Regression: bisected: AMDGPU causes Kernel Bad page state OOPS starting with kernels 5.11.x, 5.12.x, 5.13-rc

[Christian König]

Hi Luis,

adding Daniel as well.

first of all can you please create a bug report for this here: 
https://gitlab.freedesktop.org/drm/amd/-/issues This way we can better 
track issues.

Then what seems to happen is that somebody is using the TTM pages in a 
way they are not supposed to be used.

We have found a bunch of bugs in for example KVM since adding that 
commit and I have the strong suspicion that is just another one of those.

Regards,
Christian.

Am 24.05.21 um 21:25 schrieb Luís Mendes:
> Hi,
>
> AMDGPU was working fine on my armhf systems with 5.10.x and previous
> kernels and a RX550 card. Unfortunately I have only now tested kernels
> 5.11.x, 5.12.x and 5.13-rc and all are now showing problems like this
> one:
> May 10 20:23:14 picolo kernel: [   18.967626] BUG: Bad page state in
> process gnome-shell  pfn:78c08
> May 10 20:23:14 picolo kernel: [   18.973750] page:ce2e9717 refcount:2
> mapcount:1 mapping:17edced0 index:0x109e9 pfn:0x78c08
> May 10 20:23:14 picolo kernel: [   18.973763] aops:0xc0e12f54 ino:30d
>
> Full Kernel boot log is here
> https://pastebin.com/pcuUWXbj
>
> I've bisected and traced the problem to this commit:
> e93b2da9799e5cb97760969f3e1f02a5bdac29fe is the first bad commit
> commit e93b2da9799e5cb97760969f3e1f02a5bdac29fe
> Author: Christian König <christian.koenig at amd.com>
> Date:   Sat Oct 24 13:11:29 2020 +0200
>
>      drm/amdgpu: switch to new allocator v2
>
>      It should be able to handle all cases here.
>
>      v2: fix debugfs as well
>
>      Signed-off-by: Christian König <christian.koenig at amd.com>
>      Reviewed-by: Dave Airlie <airlied at redhat.com>
>      Reviewed-by: Madhav Chauhan <madhav.chauhan at amd.com>
>      Tested-by: Huang Rui <ray.huang at amd.com>
>      Link: https://patchwork.freedesktop.org/patch/397086/?series=83051&rev=1
>
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 45 ++++++++++-----------------------
>   1 file changed, 14 insertions(+), 31 deletions(-)
>
> Detailed bisect log is here:
> https://bin.privacytools.io/?a88ae63fb95fa1c1#EtrC4qxGWjmgy5C3dBzXFGqjxc7znTKULtz4cxoYFxW5
>
> Best regards,
> Luís Mendes
> Aparapi developer
> PhD Student & Researcher
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx