drm/amdkfd: page table restore through svm API

Page table restore implementation in SVM API. This is called from
the fault handler at amdgpu_vm. To update page tables through
the page fault retry IH.

Signed-off-by: Alex Sierra <alex.sierra@amd.com>
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 6d3ed19..a992fa8 100644 (file)
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1997,6 +1997,65 @@ svm_range_from_addr(struct svm_range_list *svms, unsigned long addr,
        return NULL;
 }
 
+int
+svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
+                       uint64_t addr)
+{
+       int r = 0;
+       struct mm_struct *mm = NULL;
+       struct svm_range *prange;
+       struct svm_range_list *svms;
+       struct kfd_process *p;
+
+       p = kfd_lookup_process_by_pasid(pasid);
+       if (!p) {
+               pr_debug("kfd process not founded pasid 0x%x\n", pasid);
+               return -ESRCH;
+       }
+       if (!p->xnack_enabled) {
+               pr_debug("XNACK not enabled for pasid 0x%x\n", pasid);
+               return -EFAULT;
+       }
+       svms = &p->svms;
+
+       pr_debug("restoring svms 0x%p fault address 0x%llx\n", svms, addr);
+
+       mm = get_task_mm(p->lead_thread);
+       if (!mm) {
+               pr_debug("svms 0x%p failed to get mm\n", svms);
+               r = -ESRCH;
+               goto out;
+       }
+
+       mmap_read_lock(mm);
+       mutex_lock(&svms->lock);
+       prange = svm_range_from_addr(svms, addr, NULL);
+
+       if (!prange) {
+               pr_debug("failed to find prange svms 0x%p address [0x%llx]\n",
+                        svms, addr);
+               r = -EFAULT;
+               goto out_unlock_svms;
+       }
+
+       mutex_lock(&prange->migrate_mutex);
+
+       r = svm_range_validate_and_map(mm, prange, MAX_GPU_INSTANCE, false, false);
+       if (r)
+               pr_debug("failed %d to map svms 0x%p [0x%lx 0x%lx] to gpu\n", r,
+                        svms, prange->start, prange->last);
+
+       mutex_unlock(&prange->migrate_mutex);
+out_unlock_svms:
+       mutex_unlock(&svms->lock);
+       mmap_read_unlock(mm);
+       mmput(mm);
+out:
+       kfd_unref_process(p);
+
+       return r;
+}
+
 void svm_range_list_fini(struct kfd_process *p)
 {
        struct svm_range *prange;

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
index d88926d..9a078d3 100644 (file)
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
@@ -155,6 +155,8 @@ void svm_range_vram_node_free(struct svm_range *prange);
 int svm_range_split_by_granularity(struct kfd_process *p, struct mm_struct *mm,
                               unsigned long addr, struct svm_range *parent,
                               struct svm_range *prange);
+int svm_range_restore_pages(struct amdgpu_device *adev,
+                           unsigned int pasid, uint64_t addr);
 void svm_range_add_list_work(struct svm_range_list *svms,
                             struct svm_range *prange, struct mm_struct *mm,
                             enum svm_work_list_ops op);