Hi all,
I'm writing as a bystander working with 6.1.y stable branch and possibly lacking some context with the established DRM -> stable patch flow, Cc'ing a large number of people.
The commit being reverted from 6.1.y is the one that duplicates the changes already backported to that branch with another commit. It is essentially a "similar" commit but cherry-picked at some point during the DRM development process.
The duplicate has no runtime effect but should not actually remain in the stable trees. It was already reverted [1] from 6.6/6.10/6.11 but still made its way later to 6.1.
[1]: https://lore.kernel.org/stable/20241007035711.46624-1-jsg@jsg.id.au/T/#u
At [1] Greg KH also stated that the observed problems are quite common while backporting DRM patches to stable trees. The current duplicate patch has in every sense a cosmetic impact but in other circumstances and for other patches this may have gone wrong.
So, is there any way to adjust this process?
BTW, a question to the stable-team: what Git magic (3-way-merge?) let the duplicate patch be applied successfully? The patch context in stable trees was different to that moment so should the duplicate have been expected to fail to be applied?
-- Fedor
This reverts commit 7c887efda1201110211fed8921a92a713e0b6bcd.
It is a duplicate of the change made in 6.1.105 by commit 282f0a482ee6 ("drm/amd/display: Skip Recompute DSC Params if no Stream on Link").
This is a consequence of two "similar" upstream commits existence, one of which has been cherry-picked.
Reported-by: Alexey Khoroshilov khoroshilov@ispras.ru Signed-off-by: Fedor Pchelkin pchelkin@ispras.ru --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 3 --- 1 file changed, 3 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 1acef5f3838f..855cd71f636f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -1255,9 +1255,6 @@ static bool is_dsc_need_re_compute( } }
- if (new_stream_on_link_num == 0) - return false; - if (new_stream_on_link_num == 0) return false;
On Tue, Oct 29, 2024 at 04:31:40PM +0300, Fedor Pchelkin wrote:
BTW, a question to the stable-team: what Git magic (3-way-merge?) let the duplicate patch be applied successfully? The patch context in stable trees was different to that moment so should the duplicate have been expected to fail to be applied?
Just plain git... Try it yourself :)
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 7c887efda1 Auto-merging drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c [detached HEAD 2a1c937960abd] drm/amd/display: Skip Recompute DSC Params if no Stream on Link Author: Fangzhi Zuo Jerry.Zuo@amd.com Date: Fri Jul 12 16:30:03 2024 -0400 1 file changed, 3 insertions(+)
$ git log -2 --oneline 2a1c937960abd (HEAD) drm/amd/display: Skip Recompute DSC Params if no Stream on Link 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
On Tue, 29. Oct 10:20, Sasha Levin wrote:
On Tue, Oct 29, 2024 at 04:31:40PM +0300, Fedor Pchelkin wrote:
BTW, a question to the stable-team: what Git magic (3-way-merge?) let the duplicate patch be applied successfully? The patch context in stable trees was different to that moment so should the duplicate have been expected to fail to be applied?
Just plain git... Try it yourself :)
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 7c887efda1
7c887efda1 is the commit backported to linux-6.1.y. Of course it will apply there.
What I mean is that the upstream commit for 7c887efda1 is 8151a6c13111b465dbabe07c19f572f7cbd16fef.
And cherry-picking 8151a6c13111b465dbabe07c19f572f7cbd16fef to linux-6.1.y on top of 282f0a482ee6 will not result in duplicating the change, at least with my git configuration.
I just don't understand how a duplicating if-statement could be produced in result of those cherry-pick'ings and how the content of 7c887efda1 was generated.
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee6 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 8151a6c13111b465dbabe07c19f572f7cbd16fef Auto-merging drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c HEAD detached at 282f0a482ee6 You are currently cherry-picking commit 8151a6c13111. (all conflicts fixed: run "git cherry-pick --continue") (use "git cherry-pick --skip" to skip this patch) (use "git cherry-pick --abort" to cancel the cherry-pick operation) The previous cherry-pick is now empty, possibly due to conflict resolution. If you wish to commit it anyway, use:
git commit --allow-empty
Otherwise, please use 'git cherry-pick --skip'
Auto-merging drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c [detached HEAD 2a1c937960abd] drm/amd/display: Skip Recompute DSC Params if no Stream on Link Author: Fangzhi Zuo Jerry.Zuo@amd.com Date: Fri Jul 12 16:30:03 2024 -0400 1 file changed, 3 insertions(+)
$ git log -2 --oneline 2a1c937960abd (HEAD) drm/amd/display: Skip Recompute DSC Params if no Stream on Link 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
-- Thanks, Sasha
On Tue, 29. Oct 18:12, Fedor Pchelkin wrote:
On Tue, 29. Oct 10:20, Sasha Levin wrote:
On Tue, Oct 29, 2024 at 04:31:40PM +0300, Fedor Pchelkin wrote:
BTW, a question to the stable-team: what Git magic (3-way-merge?) let the duplicate patch be applied successfully? The patch context in stable trees was different to that moment so should the duplicate have been expected to fail to be applied?
Just plain git... Try it yourself :)
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 7c887efda1
7c887efda1 is the commit backported to linux-6.1.y. Of course it will apply there.
What I mean is that the upstream commit for 7c887efda1 is 8151a6c13111b465dbabe07c19f572f7cbd16fef.
And cherry-picking 8151a6c13111b465dbabe07c19f572f7cbd16fef to linux-6.1.y on top of 282f0a482ee6 will not result in duplicating the change, at least with my git configuration.
I just don't understand how a duplicating if-statement could be produced in result of those cherry-pick'ings and how the content of 7c887efda1 was generated.
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee6 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 8151a6c13111b465dbabe07c19f572f7cbd16fef Auto-merging drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c HEAD detached at 282f0a482ee6 You are currently cherry-picking commit 8151a6c13111. (all conflicts fixed: run "git cherry-pick --continue") (use "git cherry-pick --skip" to skip this patch) (use "git cherry-pick --abort" to cancel the cherry-pick operation) The previous cherry-pick is now empty, possibly due to conflict resolution. If you wish to commit it anyway, use:
git commit --allow-emptyOtherwise, please use 'git cherry-pick --skip'
Sasha,
my concern is that maybe there is some issue with the scripts used for the preparation of backport patches.
There are two different upstream commits performing the exact same change: - 50e376f1fe3bf571d0645ddf48ad37eb58323919 - 8151a6c13111b465dbabe07c19f572f7cbd16fef
50e376f1fe3bf571d0645ddf48ad37eb58323919 was backported to stable kernels at first. After that, attempts to backport 8151a6c13111b465dbabe07c19f572f7cbd16fef to those stables should be expected to fail, no? Git would have complained about this - the patch was already applied.
It is just strange that the (exact same) change made by the commits is duplicated by backporting tools. As it is not the first case where DRM patches are involved per Greg's statement [1], I wonder if something can be done on stable-team's side to avoid such odd behavior in future.
[1]: https://lore.kernel.org/stable/20241007035711.46624-1-jsg@jsg.id.au/T/#u
-- Thanks, Fedor
On Mon, Nov 4, 2024 at 9:55 AM Fedor Pchelkin pchelkin@ispras.ru wrote:
On Tue, 29. Oct 18:12, Fedor Pchelkin wrote:
On Tue, 29. Oct 10:20, Sasha Levin wrote:
On Tue, Oct 29, 2024 at 04:31:40PM +0300, Fedor Pchelkin wrote:
BTW, a question to the stable-team: what Git magic (3-way-merge?) let the duplicate patch be applied successfully? The patch context in stable trees was different to that moment so should the duplicate have been expected to fail to be applied?
Just plain git... Try it yourself :)
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee61 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 7c887efda1
7c887efda1 is the commit backported to linux-6.1.y. Of course it will apply there.
What I mean is that the upstream commit for 7c887efda1 is 8151a6c13111b465dbabe07c19f572f7cbd16fef.
And cherry-picking 8151a6c13111b465dbabe07c19f572f7cbd16fef to linux-6.1.y on top of 282f0a482ee6 will not result in duplicating the change, at least with my git configuration.
I just don't understand how a duplicating if-statement could be produced in result of those cherry-pick'ings and how the content of 7c887efda1 was generated.
$ git checkout 282f0a482ee6 HEAD is now at 282f0a482ee6 drm/amd/display: Skip Recompute DSC Params if no Stream on Link
$ git cherry-pick 8151a6c13111b465dbabe07c19f572f7cbd16fef Auto-merging drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c HEAD detached at 282f0a482ee6 You are currently cherry-picking commit 8151a6c13111. (all conflicts fixed: run "git cherry-pick --continue") (use "git cherry-pick --skip" to skip this patch) (use "git cherry-pick --abort" to cancel the cherry-pick operation) The previous cherry-pick is now empty, possibly due to conflict resolution. If you wish to commit it anyway, use:
git commit --allow-emptyOtherwise, please use 'git cherry-pick --skip'
Sasha,
my concern is that maybe there is some issue with the scripts used for the preparation of backport patches.
There are two different upstream commits performing the exact same change:
- 50e376f1fe3bf571d0645ddf48ad37eb58323919
- 8151a6c13111b465dbabe07c19f572f7cbd16fef
There were cases where I needed to cherry-pick fixes from -next to -fixes. In the past I had not used -x when cherry-picking because I got warnings about references to commits that were not yet in mainline. I have since started using -x when cherry-picking things from -next to -fixes.
Alex
50e376f1fe3bf571d0645ddf48ad37eb58323919 was backported to stable kernels at first. After that, attempts to backport 8151a6c13111b465dbabe07c19f572f7cbd16fef to those stables should be expected to fail, no? Git would have complained about this - the patch was already applied.
It is just strange that the (exact same) change made by the commits is duplicated by backporting tools. As it is not the first case where DRM patches are involved per Greg's statement [1], I wonder if something can be done on stable-team's side to avoid such odd behavior in future.
-- Thanks, Fedor
linux-stable-mirror@lists.linaro.org
-
Alex Deucher -
Fedor Pchelkin -
Sasha Levin