Re: [Qemu-block] [PULL 0/7] Block patches for 4.1.0-rc4

[Max Reitz]

On 05.08.19 20:05, Peter Maydell wrote:
> On Mon, 5 Aug 2019 at 17:37, Max Reitz <address@hidden> wrote:
>>
>> The following changes since commit 9bb68d34dda9be60335e73e65c8fb61bca035362:
>>
>>   Merge remote-tracking branch 
>> 'remotes/philmd-gitlab/tags/edk2-next-20190803' into staging (2019-08-05 
>> 11:05:36 +0100)
>>
>> are available in the Git repository at:
>>
>>   https://github.com/XanClic/qemu.git tags/pull-block-2019-08-05
>>
>> for you to fetch changes up to 07b0851c592efe188a87259adbda26a63c61dc92:
>>
>>   block/backup: disable copy_range for compressed backup (2019-08-05 
>> 18:05:05 +0200)
>>
>> ----------------------------------------------------------------
>> Block patches for 4.1.0-rc4:
>> - Fix the backup block job when using copy offloading
>> - Fix the mirror block job when using the write-blocking copy mode
>> - Fix incremental backups after the image has been grown with the
>>   respective bitmap attached to it
>>
>> ----------------------------------------------------------------
>> Max Reitz (5):
>>   backup: Copy only dirty areas
>>   iotests: Test backup job with two guest writes
>>   iotests: Test incremental backup after truncation
>>   mirror: Only mirror granularity-aligned chunks
>>   iotests: Test unaligned blocking mirror write
>>
>> Vladimir Sementsov-Ogievskiy (2):
>>   util/hbitmap: update orig_size on truncate
>>   block/backup: disable copy_range for compressed backup
>>
>>  block/backup.c             | 15 ++++++++++++---
>>  block/mirror.c             | 29 ++++++++++++++++++++++++++++
>>  util/hbitmap.c             |  6 +++++-
>>  tests/qemu-iotests/056     | 39 ++++++++++++++++++++++++++++++++++++++
>>  tests/qemu-iotests/056.out |  4 ++--
>>  tests/qemu-iotests/124     | 38 +++++++++++++++++++++++++++++++++----
>>  tests/qemu-iotests/124.out |  4 ++--
>>  tests/qemu-iotests/151     | 25 ++++++++++++++++++++++++
>>  tests/qemu-iotests/151.out |  4 ++--
>>  9 files changed, 150 insertions(+), 14 deletions(-)
> 
> This is quite a lot of changes for rc4 -- how confident are
> you about them ? I suppose 3 out of 4 commits are updating
> the test suite...

Would dropping the test patches make it better? :-)

I am reasonably (i.e., rc4-levels of) confident that the patches don’t
break anything that wasn’t broken before.

(I’m least confident about the test patches working for everyone and
everywhere, as with all new test cases.  But it was my impression that
it’s always fine to include test case additions.)

Patch 1 is very important.  I’m very confident about it.
It fixes a silent corruption in the backup job, so I’m not too surprised
people haven’t noticed.  I would be surprised if really noone was
affected so far.

Patch 3 is not that important, but it is sufficiently simple, so I think
we should take it, even into rc4.

Patch 5 is very important for a specific mirror copying mode.  It can be
argued that nobody really uses this mode because otherwise somebody
should have noticed the corruption, because if you hit it, you will
simply lose data (as opposed to the backup case, where you will simply
get the wrong version of the data in the output image).
But that is why it’s so important.  I really don’t want anyone to hit it.
It is probably the most complicated patch here, but at any other point,
it would still be considered a simple patch.  (Just not quite trivial.)
I think it is worth taking it.

Patch 7 is actually not important.  But it’s an obvious trivial
one-liner.  I thought I might as well.

Max

signature.asc
Description: OpenPGP digital signature