Hello everyone,
it seems that GP10B support has regressed recently. With linux-next, I
need to modify device/base.c to set
.mmu = gp10b_mmu_new
for GP10B (makes sense - I guess this was left as gf100_mmu_new as a
typo) to probe. After that, running a trivial testcase (running a NOP
method in 3D class) fails with
[ 110.084649] nouveau 17000000.gpu: fifo: read fault at 0000011000
engine 06 [HOST0] client 06 [GPC0/L1_2] reas|
on 02 [PTE] on channel 1 [00f206a000 nouveau_noop_te[2413]]
|
[ 110.101423] nouveau 17000000.gpu: fifo: channel 1: killed
|
[ 110.106827] nouveau 17000000.gpu: fifo: runlist 0: scheduled for
recovery |
Submitted pushbuffer.
|
[ 110.113867] nouveau 17000000.gpu: nouveau_noop_te[2413]: channel 1
killed! |
[ 125.084858] nouveau 17000000.gpu: nouveau_noop_te[2413]: failed to
idle channel 1 [nouveau_noop_te[2413]]
I haven't managed to track this down yet. However, I did find out that
checking out the drm/nouveau directory at commit "drm/nouveau/kms/nv50:
use the correct state for base channel notifier setup" makes things work
again.
I'll continue to take look, though bisecting is a bit harder than usual
due to some other issues in Tegra186 recently, so any pointers in the
right direction would be useful :)
Thanks,
Mikko
Bisection status report: The latest commit I have gotten to work is 10842ba074e9 drm/nouveau: remove unused nouveau_fence_work() i.e. the first bad commit is d7722134b825 drm/nouveau: switch over to new memory and vmm interfaces Even with the first one some patches/hacks are needed: - in mmu/gp10b.c, in the constructor we need to select the GM200 path - the GP100 path seems to not to work - as mentioned in the first mail, we need to set .mmu = gp10b_mmu_new, - and in nouveau_mem_memory_target we need to return NVKM_MEM_TARGET_NCOH instead of NVKM_MEM_TARGET_HOST. Cheers, Mikko On 11/10/2017 11:27 PM, Mikko Perttunen wrote:> Hello everyone, > > it seems that GP10B support has regressed recently. With linux-next, I > need to modify device/base.c to set > > .mmu = gp10b_mmu_new > > for GP10B (makes sense - I guess this was left as gf100_mmu_new as a > typo) to probe. After that, running a trivial testcase (running a NOP > method in 3D class) fails with > > [ 110.084649] nouveau 17000000.gpu: fifo: read fault at 0000011000 > engine 06 [HOST0] client 06 [GPC0/L1_2] reas| > on 02 [PTE] on channel 1 [00f206a000 nouveau_noop_te[2413]] > | > [ 110.101423] nouveau 17000000.gpu: fifo: channel 1: killed > | > [ 110.106827] nouveau 17000000.gpu: fifo: runlist 0: scheduled for > recovery | > Submitted pushbuffer. | > [ 110.113867] nouveau 17000000.gpu: nouveau_noop_te[2413]: channel 1 > killed! | > [ 125.084858] nouveau 17000000.gpu: nouveau_noop_te[2413]: failed to > idle channel 1 [nouveau_noop_te[2413]] > > I haven't managed to track this down yet. However, I did find out that > checking out the drm/nouveau directory at commit "drm/nouveau/kms/nv50: > use the correct state for base channel notifier setup" makes things work > again. > > I'll continue to take look, though bisecting is a bit harder than usual > due to some other issues in Tegra186 recently, so any pointers in the > right direction would be useful :) > > Thanks, > Mikko > _______________________________________________ > Nouveau mailing list > Nouveau at lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/nouveau
Thanks to Thierry for finding this - applying
index e14643615698..00eeaaffeae5 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
@@ -2369,7 +2369,7 @@ nv13b_chipset = {
.imem = gk20a_instmem_new,
.ltc = gp100_ltc_new,
.mc = gp10b_mc_new,
- .mmu = gf100_mmu_new,
+ .mmu = gp10b_mmu_new,
.secboot = gp10b_secboot_new,
.pmu = gm20b_pmu_new,
.timer = gk20a_timer_new,
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmmgp10b.c
b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmmgp10b.c
index 3dcc6bddb32f..470a4fadc165 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmmgp10b.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmmgp10b.c
@@ -33,7 +33,7 @@ gp10b_vmm = {
{ 38, &gp100_vmm_desc_16[3], NVKM_VMM_PAGE_Sxxx },
{ 29, &gp100_vmm_desc_16[2], NVKM_VMM_PAGE_Sxxx },
{ 21, &gp100_vmm_desc_16[1], NVKM_VMM_PAGE_SxHC },
- { 16, &gp100_vmm_desc_16[0], NVKM_VMM_PAGE_SxHC },
+/* { 16, &gp100_vmm_desc_16[0], NVKM_VMM_PAGE_SxHC },*/
{ 12, &gp100_vmm_desc_12[0], NVKM_VMM_PAGE_SxHx },
{}
}
on top of next-20171121 works at least for a simple test.
Mikko
On 11/11/2017 03:02 PM, Mikko Perttunen wrote:> Bisection status report:
>
> The latest commit I have gotten to work is
>
> 10842ba074e9 drm/nouveau: remove unused nouveau_fence_work()
>
> i.e. the first bad commit is
>
> d7722134b825 drm/nouveau: switch over to new memory and vmm interfaces
>
> Even with the first one some patches/hacks are needed:
>
> - in mmu/gp10b.c, in the constructor we need to select the GM200 path -
> the GP100 path seems to not to work
>
> - as mentioned in the first mail, we need to set .mmu = gp10b_mmu_new,
>
> - and in nouveau_mem_memory_target we need to return
> NVKM_MEM_TARGET_NCOH instead of NVKM_MEM_TARGET_HOST.
>
> Cheers,
> Mikko
>
> On 11/10/2017 11:27 PM, Mikko Perttunen wrote:
>> Hello everyone,
>>
>> it seems that GP10B support has regressed recently. With linux-next, I
>> need to modify device/base.c to set
>>
>> .mmu = gp10b_mmu_new
>>
>> for GP10B (makes sense - I guess this was left as gf100_mmu_new as a
>> typo) to probe. After that, running a trivial testcase (running a NOP
>> method in 3D class) fails with
>>
>> [ 110.084649] nouveau 17000000.gpu: fifo: read fault at 0000011000
>> engine 06 [HOST0] client 06 [GPC0/L1_2] reas|
>> on 02 [PTE] on channel 1 [00f206a000 nouveau_noop_te[2413]]
>> |
>> [ 110.101423] nouveau 17000000.gpu: fifo: channel 1: killed
>> |
>> [ 110.106827] nouveau 17000000.gpu: fifo: runlist 0: scheduled for
>> recovery |
>> Submitted pushbuffer. |
>> [ 110.113867] nouveau 17000000.gpu: nouveau_noop_te[2413]: channel 1
>> killed! |
>> [ 125.084858] nouveau 17000000.gpu: nouveau_noop_te[2413]: failed to
>> idle channel 1 [nouveau_noop_te[2413]]
>>
>> I haven't managed to track this down yet. However, I did find out
that
>> checking out the drm/nouveau directory at commit
>> "drm/nouveau/kms/nv50: use the correct state for base channel
notifier
>> setup" makes things work again.
>>
>> I'll continue to take look, though bisecting is a bit harder than
>> usual due to some other issues in Tegra186 recently, so any pointers
>> in the right direction would be useful :)
>>
>> Thanks,
>> Mikko
>> _______________________________________________
>> Nouveau mailing list
>> Nouveau at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/nouveau
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/nouveau