[Intel-gfx] [PATCH 1/2] drm/i915: Check for scratch page scribbling

@@ -1737,7 +1737,7 @@ static int read_from_scratch(struct i915_gem_context *ctx,
 	return err;
 }
 
-static int check_scratch_page(struct i915_gem_context *ctx, u32 *out)
+static int check_ctx_scratch(struct i915_gem_context *ctx, u32 *out)
 {
 	struct i915_address_space *vm;
 	struct page *page;

@@ -1770,6 +1770,17 @@ static int check_scratch_page(struct i915_gem_context *ctx, u32 *out)
 	return err;
 }
 
+static void reset_ctx_scratch(struct i915_gem_context *ctx, u32 value)
+{
+	struct i915_address_space *vm = ctx_vm(ctx);
+	struct page *page = __px_page(vm->scratch[0]);
+	u32 *vaddr;
+
+	vaddr = kmap(page);
+	memset32(vaddr, value, PAGE_SIZE / sizeof(value));
+	kunmap(page);
+}
+
 static int igt_vm_isolation(void *arg)
 {
 	struct drm_i915_private *i915 = arg;

@@ -1816,11 +1827,11 @@ static int igt_vm_isolation(void *arg)
 		goto out_file;
 
 	/* Read the initial state of the scratch page */
-	err = check_scratch_page(ctx_a, &expected);
+	err = check_ctx_scratch(ctx_a, &expected);
 	if (err)
 		goto out_file;
 
-	err = check_scratch_page(ctx_b, &expected);
+	err = check_ctx_scratch(ctx_b, &expected);
 	if (err)
 		goto out_file;
 

@@ -1876,6 +1887,8 @@ static int igt_vm_isolation(void *arg)
 		count, num_engines);
 
 out_file:
+	/* As we deliberately write into scratch, cover up our tracks */
+	reset_ctx_scratch(ctx_a, expected);
 	if (igt_live_test_end(&t))
 		err = -EIO;
 	fput(file);

@@ -1265,35 +1265,6 @@ bool intel_engine_can_store_dword(struct intel_engine_cs *engine)
 	}
 }
 
-static void hexdump(struct drm_printer *m, const void *buf, size_t len)
-{
-	const size_t rowsize = 8 * sizeof(u32);
-	const void *prev = NULL;
-	bool skip = false;
-	size_t pos;
-
-	for (pos = 0; pos < len; pos += rowsize) {
-		char line[128];
-
-		if (prev && !memcmp(prev, buf + pos, rowsize)) {
-			if (!skip) {
-				drm_printf(m, "*\n");
-				skip = true;
-			}
-			continue;
-		}
-
-		WARN_ON_ONCE(hex_dump_to_buffer(buf + pos, len - pos,
-						rowsize, sizeof(u32),
-						line, sizeof(line),
-						false) >= sizeof(line));
-		drm_printf(m, "[%04zx] %s\n", pos, line);
-
-		prev = buf + pos;
-		skip = false;
-	}
-}
-
 static void intel_engine_print_registers(struct intel_engine_cs *engine,
 					 struct drm_printer *m)
 {

@@ -1450,7 +1421,7 @@ void intel_engine_dump(struct intel_engine_cs *engine,
 	}
 
 	drm_printf(m, "HWSP:\n");
-	hexdump(m, engine->status_page.addr, PAGE_SIZE);
+	i915_hexdump(m, engine->status_page.addr, PAGE_SIZE);
 
 	drm_printf(m, "Idle? %s\n", yesno(intel_engine_is_idle(engine)));
 

@@ -11,6 +11,7 @@
 #include "intel_context.h"
 #include "intel_engine_pm.h"
 #include "intel_gt.h"
+#include "intel_gtt.h"
 #include "intel_gt_clock_utils.h"
 #include "intel_gt_pm.h"
 #include "intel_gt_requests.h"

@@ -100,6 +101,9 @@ static int __gt_park(struct intel_wakeref *wf)
 	runtime_end(gt);
 	intel_gt_park_requests(gt);
 
+	if (IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))
+		check_scratch_page(gt->vm);
+
 	i915_vma_parked(gt);
 	i915_pmu_gt_parked(i915);
 	intel_rps_park(&gt->rps);

@@ -158,10 +158,49 @@ static void poison_scratch_page(struct drm_i915_gem_object *scratch)
 
 		vaddr = kmap(page);
 		memset(vaddr, val, PAGE_SIZE);
+		set_page_dirty(page); /* keep the poisoned contents */
 		kunmap(page);
 	}
 }
 
+void check_scratch_page(const struct i915_address_space *vm)
+{
+	struct drm_i915_gem_object *scratch;
+	struct sgt_iter sgt;
+	struct page *page;
+	void *vaddr;
+	u8 val;
+
+	scratch = vm->scratch[0];
+	if (!scratch || !i915_gem_object_has_struct_page(scratch))
+		return;
+
+	val = 0;
+	if (IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))
+		val = POISON_FREE;
+
+	for_each_sgt_page(page, sgt, scratch->mm.pages) {
+		vaddr = kmap(page);
+		drm_clflush_virt_range(vaddr, PAGE_SIZE);
+		if (memchr_inv(vaddr, val, PAGE_SIZE)) {
+			struct drm_printer p = drm_err_printer(__func__);
+
+			drm_err(&vm->i915->drm,
+				"%s scratch page overwitten!\n",
+				i915_is_ggtt(vm) ? "Global" : "Per-process");
+			i915_hexdump(&p, vaddr, PAGE_SIZE);
+			vaddr = NULL;
+		}
+		kunmap(page);
+		if (!vaddr)
+			break;
+	}
+
+	/* Restore the poison, so fresh errors will be detected */
+	if (!vaddr)
+		poison_scratch_page(scratch);
+}
+
 int setup_scratch_page(struct i915_address_space *vm)
 {
 	unsigned long size;

@@ -229,6 +268,9 @@ void free_scratch(struct i915_address_space *vm)
 {
 	int i;
 
+	if (IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))
+		check_scratch_page(vm);
+
 	for (i = 0; i <= vm->top; i++)
 		i915_gem_object_put(vm->scratch[i]);
 }

@@ -519,6 +519,7 @@ fill_page_dma(struct drm_i915_gem_object *p, const u64 val, unsigned int count);
 } while (0)
 
 int setup_scratch_page(struct i915_address_space *vm);
+void check_scratch_page(const struct i915_address_space *vm);
 void free_scratch(struct i915_address_space *vm);
 
 struct drm_i915_gem_object *alloc_pt_dma(struct i915_address_space *vm, int sz);

@@ -1095,35 +1095,6 @@ void i915_request_show_with_schedule(struct drm_printer *m,
 	rcu_read_unlock();
 }
 
-static void hexdump(struct drm_printer *m, const void *buf, size_t len)
-{
-	const size_t rowsize = 8 * sizeof(u32);
-	const void *prev = NULL;
-	bool skip = false;
-	size_t pos;
-
-	for (pos = 0; pos < len; pos += rowsize) {
-		char line[128];
-
-		if (prev && !memcmp(prev, buf + pos, rowsize)) {
-			if (!skip) {
-				drm_printf(m, "*\n");
-				skip = true;
-			}
-			continue;
-		}
-
-		WARN_ON_ONCE(hex_dump_to_buffer(buf + pos, len - pos,
-						rowsize, sizeof(u32),
-						line, sizeof(line),
-						false) >= sizeof(line));
-		drm_printf(m, "[%04zx] %s\n", pos, line);
-
-		prev = buf + pos;
-		skip = false;
-	}
-}
-
 static void
 print_request_ring(struct drm_printer *m, const struct i915_request *rq)
 {

@@ -1153,7 +1124,7 @@ print_request_ring(struct drm_printer *m, const struct i915_request *rq)
 		}
 		memcpy(ring + len, vaddr + head, size - len);
 
-		hexdump(m, ring, size);
+		i915_hexdump(m, ring, size);
 		kfree(ring);
 	}
 }

@@ -1195,7 +1166,7 @@ void i915_sched_show(struct drm_printer *m,
 
 		if (rq->context->lrc_reg_state) {
 			drm_printf(m, "Logical Ring Context:\n");
-			hexdump(m, rq->context->lrc_reg_state, PAGE_SIZE);
+			i915_hexdump(m, rq->context->lrc_reg_state, PAGE_SIZE);
 		}
 	}
 

@@ -49,6 +49,35 @@ __i915_printk(struct drm_i915_private *dev_priv, const char *level,
 	}
 }
 
+void i915_hexdump(struct drm_printer *m, const void *buf, size_t len)
+{
+	const size_t rowsize = 8 * sizeof(u32);
+	const void *prev = NULL;
+	bool skip = false;
+	size_t pos;
+
+	for (pos = 0; pos < len; pos += rowsize) {
+		char line[128];
+
+		if (prev && !memcmp(prev, buf + pos, rowsize)) {
+			if (!skip) {
+				drm_printf(m, "*\n");
+				skip = true;
+			}
+			continue;
+		}
+
+		WARN_ON_ONCE(hex_dump_to_buffer(buf + pos, len - pos,
+						rowsize, sizeof(u32),
+						line, sizeof(line),
+						false) >= sizeof(line));
+		drm_printf(m, "[%04zx] %s\n", pos, line);
+
+		prev = buf + pos;
+		skip = false;
+	}
+}
+
 void add_taint_for_CI(struct drm_i915_private *i915, unsigned int taint)
 {
 	__i915_printk(i915, KERN_NOTICE, "CI tainted:%#x by %pS\n",

@@ -32,6 +32,7 @@
 #include <linux/workqueue.h>
 
 struct drm_i915_private;
+struct drm_printer;
 struct timer_list;
 
 #define FDO_BUG_URL "https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs"

@@ -82,6 +83,8 @@ bool i915_error_injected(void);
 	__i915_printk(i915, i915_error_injected() ? KERN_DEBUG : KERN_ERR, \
 		      fmt, ##__VA_ARGS__)
 
+void i915_hexdump(struct drm_printer *m, const void *buf, size_t len);
+
 #if defined(GCC_VERSION) && GCC_VERSION >= 70000
 #define add_overflows_t(T, A, B) \
 	__builtin_add_overflow_p((A), (B), (T)0)