mbox series

[PULL,0/2] Block patches for 5.1.0-rc4

Message ID 20200811093505.972894-1-mreitz@redhat.com
Headers show
Series Block patches for 5.1.0-rc4 | expand

Message

Max Reitz Aug. 11, 2020, 9:35 a.m. UTC 
Hi,

There is a bug in the backup job that breaks backups from images whose
size is not aligned to the job's cluster size (i.e., qemu crashes
because of a failed assertion).  If this bug makes it into the release,
it would be a regression from 5.0.

On one hand, this is probably a rare configuration that should not
happen in practice.  On the other, it is a regression, and the fix
(patch 1) is simple.  So I think it would be good to have this in 5.1.


The following changes since commit e1d322c40524d2c544d1fcd37b267d106d16d328:

  Update version for v5.1.0-rc3 release (2020-08-05 17:37:17 +0100)

are available in the Git repository at:

  https://github.com/XanClic/qemu.git tags/pull-block-2020-08-11

for you to fetch changes up to 1f3765b652930a3b485f1a67542c2410c3774abe:

  iotests: add test for unaligned granularity bitmap backup (2020-08-11 09:29:31 +0200)

----------------------------------------------------------------
Block patches for 5.1.0-rc4:
- Fix abort when running a backup job on an image whose size is not
  aligned to the backup job's cluster size

----------------------------------------------------------------
Stefan Reiter (2):
  block/block-copy: always align copied region to cluster size
  iotests: add test for unaligned granularity bitmap backup

 block/block-copy.c         |  3 ++
 tests/qemu-iotests/304     | 60 ++++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/304.out |  2 ++
 tests/qemu-iotests/group   |  1 +
 4 files changed, 66 insertions(+)
 create mode 100755 tests/qemu-iotests/304
 create mode 100644 tests/qemu-iotests/304.out

Comments

Vladimir Sementsov-Ogievskiy May 28, 2021, 12:12 p.m. UTC | #1 
11.08.2020 12:54, Max Reitz wrote:
> On 11.08.20 11:39, Peter Maydell wrote:

>> On Tue, 11 Aug 2020 at 10:35, Max Reitz <mreitz@redhat.com> wrote:

>>>

>>> Hi,

>>>

>>> There is a bug in the backup job that breaks backups from images whose

>>> size is not aligned to the job's cluster size (i.e., qemu crashes

>>> because of a failed assertion).  If this bug makes it into the release,

>>> it would be a regression from 5.0.

>>>

>>> On one hand, this is probably a rare configuration that should not

>>> happen in practice.  On the other, it is a regression, and the fix

>>> (patch 1) is simple.  So I think it would be good to have this in 5.1.

>>

>> I'm really reluctant to have to roll an rc4...

> 

> Well, that’s the information there is on this: Regression, simple fix,

> but little relevance in practice, and late to the party.

> If, given this, you don’t want to roll an rc4, then that’s how it is.

> 


Recently bug was reproduced by accidentally starting backup with source = cdrom (image was not 64k-cluster aligned).

Fedora 33, Rhel/Centos 8 are affected.

Could this go to stable branch?

-- 
Best regards,
Vladimir