gpu: nvgpu: clear events from channel/TSG while closing

If User does not close the event fd created on channel/TSG,
it is possible that the event stays on channel/TSG
structure and reappears when channel/TSG is re-opened

This causes false failure when we try to enable some event
of channel/TSG since we do not allow enabling same event
twice on same channel/TSG

Fix this by removing all enabled events from channel/TSG
while closing it

Bug 200243092
Bug 1818654

Change-Id: I9d5ffc89f87cf4c44124f8015c2c2f0587ad2ef4
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: http://git-master/r/1237723
(cherry picked from commit 2737a5c86cf5fbfe8a04f6a87764e8ecb9b30555)
Reviewed-on: http://git-master/r/1238266
GVS: Gerrit_Virtual_Submit
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>

2 files changed, 21 insertions, 0 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
index 4458430b..f60a92b4 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -1142,6 +1142,7 @@ struct channel_gk20a *gk20a_open_new_channel(struct gk20a *g,
 {
         struct fifo_gk20a *f = &g->fifo;
         struct channel_gk20a *ch;
+        struct gk20a_event_id_data *event_id_data, *event_id_data_temp;
 
         /* compatibility with existing code */
         if (!gk20a_fifo_is_valid_runlist_id(g, runlist_id)) {
@@ -1182,6 +1183,15 @@ struct channel_gk20a *gk20a_open_new_channel(struct gk20a *g,
         ch->pid = current->pid;
         ch->tgid = current->tgid;  /* process granularity for FECS traces */
 
+        /* unhook all events created on this channel */
+        mutex_lock(&ch->event_id_list_lock);
+        list_for_each_entry_safe(event_id_data, event_id_data_temp,
+                                &ch->event_id_list,
+                                event_id_node) {
+                list_del_init(&event_id_data->event_id_node);
+        }
+        mutex_unlock(&ch->event_id_list_lock);
+
         /* By default, channel is regular (non-TSG) channel */
         ch->tsgid = NVGPU_INVALID_TSG_ID;
 
diff --git a/drivers/gpu/nvgpu/gk20a/tsg_gk20a.c b/drivers/gpu/nvgpu/gk20a/tsg_gk20a.c
index 3e83cd06..716388ec 100644
--- a/drivers/gpu/nvgpu/gk20a/tsg_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/tsg_gk20a.c
@@ -451,6 +451,7 @@ void gk20a_tsg_release(struct kref *ref)
 {
         struct tsg_gk20a *tsg = container_of(ref, struct tsg_gk20a, refcount);
         struct gk20a *g = tsg->g;
+        struct gk20a_event_id_data *event_id_data, *event_id_data_temp;
 
         if (tsg->tsg_gr_ctx) {
                 gr_gk20a_free_tsg_gr_ctx(tsg);
@@ -462,6 +463,16 @@ void gk20a_tsg_release(struct kref *ref)
         }
 
         gk20a_sched_ctrl_tsg_removed(g, tsg);
+
+        /* unhook all events created on this TSG */
+        mutex_lock(&tsg->event_id_list_lock);
+        list_for_each_entry_safe(event_id_data, event_id_data_temp,
+                                &tsg->event_id_list,
+                                event_id_node) {
+                list_del_init(&event_id_data->event_id_node);
+        }
+        mutex_unlock(&tsg->event_id_list_lock);
+
         release_used_tsg(&g->fifo, tsg);
 
         tsg->runlist_id = ~0;