gpu: nvgpu: remove reversed ordering for deadlock

In some cases, we would get deadlock issue due to there are two locks
acquisition on common clk driver's lock and nvgpu driver's locks. At
the bug, inconsistent lock ordering problem will come with one thread
gets "nvgpu lock -> clk lock" and the other thread gets "clk lock ->
nvgpu lock".

Slove the latter path with one-time initializing clk_parent entry
and use cached data afterward.

Bug 2555115

Change-Id: I31c5c2728f406307e7cfd4e555f4db0c163234d8
Signed-off-by: Jeremy Ho <jeremyh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2146727
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Aleksandr Frid <afrid@nvidia.com>
Reviewed-by: Bibek Basu <bbasu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

drivers/gpu/nvgpu/gk20a/clk_gk20a.h

@@ -95,6 +95,7 @@ struct clk_gk20a {
 	struct gk20a *g;
 #if defined(CONFIG_COMMON_CLK)
 	struct clk *tegra_clk;
+	struct clk *tegra_clk_parent;
 	struct clk_hw hw;
 #endif
 	struct pll gpc_pll;

drivers/gpu/nvgpu/os/linux/clk.c

@@ -90,7 +90,7 @@ static unsigned long nvgpu_linux_get_fmax_at_vmin_safe(struct gk20a *g)
 	 */
 	if (g->clk.tegra_clk)
 		return tegra_dvfs_get_fmax_at_vmin_safe_t(
-			clk_get_parent(g->clk.tegra_clk));
+			g->clk.tegra_clk_parent);
 
 	if (platform->maxmin_clk_id)
 		return tegra_bpmp_dvfs_get_fmax_at_vmin(
@@ -116,7 +116,7 @@ static int nvgpu_linux_predict_mv_at_hz_cur_tfloor(struct clk_gk20a *clk,
 	unsigned long rate)
 {
 	return tegra_dvfs_predict_mv_at_hz_cur_tfloor(
-				clk_get_parent(clk->tegra_clk), rate);
+				clk->tegra_clk_parent, rate);
 }
 
 static unsigned long nvgpu_linux_get_maxrate(struct gk20a *g, u32 api_domain)
@@ -125,7 +125,7 @@ static unsigned long nvgpu_linux_get_maxrate(struct gk20a *g, u32 api_domain)
 
 	switch (api_domain) {
 	case CTRL_CLK_DOMAIN_GPCCLK:
-		ret = tegra_dvfs_get_maxrate(clk_get_parent(g->clk.tegra_clk));
+		ret = tegra_dvfs_get_maxrate(g->clk.tegra_clk_parent);
 		break;
 	default:
 		nvgpu_err(g, "unknown clock: %u", api_domain);

drivers/gpu/nvgpu/os/linux/platform_gk20a_tegra.c

@@ -649,7 +649,7 @@ int gk20a_tegra_init_secure_alloc(struct gk20a_platform *platform)
 static struct clk *gk20a_clk_get(struct gk20a *g)
 {
 	if (!g->clk.tegra_clk) {
-		struct clk *clk;
+		struct clk *clk, *clk_parent;
 		char clk_dev_id[32];
 		struct device *dev = dev_from_gk20a(g);
 
@@ -661,7 +661,16 @@ static struct clk *gk20a_clk_get(struct gk20a *g)
 				  clk_dev_id);
 			return NULL;
 		}
+
+		clk_parent = clk_get_parent(clk);
+		if (IS_ERR_OR_NULL(clk_parent)) {
+			nvgpu_err(g, "fail to get tegra gpu clk parent%s/gpu\n",
+				  clk_dev_id);
+			return NULL;
+		}
+
 		g->clk.tegra_clk = clk;
+		g->clk.tegra_clk_parent = clk_parent;
 	}
 
 	return g->clk.tegra_clk;