aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
AgeCommit message (Collapse)AuthorFilesLines
2019-02-13drm/amdgpu: Add helper to wait for BO fences using a sync objectFelix Kuehling1-0/+24
Creates a temporary sync object to wait for the BO reservation. This generalizes amdgpu_vm_wait_pd. Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2019-02-01drm/amdgpu: clean up memory/GDS/GWS/OA alignment codeMarek Olšák1-4/+12
- move all adjustments into one place - specify GDS/GWS/OA alignment in basic units of the heaps - it looks like GDS alignment was 1 instead of 4 Signed-off-by: Marek Olšák <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-12-14drm/amdgpu: WARN once if amdgpu_bo_unpin is called for an unpinned BOMichel Dänzer1-1/+1
It indicates a pin/unpin imbalance bug somewhere. While the bug isn't necessarily in the call chain hitting this, it's at least one part involved. Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-12-07drm/amdgpu: Add KFD VRAM limit checkingFelix Kuehling1-1/+1
We don't want KFD processes evicting each other over VRAM usage. Therefore prevent overcommitting VRAM among KFD applications with a per-GPU limit. Also leave enough room for page tables on top of the application memory usage. Acked-by: Alex Deucher <[email protected]> Signed-off-by: Felix Kuehling <[email protected]> Reviewed-by: Eric Huang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-12-03drm/amdgpu: remove amdgpu_bo_backup_to_shadowChristian König1-47/+0
It is unused. Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-10-25dma-buf: allow reserving more than one shared fence slotChristian König1-1/+1
Let's support simultaneous submissions to multiple engines. Signed-off-by: Christian König <[email protected]> Reviewed-by: Michel Dänzer <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Reviewed-by: Huang Rui <[email protected]> Link: https://patchwork.kernel.org/patch/10626149/
2018-09-19drm/amdgpu: fix up GDS/GWS/OA shiftingChristian König1-1/+5
That only worked by pure coincident. Completely remove the shifting and always apply correct PAGE_SHIFT. Signed-off-by: Christian König <[email protected]> Acked-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amdgpu: fix shadow BO restoringChristian König1-33/+13
Don't grab the reservation lock any more and simplify the handling quite a bit. Signed-off-by: Christian König <[email protected]> Reviewed-by: Huang Rui <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amdgpu: shadow BOs don't need any alignmentChristian König1-3/+2
They aren't directly used by the hardware. Signed-off-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amdgpu: always enable shadow BOs v2Christian König1-13/+1
Even when GPU recovery is disabled we could run into a manually triggered recovery. v2: keep accidental removed comments Signed-off-by: Christian König <[email protected]> Acked-by: Emily Deng <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amdgpu: stop pipelining VM PDs/PTs movesChristian König1-1/+1
We are going to need this for recoverable page fault handling and it makes shadow handling during GPU reset much more easier. Signed-off-by: Christian König <[email protected]> Acked-by: Junwei Zhang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amd/amdgpu: Avoid fault when allocating an empty buffer objectTom St Denis1-1/+2
Signed-off-by: Tom St Denis <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-19drm/amdgpu: don't allocate zero sized kernel BOsChristian König1-0/+5
Just free the BO if the size should be zero. Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-13drm/amdgpu: use a single linked list for amdgpu_vm_bo_baseChristian König1-1/+1
Instead of the double linked list. Gets the size of amdgpu_vm_pt down to 64 bytes again. We could even reduce it down to 32 bytes, but that would require some rather extreme hacks. Signed-off-by: Christian König <[email protected]> Reviewed-by: Chunming Zhou <[email protected]> Acked-by: Felix Kuehling <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-10drm/amdgpu: manually map the shadow BOs againChristian König1-4/+1
Otherwise we won't be able to use the AGP aperture. Signed-off-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-09-10drm/amdgpu: correctly sign extend 48bit addresses v3Christian König1-1/+1
Correct sign extend the GMC addresses to 48bit. v2: sign extending turned out easier than thought. v3: clean up the defines and move them into amdgpu_gmc.h as well Signed-off-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-08-29drm/amdgpu: distinct between allocated GART space and GMC addrChristian König1-2/+0
Most of the time we only need to know if the BO has a valid GMC addr. Signed-off-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Acked-by: Felix Kuehling <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-08-27drm/amdgpu: Do not evict VRAM on APUs with disabled HIBERNATEPaul Menzel1-1/+3
Improve commit d796d844 (drm/radeon/kms: make hibernate work on IGPs) to only migrate VRAM objects if the Linux kernel is actually built with support for hibernation (suspend to disk). The better solution is to get the information, if this is suspend or hibernate, from `amdgpu_device_suspend()`, but that’s more involved, so apply the simple solution first. Link: https://bugs.freedesktop.org/show_bug.cgi?id=107277 Signed-off-by: Paul Menzel <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-31drm/amdgpu: Replace ttm_bo_unref with ttm_bo_putThomas Zimmermann1-3/+2
The function ttm_bo_put releases a reference to a TTM buffer object. The function's name is more aligned to the Linux kernel convention of naming ref-counting function _get and _put. A call to ttm_bo_unref takes the address of the TTM BO object's pointer and clears the pointer's value to NULL. This is not necessary in most cases and sometimes even worked around by the calling code. A call to ttm_bo_put only releases the reference without clearing the pointer. The current behaviour of cleaning the pointer is kept in the calling code, but should be removed if not required in a later patch. v2: * set prefix to drm/amdgpu Signed-off-by: Thomas Zimmermann <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-31drm/amdgpu: Replace ttm_bo_reference with ttm_bo_getThomas Zimmermann1-1/+1
The function ttm_bo_get acquires a reference on a TTM buffer object. The function's name is more aligned to the Linux kernel convention of naming ref-counting function _get and _put. v2: * changed prefix to drm/amdgpu Signed-off-by: Thomas Zimmermann <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-25drm/amdgpu: reduce the number of placements for a BOChristian König1-0/+2
Make struct amdgpu_bo a bit smaller. Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-25drm/amdgpu: consistenly name amdgpu_bo_ functionsChristian König1-17/+17
Just rename functions, no functional change. Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-24drm/amdgpu: Don't warn on destroying a pinned BOMichel Dänzer1-1/+1
The warning turned out to be not so useful, as BO destruction tends to be deferred to a workqueue. Also, we should be preventing any damage from this now, so not really important anymore to fix code doing this. Acked-by: Alex Deucher <[email protected]> Tested-by: Mike Lothian <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-13drm/amdgpu: Warn and update pin_size values when destroying a pinned BOMichel Dänzer1-7/+25
This shouldn't happen, but if it does, we'll get a backtrace of the caller, and update the pin_size values as needed. v2: * Check bo->pin_count instead of placement flags (Christian König) Reviewed-by: Christian König <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-13drm/amdgpu: Make pin_size values atomicMichel Dänzer1-6/+8
Concurrent execution of the non-atomic arithmetic could result in completely bogus values. v2: * Rebased on v2 of the previous patch Cc: [email protected] Bugzilla: https://bugs.freedesktop.org/106872 Reviewed-by: Christian König <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-13drm/amdgpu: Keep track of amount of pinned CPU visible VRAMMichel Dänzer1-2/+2
Instead of CPU invisible VRAM. Preparation for the following, no functional change intended. v2: * Also change amdgpu_vram_mgr_bo_invisible_size to amdgpu_vram_mgr_bo_visible_size, allowing further simplification (Christian König) Cc: [email protected] Reviewed-by: Christian König <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-10drm/amdgpu: fix kmap error handling for bo creationsJunwei Zhang1-1/+1
kmap happens after bo pin, so unpin is required on error Signed-off-by: Junwei Zhang <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-10drm/amdgpu: allocate gart memory when it's required (v3)Junwei Zhang1-6/+9
Instead of calling gart address space on every bo pin, allocates it on demand v2: fix error handling v3: drop the change on amdgpu_amdkfd_gpuvm.c, not needed. Signed-off-by: Junwei Zhang <[email protected]> Acked-by: Felix Kuehling <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-10drm/amdgpu: separate gpu address from bo pinJunwei Zhang1-11/+6
It could be got by amdgpu_bo_gpu_offset() if need Signed-off-by: Junwei Zhang <[email protected]> Reviewed-by: Michel Dänzer <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-07-04Merge v4.18-rc3 into drm-nextDave Airlie1-14/+10
Two requests have come in for a backmerge, and I've got some pull reqs on rc2, so this just makes sense. Signed-off-by: Dave Airlie <[email protected]>
2018-06-19drm/amdgpu: Refactor amdgpu_vram_mgr_bo_invisible_size helperMichel Dänzer1-4/+2
Preparation for the following fix, no functional change intended. Cc: [email protected] Signed-off-by: Michel Dänzer <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-06-19drm/amdgpu: Update pin_size values before unpinning BOMichel Dänzer1-10/+8
At least in theory, ttm_bo_validate may move the BO, in which case the pin_size accounting would be inconsistent with when the BO was pinned. Cc: [email protected] Signed-off-by: Michel Dänzer <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-06-15drm/amdgpu: Consolidate visible vs. real vram check v2.Andrey Grodzovsky1-1/+1
Move all instnaces of this check into a function in amdgpu_gmc.h Rename the original function to a more proper name. v2: Add more places to cleanup. Reviewed-by: Christian König <[email protected]> Signed-off-by: Andrey Grodzovsky <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-06-15drm/amdgpu: Fix-ups for amdgpu_object.c documentationMichel Dänzer1-23/+55
* Fix format of return value descriptions * Document all parameters of amdgpu_bo_free_kernel * Document amdgpu_bo_get_preferred_pin_domain Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Michel Dänzer <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-06-15drm/amdgpu: add kernel doc for amdgpu_object.cSamuel Li1-0/+269
Document the amdgpu buffer object API. v2: Add a DOC section and some more clarification. v3: Add some clarification and fix a spelling. Suggested-by: Alex Deucher <[email protected]> Signed-off-by: Samuel Li <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-29drm/amdgpu: Add helper function to get buffer domainDeepak Sharma1-5/+12
Move logic of getting supported domain to a helper function Signed-off-by: Deepak Sharma <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: set ttm bo priority before initializationJunwei Zhang1-3/+2
Signed-off-by: Junwei Zhang <[email protected]> Reviewed-by: David Zhou <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: Enable scatter gather display supportSamuel Li1-0/+9
Enables sg display if vram size <= THRESHOLD(256M); otherwise still use vram as display buffer. This patch fixed some potention issues introduced by change "allow framebuffer in GART memory as well" due to CZ/ST hardware limitation. v2: Change default setting to auto. v3: Move some logic from amdgpu_display_framebuffer_domains() to pin function, suggested by Christian. v4: Split into several patches. v5: Drop module parameter for now. Signed-off-by: Samuel Li <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: Remove VRAM from shared bo domains.Samuel Li1-2/+6
This fixes an issue introduced by change "allow framebuffer in GART memory as well" which could lead to a shared buffer ending up pinned in vram. Use GTT if it is included in the domain, otherwise return an error. Signed-off-by: Samuel Li <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: handle domain mask checking v2Chunming Zhou1-8/+1
if domain is illegal, we should return error. v2: remove duplicated domain checking. Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: set preferred_domain independent of fallback handlingChunming Zhou1-6/+9
When GEM needs to fallback to GTT for VRAM BOs we still want the preferred domain to be untouched so that the BO has a cance to move back to VRAM in the future. Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: use amdgpu_bo_param for amdgpu_bo_create v2Chunming Zhou1-29/+29
After that, we can easily add new parameter when need. v2: a) rebase. b) Initialize struct amdgpu_bo_param, future new member could only be used in some one case, but all member should have its own initial value. Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Huang Rui <[email protected]> (v1) Reviewed-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Cc: [email protected] Cc: [email protected] Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: add amdgpu_bo_paramChunming Zhou1-33/+42
amdgpu_bo_create has too many parameters, and used in too many places. Collect them to one structure. Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Huang Rui <[email protected]> Reviewed-by: Christian König <[email protected]> Reviewed-by: Junwei Zhang <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: revert "Don't change preferred domian when fallback GTT v6"Christian König1-24/+13
This reverts commit 7d1ca1325260a9e9329b10a21e3692e6f188936f. Makes fallback handling to complicated. This is just a feature for the GEM interface and shouldn't leak into the core BO create function. The intended change to preserve the preferred domains is implemented in a follow up patch. Signed-off-by: Christian König <[email protected]> Acked-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: revert "add new bo flag that indicates BOs don't need fallback (v2)"Christian König1-4/+1
This reverts commit 6f51d28bfe8e1a676de5cd877639245bed3cc818. Makes fallback handling to complicated. This is just a feature for the GEM interface and shouldn't leak into the core BO create function. Signed-off-by: Christian König <[email protected]> Acked-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-05-15drm/amdgpu: Free VGA stolen memory as soon as possible.Andrey Grodzovsky1-0/+7
Reserved VRAM is used to avoid overriding pre OS FB. Once our display stack takes over we don't need the reserved VRAM anymore. v2: Remove comment, we know actually why we need to reserve the stolen VRAM. Fix return type for amdgpu_ttm_late_init. v3: Return 0 in amdgpu_bo_late_init, rebase on changes to previous patch v4: rebase v5: For GMC9 reserve always just 9M and keep the stolem memory around until GART table curruption on S3 resume is resolved. Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Andrey Grodzovsky <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-04-11drm/amdgpu: add new bo flag that indicates BOs don't need fallback (v2)Chunming Zhou1-1/+4
user cases: 1. KFD wraps amdgpu_bo_create, they have no fallback case which is different with amdgpu_gem_object_create. since upstream branch has no amdgpu_amdkfd_gpuvm.c, which need KFD guys add this flag to __alloc_memory_of_gpu: + flags |= AMDGPU_GEM_CREATE_NO_FALLBACK; 2. UMD can specify this flag for their allocation as well if they like. v2: squash in merge conflict fix (Chunming) Signed-off-by: Chunming Zhou <[email protected]> Acked-by: Felix Kuehling <[email protected]> Cc: [email protected] Signed-off-by: Alex Deucher <[email protected]>
2018-04-11drm/amdgpu: Don't change preferred domian when fallback GTT v6Chunming Zhou1-13/+24
v2: add sanity checking v3: make code open v4: also handle visible to invisible fallback v5: Since two fallback cases, re-use goto retry v6: avoid bo is unref when retry, and only user BO can fallback Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Christian König <[email protected]> (v5) Acked-by: Felix Kuehling <[email protected]> Cc: [email protected] Signed-off-by: Alex Deucher <[email protected]>
2018-04-03Revert "drm/amdgpu: Don't change preferred domian when fallback GTT v5"Chunming Zhou1-15/+3
This reverts commit bbaf1871ea1d7b7021e350a3e8bb1ec8408c2030. Felix reports this commit causes regression. Signed-off-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
2018-04-03drm/amdgpu: fix typo of domain fallbackChunming Zhou1-1/+1
Signed-off-by: Chunming Zhou <[email protected]> Reviewed-by: Christian König <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
Powered by cgit, Gruvboxed by Blaster4385.