gpu: nvgpu: Fix crash on read fail of mc_boot_0_r

This CL handles
- erroneous use of boot_0 function pointer
before being assigned in __nvgpu_check_gpu_state
- And proper handling of error returned from gk20a_readl
in gk20a_mc_boot_0
With these fixes crash is not seen in case mc_boot_0 read
returns 0 in gk20a_mc_boot_0
- And also this handles the recursion caused by mc.boot_0()
calling nvgpu_readl and nvgpu_readl in turn
calling mc.boot_0 in case of read failure

Bug 2010966

Change-Id: Ia087811c67d88948b7fc5fff35e0fabc6ea91989
Signed-off-by: Supriya <ssharatkumar@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1616274
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

drivers/gpu/nvgpu/common/linux/io.c

@@ -32,6 +32,16 @@ void nvgpu_writel(struct gk20a *g, u32 r, u32 v)
 }
 
 u32 nvgpu_readl(struct gk20a *g, u32 r)
+{
+	u32 v = __nvgpu_readl(g, r);
+
+	if (v == 0xffffffff)
+		__nvgpu_check_gpu_state(g);
+
+	return v;
+}
+
+u32 __nvgpu_readl(struct gk20a *g, u32 r)
 {
 	struct nvgpu_os_linux *l = nvgpu_os_linux_from_gk20a(g);
 	u32 v = 0xffffffff;
@@ -41,8 +51,6 @@ u32 nvgpu_readl(struct gk20a *g, u32 r)
 		gk20a_dbg(gpu_dbg_reg, "r=0x%x v=0x%x (failed)", r, v);
 	} else {
 		v = readl(l->regs + r);
-		if (v == 0xffffffff)
-			__nvgpu_check_gpu_state(g);
 		gk20a_dbg(gpu_dbg_reg, "r=0x%x v=0x%x", r, v);
 	}
 

drivers/gpu/nvgpu/gk20a/gk20a.c

@@ -54,11 +54,17 @@
 
 void __nvgpu_check_gpu_state(struct gk20a *g)
 {
-	u32 boot_0 = g->ops.mc.boot_0(g, NULL, NULL, NULL);
+	u32 boot_0 = 0xffffffff;
 
+	if (!g->ops.mc.boot_0) {
+		nvgpu_err(g, "Can't determine GPU state, mc.boot_0 unset");
+		return;
+	}
+
+	boot_0 = g->ops.mc.boot_0(g, NULL, NULL, NULL);
 	if (boot_0 == 0xffffffff) {
-		pr_err("nvgpu: GPU has disappeared from bus!!\n");
-		pr_err("nvgpu: Rebooting system!!\n");
+		nvgpu_err(g, "GPU has disappeared from bus!!");
+		nvgpu_err(g, "Rebooting system!!");
 		kernel_restart(NULL);
 	}
 }

drivers/gpu/nvgpu/gk20a/mc_gk20a.c

@@ -200,7 +200,10 @@ void gk20a_mc_reset(struct gk20a *g, u32 units)
 
 u32 gk20a_mc_boot_0(struct gk20a *g, u32 *arch, u32 *impl, u32 *rev)
 {
-	u32 val = gk20a_readl(g, mc_boot_0_r());
+	u32 val = __nvgpu_readl(g, mc_boot_0_r());
+
+	if (val == 0xffffffff)
+		return val;
 
 	if (arch)
 		*arch = mc_boot_0_architecture_v(val) <<

drivers/gpu/nvgpu/include/nvgpu/io.h

@@ -40,6 +40,7 @@ struct gk20a;
 
 void nvgpu_writel(struct gk20a *g, u32 r, u32 v);
 u32 nvgpu_readl(struct gk20a *g, u32 r);
+u32 __nvgpu_readl(struct gk20a *g, u32 r);
 void nvgpu_writel_check(struct gk20a *g, u32 r, u32 v);
 void nvgpu_bar1_writel(struct gk20a *g, u32 b, u32 v);
 u32 nvgpu_bar1_readl(struct gk20a *g, u32 b);