Display corruption with Kernel 5.10 and 5.12

[Roland]

Thanks for testing. I don't see any important changes in the changelog of the alpha4.

I got now the display corruption with alpha3, after 2 days of testing. Will now continue with alpha2...

[Roland]

I have been using alpha2 during several days but have not seen the screen corruption... Is there in alpha3 any change which could potentially be the source of this issue?

[xeno74]

I have been using alpha2 during several days but have not seen the screen corruption... Is there in alpha3 any change which could potentially be the source of this issue?

I think the issue is somewhere in the PowerPC updates 5.10-1.

Hi All,

Here is the third alpha of kernel 5.10 with the first PowerPC updates 5.10-1 for testing.

Download: linux-image-5.10-alpha3-X1000_X5000.tar.gz



Please test the kernels.

Thanks,
Christian

Please look in the PowerPC updates 5.10-1.

We can do a git bisect but this is really a lot of work for us. We have to compile at least 12 kernels and you have to test them because I don't have these issues. And your test per kernel takes a long time.

[Roland]

I think the issue is somewhere in the PowerPC updates 5.10-1.

We can do a git bisect but this is really a lot of work for us. We have to compile at least 12 kernels and you have to test them because I don't have these issues. And your test per kernel takes a long time.

Is there any other choise to find the origin of this issue...? Are these test kernels in other respect functional? Alpha2 has been so far totally stable, no other issues. For me there is not much extra work if I can just use them until the problem appears, or does not.

[xeno74]

Is there any other choise to find the origin of this issue...?

Yes, please look in the PowerPC updates 5.10-1. Maybe you find a suspicious commit. If you have found a suspicious commit, then I can try to remove the changes.

Are these test kernels in other respect functional?

Yes, I tested them intensiveley and other tested these intensively too. (See test thread)

[Roland]

Yes, please look in the PowerPC updates 5.10-1. Maybe you find a suspicious commit. If you have found a suspicious commit, then I can try to remove the changes.

I have not found anything obvious but any change which has something to do with memory is suspicious to me:

- Addition of a filter for RTAS (firmware) calls done via sys_rtas(),
to prevent root from overwriting kernel memory.

But I'm afraid we need to do it through the long path...

Alpha2 has still worked without problems (I have run several livestreams etc. with it), so it is quite sure the issue appeared in alpha3.

[xeno74]

But I'm afraid we need to do it through the long path...

OK, let's do it ...

1. Code: Select all

   git bisect start

2. Code: Select all

   git bisect good bbf5c979011a099af5dc76498918ed7df445635b

   (Good: Linux 5.9)

3. Code: Select all

   git bisect bad 3650b228f83adda7e5ee532e2b90429c03f7b9ec

   (Bad: Linux 5.10-rc1)
   Output:

   Code: Select all

   [c48b75b7271db23c1b2d1204d6e8496d91f27711] Merge tag 'sound-5.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

Download: uImage-rb1

Please test it.

[Roland]

Download: uImage-rb1

I tested rb1 today with intensive and various streaming/browser sessions... No sign of problems. I think we could already test the next set...

[xeno74]

I tested rb1 today with intensive and various streaming/browser sessions... No sign of problems. I think we could already test the next set...

OK, git bisect good

Output:

Code: Select all

[847d4287a0c6709fd1ce24002b96d404a6da8b5b] Merge tag 's390-5.10-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Download: uImage-rb2

Please test it.

[Roland]

Download: uImage-rb2

Testing it now... But can you tell me how this bisect method exactly goes? If the rb1 was ok, should the issue now be seen with rb2?