Skip to contentSkip to conversation

ACPI / APEI: fix the regression of synchronous external aborts occur in user-mode

PATCH Superseded Xiaofei Tan 3 participants 9 messages Dec 10, 2020 – Jun 8, 2021

Subsystems (2)

v5 of 7 Superseded Latest is v7 →

lore
Share & tools
lore
#0 lore raw

PATCH v5 1/1 ACPI / APEI: fix the regression of synchronous external aborts occur in user-mode

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +59 -17
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
 	gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-				       int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
 	unsigned long pfn;
-	int flags = -1;
-	int sec_sev = ghes_severity(gdata->error_severity);
-	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
 	if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
 		return false;
 
-	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return false;
-
-	pfn = mem_err->physical_addr >> PAGE_SHIFT;
+	pfn = PHYS_PFN(physical_addr);
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
-		mem_err->physical_addr);
+		physical_addr);
 		return false;
 	}
 
+	memory_failure_queue(pfn, flags);
+	return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+				       int sev)
+{
+	int flags = -1;
+	int sec_sev = ghes_severity(gdata->error_severity);
+	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+		return false;
+
 	/* iff following two events can be handled properly by now */
 	if (sec_sev == GHES_SEV_CORRECTED &&
 	    (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
 	if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
 		flags = 0;
 
-	if (flags != -1) {
-		memory_failure_queue(pfn, flags);
-		return true;
-	}
+	if (flags != -1)
+		return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
 	return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;
+		}
+
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;
+	}
+
+	return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
-			struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-			log_arm_hw_error(err);
+			queued = ghes_handle_arm_hw_error(gdata, sev);
 		} else {
 			void *err = acpi_hest_get_payload(gdata);
 
Signature · 1 line
2.8.1
author
#1 lore raw

@James
Hi James, please help to review this patch. Thank you very much. :)

Quoted 155 lines · Xiaofei Tan

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +59 -17
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
 	gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-				       int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
 	unsigned long pfn;
-	int flags = -1;
-	int sec_sev = ghes_severity(gdata->error_severity);
-	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
 	if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
 		return false;
 
-	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return false;
-
-	pfn = mem_err->physical_addr >> PAGE_SHIFT;
+	pfn = PHYS_PFN(physical_addr);
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
-		mem_err->physical_addr);
+		physical_addr);
 		return false;
 	}
 
+	memory_failure_queue(pfn, flags);
+	return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+				       int sev)
+{
+	int flags = -1;
+	int sec_sev = ghes_severity(gdata->error_severity);
+	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+		return false;
+
 	/* iff following two events can be handled properly by now */
 	if (sec_sev == GHES_SEV_CORRECTED &&
 	    (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
 	if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
 		flags = 0;
 
-	if (flags != -1) {
-		memory_failure_queue(pfn, flags);
-		return true;
-	}
+	if (flags != -1)
+		return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
 	return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;
+		}
+
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;
+	}
+
+	return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
-			struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-			log_arm_hw_error(err);
+			queued = ghes_handle_arm_hw_error(gdata, sev);
 		} else {
 			void *err = acpi_hest_get_payload(gdata);
 
#2 lore raw
Quoted 3 lines · tanxiaofei

@James
Hi James, please help to review this patch. Thank you very much. :)

James, Boris, any comments?

Quoted 157 lines · Xiaofei Tan
Quoted 155 lines

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +59 -17
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
      gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-                                    int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
      unsigned long pfn;
-     int flags = -1;
-     int sec_sev = ghes_severity(gdata->error_severity);
-     struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
      if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
              return false;
 
-     if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-             return false;
-
-     pfn = mem_err->physical_addr >> PAGE_SHIFT;
+     pfn = PHYS_PFN(physical_addr);
      if (!pfn_valid(pfn)) {
              pr_warn_ratelimited(FW_WARN GHES_PFX
              "Invalid address in generic error data: %#llx\n",
-             mem_err->physical_addr);
+             physical_addr);
              return false;
      }
 
+     memory_failure_queue(pfn, flags);
+     return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+                                    int sev)
+{
+     int flags = -1;
+     int sec_sev = ghes_severity(gdata->error_severity);
+     struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+     if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+             return false;
+
      /* iff following two events can be handled properly by now */
      if (sec_sev == GHES_SEV_CORRECTED &&
          (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
      if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
              flags = 0;
 
-     if (flags != -1) {
-             memory_failure_queue(pfn, flags);
-             return true;
-     }
+     if (flags != -1)
+             return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
      return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+     struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+     struct cper_arm_err_info *err_info;
+     bool queued = false;
+     int sec_sev, i;
+
+     log_arm_hw_error(err);
+
+     sec_sev = ghes_severity(gdata->error_severity);
+     if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+             return false;
+
+     err_info = (struct cper_arm_err_info *) (err + 1);
+     for (i = 0; i < err->err_info_num; i++, err_info++) {
+             bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+             bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+             /*
+              * The field (err_info->error_info & BIT(26)) is fixed to set to
+              * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+              * firmware won't mix corrected errors in an uncorrected section,
+              * and don't filter out 'corrected' error here.
+              */
+             if (!is_cache || !has_pa) {
+                     pr_warn_ratelimited(FW_WARN GHES_PFX
+                     "Unhandled processor error type %s\n",
+                     err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+                     cper_proc_error_type_strs[err_info->type] : "unknown error");
+                     continue;
+             }
+
+             if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+                     queued = true;
+     }
+
+     return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
                      ghes_handle_aer(gdata);
              }
              else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
-                     struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-                     log_arm_hw_error(err);
+                     queued = ghes_handle_arm_hw_error(gdata, sev);
              } else {
                      void *err = acpi_hest_get_payload(gdata);
 
#3 lore raw

Hi James, Boris

a friendly ping..
any comments for this ?

Quoted 167 lines · Rafael J. Wysocki
Quoted 3 lines · tanxiaofei

@James
Hi James, please help to review this patch. Thank you very much. :)

James, Boris, any comments?

Quoted 148 lines · Xiaofei Tan
Quoted 147 lines

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +58 -14
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
      gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-                                    int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
      unsigned long pfn;
-     int flags = -1;
-     int sec_sev = ghes_severity(gdata->error_severity);
-     struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
      if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
              return false;
 
-     if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-             return false;
-
-     pfn = mem_err->physical_addr >> PAGE_SHIFT;
+     pfn = PHYS_PFN(physical_addr);
      if (!pfn_valid(pfn)) {
              pr_warn_ratelimited(FW_WARN GHES_PFX
              "Invalid address in generic error data: %#llx\n",
-             mem_err->physical_addr);
+             physical_addr);
              return false;
      }
 
+     memory_failure_queue(pfn, flags);
+     return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+                                    int sev)
+{
+     int flags = -1;
+     int sec_sev = ghes_severity(gdata->error_severity);
+     struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+     if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+             return false;
+
      /* iff following two events can be handled properly by now */
      if (sec_sev == GHES_SEV_CORRECTED &&
          (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
      if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
              flags = 0;
 
-     if (flags != -1) {
-             memory_failure_queue(pfn, flags);
-             return true;
-     }
+     if (flags != -1)
+             return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
      return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+     struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+     struct cper_arm_err_info *err_info;
+     bool queued = false;
+     int sec_sev, i;
+
+     log_arm_hw_error(err);
+
+     sec_sev = ghes_severity(gdata->error_severity);
+     if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+             return false;
+
+     err_info = (struct cper_arm_err_info *) (err + 1);
+     for (i = 0; i < err->err_info_num; i++, err_info++) {
+             bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+             bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+             /*
+              * The field (err_info->error_info & BIT(26)) is fixed to set to
+              * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+              * firmware won't mix corrected errors in an uncorrected section,
+              * and don't filter out 'corrected' error here.
+              */
+             if (!is_cache || !has_pa) {
+                     pr_warn_ratelimited(FW_WARN GHES_PFX
+                     "Unhandled processor error type %s\n",
+                     err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+                     cper_proc_error_type_strs[err_info->type] : "unknown error");
+                     continue;
+             }
+
+             if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+                     queued = true;
+     }
+
+     return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
                      ghes_handle_aer(gdata);
              }
              else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
Quoted 9 lines

drivers/acpi/apei/ghes.c static bool ghes_do_proc(struct ghes *ghes, · line 655

Quoted 8 lines
drivers/acpi/apei/ghes.c +1 -3
-                     struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-                     log_arm_hw_error(err);
+                     queued = ghes_handle_arm_hw_error(gdata, sev);
              } else {
                      void *err = acpi_hest_get_payload(gdata);
 

.

#4 lore raw
In reply to Xiaofei Tan

ping..

Quoted 155 lines · Xiaofei Tan

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +59 -17
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
 	gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-				       int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
 	unsigned long pfn;
-	int flags = -1;
-	int sec_sev = ghes_severity(gdata->error_severity);
-	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
 	if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
 		return false;
 
-	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return false;
-
-	pfn = mem_err->physical_addr >> PAGE_SHIFT;
+	pfn = PHYS_PFN(physical_addr);
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
-		mem_err->physical_addr);
+		physical_addr);
 		return false;
 	}
 
+	memory_failure_queue(pfn, flags);
+	return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+				       int sev)
+{
+	int flags = -1;
+	int sec_sev = ghes_severity(gdata->error_severity);
+	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+		return false;
+
 	/* iff following two events can be handled properly by now */
 	if (sec_sev == GHES_SEV_CORRECTED &&
 	    (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
 	if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
 		flags = 0;
 
-	if (flags != -1) {
-		memory_failure_queue(pfn, flags);
-		return true;
-	}
+	if (flags != -1)
+		return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
 	return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;
+		}
+
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;
+	}
+
+	return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
-			struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-			log_arm_hw_error(err);
+			queued = ghes_handle_arm_hw_error(gdata, sev);
 		} else {
 			void *err = acpi_hest_get_payload(gdata);
 
#5 lore raw
In reply to Xiaofei Tan

ping..

Quoted 155 lines · Xiaofei Tan

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA). It is not enough. Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either cache or memory errors could generate SEA when consumed by a processor.

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

---
Changes since v4:
- 1. Change the patch name from " ACPI / APEI: do memory failure on the physical address reported by ARM processor error section" to this more proper one.
- 2. Add a comment in the code to tell why not filter out corrected error in an uncorrected section.

Changes since v3:
- Print unhandled error following James Morse's advice.

Changes since v2:
- Updated commit log

---
 drivers/acpi/apei/ghes.c | 76 +++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 59 insertions(+), 17 deletions(-)
drivers/acpi/apei/ghes.c +59 -17
index fce7ade..0893968 100644
@@ -441,28 +441,35 @@ static void ghes_kick_task_work(struct callback_head *head)
 	gen_pool_free(ghes_estatus_pool, (unsigned long)estatus_node, node_len);
 }
 
-static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
-				       int sev)
+static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 {
 	unsigned long pfn;
-	int flags = -1;
-	int sec_sev = ghes_severity(gdata->error_severity);
-	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
 	if (!IS_ENABLED(CONFIG_ACPI_APEI_MEMORY_FAILURE))
 		return false;
 
-	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return false;
-
-	pfn = mem_err->physical_addr >> PAGE_SHIFT;
+	pfn = PHYS_PFN(physical_addr);
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
-		mem_err->physical_addr);
+		physical_addr);
 		return false;
 	}
 
+	memory_failure_queue(pfn, flags);
+	return true;
+}
+
+static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+				       int sev)
+{
+	int flags = -1;
+	int sec_sev = ghes_severity(gdata->error_severity);
+	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+
+	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
+		return false;
+
 	/* iff following two events can be handled properly by now */
 	if (sec_sev == GHES_SEV_CORRECTED &&
 	    (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED))
@@ -470,14 +477,51 @@ static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
 	if (sev == GHES_SEV_RECOVERABLE && sec_sev == GHES_SEV_RECOVERABLE)
 		flags = 0;
 
-	if (flags != -1) {
-		memory_failure_queue(pfn, flags);
-		return true;
-	}
+	if (flags != -1)
+		return ghes_do_memory_failure(mem_err->physical_addr, flags);
 
 	return false;
 }
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
+
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;
+		}
+
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;
+	}
+
+	return queued;
+}
+
 /*
  * PCIe AER errors need to be sent to the AER driver for reporting and
  * recovery. The GHES severities map to the following AER severities and
@@ -605,9 +649,7 @@ static bool ghes_do_proc(struct ghes *ghes,
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
-			struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
-
-			log_arm_hw_error(err);
+			queued = ghes_handle_arm_hw_error(gdata, sev);
 		} else {
 			void *err = acpi_hest_get_payload(gdata);
 
Reviewed-by on v5 1/1
#6 lore raw
In reply to Xiaofei Tan

Hi Xiaofei Tan,

Sorry for the delayed response,
this still applies and builds to v5.13-rc4.

Quoted 5 lines · Xiaofei Tan

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

If you describe it the other way round, it would be clearer what the problem here is. Something like:

| Before commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise
| with APEI's irq work"), do_sea() would unconditionally signal the affected task
| from the arch code. Since that change, the GHES driver sends the signals,.
| This exposes a problem as errors the GHES driver doesn't understand are silently
| ignored.
Quoted 2 lines

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA).

(you're starting to confuse me! - I went and checked before I realised you were talking to me, not describing the code...)

Quoted 7 lines

It is not enough. > Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to
use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce
compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either
cache or memory errors could generate SEA when consumed by a processor.

I think you just need to say:
| Existing firmware on Kunpeng9xx systems reports cache errors with the 'ARM Processor | Error' CPER records.

Could you add something about why the silent-ignore is a problem? Do the errors get taken again? Does user-space get stuck in this loop?

Quoted 13 lines

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

drivers/acpi/apei/ghes.c +5 -0
index fce7ade..0893968 100644
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 500

Quoted 10 lines
drivers/acpi/apei/ghes.c +10 -0
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {

err_info has a version and a length, so its expected to be made bigger at some point. It would be better to use the length instead of 'err_info++', or at least to break out of the loop if a length > sizeof(*err_info) is seen.

With that:

The following nits would make this easier to read:

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 509

Quoted 9 lines
drivers/acpi/apei/ghes.c +8 -0
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
 
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */

(Nothing reads err_info->error_info, I guess this is a warning to the next person to touch this)

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 515

Quoted 6 lines
drivers/acpi/apei/ghes.c +6 -0
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;

This is hard to read. The convention is to indent the extra lines to the relevant '('. e.g.:
| pr_warn_ratelimited(FW_WARN GHES_PFX
| "Unhandled processor error type %s\n",

You could make it shorter by working out the error_type string earlier e.g.:

|		char *error_type = "unknown_error";
|			
|		if  (err_info->type < ARRAY_SIZE(cper_proc_error_type_strs)
|			error_type = cper_proc_error_type_strs[err_info->type];

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 519

Quoted 4 lines
drivers/acpi/apei/ghes.c +3 -0
+		}
 
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;

| if (it_returned_true())
| queued = true;

Looks funny, and if you moved this earlier, your pr_warn_ratelimted() would have an extra level of indentation to play with.
i.e.:

|		if (is_cache && has_pa) {
|			queued = ghes_do_memory_failure(err_info->physical_fault_addr, 0);
|			continue;
|		}

Thanks,

James

#7 lore raw

Hi James,

Quoted 3 lines · James Morse

Hi Xiaofei Tan,

Sorry for the delayed response,

It's okay.

Quoted 1 line

this still applies and builds to v5.13-rc4.

Yes, that's why i hadn't resend it.

Quoted 8 lines · Xiaofei Tan
Quoted 5 lines

After the commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise with APEI's irq work") applied, do_sea() return directly for user-mode if apei_claim_sea() handled any error record. Therefore, each error record reported by the user-mode SEA must be effectively processed in APEI GHES driver.

Quoted 8 lines

If you describe it the other way round, it would be clearer what the problem here is. Something like:

| Before commit 8fcc4ae6faf8 ("arm64: acpi: Make apei_claim_sea() synchronise
| with APEI's irq work"), do_sea() would unconditionally signal the affected task
| from the arch code. Since that change, the GHES driver sends the signals,.
| This exposes a problem as errors the GHES driver doesn't understand are silently
| ignored.

Yes, it's clearer. Only one little change to the last sentence. Change to:
This exposes a problem as errors the GHES driver doesn't understand or doesn't handle effectively are silently ignored.

Quoted 15 lines
Quoted 3 lines

Currently, GHES driver only processes Memory Error Section.(Ignore PCIe Error Section, as it has nothing to do with SEA).

(you're starting to confuse me! - I went and checked before I realised you were talking to me, not describing the code...)

Quoted 7 lines

It is not enough. > Because ARM Processor Error could also be used for SEA in some hardware platforms, such as Kunpeng9xx series. We can't ask them to switch to
use Memory Error Section for two reasons:
1)The server was delivered to customers, and it will introduce
compatibility issue.
2)It make sense to use ARM Processor Error Section. Because either
cache or memory errors could generate SEA when consumed by a processor.

Quoted 4 lines

I think you just need to say:
| Existing firmware on Kunpeng9xx systems reports cache errors with the 'ARM Processor | Error' CPER records.

OK

Quoted 2 lines

Could you add something about why the silent-ignore is a problem?

OK

Quoted 1 line

Do the errors get taken again?Does user-space get stuck in this loop?

Yes, the process will repeat endlessly.

Quoted 19 lines
Quoted 19 lines

Do memory failure handling for ARM Processor Error Section just like for Memory Error Section.

drivers/acpi/apei/ghes.c +9 -0
index fce7ade..0893968 100644
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+	struct cper_arm_err_info *err_info;
+	bool queued = false;
+	int sec_sev, i;
+
+	log_arm_hw_error(err);
+
Quoted 8 lines

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 500

Quoted 6 lines
drivers/acpi/apei/ghes.c +6 -0
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+		return false;
+
+	err_info = (struct cper_arm_err_info *) (err + 1);
+	for (i = 0; i < err->err_info_num; i++, err_info++) {

err_info has a version and a length, so its expected to be made bigger at some point.

Yes, but the table "ARM Processor Error Section" fixed the length of processor error information structure to fixed 32 bytes. Then the description of the UEFI spec need to update.

Quoted 3 lines

It would be better to use the length instead of 'err_info++', or at least to break out of the loop if a length > sizeof(*err_info) is seen.

OK. Maybe check length != sizeof(*err_info) is better. I will add this.

	if (err_info->length != sizeof(struct cper_arm_err_info)) {
		pr_warn_ratelimited(FW_WARN GHES_PFX
				    "Error info length %d is invalid\n",
				    err_info->length);
		break;
	}
Quoted 9 lines

With that:

The following nits would make this easier to read:

Quoted 2 lines
drivers/acpi/apei/ghes.c +2 -0
+		bool is_cache = (err_info->type == CPER_ARM_CACHE_ERROR);
+		bool has_pa = (err_info->validation_bits & CPER_ARM_INFO_VALID_PHYSICAL_ADDR);
Quoted 9 lines

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 509

Quoted 6 lines
drivers/acpi/apei/ghes.c +6 -0
+		/*
+		 * The field (err_info->error_info & BIT(26)) is fixed to set to
+		 * 1 in some old firmware of HiSilicon Kunpeng920. We assume that
+		 * firmware won't mix corrected errors in an uncorrected section,
+		 * and don't filter out 'corrected' error here.
+		 */

(Nothing reads err_info->error_info, I guess this is a warning to the next person to touch this)

Yes

Quoted 13 lines

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 515

Quoted 7 lines
drivers/acpi/apei/ghes.c +6 -0
+		if (!is_cache || !has_pa) {
+			pr_warn_ratelimited(FW_WARN GHES_PFX
+			"Unhandled processor error type %s\n",
+			err_info->type < ARRAY_SIZE(cper_proc_error_type_strs) ?
+			cper_proc_error_type_strs[err_info->type] : "unknown error");
+			continue;

This is hard to read. The convention is to indent the extra lines to the relevant '('. e.g.:
| pr_warn_ratelimited(FW_WARN GHES_PFX
| "Unhandled processor error type %s\n",

Quoted 7 lines

You could make it shorter by working out the error_type string earlier e.g.:

|		char *error_type = "unknown_error";
|			
|		if  (err_info->type < ARRAY_SIZE(cper_proc_error_type_strs)
|			error_type = cper_proc_error_type_strs[err_info->type];

OK, I will do this change.

Quoted 9 lines

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 519

Quoted 5 lines
drivers/acpi/apei/ghes.c +3 -0
+		}
 
+		if (ghes_do_memory_failure(err_info->physical_fault_addr, 0))
+			queued = true;

| if (it_returned_true())
| queued = true;

Quoted 8 lines

Looks funny, and if you moved this earlier, your pr_warn_ratelimted() would have an extra level of indentation to play with.
i.e.:

|		if (is_cache && has_pa) {
|			queued = ghes_do_memory_failure(err_info->physical_fault_addr, 0);
|			continue;
|		}

Right.

Quoted 7 lines

Thanks,

James

.

#8 lore raw

Hi James,

Quoted 3 lines · Xiaofei Tan

Hi James,

On 2021/6/4 22:19, James Morse wrote:

...

Quoted 32 lines
Quoted 14 lines
Quoted 14 lines
drivers/acpi/apei/ghes.c +9 -0
index fce7ade..0893968 100644
 
+static bool ghes_handle_arm_hw_error(struct acpi_hest_generic_data *gdata, int sev)
+{
+    struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+    struct cper_arm_err_info *err_info;
+    bool queued = false;
+    int sec_sev, i;
+
+    log_arm_hw_error(err);
+
Quoted 8 lines

drivers/acpi/apei/ghes.c static bool ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, · line 500

Quoted 6 lines
drivers/acpi/apei/ghes.c +6 -0
+    sec_sev = ghes_severity(gdata->error_severity);
+    if (sev != GHES_SEV_RECOVERABLE || sec_sev != GHES_SEV_RECOVERABLE)
+        return false;
+
+    err_info = (struct cper_arm_err_info *) (err + 1);
+    for (i = 0; i < err->err_info_num; i++, err_info++) {

err_info has a version and a length, so its expected to be made bigger at some point.

Yes, but the table "ARM Processor Error Section" fixed the length of processor error information structure to fixed 32 bytes. Then the description of the UEFI spec need to update.

Quoted 3 lines

It would be better to use the length instead of 'err_info++', or at least to break out of the loop if a length > sizeof(*err_info) is seen.

Quoted 9 lines

OK. Maybe check length != sizeof(*err_info) is better. I will add this.

    if (err_info->length != sizeof(struct cper_arm_err_info)) {
        pr_warn_ratelimited(FW_WARN GHES_PFX
                    "Error info length %d is invalid\n",
                    err_info->length);
        break;
    }

I considered this more carefully. It should be better to use the length instead of 'err_info', than just break out. Because if we want to expand "ARM Processor Error Infomation Structure", the proper way is just add some new member at the end of the structure. Then we have a chance to work well for the situation of new firmware + old kernel.

Quoted 24 lines
Quoted 10 lines

With that:

The following nits would make this easier to read:

Right.

Quoted 7 lines

Thanks,

James

.

.