diff options
| author | Lucas De Marchi <lucas.demarchi@intel.com> | 2024-02-20 09:26:12 -0800 |
|---|---|---|
| committer | Lucas De Marchi <lucas.demarchi@intel.com> | 2024-02-20 09:57:17 -0800 |
| commit | fbb944086f2fa36c633be71cfcb38ce9f37eb90e (patch) | |
| tree | 78428083a0de1e910a2eb025102b462bebdb8bfc /drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c | |
| parent | 0f688c0eb63a643ef0568b29b12cefbb23181e1a (diff) | |
| parent | 9ac4beb7578a88baa4f7e6a59eeb5be79d7b011a (diff) | |
| download | lwn-fbb944086f2fa36c633be71cfcb38ce9f37eb90e.tar.gz lwn-fbb944086f2fa36c633be71cfcb38ce9f37eb90e.zip | |
Merge drm/drm-next into drm-xe-next
Bring changes from drm-misc-next that got merged in drm-next back to
drm-xe so they can be used for additional features.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index 73b8cca35bab..c623e23049d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -121,6 +121,7 @@ int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev) struct amdgpu_bo_param bp; dma_addr_t dma_addr; struct page *p; + unsigned long x; int ret; if (adev->gart.bo != NULL) @@ -130,6 +131,10 @@ int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev) if (!p) return -ENOMEM; + /* assign pages to this device */ + for (x = 0; x < (1UL << order); x++) + p[x].mapping = adev->mman.bdev.dev_mapping; + /* If the hardware does not support UTCL2 snooping of the CPU caches * then set_memory_wc() could be used as a workaround to mark the pages * as write combine memory. @@ -223,6 +228,7 @@ void amdgpu_gart_table_ram_free(struct amdgpu_device *adev) unsigned int order = get_order(adev->gart.table_size); struct sg_table *sg = adev->gart.bo->tbo.sg; struct page *p; + unsigned long x; int ret; ret = amdgpu_bo_reserve(adev->gart.bo, false); @@ -234,6 +240,8 @@ void amdgpu_gart_table_ram_free(struct amdgpu_device *adev) sg_free_table(sg); kfree(sg); p = virt_to_page(adev->gart.ptr); + for (x = 0; x < (1UL << order); x++) + p[x].mapping = NULL; __free_pages(p, order); adev->gart.ptr = NULL; |