gpu: nvgpu: protect clk_arb init with mutex.

g->clk_arb is currently initialized as a part of gk20a_finalize_poweron().
Any subsequent call to gk20a_finalize_poweron reinitializes the clk_arb
and leading to memory leaks. This is resolved by protecting the
g->clk_arb initialization with a mutex clk_arb_enable_lock in struct
gk20a. We skip initializing the g->clk_arb if its not NULL.

Bug 2061372

Change-Id: I59158e0a5e4c827fdbd6d9ea2d04c78d0986347a
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1811650
Reviewed-by: Deepak Nibade <dnibade@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

5 files changed, 22 insertions, 1 deletions

diff --git a/drivers/gpu/nvgpu/clk/clk_arb.c b/drivers/gpu/nvgpu/clk/clk_arb.c
index 983a82f9..8543196b 100644
--- a/drivers/gpu/nvgpu/clk/clk_arb.c
+++ b/drivers/gpu/nvgpu/clk/clk_arb.c
@@ -750,12 +750,20 @@ error_check:
 
 int nvgpu_clk_arb_init_arbiter(struct gk20a *g)
 {
+        int err = 0;
+
         if (!g->ops.clk.support_clk_freq_controller ||
                 !g->ops.clk_arb.get_arbiter_clk_domains) {
                 return 0;
         }
 
-        return g->ops.clk_arb.arbiter_clk_init(g);
+        nvgpu_mutex_acquire(&g->clk_arb_enable_lock);
+
+        err = g->ops.clk_arb.arbiter_clk_init(g);
+
+        nvgpu_mutex_release(&g->clk_arb_enable_lock);
+
+        return err;
 }
 
 void nvgpu_clk_arb_send_thermal_alarm(struct gk20a *g)
@@ -783,10 +791,14 @@ void nvgpu_clk_arb_cleanup_arbiter(struct gk20a *g)
 {
         struct nvgpu_clk_arb *arb = g->clk_arb;
 
+        nvgpu_mutex_acquire(&g->clk_arb_enable_lock);
+
         if (arb) {
                 nvgpu_clk_arb_worker_deinit(g);
                 g->ops.clk_arb.clk_arb_cleanup(g->clk_arb);
         }
+
+        nvgpu_mutex_release(&g->clk_arb_enable_lock);
 }
 
 int nvgpu_clk_arb_init_session(struct gk20a *g,
diff --git a/drivers/gpu/nvgpu/gk20a/gk20a.h b/drivers/gpu/nvgpu/gk20a/gk20a.h
index bdf3a168..f62dfb94 100644
--- a/drivers/gpu/nvgpu/gk20a/gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/gk20a.h
@@ -1588,6 +1588,8 @@ struct gk20a {
 
         struct nvgpu_clk_arb *clk_arb;
 
+        struct nvgpu_mutex clk_arb_enable_lock;
+
         struct gk20a_ce_app ce_app;
 
         bool ltc_intr_en_illegal_compstat;
diff --git a/drivers/gpu/nvgpu/gp106/clk_arb_gp106.c b/drivers/gpu/nvgpu/gp106/clk_arb_gp106.c
index 860344d0..001f2bfc 100644
--- a/drivers/gpu/nvgpu/gp106/clk_arb_gp106.c
+++ b/drivers/gpu/nvgpu/gp106/clk_arb_gp106.c
@@ -121,6 +121,10 @@ int gp106_init_clk_arbiter(struct gk20a *g)
 
         clk_arb_dbg(g, " ");
 
+        if (g->clk_arb != NULL) {
+                return 0;
+        }
+
         arb = nvgpu_kzalloc(g, sizeof(struct nvgpu_clk_arb));
         if (!arb)
                 return -ENOMEM;
diff --git a/drivers/gpu/nvgpu/os/linux/driver_common.c b/drivers/gpu/nvgpu/os/linux/driver_common.c
index 16af2cb0..b81cc90c 100644
--- a/drivers/gpu/nvgpu/os/linux/driver_common.c
+++ b/drivers/gpu/nvgpu/os/linux/driver_common.c
@@ -65,6 +65,7 @@ static void nvgpu_init_vars(struct gk20a *g)
         nvgpu_mutex_init(&g->poweroff_lock);
         nvgpu_mutex_init(&g->ctxsw_disable_lock);
         nvgpu_mutex_init(&g->tpc_pg_lock);
+        nvgpu_mutex_init(&g->clk_arb_enable_lock);
 
         l->regs_saved = l->regs;
         l->bar1_saved = l->bar1;
diff --git a/drivers/gpu/nvgpu/os/linux/module.c b/drivers/gpu/nvgpu/os/linux/module.c
index cb501683..d226ceeb 100644
--- a/drivers/gpu/nvgpu/os/linux/module.c
+++ b/drivers/gpu/nvgpu/os/linux/module.c
@@ -1310,6 +1310,8 @@ int nvgpu_remove(struct device *dev, struct class *class)
         if (platform->remove)
                 platform->remove(dev);
 
+        nvgpu_mutex_destroy(&g->clk_arb_enable_lock);
+
         nvgpu_log_fn(g, "removed");
 
         return err;