[PATCH] drm/amdkfd: Remove arbitrary timeout for hmm_range_fault

Philip Yang Philip.Yang at amd.com
Wed May 1 22:56:55 UTC 2024


On system with khugepaged enabled and user cases with THP buffer, the
hmm_range_fault may takes > 15 seconds to return -EBUSY, the arbitrary
timeout value is not accurate, cause memory allocation failure.

Remove the arbitrary timeout value, return EAGAIN to application if
hmm_range_fault return EBUSY, then userspace libdrm and Thunk will call
ioctl again.

Change EAGAIN to debug message as this is not error.

Signed-off-by: Philip Yang <Philip.Yang at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c |  5 ++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c          | 12 +++---------
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c             |  5 +----
 3 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 54198c3928c7..02696c2102f1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1087,7 +1087,10 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr,
 
 	ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages, &range);
 	if (ret) {
-		pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
+		if (ret == -EAGAIN)
+			pr_debug("Failed to get user pages, try again\n");
+		else
+			pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
 		goto unregister_out;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
index 431ec72655ec..e36fede7f74c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
@@ -202,20 +202,12 @@ int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
 		pr_debug("hmm range: start = 0x%lx, end = 0x%lx",
 			hmm_range->start, hmm_range->end);
 
-		/* Assuming 64MB takes maximum 1 second to fault page address */
-		timeout = max((hmm_range->end - hmm_range->start) >> 26, 1UL);
-		timeout *= HMM_RANGE_DEFAULT_TIMEOUT;
-		timeout = jiffies + msecs_to_jiffies(timeout);
+		timeout = jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
 
 retry:
 		hmm_range->notifier_seq = mmu_interval_read_begin(notifier);
 		r = hmm_range_fault(hmm_range);
 		if (unlikely(r)) {
-			schedule();
-			/*
-			 * FIXME: This timeout should encompass the retry from
-			 * mmu_interval_read_retry() as well.
-			 */
 			if (r == -EBUSY && !time_after(jiffies, timeout))
 				goto retry;
 			goto out_free_pfns;
@@ -247,6 +239,8 @@ int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
 out_free_range:
 	kfree(hmm_range);
 
+	if (r == -EBUSY)
+		r = -EAGAIN;
 	return r;
 }
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 94f83be2232d..e7040f809f33 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1670,11 +1670,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm,
 						       readonly, owner, NULL,
 						       &hmm_range);
 			WRITE_ONCE(p->svms.faulting_task, NULL);
-			if (r) {
+			if (r)
 				pr_debug("failed %d to get svm range pages\n", r);
-				if (r == -EBUSY)
-					r = -EAGAIN;
-			}
 		} else {
 			r = -EFAULT;
 		}
-- 
2.43.2



More information about the amd-gfx mailing list