Revert "hsaKmtCloseKFD destroy objects in all apertures"

This reverts commit bb4291dcac.

This change causes KFDTest failed on gfx803. The first hsaKmtCreateEvent
call allocate system memory for events_page because global variable
events_page is NULL. And this events page vm address should not be freed
until the process exit.

The change to destrory objects in hsaKmtCloseKFD removes events page. As
a result, KFDTest call hsaKmtOpenKFD again and then allocate memory will
get same events_page vm address on gfx803, and map this vm failed because
the vm conflict with events_page mapping.

KFDTest passed on VG10, gfx906 because allocate memory get different vm
address. hsaKmtCreateEvent still works fine as the driver keeps the
events page mapping of the process.

We should only destroy objects in fork cloned child process regardless
if gpu_vm is NULL or not.

Change-Id: I174ef65321cbd6074c855c2021318fe961c8c72c
Signed-off-by: Philip Yang <Philip.Yang@amd.com>


[ROCm/ROCR-Runtime commit: 7fc6a9f7c2]

projects/rocr-runtime/src/fmm.c

@@ -2345,37 +2345,14 @@ sysfs_parse_failed:
 	return ret;
 }
 
-static void fmm_clear_aperture(manageable_aperture_t *app)
-{
-	rbtree_node_t *n;
-
-	while ((n = rbtree_node_any(&app->tree, MID)))
-		vm_remove_object(app, vm_object_entry(n, 0));
-
-	while (app->vm_ranges)
-		vm_remove_area(app, app->vm_ranges);
-}
-
 void fmm_destroy_process_apertures(void)
 {
-	unsigned int i;
-
 	release_mmio();
-
-	fmm_clear_aperture(&cpuvm_aperture);
-	fmm_clear_aperture(&svm.apertures[SVM_DEFAULT]);
-	fmm_clear_aperture(&svm.apertures[SVM_COHERENT]);
-
 	if (gpu_mem) {
-		for (i = 0; i < gpu_mem_count; i++) {
-			fmm_clear_aperture(&gpu_mem[i].gpuvm_aperture);
-			fmm_clear_aperture(&gpu_mem[i].scratch_physical);
-		}
-
 		free(gpu_mem);
 		gpu_mem = NULL;
-		gpu_mem_count = 0;
 	}
+	gpu_mem_count = 0;
 }
 
 HSAKMT_STATUS fmm_get_aperture_base_and_limit(aperture_type_e aperture_type, HSAuint32 gpu_id,
@@ -3554,6 +3531,18 @@ HSAKMT_STATUS fmm_set_mem_user_data(const void *mem, void *usr_data)
 	return HSAKMT_STATUS_SUCCESS;
 }
 
+static void fmm_clear_aperture(manageable_aperture_t *app)
+{
+	rbtree_node_t *n;
+
+	while ((n = rbtree_node_any(&app->tree, MID)))
+		vm_remove_object(app, vm_object_entry(n, 0));
+
+	while (app->vm_ranges)
+		vm_remove_area(app, app->vm_ranges);
+
+}
+
 /* This is a special funcion that should be called only from the child process
  * after a fork(). This will clear all vm_objects and mmaps duplicated from
  * the parent.
@@ -3570,7 +3559,19 @@ void fmm_clear_all_mem(void)
 			drm_render_fds[i] = 0;
 		}
 
-	fmm_destroy_process_apertures();
+	/* Nothing is initialized. */
+	if (!gpu_mem)
+		return;
+
+	fmm_clear_aperture(&cpuvm_aperture);
+
+	for (i = 0; i < gpu_mem_count; i++) {
+		fmm_clear_aperture(&gpu_mem[i].gpuvm_aperture);
+		fmm_clear_aperture(&gpu_mem[i].scratch_physical);
+	}
+
+	fmm_clear_aperture(&svm.apertures[SVM_DEFAULT]);
+	fmm_clear_aperture(&svm.apertures[SVM_COHERENT]);
 
 	if (dgpu_shared_aperture_limit) {
 		/* Use the same dgpu range as the parent. If failed, then set
@@ -3596,4 +3597,7 @@ void fmm_clear_all_mem(void)
 
 	all_gpu_id_array_size = 0;
 	all_gpu_id_array = NULL;
+
+	gpu_mem_count = 0;
+	free(gpu_mem);
 }