drm/amdkfd: optimize svm range evict

It is to avoid unnecessary queue eviction when range
is not mapped to gpu.

Signed-off-by: Eric Huang <jinhuieric.huang@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 a2d103ce26c764cfdf8cd1e489eb92eb65649d5e..b0c1c7172a45613b33481caaf37d9ec8b2f96ed3 100644 (file)
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1778,8 +1778,12 @@ svm_range_evict(struct svm_range *prange, struct mm_struct *mm,
        if (!p->xnack_enabled ||
            (prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED)) {
                int evicted_ranges;
+               bool mapped = prange->mapped_to_gpu;
 
                list_for_each_entry(pchild, &prange->child_list, child_list) {
+                       if (!pchild->mapped_to_gpu)
+                               continue;
+                       mapped = true;
                        mutex_lock_nested(&pchild->lock, 1);
                        if (pchild->start <= last && pchild->last >= start) {
                                pr_debug("increment pchild invalid [0x%lx 0x%lx]\n",
@@ -1789,6 +1793,9 @@ svm_range_evict(struct svm_range *prange, struct mm_struct *mm,
                        mutex_unlock(&pchild->lock);
                }
 
+               if (!mapped)
+                       return r;
+
                if (prange->start <= last && prange->last >= start)
                        atomic_inc(&prange->invalid);
 
@@ -3345,7 +3352,8 @@ svm_range_set_attr(struct kfd_process *p, struct mm_struct *mm,
                        goto out_unlock_range;
 
                if (migrated && (!p->xnack_enabled ||
-                   (prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED))) {
+                   (prange->flags & KFD_IOCTL_SVM_FLAG_GPU_ALWAYS_MAPPED)) &&
+                   prange->mapped_to_gpu) {
                        pr_debug("restore_work will update mappings of GPUs\n");
                        mutex_unlock(&prange->migrate_mutex);
                        continue;