gpu: nvgpu: protect recovery with engines_reset_mutex

Rename gr_reset_mutex to engines_reset_mutex and acquire it
before initiating recovery. Recovery running in parallel with
engine reset is not recommended.

On hitting engine reset, h/w drops the ctxsw_status to INVALID in
fifo_engine_status register. Also while the engine is held in reset
h/w passes busy/idle straight through. fifo_engine_status registers
are correct in that there is no context switch outstanding
as the CTXSW is aborted when reset is asserted.

Use deferred_reset_mutex to protect deferred_reset_pending variable
If deferred_reset_pending is true then acquire engines_reset_mutex
and call gk20a_fifo_deferred_reset.
gk20a_fifo_deferred_reset would also check the value of
deferred_reset_pending before initiating reset process

Bug 2092051
Bug 2429295
Bug 2484211
Bug 1890287

Change-Id: I47de669a6203e0b2e9a8237ec4e4747339b9837c
Signed-off-by: Seema Khowala <seemaj@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2022373
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
(cherry-picked from cb91bf1e13740023903282d1c2271d9154e940ba
in dev-main)
Reviewed-on: https://git-master.nvidia.com/r/2024901
GVS: Gerrit_Virtual_Submit
Reviewed-by: Bibek Basu <bbasu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 10 insertions, 9 deletions

diff --git a/drivers/gpu/nvgpu/common/fifo/channel.c b/drivers/gpu/nvgpu/common/fifo/channel.c
index d30b8ded..4bea032a 100644
--- a/drivers/gpu/nvgpu/common/fifo/channel.c
+++ b/drivers/gpu/nvgpu/common/fifo/channel.c
@@ -308,6 +308,7 @@ static void gk20a_free_channel(struct channel_gk20a *ch, bool force)
         struct dbg_session_data *session_data, *tmp_s;
         struct dbg_session_channel_data *ch_data, *tmp;
         int err;
+        bool deferred_reset_pending;
 
         nvgpu_log_fn(g, " ");
 
@@ -381,17 +382,17 @@ static void gk20a_free_channel(struct channel_gk20a *ch, bool force)
 
         /* if engine reset was deferred, perform it now */
         nvgpu_mutex_acquire(&f->deferred_reset_mutex);
-        if (g->fifo.deferred_reset_pending) {
+        deferred_reset_pending = g->fifo.deferred_reset_pending;
+        nvgpu_mutex_release(&f->deferred_reset_mutex);
+
+        if (deferred_reset_pending) {
                 nvgpu_log(g, gpu_dbg_intr | gpu_dbg_gpu_dbg, "engine reset was"
-                           " deferred, running now");
-                /* if lock is already taken, a reset is taking place
-                so no need to repeat */
-                if (nvgpu_mutex_tryacquire(&g->fifo.gr_reset_mutex)) {
-                        gk20a_fifo_deferred_reset(g, ch);
-                        nvgpu_mutex_release(&g->fifo.gr_reset_mutex);
-                }
+                                " deferred, running now");
+                nvgpu_mutex_acquire(&g->fifo.engines_reset_mutex);
+                gk20a_fifo_deferred_reset(g, ch);
+                nvgpu_mutex_release(&g->fifo.engines_reset_mutex);
         }
-        nvgpu_mutex_release(&f->deferred_reset_mutex);
+
 
         if (!gk20a_channel_as_bound(ch)) {
                 goto unbind;