On 3/16/20 10:52 AM, Christoph Hellwig wrote:

No driver has actually used properly wire up and support this feature. There is various code related to it in nouveau, but as far as I can tell it never actually got turned on, and the only changes since the initial commit are global cleanups.

This is not actually true. OpenCL 2.x does support SVM with nouveau and device private memory via clEnqueueSVMMigrateMem(). Also, Ben Skeggs has accepted a set of patches to map GPU memory after being migrated and this change would conflict with that.

Signed-off-by: Christoph Hellwig hch@lst.de

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 1 - drivers/gpu/drm/nouveau/nouveau_dmem.c | 37 ------------------------- drivers/gpu/drm/nouveau/nouveau_dmem.h | 2 -- drivers/gpu/drm/nouveau/nouveau_svm.c | 3 -- include/linux/hmm.h | 2 -- mm/hmm.c | 28 ------------------- 6 files changed, 73 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index dee446278417..90821ce5e6ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -776,7 +776,6 @@ struct amdgpu_ttm_tt { static const uint64_t hmm_range_flags[HMM_PFN_FLAG_MAX] = { (1 << 0), /* HMM_PFN_VALID */ (1 << 1), /* HMM_PFN_WRITE */

• 0 /* HMM_PFN_DEVICE_PRIVATE */ };

  static const uint64_t hmm_range_values[HMM_PFN_VALUE_MAX] = {

diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.c b/drivers/gpu/drm/nouveau/nouveau_dmem.c index 7605c4c48985..42808efceaf2 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dmem.c +++ b/drivers/gpu/drm/nouveau/nouveau_dmem.c @@ -671,40 +671,3 @@ nouveau_dmem_migrate_vma(struct nouveau_drm *drm, out: return ret; }

-static inline bool -nouveau_dmem_page(struct nouveau_drm *drm, struct page *page) -{

• return is_device_private_page(page) && drm->dmem == page_to_dmem(page);

-}

-void -nouveau_dmem_convert_pfn(struct nouveau_drm *drm,

• 	 struct hmm_range *range)
  

-{

• unsigned long i, npages;
• npages = (range->end - range->start) >> PAGE_SHIFT;
• for (i = 0; i < npages; ++i) {

• struct page *page;
  

• uint64_t addr;
  

• page = hmm_device_entry_to_page(range, range->pfns[i]);
  

• if (page == NULL)
  

• 	continue;
  

• if (!(range->pfns[i] & range->flags[HMM_PFN_DEVICE_PRIVATE])) {
  

• 	continue;
  

• }
  

• if (!nouveau_dmem_page(drm, page)) {
  

• 	WARN(1, "Some unknown device memory !\n");
  

• 	range->pfns[i] = 0;
  

• 	continue;
  

• }
  

• addr = nouveau_dmem_page_addr(page);
  

• range->pfns[i] &= ((1UL << range->pfn_shift) - 1);
  

• range->pfns[i] |= (addr >> PAGE_SHIFT) << range->pfn_shift;
  

• }

-} diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.h b/drivers/gpu/drm/nouveau/nouveau_dmem.h index 92394be5d649..1ac620b3d4fb 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dmem.h +++ b/drivers/gpu/drm/nouveau/nouveau_dmem.h @@ -38,8 +38,6 @@ int nouveau_dmem_migrate_vma(struct nouveau_drm *drm, unsigned long start, unsigned long end);

-void nouveau_dmem_convert_pfn(struct nouveau_drm *drm,

• 	      struct hmm_range *range);
  

  #else /* IS_ENABLED(CONFIG_DRM_NOUVEAU_SVM) */ static inline void nouveau_dmem_init(struct nouveau_drm *drm) {} static inline void nouveau_dmem_fini(struct nouveau_drm *drm) {}

diff --git a/drivers/gpu/drm/nouveau/nouveau_svm.c b/drivers/gpu/drm/nouveau/nouveau_svm.c index df9bf1fd1bc0..7e0376dca137 100644 --- a/drivers/gpu/drm/nouveau/nouveau_svm.c +++ b/drivers/gpu/drm/nouveau/nouveau_svm.c @@ -367,7 +367,6 @@ static const u64 nouveau_svm_pfn_flags[HMM_PFN_FLAG_MAX] = { [HMM_PFN_VALID ] = NVIF_VMM_PFNMAP_V0_V, [HMM_PFN_WRITE ] = NVIF_VMM_PFNMAP_V0_W,

• [HMM_PFN_DEVICE_PRIVATE] = NVIF_VMM_PFNMAP_V0_VRAM, };

  static const u64

@@ -558,8 +557,6 @@ static int nouveau_range_fault(struct nouveau_svmm *svmm, break; }

• nouveau_dmem_convert_pfn(drm, &range);
• svmm->vmm->vmm.object.client->super = true; ret = nvif_object_ioctl(&svmm->vmm->vmm.object, data, size, NULL); svmm->vmm->vmm.object.client->super = false;

diff --git a/include/linux/hmm.h b/include/linux/hmm.h index 4bf8d6997b12..5e6034f105c3 100644 --- a/include/linux/hmm.h +++ b/include/linux/hmm.h @@ -74,7 +74,6 @@

• Flags:
• HMM_PFN_VALID: pfn is valid. It has, at least, read permission.
• HMM_PFN_WRITE: CPU page table has write permission set

• • HMM_PFN_DEVICE_PRIVATE: private device memory (ZONE_DEVICE)
  • The driver provides a flags array for mapping page protections to device
  • PTE bits. If the driver valid bit for an entry is bit 3,

@@ -86,7 +85,6 @@ enum hmm_pfn_flag_e { HMM_PFN_VALID = 0, HMM_PFN_WRITE,

• HMM_PFN_DEVICE_PRIVATE, HMM_PFN_FLAG_MAX };

diff --git a/mm/hmm.c b/mm/hmm.c index 180e398170b0..3d10485bf323 100644 --- a/mm/hmm.c +++ b/mm/hmm.c @@ -118,15 +118,6 @@ static inline void hmm_pte_need_fault(const struct hmm_vma_walk *hmm_vma_walk, /* We aren't ask to do anything ... */ if (!(pfns & range->flags[HMM_PFN_VALID])) return;

• /* If this is device memory then only fault if explicitly requested */

• if ((cpu_flags & range->flags[HMM_PFN_DEVICE_PRIVATE])) {

• /* Do we fault on device memory ? */
  

• if (pfns & range->flags[HMM_PFN_DEVICE_PRIVATE]) {
  

• 	*write_fault = pfns & range->flags[HMM_PFN_WRITE];
  

• 	*fault = true;
  

• }
  

• return;
  

• }

  /* If CPU page table is not valid then we need to fault */ *fault = !(cpu_flags & range->flags[HMM_PFN_VALID]);

@@ -259,25 +250,6 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr, if (!pte_present(pte)) { swp_entry_t entry = pte_to_swp_entry(pte);

• /*
  

•  * This is a special swap entry, ignore migration, use
  

•  * device and report anything else as error.
  

•  */
  

• if (is_device_private_entry(entry)) {
  

• 	cpu_flags = range->flags[HMM_PFN_VALID] |
  

• 		range->flags[HMM_PFN_DEVICE_PRIVATE];
  

• 	cpu_flags |= is_write_device_private_entry(entry) ?
  

• 		range->flags[HMM_PFN_WRITE] : 0;
  

• 	hmm_pte_need_fault(hmm_vma_walk, orig_pfn, cpu_flags,
  

• 			   &fault, &write_fault);
  

• 	if (fault || write_fault)
  

• 		goto fault;
  

• 	*pfn = hmm_device_entry_from_pfn(range,
  

• 			    swp_offset(entry));
  

• 	*pfn |= cpu_flags;
  

• 	return 0;
  

• }
  

• hmm_pte_need_fault(hmm_vma_walk, orig_pfn, 0, &fault, &write_fault); if (!fault && !write_fault)