drm/amdgpu: Update usage for bad page threshold

The driver's behavior varies based on
the configuration of amdgpu_bad_page_threshold setting

Signed-off-by: Hawking Zhang <[email protected]>
Reviewed-by: Tao Zhou <[email protected]>

Author: Hawking Zhang

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -980,7 +980,7 @@ module_param_named(reset_method, amdgpu_reset_method, int, 0644);
  * result in the GPU entering bad status when the number of total
  * faulty pages by ECC exceeds the threshold value.
  */
-MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = driver sets threshold)");
+MODULE_PARM_DESC(bad_page_threshold, "Bad page threshold(-1 = ignore threshold (default value), 0 = disable bad page retirement, -2 = threshold determined by a formula, 0 < threshold < max records, user-defined threshold)");
 module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444);
 
 MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)");

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c

@@ -2882,31 +2882,29 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev,
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
 	/*
-	 * Justification of value bad_page_cnt_threshold in ras structure
-	 *
-	 * Generally, 0 <= amdgpu_bad_page_threshold <= max record length
-	 * in eeprom or amdgpu_bad_page_threshold == -2, introduce two
-	 * scenarios accordingly.
-	 *
-	 * Bad page retirement enablement:
-	 *    - If amdgpu_bad_page_threshold = -2,
-	 *      bad_page_cnt_threshold = typical value by formula.
-	 *
-	 *    - When the value from user is 0 < amdgpu_bad_page_threshold <
-	 *      max record length in eeprom, use it directly.
-	 *
-	 * Bad page retirement disablement:
-	 *    - If amdgpu_bad_page_threshold = 0, bad page retirement
-	 *      functionality is disabled, and bad_page_cnt_threshold will
-	 *      take no effect.
+	 * amdgpu_bad_page_threshold is used to config
+	 * the threshold for the number of bad pages.
+	 * -1:  Threshold is set to default value
+	 *      Driver will issue a warning message when threshold is reached
+	 *      and continue runtime services.
+	 * 0:   Disable bad page retirement
+	 *      Driver will not retire bad pages
+	 *      which is intended for debugging purpose.
+	 * -2:  Threshold is determined by a formula
+	 *      that assumes 1 bad page per 100M of local memory.
+	 *      Driver will continue runtime services when threhold is reached.
+	 * 0 < threshold < max number of bad page records in EEPROM,
+	 *      A user-defined threshold is set
+	 *      Driver will halt runtime services when this custom threshold is reached.
 	 */
-
-	if (amdgpu_bad_page_threshold < 0) {
+	if (amdgpu_bad_page_threshold == -2) {
 		u64 val = adev->gmc.mc_vram_size;
 
 		do_div(val, RAS_BAD_PAGE_COVER);
 		con->bad_page_cnt_threshold = min(lower_32_bits(val),
 						  max_count);
+	} else if (amdgpu_bad_page_threshold == -1) {
+		con->bad_page_cnt_threshold = ((con->reserved_pages_in_bytes) >> 21) << 4;
 	} else {
 		con->bad_page_cnt_threshold = min_t(int, max_count,
 						    amdgpu_bad_page_threshold);
@@ -3654,8 +3652,10 @@ static void amdgpu_ras_init_reserved_vram_size(struct amdgpu_device *adev)
 	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
 	case IP_VERSION(13, 0, 2):
 	case IP_VERSION(13, 0, 6):
+		con->reserved_pages_in_bytes = AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT;
+		break;
 	case IP_VERSION(13, 0, 14):
-		con->reserved_pages_in_bytes = AMDGPU_RAS_RESERVED_VRAM_SIZE;
+		con->reserved_pages_in_bytes = (AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT << 1);
 		break;
 	default:
 		break;

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h

@@ -65,7 +65,7 @@ struct amdgpu_iv_entry;
 
 /* Reserve 8 physical dram row for possible retirement.
  * In worst cases, it will lose 8 * 2MB memory in vram domain */
-#define AMDGPU_RAS_RESERVED_VRAM_SIZE	(16ULL << 20)
+#define AMDGPU_RAS_RESERVED_VRAM_SIZE_DEFAULT	(16ULL << 20)
 /* The high three bits indicates socketid */
 #define AMDGPU_RAS_GET_FEATURES(val)  ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK)
 

drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c

@@ -557,16 +557,16 @@ bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev)
 			return false;
 
 	if (con->eeprom_control.tbl_hdr.header == RAS_TABLE_HDR_BAD) {
-		if (amdgpu_bad_page_threshold == -1) {
+		if ((amdgpu_bad_page_threshold == -1) ||
+		    (amdgpu_bad_page_threshold == -2)) {
 			dev_warn(adev->dev, "RAS records:%d exceed threshold:%d",
 				con->eeprom_control.ras_num_recs, con->bad_page_cnt_threshold);
 			dev_warn(adev->dev,
-				"But GPU can be operated due to bad_page_threshold = -1.\n");
+				 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n");
 			return false;
 		} else {
-			dev_warn(adev->dev, "This GPU is in BAD status.");
-			dev_warn(adev->dev, "Please retire it or set a larger "
-				 "threshold value when reloading driver.\n");
+			dev_warn(adev->dev,
+				 "Please consider adjusting the customized threshold\n");
 			return true;
 		}
 	}
@@ -750,7 +750,8 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
 			control->tbl_rai.health_percent = 0;
 		}
 
-		if (amdgpu_bad_page_threshold != -1)
+		if ((amdgpu_bad_page_threshold != -1) &&
+		    (amdgpu_bad_page_threshold != -2))
 			ras->is_rma = true;
 
 		/* ignore the -ENOTSUPP return value */
@@ -1385,8 +1386,9 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 
 		res = __verify_ras_table_checksum(control);
 		if (res)
-			DRM_ERROR("RAS table incorrect checksum or error:%d\n",
-				  res);
+			dev_err(adev->dev,
+				"RAS table incorrect checksum or error:%d\n",
+				res);
 
 		/* Warn if we are at 90% of the threshold or above
 		 */
@@ -1404,8 +1406,9 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 
 		res = __verify_ras_table_checksum(control);
 		if (res)
-			DRM_ERROR("RAS Table incorrect checksum or error:%d\n",
-				  res);
+			dev_err(adev->dev,
+				"RAS Table incorrect checksum or error:%d\n",
+				res);
 		if (ras->bad_page_cnt_threshold > control->ras_num_recs) {
 			/* This means that, the threshold was increased since
 			 * the last time the system was booted, and now,
@@ -1421,17 +1424,18 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
 			res = amdgpu_ras_eeprom_correct_header_tag(control,
 								   RAS_TABLE_HDR_VAL);
 		} else {
-			dev_err(adev->dev, "RAS records:%d exceed threshold:%d",
-				control->ras_num_recs, ras->bad_page_cnt_threshold);
-			if (amdgpu_bad_page_threshold == -1) {
-				dev_warn(adev->dev, "GPU will be initialized due to bad_page_threshold = -1.");
+			dev_warn(adev->dev,
+				 "RAS records:%d exceed threshold:%d",
+				 control->ras_num_recs, ras->bad_page_cnt_threshold);
+			if ((amdgpu_bad_page_threshold == -1) ||
+			    (amdgpu_bad_page_threshold == -2)) {
 				res = 0;
+				dev_warn(adev->dev,
+					 "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n");
 			} else {
 				ras->is_rma = true;
-				dev_err(adev->dev,
-					"RAS records:%d exceed threshold:%d, "
-					"GPU will not be initialized. Replace this GPU or increase the threshold",
-					control->ras_num_recs, ras->bad_page_cnt_threshold);
+				dev_warn(adev->dev,
+					 "User defined threshold is set, runtime service will be halt when threshold is reached\n");
 			}
 		}
 	} else {