aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
AgeCommit message (Collapse)AuthorFilesLines
2016-07-14drm/amdgpu: trace need_flush in grab_vm as wellChristian König1-2/+2
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-14drm/amdgpu: check flush fence context instead of same ring v2Christian König1-3/+6
Otherwise we can run into the following situation: 1. Process A grabs ID 1 for ring 0. 2. Process B grabs ID 1 for ring 0. 3. Process A grabs ID 1 for ring 1. 4. Process A tries to reuse ID1 for ring 0 but things he doesn't need to flush. v2: check the context of the flush fence instead of messing with the owner field. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: recovery hw jobs when gpu reset V3Chunming Zhou1-1/+2
V3: directly use pd_addr. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: abstract amdgpu_vm_is_gpu_resetChunming Zhou1-1/+8
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: add a bool to specify if needing vm flush V2Chunming Zhou1-36/+31
which avoids job->vm_pd_addr be changed. V2: pass job structure to amdgpu_vm_grab_id and amdgpu_vm_flush directly. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: must update page table after gpu resetChunming Zhou1-0/+1
Record the gpu reset count in vmid to identify if gpu reset happened. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: must update page table after gpu resetChunming Zhou1-0/+3
Record the gpu reset count in vmid to identify if gpu reset happened. Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: validate VM PTs only on evictionChristian König1-1/+12
We don't need to validate them again if the eviction counter didn't changed. Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: disable compute pipeline sync workaround when using fixed fwAlex Deucher1-1/+27
No need to stall the pipe when we are using firmware with the fix. Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: save the PD addr before scheduling the jobChristian König1-4/+2
When we pipeline evictions the page directory could already be moving somewhere else when grab_id is called. Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: sync to buffer moves before VM updatesChristian König1-4/+19
Otherwise we could update the VM page tables while the move is only scheduled. Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: Initialize the variables in a straight-forward wayAlex Xie1-6/+28
Initialize the variable in a straight-forward way instead of hiding the initialization inside the loop. This can also reduce one function call. Signed-off-by: Alex Xie <AlexBin.Xie@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: Add comment to describe the purpose of one difficult if statementAlex Xie1-4/+7
Use == instead of != in the if statement to make code easier understood Signed-off-by: Alex Xie <AlexBin.Xie@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: Change some variable names to make code easier understoodAlex Xie1-16/+16
Add comment to describe some variables otherwise. Signed-off-by: Alex Xie <AlexBin.Xie@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: remove now unnecessary checksChristian König1-20/+11
vm_flush() now comes directly after vm_grab_id(). Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: use a fence array for VMID managementChristian König1-58/+97
Just wait for any fence to become available, instead of waiting for the last entry of the LRU. Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: reuse VMIDs assigned to a VM only if there is also a free oneChristian König1-54/+59
This fixes a fairness problem with the GPU scheduler. VM having lot of jobs could previously starve VM with less jobs. Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: prefer VMIDs idle on the current ringChristian König1-15/+9
Prefer to use a VMIDs which are idle on the ring we want to submit to. This also removes bubbling idle VMIDs up on the LRU, which is actually not beneficial. Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-07-07drm/amdgpu: add optional ring to amdgpu_sync_is_idleChristian König1-2/+2
Check if the sync object is idle depending on the ring a submission works with. Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: Encapsulate some VM table update parameters (v2)Harish Kasiviswanathan1-49/+62
Bundle some VM table parameters into amdgpu_vm_update_params structure, so that number of function parameters can be reduced. Only structural change, no logic change. v2: agd: squash in fix from Harish Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: remove define for reserved client IDChristian König1-1/+1
Just set it to zero instead. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: remove owner cleanup v2Christian König1-13/+0
The client ID is now unique, so no need to resert the owner fields any more. v2: remove unused variables as well Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> (v1) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: make the VMID owner always 64bitChristian König1-2/+2
Otherwise we could (in theory) run into problems on 32bit systems. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: add pipeline sync for compute jobChunming Zhou1-1/+2
hardware ring is async processed, the job is executed in parallel. In some case, this will result vm fault, like jobs with different vmids. This works around a CPC hw issue which will eventually be fixed in fw. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: keep vm in job instead of ib (v2)Monk Liu1-1/+2
ib.vm is a legacy way to get vm, after scheduler implemented vm should be get from job, and all ibs from one job share the same vm, no need to keep ib.vm just move vm field to job. this patch as well add job as paramter to ib_schedule so it can get vm from job->vm. v2: agd: sqaush in: drm/amdgpu: check if ring emit_vm_flush exists in vm flush No vm flush on engines that don't support VM. bug: https://bugs.freedesktop.org/show_bug.cgi?id=95195 Signed-off-by: Monk Liu <Monk.Liu@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: make vmid owner be client_idChunming Zhou1-3/+3
Using the pointer is not adequate. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: add client id for every vmChunming Zhou1-0/+2
This adds a unique id for each vm client so we can properly track them. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-11drm/amdgpu: fix wrong release of vmid ownerChunming Zhou1-5/+9
The release of the vmid owner was not handled correctly. We need to take the lock and walk the lru list. Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: fix error checking when reuse vmid on same ringChunming Zhou1-2/+2
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: only update last_flush when vmid doesn't have other new ownerChunming Zhou1-5/+10
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: reuse VMIDs already assigned to a processChristian König1-31/+46
If we don't need to flush we can easily use another VMID already assigned to the process. Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: add a fence after the VM flushChristian König1-5/+21
This way we can track when the flush is done. Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: use a sync object for VMID fences v2Christian König1-25/+28
v2: rebase & cleanup This way we can store more than one fence as user for each VMID. Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> (v1) Reviewed-by: Chunming Zhou <david1.zhou@amd.com> (v1) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-04drm/amdgpu: merge VM manager and VM context ID structureChristian König1-68/+71
No need to have two of them any more. Signed-off-by: Christian König <christian.koenig@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-02drm/amdgpu: use BO pages instead of GART arrayChristian König1-13/+16
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-02drm/amdgpu: change parameter passing in the VM codeChristian König1-46/+54
Make it more flexible by passing src and page addresses directly instead of the structures they contain. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-05-02drm/amdgpu: improve vmid assigment V2Chunming Zhou1-0/+14
V2: the signaled items on the LRU maintain their order Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-09drm/amdgpu: Revert "add mutex for ba_va->valids/invalids"Christian König1-11/+6
Not needed any more because we need to protect the elements on the list anyway. This reverts commit 38bf516c75b4ef0f5c716e05fa9baab7c52d6c39. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-09drm/amdgpu: Revert "add lock for interval tree in vm"Christian König1-14/+2
Not needed any more because we need to protect the elements on the list anyway. This reverts commit fe237ed7efec8ac147a4572fdf81173a7f8ddda7. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-09drm/amdgpu: Revert "add spin lock to protect freed list in vm (v3)"Christian König1-13/+3
Not needed any more because we need to protect the elements on the list anyway. This reverts commit dae6ecf9e6c9b677e577826c3ac665c6dd9c490b. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-09drm/amdgpu: Fix two bugs in amdgpu_vm_bo_split_mappingFelix Kuehling1-2/+2
Off-by-one: last is inclusive, so the maximum is start + max_size - 1 Wrong unit: addr is in bytes, max_size is in pages Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-08drm/amdgpu: move get_user_pages out of amdgpu_ttm_tt_pin_userptr v6Christian König1-0/+2
That avoids lock inversion between the BO reservation lock and the anon_vma lock. v2: * Changed amdgpu_bo_list_entry.user_pages to an array of pointers * Lock mmap_sem only for get_user_pages * Added invalidation of unbound userpointer BOs * Fixed memory leak and page reference leak v3 (chk): * Revert locking mmap_sem only for_get user_pages * Revert adding invalidation of unbound userpointer BOs * Sanitize and fix error handling v4 (chk): * Init userpages pointer everywhere. * Fix error handling when get_user_pages() fails. * Add invalidation of unbound userpointer BOs again. v5 (chk): * Add maximum number of tries. v6 (chk): * Fix error handling when we run out of tries. Signed-off-by: Christian König <christian.koenig@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> (v4) Acked-by: Alex Deucher <alexander.deucher@amd.com>
2016-03-08drm/amdgpu: if a GDS switch is needed emit a pipeline sync as wellChristian König1-10/+12
Otherwise we might change the GDS settings while they are still in use. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
2016-03-08drm/amdgpu: split pipeline sync and vm flushChristian König1-0/+2
This allows us to use the pipeline sync for other tasks as well. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
2016-03-08drm/amdgpu: switch the GDS only on demand v2Christian König1-2/+41
Switching the GDS space to often seems to be problematic. This patch together with the following can avoid VM faults on context switch. v2: extend commit message a bit Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1) Reviewed-by: Chunming Zhou <david1.zhou@amd.com> (v1)
2016-03-08drm/amdgpu: move the GDS switch into vm flush as wellChristian König1-5/+13
After all it's an operation on the VMID. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
2016-03-08drm/amdgpu: sync to the active user on reusing a VMIDChristian König1-0/+7
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-02-29drm/amdgpu: trace the pd_addr in vm_grab_id as wellChristian König1-2/+3
Makes matching it to the flushes much easier. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-02-29drm/amdgpu: fix VM faults caused by vm_grab_id() v4Christian König1-55/+60
The owner must be per ring as long as we don't support sharing VMIDs per process. Also move the assigned VMID and page directory address into the IB structure. v3: assign the VMID to all IBs, not just the first one. v4: use correct pointer for owner Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Chunming Zhou <david1.zhou@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2016-02-12drm/amdgpu: use SDMA round robin for VM updates v3Christian König1-6/+18
Distribute the load on both rings. v2: use a loop for the initialization v3: agd: rebase on upstream Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Powered by cgit, Gruvboxed by Blaster4385.