Fix broken Raven system before Linux 4.18 release
Christian König
christian.koenig at amd.com
Mon Jul 30 13:30:58 UTC 2018
Hi Paul,
well that is rather odd, cause the patch you pointed out is just a code
cleanup without functional change.
Can you double check your bisect result again?
Thanks,
Christian.
Am 30.07.2018 um 14:35 schrieb Paul Menzel:
> Dear Linux folks,
>
>
> On 07/04/18 12:43, Christian König wrote:
>> Am 04.07.2018 um 12:16 schrieb Huang Rui:
>>> BO ptr already be initialized at definition, we needn't use the complicated
>>> reference.
>>>
>>> Signed-off-by: Huang Rui <ray.huang at amd.com>
>> There is a typo in the subject line, with that fixed the patch is
>> Reviewed-by: Christian König <christian.koenig at amd.com>.
>>> ---
>>> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 7 +++----
>>> 1 file changed, 3 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>> index 712af5c..b53562b 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>> @@ -1645,18 +1645,17 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
>>> uint64_t flags;
>>> int r;
>>> - if (clear || !bo_va->base.bo) {
>>> + if (clear || !bo) {
>>> mem = NULL;
>>> nodes = NULL;
>>> exclusive = NULL;
>>> } else {
>>> struct ttm_dma_tt *ttm;
>>> - mem = &bo_va->base.bo->tbo.mem;
>>> + mem = &bo->tbo.mem;
>>> nodes = mem->mm_node;
>>> if (mem->mem_type == TTM_PL_TT) {
>>> - ttm = container_of(bo_va->base.bo->tbo.ttm,
>>> - struct ttm_dma_tt, ttm);
>>> + ttm = container_of(bo->tbo.ttm, struct ttm_dma_tt, ttm);
>>> pages_addr = ttm->dma_address;
>>> }
>>> exclusive = reservation_object_get_excl(bo->tbo.resv);
> I didn’t see the v2 message in my INBOX, but this commit 7c4e581a
> (drm/amdgpu: simplify the bo reference on amdgpu_bo_update) should be applied
> to current Linux 4.18-rc7 series, best before the Linux 4.18 release next
> Sunday. Otherwise the screen is not updated when switching from EFI frame
> buffer to amdgpufb, and GDM doesn’t start on the MSI B350M MORTAR (MS-7A37).
> The problem is present in 4.18-rc1, and I bisected the branch
> *amd-staging-drm-next* (with the meaning of good and bad reversed).
>
> ```
> $ git bisect log
> git bisect start
> # bad: [27f0996a5a6a5052a60a76541f434cc7b2220489] drm/amdgpu: clean up the superfluous space and align the comment text for amdgpu_ttm
> git bisect bad 27f0996a5a6a5052a60a76541f434cc7b2220489
> # good: [ce397d215ccd07b8ae3f71db689aedb85d56ab40] Linux 4.18-rc1
> git bisect good ce397d215ccd07b8ae3f71db689aedb85d56ab40
> # good: [96157892c82a6274b07ad884aeeb9d7550540878] drm/amd/display: move dcn1 dispclk programming to dccg
> git bisect good 96157892c82a6274b07ad884aeeb9d7550540878
> # good: [617b0eb160a87ef4a61d6770f360c5e27e01c5ee] drm/amd/pp: fix semicolon.cocci warnings
> git bisect good 617b0eb160a87ef4a61d6770f360c5e27e01c5ee
> # bad: [68cf3a51430c7e6dd2bf7642aac6db1042f7955d] amdgpu: Do not evict VRAM
> git bisect bad 68cf3a51430c7e6dd2bf7642aac6db1042f7955d
> # good: [de5d27d291047e1791a237f8a1c90215bb4db9ce] drm/amd/display: properly turn autocal off
> git bisect good de5d27d291047e1791a237f8a1c90215bb4db9ce
> # good: [c20182735c2fb85df0e039192ae3b46d9b6a2c68] drm/amdgpu: remove internal/unused kernel module parameters
> git bisect good c20182735c2fb85df0e039192ae3b46d9b6a2c68
> # bad: [ca5401ce7eb8ad6f3696a1afeed6dfd4a20d994f] drm/amdgpu: use drm_fb helper for console_(un)lock
> git bisect bad ca5401ce7eb8ad6f3696a1afeed6dfd4a20d994f
> # bad: [829e05edce80cf43a982eb4c1ce547a921e5afd2] drm/amdgpu: clean up UVD instance handling v2
> git bisect bad 829e05edce80cf43a982eb4c1ce547a921e5afd2
> # bad: [97b89da992d5efd4239904365162081a6e2394f2] drm/amd/amdgpu: creating two I2S instances for stoney/cz (v2)
> git bisect bad 97b89da992d5efd4239904365162081a6e2394f2
> # bad: [3f9da96fa9d2aaaca452d7396dbd84eaa9ac6766] drm/amdgpu/pm: Remove VLA usage
> git bisect bad 3f9da96fa9d2aaaca452d7396dbd84eaa9ac6766
> ```
>
>
> Kind regards,
>
> Paul
More information about the amd-gfx
mailing list