drm/amdkfd: fix some race conditions in vram buffer alloc/free of svm code

This patch fixes:
1: ref number of prange's svm_bo got decreased by an async call from hmm. When
wait svm_bo of prange got released we shoul also wait prang->svm_bo become NULL,
otherwise prange->svm_bo may be set to null after allocate new vram buffer.

2: During waiting svm_bo of prange got released in a while loop should reschedule
current task to give other tasks oppotunity to run, specially the the workque
task that handles svm_bo ref release, otherwise we may enter to softlock.

Signed-off-by: Xiaogang.Chen <xiaogang.chen@amd.com>
Reviewed-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 136a7a969d4809a643cd706dfa78b20963206769..56a2afdfd3d4f359c0d6f04cb0b99aeab5a22cab 100644 (file)
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -504,11 +504,11 @@ svm_range_validate_svm_bo(struct kfd_node *node, struct svm_range *prange)
 
        /* We need a new svm_bo. Spin-loop to wait for concurrent
         * svm_range_bo_release to finish removing this range from
-        * its range list. After this, it is safe to reuse the
-        * svm_bo pointer and svm_bo_list head.
+        * its range list and set prange->svm_bo to null. After this,
+        * it is safe to reuse the svm_bo pointer and svm_bo_list head.
         */
-       while (!list_empty_careful(&prange->svm_bo_list))
-               ;
+       while (!list_empty_careful(&prange->svm_bo_list) || prange->svm_bo)
+               cond_resched();
 
        return false;
 }