summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdkfd/kfd_process.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdkfd/kfd_process.c')
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_process.c33
1 files changed, 25 insertions, 8 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index 1067afdb456e..7c0c24732481 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -35,6 +35,7 @@
#include <linux/pm_runtime.h>
#include "amdgpu_amdkfd.h"
#include "amdgpu.h"
+#include "amdgpu_reset.h"
struct mm_struct;
@@ -838,6 +839,14 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
return ERR_PTR(-EINVAL);
}
+ /* If the process just called exec(3), it is possible that the
+ * cleanup of the kfd_process (following the release of the mm
+ * of the old process image) is still in the cleanup work queue.
+ * Make sure to drain any job before trying to recreate any
+ * resource for this process.
+ */
+ flush_workqueue(kfd_process_wq);
+
/*
* take kfd processes mutex before starting of process creation
* so there won't be a case where two threads of the same process
@@ -858,14 +867,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
if (process) {
pr_debug("Process already found\n");
} else {
- /* If the process just called exec(3), it is possible that the
- * cleanup of the kfd_process (following the release of the mm
- * of the old process image) is still in the cleanup work queue.
- * Make sure to drain any job before trying to recreate any
- * resource for this process.
- */
- flush_workqueue(kfd_process_wq);
-
process = create_process(thread);
if (IS_ERR(process))
goto out;
@@ -1133,6 +1134,17 @@ static void kfd_process_remove_sysfs(struct kfd_process *p)
p->kobj = NULL;
}
+/*
+ * If any GPU is ongoing reset, wait for reset complete.
+ */
+static void kfd_process_wait_gpu_reset_complete(struct kfd_process *p)
+{
+ int i;
+
+ for (i = 0; i < p->n_pdds; i++)
+ flush_workqueue(p->pdds[i]->dev->adev->reset_domain->wq);
+}
+
/* No process locking is needed in this function, because the process
* is not findable any more. We must assume that no other thread is
* using it any more, otherwise we couldn't safely free the process
@@ -1147,6 +1159,11 @@ static void kfd_process_wq_release(struct work_struct *work)
kfd_process_dequeue_from_all_devices(p);
pqm_uninit(&p->pqm);
+ /*
+ * If GPU in reset, user queues may still running, wait for reset complete.
+ */
+ kfd_process_wait_gpu_reset_complete(p);
+
/* Signal the eviction fence after user mode queues are
* destroyed. This allows any BOs to be freed without
* triggering pointless evictions or waiting for fences.