What about
"Enable recursive locking at swapout time to make it possible to swap
out BOs that share the same reservation object."
Is "per VM BOs" an AMD specific name? In that case, I'd avoid using it
in the TTM code since most people have no idea what they are and why the
need specific tre
Roger,
5 out of 7 patches in this series are completely lacking a commit log
message, and thats not OK. Really.
https://www.kernel.org/doc/html/v4.12/process/submitting-patches.html#describe-your-changes
I'll review these, but IIRC the no_wait in the memory accounting code is
different in th
-Original Message-
From: Christian König [mailto:ckoenig.leichtzumer...@gmail.com]
Sent: Wednesday, December 20, 2017 9:36 PM
To: He, Roger ; amd-gfx@lists.freedesktop.org;
dri-de...@lists.freedesktop.org
Subject: Re: [PATCH 3/7] drm/ttm: use an operation ctx for
ttm_mem_global_alloc_p
I agree using fence_context is same efficiency with original client id,
since vm only has one sdma entity.
But PASID seems be more proper with per-VM-per-process concept, the
comment also said that.
Regards,
David Zhou
On 2017年12月20日 21:21, Christian König wrote:
Use the fence context from t
Reviewed-by: Chunming Zhou
-Original Message-
From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of
Christian K?nig
Sent: Wednesday, December 20, 2017 9:21 PM
To: amd-gfx@lists.freedesktop.org
Subject: [PATCH 4/5] drm/amdgpu: rename vm_id to vmid
sed -i "s/vm_id/vmid
Looks very good, Reviewed-by: Chunming Zhou
-Original Message-
From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of
Christian K?nig
Sent: Wednesday, December 20, 2017 9:21 PM
To: amd-gfx@lists.freedesktop.org
Subject: [PATCH 3/5] drm/amdgpu: separate VMID and PASID h
Hi Dave,
More stuff for 4.16:
- Further radeon cleanup after dropping kfd support
- Fix S3 on raven
- Further GPU reset clean ups and fixes
- Move the amd gpu scheduler into common code so it can be shared
with other drivers, notably etnaviv
- Various ttm fixes
- Support for 2+1 level GPU page t
Am 20.12.2017 um 20:43 schrieb Daniel Vetter:
On Wed, Dec 20, 2017 at 6:20 PM, Li, Samuel wrote:
Ping... can someone please review this patch?
Might be simpler to implement your own dma-buf backend instead of
going through the drm_prime midlayer. That was mostly written to give
nvidia a set of
On Wed, Dec 20, 2017 at 6:20 PM, Li, Samuel wrote:
> Ping... can someone please review this patch?
Might be simpler to implement your own dma-buf backend instead of
going through the drm_prime midlayer. That was mostly written to give
nvidia a set of non-EXPORT_GPL symbols to support dma-buf. Or
Ping... can someone please review this patch?
Samuel Li
> -Original Message-
> From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf
> Of Samuel Li
> Sent: Friday, December 15, 2017 11:28 AM
> To: dri-de...@lists.freedesktop.org; amd-gfx@lists.freedesktop.org
> Cc: Koen
On 2017-12-20 03:49 PM, Christian König wrote:
> Am 20.12.2017 um 15:41 schrieb Michel Dänzer:
>> On 2017-12-20 02:21 PM, Christian König wrote:
>>> TTM tries to allocate coherent memory in chunks of 2MB first to improve
>>> TLB efficiency and falls back to allocating 4K pages if that fails.
>>>
>>
Am 20.12.2017 um 15:41 schrieb Michel Dänzer:
On 2017-12-20 02:21 PM, Christian König wrote:
TTM tries to allocate coherent memory in chunks of 2MB first to improve
TLB efficiency and falls back to allocating 4K pages if that fails.
Suppress the warning when the 2MB allocations fails since ther
On 2017-12-20 02:21 PM, Christian König wrote:
> TTM tries to allocate coherent memory in chunks of 2MB first to improve
> TLB efficiency and falls back to allocating 4K pages if that fails.
>
> Suppress the warning when the 2MB allocations fails since there is a
> valid fall back path.
>
> Signe
On 2017-12-20 02:21 PM, Christian König wrote:
> They don't work 100% correctly at the moment.
>
> Signed-off-by: Christian König
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 6 +-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ob
Am 20.12.2017 um 11:34 schrieb Roger He:
Change-Id: I803ea52d11e5c06add0dffab836c3aecc00b56dd
Signed-off-by: Roger He
Commit message! And please double check the coding style of
ast_ttm_tt_populate.
With that fixed that patch is Reviewed-by: Christian König
.
Regards,
Christian.
---
Impact: make symbol static.
Fix this sparse warning:
drivers/gpu/drm/radeon/radeon_dp_mst.c:332:38: warning: symbol 'mst_cbs' was
not declared. Should it be static?
Signed-off-by: Dmitry Rozhkov
---
drivers/gpu/drm/radeon/radeon_dp_mst.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Commit message!
Am 20.12.2017 um 11:34 schrieb Roger He:
Change-Id: I4104a12e09a374b6477a0dd5a8fce26dce27a746
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_memory.c | 15 ---
drivers/gpu/drm/ttm/ttm_page_alloc.c | 6 +-
drivers/gpu/drm/ttm/ttm_page_alloc_d
Commit message needed! Something like:
Forward the operation context to ttm_mem_global_alloc as well.
Am 20.12.2017 um 11:34 schrieb Roger He:
Change-Id: I5279b5cd3560c4082b00f822219575a5f9c3808a
Signed-off-by: Roger He
With the commit message fixed, patch is Reviewed-by: Christian König
.
Am 20.12.2017 um 11:34 schrieb Roger He:
then remove superfluous functions
We need a better commit message. Something like:
Remove the extra indirection, cause we have only one implementation anyway.
Change-Id: Iea020f0e30a239e0265e7a1500168c7d7f819bd9
Signed-off-by: Roger He
With the co
They don't work 100% correctly at the moment.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 6 +-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 4fee
Use the fence context from the scheduler entity.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c | 8
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 2 --
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 4
3 files changed, 4 insertions(+), 10 deletions(-)
diff --gi
sed -i "s/vm_id/vmid/g" drivers/gpu/drm/amd/amdgpu/*.c
sed -i "s/vm_id/vmid/g" drivers/gpu/drm/amd/amdgpu/*.h
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 6 +++---
drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c| 8 +++
drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
Move both into the new files amdgpu_ids.[ch]. No functional change.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/Makefile | 3 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c | 4 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c | 4 +-
drivers/gpu/d
TTM tries to allocate coherent memory in chunks of 2MB first to improve
TLB efficiency and falls back to allocating 4K pages if that fails.
Suppress the warning when the 2MB allocations fails since there is a
valid fall back path.
Signed-off-by: Christian König
---
lib/swiotlb.c | 8 +---
1
extract this function since eviction and swapout share same logic
Change-Id: I80a475a93fceed8d66d74a1832c815a0756341ac
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_bo.c | 29 +++--
1 file changed, 19 insertions(+), 10 deletions(-)
diff --git a/drivers/gpu/drm/ttm/
Change-Id: I1e87954564f38ad298bf6e4ff88c9f26f291a62d
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_bo.c | 15 +++
drivers/gpu/drm/ttm/ttm_memory.c | 12
include/drm/ttm/ttm_bo_api.h | 3 ++-
3 files changed, 21 insertions(+), 9 deletions(-)
diff --git a/dr
Change-Id: I42a7df8c50e1ce3b527ee9cb78809f8e58136f07
Signed-off-by: Roger He
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 2 +-
drivers/gpu/drm/nouveau/nouveau_bo.c| 2 +-
drivers/gpu/drm/radeon/radeon_ttm.c | 2 +-
drivers/gpu/drm/ttm/ttm_bo.c| 2 +-
drivers/gpu/drm/ttm/ttm_
Change-Id: I803ea52d11e5c06add0dffab836c3aecc00b56dd
Signed-off-by: Roger He
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 7 ---
drivers/gpu/drm/ast/ast_ttm.c| 5 +++--
drivers/gpu/drm/cirrus/cirrus_ttm.c | 5 +++--
drivers/gpu/drm/nouveau/nouveau_bo.c | 8
Change-Id: I4104a12e09a374b6477a0dd5a8fce26dce27a746
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_memory.c | 15 ---
drivers/gpu/drm/ttm/ttm_page_alloc.c | 6 +-
drivers/gpu/drm/ttm/ttm_page_alloc_dma.c | 8 ++--
include/drm/ttm/ttm_memory.h |
then remove superfluous functions
Change-Id: Iea020f0e30a239e0265e7a1500168c7d7f819bd9
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_bo.c | 21 +++-
drivers/gpu/drm/ttm/ttm_memory.c | 12 ++-
include/drm/ttm/ttm_bo_api.h | 1 +
include/drm/ttm/ttm_bo_driver.h | 1 -
Change-Id: I5279b5cd3560c4082b00f822219575a5f9c3808a
Signed-off-by: Roger He
---
drivers/gpu/drm/ttm/ttm_bo.c| 2 +-
drivers/gpu/drm/ttm/ttm_memory.c| 15 +--
drivers/gpu/drm/ttm/ttm_object.c| 13 ++---
drivers/gpu/drm/vmwgf
Am 20.12.2017 um 03:44 schrieb Yu, Xiangliang:
This patch releases the mutex held soon before entering the initialization
function in case the device doesn't support mxgpu.
Signed-off-by: José Pekkarinen
---
drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c | 1 +
1 file changed, 1 insertion(+)
diff --
32 matches
Mail list logo
