speed to 1, which should make it stop after the block job has processed
> the first block of data.
> 
> I won't oppose this patch, because if it fixes things for you, that's
> good. But I don't think it should be necessary.

We don't generally change test cases when they fail. Making a test case
pass doesn't "fix things" per se. It only helps when the failure is a
false positive.

In this case, it looks like there might be a problem with block job
throttling, so maybe we should look into that before changing the test?

Kevin

> > Once we actually move enough data around for the block job not to
> > complete prematurely, the test will still fail because the offset value
> > in the BLOCK_JOB_CANCELLED event will vary. Since this is more or less
> > inherent to the nature of this event, we just replace it with a fixed
> > value globally (in _filter_qmp), the same way we handle timestamps.
> > 
> > Signed-off-by: Sascha Silbe <address@hidden>
> > Reviewed-by: Bo Tu <address@hidden>
> > ---
> >  tests/qemu-iotests/141           | 11 ++++++-----
> >  tests/qemu-iotests/141.out       | 24 ++++++++++++------------
> >  tests/qemu-iotests/common.filter |  1 +
> >  3 files changed, 19 insertions(+), 17 deletions(-)
>

pgpqhGWL1TdCh.pgp
Description: PGP signature