Re: [PULL v2 0/9] Block jobs & NBD patches

[Vladimir Sementsov-Ogievskiy]

On 7/1/22 20:02, John Snow wrote:
> On Wed, Jun 29, 2022 at 7:18 PM Richard Henderson
> <richard.henderson@linaro.org> wrote:
> > On 6/29/22 13:45, Vladimir Sementsov-Ogievskiy wrote:
> > > The following changes since commit ad4c7f529a279685da84297773b4ec8080153c2d:
> > >
> > >     Merge tag 'pull-semi-20220628' of https://gitlab.com/rth7680/qemu into 
> > > staging (2022-06-28 10:24:31 +0530)
> > >
> > > are available in the Git repository at:
> > >
> > >     https://gitlab.com/vsementsov/qemu.git tags/pull-block-2022-06-14-v2
> > >
> > > for you to fetch changes up to 1b8f777673985af366de099ad4e41d334b36fb12:
> > >
> > >     block: use 'unsigned' for in_flight field on driver state (2022-06-29 
> > > 10:57:02 +0300)
> > >
> > > ----------------------------------------------------------------
> > > Block jobs & NBD patches
> > >
> > > v2: - add arguments to QEMUMachine constructor in test, to make it work
> > >         on arm in gitlab pipeline
> > >       - use bdrv_inc_in_flight() / bdrv_dec_in_flight() instead of direct
> > >         manipulation with bs->in_flight
> >
> > Applied, thanks.  Please update https://wiki.qemu.org/ChangeLog/7.1 as 
> > appropriate.
> >
> >
> > r~
> >
> >
> > > - add new options for copy-before-write filter
> > > - new trace points for NBD
> > > - prefer unsigned type for some 'in_flight' fields
> > >
> > > Denis V. Lunev (2):
> > >     nbd: trace long NBD operations
> > >     block: use 'unsigned' for in_flight field on driver state
> > >
> > > Vladimir Sementsov-Ogievskiy (7):
> > >     block/copy-before-write: refactor option parsing
> > >     block/copy-before-write: add on-cbw-error open parameter
> > >     iotests: add copy-before-write: on-cbw-error tests
> > >     util: add qemu-co-timeout
> > >     block/block-copy: block_copy(): add timeout_ns parameter
> > >     block/copy-before-write: implement cbw-timeout option
> > >     iotests: copy-before-write: add cases for cbw-timeout option
> > >
> > >    block/block-copy.c                            |  33 ++-
> > >    block/copy-before-write.c                     | 110 ++++++---
> > >    block/mirror.c                                |   2 +-
> > >    block/nbd.c                                   |   8 +-
> > >    block/trace-events                            |   2 +
> > >    include/block/block-copy.h                    |   4 +-
> > >    include/qemu/coroutine.h                      |  13 ++
> > >    nbd/client-connection.c                       |   2 +
> > >    nbd/trace-events                              |   3 +
> > >    qapi/block-core.json                          |  31 ++-
> > >    tests/qemu-iotests/pylintrc                   |   5 +
> > >    tests/qemu-iotests/tests/copy-before-write    | 216 ++++++++++++++++++
> > >    .../qemu-iotests/tests/copy-before-write.out  |   5 +
> > >    util/meson.build                              |   1 +
> > >    util/qemu-co-timeout.c                        |  89 ++++++++
> > >    15 files changed, 482 insertions(+), 42 deletions(-)
> > >    create mode 100755 tests/qemu-iotests/tests/copy-before-write
> > >    create mode 100644 tests/qemu-iotests/tests/copy-before-write.out
> > >    create mode 100644 util/qemu-co-timeout.c
>
> iotests: copy-before-write: add cases for cbw-timeout option
>
> - This is causing the FreeBSD VM tests to regress for me, because the
> new iotest is failing there. Haven't diagnosed further yet, but I will
> update this thread if I get better info.

Like other problems around this test (I had a hard debugging session for the 
problem that reproduces only on gitlab pipline :/, it may relate to the fact 
that I use QEMUMachine class directly and avoid -accel qtest. Also, to fix test 
on ARM, I've added -machine none.


--
Best regards,
Vladimir