gpu: nvgpu: fix cleanup path in css_gr_create_client_data()

In css_gr_create_client_data(), we call css_gr_free_client_data() in case
cur->perfmon_start is zero

css_gr_free_client_data() deletes the node from list, and performs some
cleanup if cur->perfmon_start is non-zero

Hence it is incorrect to call css_gr_free_client_data() in clean up path
since node is not added to the list, and cur->perfmon_start is zero anyways

Fix the cleanup path, and remove call to css_gr_free_client_data()

Jira NVGPU-397

Change-Id: I9cb1abf4d6a02716156c9ea2d4a7963e1c92a531
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1606977
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 2 insertions, 12 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
index afba2496..dcd5b074 100644
--- a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
@@ -455,8 +455,6 @@ static int css_gr_create_client_data(struct gk20a *g,
                         u32 perfmon_count,
                         struct gk20a_cs_snapshot_client *cur)
 {
-        int ret = 0;
-
         memset(cur->snapshot, 0, sizeof(*cur->snapshot));
         cur->snapshot->start = sizeof(*cur->snapshot);
         /* we should be ensure that can fit all fifo entries here */
@@ -475,21 +473,13 @@ static int css_gr_create_client_data(struct gk20a *g,
         if (cur->perfmon_count && g->ops.css.allocate_perfmon_ids) {
                 cur->perfmon_start = g->ops.css.allocate_perfmon_ids(data,
                                                         cur->perfmon_count);
-                if (!cur->perfmon_start) {
-                        ret = -ENOENT;
-                        goto failed;
-                }
+                if (!cur->perfmon_start)
+                        return -ENOENT;
         }
 
         nvgpu_list_add_tail(&cur->list, &data->clients);
 
         return 0;
-
-failed:
-        if (cur)
-                css_gr_free_client_data(g, data, cur);
-
-        return ret;
 }