- 05 May, 2016 40 commits
-
-
Alex Deucher authored
They can be shared with other asics with minor modifications. Acked-by:
Tom St Denis <tom.stdenis@amd.com> Signed-off-by:
Alex Deucher <alexander.deucher@amd.com>
-
Alex Deucher authored
Different asics tend to have different ways to interact with the RLC. This just covers enter/exit of safe mode for updating CG and PG state, but could be extended to cover other RLC operations in the future if necessary. Acked-by:
-
Alex Deucher authored
Acked-by:
-
Alex Deucher authored
So it can be shared with CZ/ST. Acked-by:
-
Alex Deucher authored
We were already enabling these CG features, this uses the standard interface for doing so. Acked-by:
-
Alex Deucher authored
We were already enabling these CG features, this uses the standard interface for doing so. Acked-by:
-
Alex Deucher authored
We were already enabling these CG features, this uses the standard interface for doing so. Acked-by:
-
Alex Deucher authored
We were already enabling these CG features, this uses the standard interface for doing so. Acked-by:
-
Alex Deucher authored
Acked-by:
-
Christian König authored
This allows fine grained control for the driver where to add a BO into the LRU. v2: fix typo in comment Reviewed-by:
Sinclair Yeh <syeh@vmware.com> Signed-off-by:
Christian König <christian.koenig@amd.com> Reviewed-by:
-
Christian König authored
Useful for driver specific LRU handling. v2: fix typo in comment Reviewed-by:
-
Christian König authored
Not used any more. Reviewed-by:
-
Christian König authored
Not used any more. Reviewed-by:
-
Christian König authored
Not used any more. Reviewed-by:
-
Christian König authored
When we use an extern reservation object that otherwise waits for every fence registered with it. Reviewed-by:
-
Christian König authored
No functional change. Signed-off-by:
-
Christian König authored
Instead of hard coding just another name in the ring code. v2: squash in Tom's rebase fix Signed-off-by:
-
Christian König authored
Those are way too large. Signed-off-by:
-
Christian König authored
Those are way too large. Signed-off-by:
-
Christian König authored
Instead of specifying the total ring size calculate that from the maximum number of dw a submission can have and the number of concurrent submissions. This fixes UVD with 8 concurrent submissions or more. Signed-off-by:
-
Eric Huang authored
Signed-off-by:
Eric Huang <JinHuiEric.Huang@amd.com> Reviewed-by:
-
Eric Huang authored
Signed-off-by:
-
Felix Kuehling authored
TTM BO accounting is out of sync with how memory is really allocated in ttm[_dma]_tt_alloc_page_directory. This resulted in excessive estimated overhead with many small allocations. ttm_dma_tt_alloc_page_directory makes a single allocation for three arrays: pages, DMA and CPU addresses. It uses drm_calloc_large, which uses kmalloc internally for allocations smaller than PAGE_SIZE. ttm_round_pot should be a good approximation of its memory usage both above and below PAGE_SIZE. Reviewed-by:
Thomas Hellstrom <thellstrom@vmware.com> Signed-off-by:
Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by:
Monk Liu <monk.liu@amd.com> Reviewed-by:
-
Dave Airlie authored
This is just a type-safety things to avoid everyone taking void *, it doesn't change anything. v2: agd5f: split out the dal changes into a separate patch. Signed-off-by:
Dave Airlie <airlied@redhat.com> Signed-off-by:
-
Arindam Nath authored
Change History -------------- v2: - Make firmware version check correctly. Firmware versions >= 1.80 should all support 40 UVD instances. - Replace AMDGPU_MAX_UVD_HANDLES with max_handles variable. v1: - The firmware can handle upto 40 UVD sessions. Signed-off-by:
Arindam Nath <arindam.nath@amd.com> Signed-off-by:
Ayyappa Chandolu <ayyappa.chandolu@amd.com> Reviewed-by:
-
Nils Wallménius authored
These tables were initialized on stack on each call, avoid that and save a little bit of text size. Reviewed-by:
Nils Wallménius <nils.wallmenius@gmail.com> Signed-off-by:
-
Nils Wallménius authored
Also adjust phm_construct_table to take a const pointer Reviewed-by:
-
Nils Wallménius authored
As these arrays were of pointer to pointer type, they were pointer to pointer to const. Make them pointer to const pointer to const. Reviewed-by:
-
Nils Wallménius authored
All these are compile time constand and the drm_debugfs_create/remove_files functions take a const pointer argument. Reviewed-by:
-
Nils Wallménius authored
This marks the struct amdgpu_sched_ops const and adjusts amd_sched_init to take a const pointer for the ops param. The ops member of struct amd_gpu_scheduler is also changed to const. Reviewed-by:
-
Nils Wallménius authored
This patch marks some compile-time constant tables 'const'. The tables marked in this patch are the low hanging fruit where little other changes were necesary to avoid casting away constness etc. Also mark some tables that are private to a file as static. Reviewed-by:
-
Alex Deucher authored
Stale ifdef. Reviewed-by:
Harry Wentland <harry.wentland@amd.com> Signed-off-by:
-
Dave Airlie authored
This adds support to the command parser for the set append counter packet3, this is required to support atomic counters on evergreen/cayman GPUs. v2: fixup some of the hardcoded numbers with real register names (Christian) Reviewed-by:
-
Arindam Nath authored
Signed-off-by:
Leo Liu <leo.liu@amd.com> Signed-off-by:
-
Arindam Nath authored
Signed-off-by:
-
Christian König authored
If we don't need to flush we can easily use another VMID already assigned to the process. Signed-off-by:
Chunming Zhou <david1.zhou@amd.com> Signed-off-by:
-
Christian König authored
This way we can track when the flush is done. Signed-off-by:
-
Christian König authored
v2: rebase & cleanup This way we can store more than one fence as user for each VMID. Signed-off-by:
-
Christian König authored
No need to have two of them any more. Signed-off-by:
-
Andrey Grodzovsky authored
Also add some pflip debug prints. This change allows us to wait on pflip status until the new surface address is actually submitted to the register. This reverts ed3020e923240829dcdfd3343f6e91dc02c63775 drm/amdgpu: Move MMIO flip out of spinlocked region The original change assumed DAL will aquire locks inside DAL implemetion of page_flip callback which eventaully didn't happen. This moves the flip before status update which makes sense for the non-DAL code pathes as well. Signed-off-by:
Andrey Grodzovsky <Andrey.Grodzovsky@amd.com> Reviewed-by:
-
