drm/amdgpu: replace get_user_pages with HMM mirror helpers

Use HMM helper function hmm_vma_fault() to get physical pages backing
userptr and start CPU page table update track of those pages. Then use
hmm_vma_range_done() to check if those pages are updated before
amdgpu_cs_submit for gfx or before user queues are resumed for kfd.

If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart
from scratch, for kfd, restore worker is rescheduled to retry.

HMM simplify the CPU page table concurrent update check, so remove
guptasklock, mmu_invalidations, last_set_pages fields from
amdgpu_ttm_tt struct.

HMM does not pin the page (increase page ref count), so remove related
operations like release_pages(), put_page(), mark_page_dirty().

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Philip Yang 2018-12-13 15:35:28 -05:00 committed by Alex Deucher
parent 8dd69e69f4
commit 915d3eecfa
9 changed files with 183 additions and 279 deletions

View file

@ -329,26 +329,24 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
r = amdgpu_bo_reserve(bo, true);
if (r)
goto free_pages;
goto user_pages_done;
amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
amdgpu_bo_unreserve(bo);
if (r)
goto free_pages;
goto user_pages_done;
}
r = drm_gem_handle_create(filp, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_put_unlocked(gobj);
if (r)
return r;
goto user_pages_done;
args->handle = handle;
return 0;
free_pages:
release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages);
user_pages_done:
if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE)
amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
release_object:
drm_gem_object_put_unlocked(gobj);