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
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 -- 2.26.2
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... thanks -- PMM
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. Max
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
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. > > > 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 > Applied, thanks. Please update the changelog at https://wiki.qemu.org/ChangeLog/5.2 for any user-visible changes. -- PMM
© 2016 - 2024 Red Hat, Inc.