[PATCH for-5.0 v2 0/4] bug fixes extracted from larger qcow2 zero bit work

Eric Blake posted 4 patches 2 weeks ago
Test docker-mingw@fedora passed
Test docker-quick@centos7 passed
Test checkpatch passed
Test FreeBSD passed
Test asan passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20200324174233.1622067-1-eblake@redhat.com
Maintainers: Max Reitz <mreitz@redhat.com>, Kevin Wolf <kwolf@redhat.com>, Liu Yuan <namei.unix@gmail.com>
docs/interop/qcow2.txt     |  3 ++-
block/qcow2.c              | 29 +++++++++++++++++++++++------
block/sheepdog.c           |  2 ++
tests/qemu-iotests/031.out |  8 ++++----
tests/qemu-iotests/036     |  6 ++++--
tests/qemu-iotests/036.out |  4 ++--
tests/qemu-iotests/061     |  6 ++++--
tests/qemu-iotests/061.out | 14 +++++++-------
8 files changed, 48 insertions(+), 24 deletions(-)

[PATCH for-5.0 v2 0/4] bug fixes extracted from larger qcow2 zero bit work

Posted by Eric Blake 2 weeks ago
My proposal [1] to add an autoclear all-zero-content bit to the qcow2
format has now stalled into 5.1 territory, but several patches in my
initial proposal are uncontroversial and obvious bug fixes worth
having in 5.0.

[1] https://lists.gnu.org/archive/html/qemu-devel/2020-01/msg08075.html

Eric Blake (4):
  qcow2: Comment typo fixes
  qcow2: List autoclear bit names in header
  qcow2: Avoid feature name extension on small cluster size
  sheepdog: Consistently set bdrv_has_zero_init_truncate

 docs/interop/qcow2.txt     |  3 ++-
 block/qcow2.c              | 29 +++++++++++++++++++++++------
 block/sheepdog.c           |  2 ++
 tests/qemu-iotests/031.out |  8 ++++----
 tests/qemu-iotests/036     |  6 ++++--
 tests/qemu-iotests/036.out |  4 ++--
 tests/qemu-iotests/061     |  6 ++++--
 tests/qemu-iotests/061.out | 14 +++++++-------
 8 files changed, 48 insertions(+), 24 deletions(-)

-- 
2.26.0.rc2


Re: [PATCH for-5.0 v2 0/4] bug fixes extracted from larger qcow2 zero bit work

Posted by Max Reitz 1 week ago
On 24.03.20 18:42, Eric Blake wrote:
> My proposal [1] to add an autoclear all-zero-content bit to the qcow2
> format has now stalled into 5.1 territory, but several patches in my
> initial proposal are uncontroversial and obvious bug fixes worth
> having in 5.0.
> 
> [1] https://lists.gnu.org/archive/html/qemu-devel/2020-01/msg08075.html
> 
> Eric Blake (4):
>   qcow2: Comment typo fixes
>   qcow2: List autoclear bit names in header
>   qcow2: Avoid feature name extension on small cluster size
>   sheepdog: Consistently set bdrv_has_zero_init_truncate

OK, so my only real gripe was with the question whether we could make
patch 3 decide dynamically when to include or not include the feature
name table.  There’s little difference in practice: Right now, we could
probably get away with including it in images with 4k clusters, then it
would automatically switch to 8k clusters minimum (which is the limit
chosen in patch 3 as-is), and at some theoretical point in the far, far
future it would switch 16k clusters minimum.

I don’t think anyone really cares about whether the feature name table
is in images with 4k clusters or not, and I think we still have a couple
of decades before we the table gets too big for images with 8k clusters
(and we probably won’t be using qcow2 then anymore).

So tl;dr: There’s no practical benefit of deciding dynamically, hence
I’m taking this series as-is:

https://git.xanclic.moe/XanClic/qemu/commits/branch/block

Thanks!

Max