Am 14.01.20 um 05:20 schrieb Huang Rui:
On Mon, Jan 13, 2020 at 11:56:46PM +0800, Huang Rui wrote:
On Mon, Jan 13, 2020 at 06:16:21PM +0800, chen gong wrote:
Reading some registers by mmio will result in hang when GPU is in
"gfxoff" state.
This problem can be solved by GPU in "ring command
Hi
Am 13.01.20 um 19:52 schrieb Alex Deucher:
> On Fri, Jan 10, 2020 at 4:21 AM Thomas Zimmermann wrote:
>>
>> The callback struct drm_driver.get_scanout_position() is deprecated in
>> favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
>> amdgpu over.
>>
>
> I would prefer to
On Mon, Jan 13, 2020 at 11:56:46PM +0800, Huang Rui wrote:
> On Mon, Jan 13, 2020 at 06:16:21PM +0800, chen gong wrote:
> > Reading some registers by mmio will result in hang when GPU is in
> > "gfxoff" state.
> >
> > This problem can be solved by GPU in "ring command packages" way.
> >
> >
[Why]
PM4 packet size for flush message was oversized.
[How]
Packet size adjusted to allocate flush + fence packets.
Change-Id: I9a577d2118398b3139011829de12789b2a577a19
Signed-off-by: Alex Sierra
---
drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 2 +-
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 2 +-
[AMD Official Use Only - Internal Distribution Only]
Yes, both gfx_v10_0_ring_emit_fence and gfx_v9_0_ring_emit_fence have 8 ring
writes. Plus 2 of the flush.
Regards,
Alejandro S.
-Original Message-
From: Kuehling, Felix
Sent: Monday, January 13, 2020 6:55 PM
To: Sierra Guiza,
[AMD Official Use Only - Internal Distribution Only]
Reviewed-by: Hawking Zhang
Regards,
Hawking
-Original Message-
From: amd-gfx On Behalf Of shaoyunl
Sent: Tuesday, January 14, 2020 00:20
To: amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/amdgpu: check rlc_g firmware pointer
[AMD Public Use]
Thanks for your time and hope you get well soon!
-Original Message-
From: Lyude Paul
Sent: Tuesday, January 14, 2020 1:59 AM
To: Lin, Wayne ; dri-de...@lists.freedesktop.org;
amd-gfx@lists.freedesktop.org
Cc: Kazlauskas, Nicholas ; Wentland, Harry
; Zuo, Jerry
Yes, 'ssize_t' is fine. But 'count' comes with type of 'size_t'.
Although it should not bring any true problem. I'm not sure whether there will
be compile warnings.
Regards,
Evan
> -Original Message-
> From: Alex Deucher
> Sent: Tuesday, January 14, 2020 4:27 AM
> To: Quan, Evan
> Cc:
I noticed that the invalidate_tlbs_size in patch 3 was also wrong. That
should only be 2 dwords, not 12. The code here should do
amdgpu_ring_alloc(ring, kiq->pmf->invalidate_tlbs_size + 8);
I think 12 was too much in the original code. Flush + fence should only
be 10 dwords, unless I
[AMD Official Use Only - Internal Distribution Only]
I just pushed the series, but I'll go ahead and create a new patch for this.
The .invalidate_tlbs_size, is it based on dword size? Currently is 12, should I
need to drop it to 8 then?
-Original Message-
From: Kuehling, Felix
Sent:
On 2020-01-13 3:26 p.m., Alex Sierra wrote:
tlbs invalidate pointer function added to kiq_pm4_funcs struct.
This way, tlb flush can be done through kiq member.
TLBs invalidatation implemented for gfx9 and gfx10.
Change-Id: I1b77b364f3ae0038ff3e70e869be5f2ef6e6d293
Signed-off-by: Alex Sierra
Sorry, I already said, Reviewed-by, but realized there was one more
problem. If you haven't submitted yet, please fix that first. Otherwise,
please make it a follow-up patch. See inline ...
On 2020-01-13 3:26 p.m., Alex Sierra wrote:
This can be used directly from amdgpu and amdkfd to
The series is
Reviewed-by: Felix Kuehling
On 2020-01-13 3:26 p.m., Alex Sierra wrote:
tlbs invalidate pointer function added to kiq_pm4_funcs struct.
This way, tlb flush can be done through kiq member.
TLBs invalidatation implemented for gfx9 and gfx10.
Change-Id:
On Sun, Jan 12, 2020 at 10:35 PM Quan, Evan wrote:
>
> The "count" from amdgpu_set_pp_feature_status() seems with type of size_t.
> Then assignment "count = -EINVAL" may be improper.
It's fine. ssize_t is signed. We've been doing it long before this patch.
Alex
> With that confirmed, the
tlbs invalidate pointer function added to kiq_pm4_funcs struct.
This way, tlb flush can be done through kiq member.
TLBs invalidatation implemented for gfx9 and gfx10.
Change-Id: I1b77b364f3ae0038ff3e70e869be5f2ef6e6d293
Signed-off-by: Alex Sierra
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h |
This can be used directly from amdgpu and amdkfd to invalidate
TLB through pasid.
It supports gmc v7, v8, v9 and v10.
Change-Id: I6563a8eba2e42d1a67fa2547156c20da41d1e490
Signed-off-by: Alex Sierra
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h | 6 ++
drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c | 74
[Why]
There are HW-indpendent functions that enables and disables kcq. These
functions use
the kiq_pm4_funcs implementation.
[How]
Local kcq enable and disable functions removed and replace it by the generic kcq
enable under amdgpu_gfx
Change-Id: I7709bdba93742c234941a5936c82eb67e346077c
On Fri, Jan 10, 2020 at 4:22 AM Thomas Zimmermann wrote:
>
> VBLANK callbacks in struct drm_driver are deprecated in favor of
> their equivalents in struct drm_crtc_funcs. Convert radeon over.
>
> Signed-off-by: Thomas Zimmermann
Reviewed-by: Alex Deucher
> ---
>
On Fri, Jan 10, 2020 at 4:22 AM Thomas Zimmermann wrote:
>
> VBLANK callbacks in struct drm_driver are deprecated in favor of
> their equivalents in struct drm_crtc_funcs. Convert amdgpu over.
I think I'd prefer to just update the signatures of the relevant
functions rather than wrapping them.
I replied with one question about patch 4. Other than that the series is
Reviewed-by: Felix Kuehling
On 2020-01-13 9:40 a.m., Christian König wrote:
Entirely unused.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 1 -
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
> -Original Message-
> From: amd-gfx On Behalf Of Alex
> Deucher
> Sent: 2020/January/13, Monday 1:42 PM
> To: amd-gfx list
> Cc: Deucher, Alexander
> Subject: Re: [PATCH] drm/amdgpu/df3.6: remove unused variable
>
> Ping?
>
> Alex
>
> On Fri, Jan 10, 2020 at 7:22 PM Alex Deucher
On Fri, Jan 10, 2020 at 4:22 AM Thomas Zimmermann wrote:
>
> The callback struct drm_driver.get_scanout_position() is deprecated in
> favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
> radeon over.
>
I'd prefer to just change the signature of
radeon_get_crtc_scanoutpos() to
On Fri, Jan 10, 2020 at 4:21 AM Thomas Zimmermann wrote:
>
> The callback struct drm_driver.get_scanout_position() is deprecated in
> favor of struct drm_crtc_helper_funcs.get_scanout_position(). Convert
> amdgpu over.
>
I would prefer to just change the signature of
Ping?
Alex
On Fri, Jan 10, 2020 at 7:22 PM Alex Deucher wrote:
>
> Unused so drop it.
>
> Signed-off-by: Alex Deucher
> ---
> drivers/gpu/drm/amd/amdgpu/df_v3_6.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c
>
Hey! Haven't taken a look at this patch yet but just wanted to let you know
I'm going to try to get to most of the stuff you've got pending for me. I came
down with a really nasty cold last week so sorry if you've had any other
patches waiting until now!
On Mon, 2020-01-13 at 17:36 +0800, Wayne
On 2020-01-13 9:40 a.m., Christian König wrote:
For unlocked page table updates we need to be able
to sync to fences of a specific VM.
Signed-off-by: Christian König
---
.../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 6 +--
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c| 8 +--
On 01/11, Christian König wrote:
> Am 10.01.20 um 22:30 schrieb Harry Wentland:
> > On 2020-01-10 4:16 p.m., Harry Wentland wrote:
> > > On 2020-01-10 1:47 p.m., Liu, Zhan wrote:
> > > >
> > > > > -Original Message-
> > > > > From: amd-gfx On Behalf Of
> > > > > Christian König
> > > > >
On Mon, Jan 13, 2020 at 08:20:42AM -0500, mikita.lip...@amd.com wrote:
> From: Mikita Lipski
>
> [why]
> Fix compilation warnings on i386 architecture:
> undefined reference to `__udivdi3'
> [how]
> Switch DIV_ROUND_UP to DIV64_U64_ROUND_UP
>
> Reported-by: Randy Dunlap
> Signed-off-by: Mikita
On 2020-01-11 1:39 p.m., Alex Sierra wrote:
Functions implemented from kiq_pm4_funcs struct members
for gfx_v9 version.
Change-Id: I8fd3e160c4bd58f19d35d29e39517db967063afe
Signed-off-by: Alex Sierra
Reviewed-by: Felix Kuehling
---
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 115
On 2020-01-11 1:39 p.m., Alex Sierra wrote:
[Why]
TLB flush method has been deprecated using kfd2kgd interface.
This implementation is now on the amdgpu_amdkfd API.
[How]
TLB flush functions now implemented in amdgpu_amdkfd.
Change-Id: Ic51cccdfe6e71288d78da772b6e1b6ced72f8ef7
Signed-off-by:
Can you include kcq_disable in the patch as well?
Thanks,
Felix
On 2020-01-11 1:39 p.m., Alex Sierra wrote:
[Why]
There's a HW-indpendent function that enables kcq. This function uses
the kiq_pm4_funcs implementation.
[How]
Local kcq enable function removed and replace it by the generic kcq
On 2020-01-11 1:39 p.m., Alex Sierra wrote:
This can be used directly from amdgpu and amdkfd to invalidate
TLB through pasid.
It supports gmc v7, v8, v9 and v10.
Change-Id: I6563a8eba2e42d1a67fa2547156c20da41d1e490
Signed-off-by: Alex Sierra
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h | 6
On 2020-01-13 8:20 a.m., mikita.lip...@amd.com wrote:
> From: Mikita Lipski
>
> [why]
> Fix compilation warnings on i386 architecture:
> undefined reference to `__udivdi3'
> [how]
> Switch DIV_ROUND_UP to DIV64_U64_ROUND_UP
>
> Reported-by: Randy Dunlap
> Signed-off-by: Mikita Lipski
On 2020-01-11 1:39 p.m., Alex Sierra wrote:
tlbs invalidate pointer function added to kiq_pm4_funcs struct.
This way, tlb flush can be done through kiq member.
TLBs invalidatation implemented for gfx9 and gfx10.
Change-Id: I1b77b364f3ae0038ff3e70e869be5f2ef6e6d293
Signed-off-by: Alex Sierra
On 2020-01-08 11:51 p.m., Alex Deucher wrote:
If a user specifies a non-existant sdma instance,
warn and return the offset for instance 0.
Signed-off-by: Alex Deucher
Reviewed-by: Felix Kuehling
---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c | 7 +--
1 file changed, 5
ping.
On 2020-01-10 1:33 p.m., shaoyunl wrote:
In SRIOV, rlc_g firmware is loaded by host, guest driver won't load it which
will
cause the rlc_fw pointer is null
Change-Id: Id16f65171dd427d623af4c5bc75f674019e63dec
Signed-off-by: shaoyunl
---
drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 9
On Mon, Jan 13, 2020 at 06:16:22PM +0800, chen gong wrote:
> Reading CP_MEM_SLP_CNTL register with RREG32_SOC15 macro will lead to
> hang when GPU is in "gfxoff" state.
>
> Signed-off-by: chen gong
> ---
> drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 2 +-
> 1 file changed, 1 insertion(+), 1
On Mon, Jan 13, 2020 at 06:16:21PM +0800, chen gong wrote:
> Reading some registers by mmio will result in hang when GPU is in
> "gfxoff" state.
>
> This problem can be solved by GPU in "ring command packages" way.
>
> Signed-off-by: chen gong
Acked-by: Huang Rui
> ---
>
For HMM we need to rework how VM synchronization works, so instead of the filp
use VM as job owner.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
Explicitly sync VM updates to the moving fence in PDs and PTs.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_cpu.c | 7 +++
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c | 5 +
2 files changed, 12 insertions(+)
diff --git
If provided we only sync to the BOs reservation
object and no longer to the root PD.
v2: update comment, cleanup amdgpu_bo_sync_wait_resv
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 37 +++-
drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 3
The root PD can also contain huge PDEs.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 25 +
1 file changed, 5 insertions(+), 20 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index
Entirely unused.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 1 -
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 1 -
drivers/gpu/drm/amd/amdgpu/amdgpu_job.h | 1 -
3 files changed, 3 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
For unlocked page table updates we need to be able
to sync to fences of a specific VM.
Signed-off-by: Christian König
---
.../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 6 +--
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c| 8 +--
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c| 3 +-
Make sure that we never allocate a page table for an invalidation operation.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 22 +-
1 file changed, 13 insertions(+), 9 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
We need to update page tables without any lock held.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
From: Mikita Lipski
[why]
Fix compilation warnings on i386 architecture:
undefined reference to `__udivdi3'
[how]
Switch DIV_ROUND_UP to DIV64_U64_ROUND_UP
Reported-by: Randy Dunlap
Signed-off-by: Mikita Lipski
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 2 +-
1 file
[AMD Official Use Only - Internal Distribution Only]
FWIW:
https://patchwork.freedesktop.org/patch/347923/
Alex
From: amd-gfx on behalf of Yong Zhao
Sent: Friday, January 10, 2020 7:40 PM
To: amd-gfx@lists.freedesktop.org
Cc: Zhao, Yong
Subject: [PATCH]
Am 11.01.20 um 19:39 schrieb Alex Sierra:
[Why]
kfd2kgd interface will be deprecated. This removal only covers TLB
invalidation for now. They have been replaced in amdgpu_amdkfd API.
[How]
TLB invalidate functions removed from the different amdkfd_gfx_v*
versions.
Change-Id:
Am 11.01.20 um 19:39 schrieb Alex Sierra:
[Why]
TLB flush method has been deprecated using kfd2kgd interface.
This implementation is now on the amdgpu_amdkfd API.
[How]
TLB flush functions now implemented in amdgpu_amdkfd.
Change-Id: Ic51cccdfe6e71288d78da772b6e1b6ced72f8ef7
Signed-off-by:
Am 11.01.20 um 19:39 schrieb Alex Sierra:
This can be used directly from amdgpu and amdkfd to invalidate
TLB through pasid.
It supports gmc v7, v8, v9 and v10.
Change-Id: I6563a8eba2e42d1a67fa2547156c20da41d1e490
Signed-off-by: Alex Sierra
Flushing by pasid is in principle racy, but I don't
Am 11.01.20 um 19:39 schrieb Alex Sierra:
[Why]
There's a HW-indpendent function that enables kcq. This function uses
the kiq_pm4_funcs implementation.
[How]
Local kcq enable function removed and replace it by the generic kcq
enable under amdgpu_gfx
Change-Id:
Am 11.01.20 um 19:39 schrieb Alex Sierra:
tlbs invalidate pointer function added to kiq_pm4_funcs struct.
This way, tlb flush can be done through kiq member.
TLBs invalidatation implemented for gfx9 and gfx10.
Change-Id: I1b77b364f3ae0038ff3e70e869be5f2ef6e6d293
Signed-off-by: Alex Sierra
Am 11.01.20 um 19:39 schrieb Alex Sierra:
Functions implemented from kiq_pm4_funcs struct members
for gfx_v9 version.
Change-Id: I8fd3e160c4bd58f19d35d29e39517db967063afe
Signed-off-by: Alex Sierra
Acked-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 115
Am 11.01.20 um 19:39 schrieb Alex Sierra:
[Why]
Avoid reclaim filesystem while eviction lock is held called from
MMU notifier.
[How]
Setting PF_MEMALLOC_NOFS flags while eviction mutex is locked.
Using memalloc_nofs_save / memalloc_nofs_restore API.
Change-Id:
Am 12.01.20 um 02:25 schrieb Nirmoy Das:
This patch adds required fields to drm_sched_job and drm_gpu_scheduler
structure to cumulatively calculate amount of time a drm_gpu_scheduler
spend on serving a job.
Using least used drm scheduler to choose a run queue
improves
Reading some registers by mmio will result in hang when GPU is in
"gfxoff" state.
This problem can be solved by GPU in "ring command packages" way.
Signed-off-by: chen gong
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h| 3 +++
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 +-
2 files
Reading CP_MEM_SLP_CNTL register with RREG32_SOC15 macro will lead to
hang when GPU is in "gfxoff" state.
Signed-off-by: chen gong
---
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
[Why]
Noticed this while testing MST with the 4 ports MST hub from
StarTech.com. Sometimes can't light up monitors normally and get the
error message as 'sideband msg build failed'.
Look into aux transactions, found out that source sometimes will send
out another down request before receiving the
On Sat, Jan 11, 2020 at 07:08:42AM +0800, Kuehling, Felix wrote:
> On 2020-01-10 1:37 a.m., Huang Rui wrote:
> > To align with gfx v9, we use the map_queues packet to load hiq MQD.
> >
> > Signed-off-by: Huang Rui
>
> Please make sure you test this on a GFXv10 GPU.
>
> Reviewed-by: Felix
60 matches
Mail list logo
