gpu: nvgpu: fix pid retrieval on tsg reset

On engine reset, an event is generate for FECS
trace. In case a TSG context is currentlu loaded
on GR engine, we retrieve the pid of the TSG from
the first channel in the ch_list. Fixed invalid
invocation of list_entry that led to crash.

Bug 200193891

Change-Id: I79358bbb6685748cde68396ce220ab7b660d414d
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: http://git-master/r/1154811
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Richard Zhao <rizhao@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
Tested-by: Terje Bergstrom <tbergstrom@nvidia.com>

1 files changed, 5 insertions, 3 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c b/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
index 77e93458..3682d904 100644
--- a/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/ctxsw_trace_gk20a.c
@@ -686,11 +686,13 @@ void gk20a_ctxsw_trace_tsg_reset(struct gk20a *g, struct tsg_gk20a *tsg)
                 return;
 
         mutex_lock(&tsg->ch_list_lock);
-        ch = list_entry(&tsg->ch_list, struct channel_gk20a, ch_entry);
+        if (!list_empty(&tsg->ch_list)) {
+                ch = list_entry(tsg->ch_list.next,
+                                struct channel_gk20a, ch_entry);
+                entry.pid = ch->pid;
+        }
         mutex_unlock(&tsg->ch_list_lock);
 
-        entry.pid = ch->pid;
-
         gk20a_ctxsw_trace_write(g, &entry);
         gk20a_ctxsw_trace_wake_up(g, 0);
 #endif