[PATCH 0/6] rasdaemon:add logging of HiSilicon HIP08 non-standard H/W errors and changes in the error decoding code

[PATCH 0/6] rasdaemon:add logging of HiSilicon HIP08 non-standard H/W errors and changes in the error decoding code

[Shiju Jose]

This patch set add few changes in the non-standard error decoding code and
logging for the HiSilicon HIP08 non-standard H/W errors.

Shiju Jose (6):
  rasdaemon:print non-standard error data if not decoded
  rasdaemon: rearrange HiSilicon HIP07 decoding function table
  rasdaemon: update iteration logic for the non-standard error decoding
    functions
  rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM
    format1
  rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM
    format2
  rasdaemon:add logging HiSilicon HIP08 PCIe local errors

 Makefile.am                |   2 +-
 non-standard-hisi_hip07.c  |  36 +-
 non-standard-hisi_hip08.c  | 855 +++++++++++++++++++++++++++++++++++++++++++++
 ras-non-standard-handler.c |  36 +-
 ras-non-standard-handler.h |   8 +-
 ras-record.c               |  30 +-
 ras-record.h               |  13 +
 7 files changed, 932 insertions(+), 48 deletions(-)
 create mode 100644 non-standard-hisi_hip08.c

-- 
1.9.1

[PATCH 1/6] rasdaemon:print non-standard error data if not decoded

[Shiju Jose]

This patch change printing non-standard error data
only if not decoded.

Suggested-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-non-standard-handler.c | 31 +++++++++++++++++--------------
 1 file changed, 17 insertions(+), 14 deletions(-)

diff --git a/ras-non-standard-handler.c b/ras-non-standard-handler.c
index 21e6a76..d343a2a 100644
--- a/ras-non-standard-handler.c
+++ b/ras-non-standard-handler.c
@@ -160,20 +160,6 @@ int ras_non_standard_event_handler(struct trace_seq *s,
 	ev.error = pevent_get_field_raw(s, event, "buf", record, &len, 1);
 	if(!ev.error)
 		return -1;
-	len = ev.length;
-	i = 0;
-	line_count = 0;
-	trace_seq_printf(s, " error:\n  %08x: ", i);
-	while(len >= 4) {
-		print_le_hex(s, ev.error, i);
-		i+=4;
-		len-=4;
-		if(++line_count == 4) {
-			trace_seq_printf(s, "\n  %08x: ", i);
-			line_count = 0;
-		} else
-			trace_seq_printf(s, " ");
-	}
 
 	for (count = 0; count < dec_tab_count && !dec_done; count++) {
 		dec_tab = ns_dec_tab[count];
@@ -187,6 +173,23 @@ int ras_non_standard_event_handler(struct trace_seq *s,
 		}
 	}
 
+	if (!dec_done) {
+		len = ev.length;
+		i = 0;
+		line_count = 0;
+		trace_seq_printf(s, " error:\n  %08x: ", i);
+		while (len >= 4) {
+			print_le_hex(s, ev.error, i);
+			i += 4;
+			len -= 4;
+			if (++line_count == 4) {
+				trace_seq_printf(s, "\n  %08x: ", i);
+				line_count = 0;
+			} else
+				trace_seq_printf(s, " ");
+		}
+	}
+
 	/* Insert data into the SGBD */
 #ifdef HAVE_SQLITE3
 	ras_store_non_standard_record(ras, &ev);
-- 
1.9.1

[PATCH 2/6] rasdaemon: rearrange HiSilicon HIP07 decoding function table

[Shiju Jose]

This patch rearranges the decoding function table for the
HiSilicon HIP07 non-standard errors.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 non-standard-hisi_hip07.c | 26 ++++++++++++--------------
 1 file changed, 12 insertions(+), 14 deletions(-)

diff --git a/non-standard-hisi_hip07.c b/non-standard-hisi_hip07.c
index 3e9dabd..19a5c47 100644
--- a/non-standard-hisi_hip07.c
+++ b/non-standard-hisi_hip07.c
@@ -24,20 +24,6 @@
 #define HISI_SAS_VALID_ERR_TYPE       BIT(2)
 #define HISI_SAS_VALID_AXI_ERR_INFO   BIT(3)
 
-static int decode_hip07_sas_error(struct trace_seq *s, const void *error);
-static int decode_hip07_hns_error(struct trace_seq *s, const void *error);
-
-struct ras_ns_dec_tab hisi_ns_dec_tab[] = {
-	{
-		.sec_type = "daffd8146eba4d8c8a91bc9bbf4aa301",
-		.decode = decode_hip07_sas_error,
-	},
-	{
-		.sec_type = "fbc2d923ea7a453dab132949f5af9e53",
-		.decode = decode_hip07_hns_error,
-	},
-};
-
 struct hisi_sas_err_sec {
 	uint64_t   val_bits;
 	uint64_t   physical_addr;
@@ -138,6 +124,18 @@ static int decode_hip07_hns_error(struct trace_seq *s, const void *error)
 {
 	return 0;
 }
+
+struct ras_ns_dec_tab hisi_ns_dec_tab[] = {
+	{
+		.sec_type = "daffd8146eba4d8c8a91bc9bbf4aa301",
+		.decode = decode_hip07_sas_error,
+	},
+	{
+		.sec_type = "fbc2d923ea7a453dab132949f5af9e53",
+		.decode = decode_hip07_hns_error,
+	},
+};
+
 __attribute__((constructor))
 static void hip07_init(void)
 {
-- 
1.9.1

[PATCH 3/6] rasdaemon: update iteration logic for the non-standard error decoding functions

[Shiju Jose]

This patch updates the iteration logic for the non-standard
error decoding functions.

Suggested-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 non-standard-hisi_hip07.c  | 2 +-
 ras-non-standard-handler.c | 2 +-
 ras-non-standard-handler.h | 1 -
 3 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/non-standard-hisi_hip07.c b/non-standard-hisi_hip07.c
index 19a5c47..bb2576e 100644
--- a/non-standard-hisi_hip07.c
+++ b/non-standard-hisi_hip07.c
@@ -134,11 +134,11 @@ struct ras_ns_dec_tab hisi_ns_dec_tab[] = {
 		.sec_type = "fbc2d923ea7a453dab132949f5af9e53",
 		.decode = decode_hip07_hns_error,
 	},
+	{ /* sentinel */ }
 };
 
 __attribute__((constructor))
 static void hip07_init(void)
 {
-	hisi_ns_dec_tab[0].len = ARRAY_SIZE(hisi_ns_dec_tab);
 	register_ns_dec_tab(hisi_ns_dec_tab);
 }
diff --git a/ras-non-standard-handler.c b/ras-non-standard-handler.c
index d343a2a..392bb27 100644
--- a/ras-non-standard-handler.c
+++ b/ras-non-standard-handler.c
@@ -163,7 +163,7 @@ int ras_non_standard_event_handler(struct trace_seq *s,
 
 	for (count = 0; count < dec_tab_count && !dec_done; count++) {
 		dec_tab = ns_dec_tab[count];
-		for (i = 0; i < dec_tab[0].len; i++) {
+		for (i = 0; dec_tab[i].decode; i++) {
 			if (uuid_le_cmp(ev.sec_type,
 					dec_tab[i].sec_type) == 0) {
 				dec_tab[i].decode(s, ev.error);
diff --git a/ras-non-standard-handler.h b/ras-non-standard-handler.h
index b9e9fb1..b2c9743 100644
--- a/ras-non-standard-handler.h
+++ b/ras-non-standard-handler.h
@@ -23,7 +23,6 @@
 typedef struct ras_ns_dec_tab {
 	const char *sec_type;
 	int (*decode)(struct trace_seq *s, const void *err);
-	size_t len;
 } *p_ns_dec_tab;
 
 int ras_non_standard_event_handler(struct trace_seq *s,
-- 
1.9.1

[PATCH 4/6] rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM format1

[Shiju Jose]

This patch adds logging the HiSilicon HIP08 H/W errors reported
in the non-standard OEM format1.
These errors are from the H/W modules MN, PLL, SLLC, AA, SIOE,
POE, DISP, LPC, SAS and SATA.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 Makefile.am                |   2 +-
 non-standard-hisi_hip07.c  |   8 +-
 non-standard-hisi_hip08.c  | 332 +++++++++++++++++++++++++++++++++++++++++++++
 ras-non-standard-handler.c |   3 +-
 ras-non-standard-handler.h |   7 +-
 ras-record.c               |  30 ++--
 ras-record.h               |  13 ++
 7 files changed, 378 insertions(+), 17 deletions(-)
 create mode 100644 non-standard-hisi_hip08.c

diff --git a/Makefile.am b/Makefile.am
index f036ffd..3d89672 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -49,7 +49,7 @@ if WITH_ABRT_REPORT
    rasdaemon_SOURCES += ras-report.c
 endif
 if WITH_HISI_NS_DECODE
-   rasdaemon_SOURCES += non-standard-hisi_hip07.c
+   rasdaemon_SOURCES += non-standard-hisi_hip07.c non-standard-hisi_hip08.c
 endif
 rasdaemon_LDADD = -lpthread $(SQLITE3_LIBS) libtrace/libtrace.a
 
diff --git a/non-standard-hisi_hip07.c b/non-standard-hisi_hip07.c
index bb2576e..7f58fb3 100644
--- a/non-standard-hisi_hip07.c
+++ b/non-standard-hisi_hip07.c
@@ -87,7 +87,9 @@ static char *sas_axi_err_type(int etype)
 	return "unknown error";
 }
 
-static int decode_hip07_sas_error(struct trace_seq *s, const void *error)
+static int decode_hip07_sas_error(struct ras_events *ras,
+				  struct ras_ns_dec_tab *dec_tab,
+				  struct trace_seq *s, const void *error)
 {
 	char buf[1024];
 	char *p = buf;
@@ -120,7 +122,9 @@ static int decode_hip07_sas_error(struct trace_seq *s, const void *error)
 	return 0;
 }
 
-static int decode_hip07_hns_error(struct trace_seq *s, const void *error)
+static int decode_hip07_hns_error(struct ras_events *ras,
+				  struct ras_ns_dec_tab *dec_tab,
+				  struct trace_seq *s, const void *error)
 {
 	return 0;
 }
diff --git a/non-standard-hisi_hip08.c b/non-standard-hisi_hip08.c
new file mode 100644
index 0000000..240e832
--- /dev/null
+++ b/non-standard-hisi_hip08.c
@@ -0,0 +1,332 @@
+/*
+ * Copyright (c) 2019 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "ras-record.h"
+#include "ras-logger.h"
+#include "ras-report.h"
+#include "ras-non-standard-handler.h"
+
+/* HISI OEM error definitions */
+/* HISI OEM format1 error definitions */
+#define HISI_OEM_MODULE_ID_MN	0
+#define HISI_OEM_MODULE_ID_PLL	1
+#define HISI_OEM_MODULE_ID_SLLC	2
+#define HISI_OEM_MODULE_ID_AA	3
+#define HISI_OEM_MODULE_ID_SIOE	4
+#define HISI_OEM_MODULE_ID_POE	5
+#define HISI_OEM_MODULE_ID_DISP	8
+#define HISI_OEM_MODULE_ID_LPC	9
+#define HISI_OEM_MODULE_ID_SAS	15
+#define HISI_OEM_MODULE_ID_SATA	16
+
+#define HISI_OEM_VALID_SOC_ID		BIT(0)
+#define HISI_OEM_VALID_SOCKET_ID	BIT(1)
+#define HISI_OEM_VALID_NIMBUS_ID	BIT(2)
+#define HISI_OEM_VALID_MODULE_ID	BIT(3)
+#define HISI_OEM_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_OEM_VALID_ERR_SEVERITY	BIT(5)
+
+#define HISI_OEM_TYPE1_VALID_ERR_MISC_0	BIT(6)
+#define HISI_OEM_TYPE1_VALID_ERR_MISC_1	BIT(7)
+#define HISI_OEM_TYPE1_VALID_ERR_MISC_2	BIT(8)
+#define HISI_OEM_TYPE1_VALID_ERR_MISC_3	BIT(9)
+#define HISI_OEM_TYPE1_VALID_ERR_MISC_4	BIT(10)
+#define HISI_OEM_TYPE1_VALID_ERR_ADDR	BIT(11)
+
+struct hisi_oem_type1_err_sec {
+	uint32_t   val_bits;
+	uint8_t    version;
+	uint8_t    soc_id;
+	uint8_t    socket_id;
+	uint8_t    nimbus_id;
+	uint8_t    module_id;
+	uint8_t    sub_module_id;
+	uint8_t    err_severity;
+	uint8_t    reserv;
+	uint32_t   err_misc_0;
+	uint32_t   err_misc_1;
+	uint32_t   err_misc_2;
+	uint32_t   err_misc_3;
+	uint32_t   err_misc_4;
+	uint64_t   err_addr;
+};
+
+enum hisi_oem_data_type {
+	hisi_oem_data_type_int,
+	hisi_oem_data_type_int64,
+	hisi_oem_data_type_text,
+};
+
+enum {
+	hip08_oem_type1_field_id,
+	hip08_oem_type1_field_version,
+	hip08_oem_type1_field_soc_id,
+	hip08_oem_type1_field_socket_id,
+	hip08_oem_type1_field_nimbus_id,
+	hip08_oem_type1_field_module_id,
+	hip08_oem_type1_field_sub_module_id,
+	hip08_oem_type1_field_err_sev,
+	hip08_oem_type1_field_err_misc_0,
+	hip08_oem_type1_field_err_misc_1,
+	hip08_oem_type1_field_err_misc_2,
+	hip08_oem_type1_field_err_misc_3,
+	hip08_oem_type1_field_err_misc_4,
+	hip08_oem_type1_field_err_addr,
+};
+
+/* helper functions */
+static char *err_severity(uint8_t err_sev)
+{
+	switch (err_sev) {
+	case 0: return "recoverable";
+	case 1: return "fatal";
+	case 2: return "corrected";
+	case 3: return "none";
+	}
+	return "unknown";
+}
+
+static char *oem_type1_module_name(uint8_t module_id)
+{
+	switch (module_id) {
+	case HISI_OEM_MODULE_ID_MN: return "MN";
+	case HISI_OEM_MODULE_ID_PLL: return "PLL";
+	case HISI_OEM_MODULE_ID_SLLC: return "SLLC";
+	case HISI_OEM_MODULE_ID_AA: return "AA";
+	case HISI_OEM_MODULE_ID_SIOE: return "SIOE";
+	case HISI_OEM_MODULE_ID_POE: return "POE";
+	case HISI_OEM_MODULE_ID_DISP: return "DISP";
+	case HISI_OEM_MODULE_ID_LPC: return "LPC";
+	case HISI_OEM_MODULE_ID_SAS: return "SAS";
+	case HISI_OEM_MODULE_ID_SATA: return "SATA";
+	}
+	return "unknown";
+}
+
+#ifdef HAVE_SQLITE3
+static const struct db_fields hip08_oem_type1_event_fields[] = {
+	{ .name = "id",			.type = "INTEGER PRIMARY KEY" },
+	{ .name = "version",		.type = "INTEGER" },
+	{ .name = "soc_id",		.type = "INTEGER" },
+	{ .name = "socket_id",		.type = "INTEGER" },
+	{ .name = "nimbus_id",		.type = "INTEGER" },
+	{ .name = "module_id",		.type = "TEXT" },
+	{ .name = "sub_module_id",	.type = "INTEGER" },
+	{ .name = "err_severity",	.type = "TEXT" },
+	{ .name = "err_misc_0",		.type = "INTEGER" },
+	{ .name = "err_misc_1",		.type = "INTEGER" },
+	{ .name = "err_misc_2",		.type = "INTEGER" },
+	{ .name = "err_misc_3",		.type = "INTEGER" },
+	{ .name = "err_misc_4",		.type = "INTEGER" },
+	{ .name = "err_addr",		.type = "INTEGER" },
+};
+
+static const struct db_table_descriptor hip08_oem_type1_event_tab = {
+	.name = "hip08_oem_type1_event",
+	.fields = hip08_oem_type1_event_fields,
+	.num_fields = ARRAY_SIZE(hip08_oem_type1_event_fields),
+};
+
+static void record_vendor_data(struct ras_ns_dec_tab *dec_tab,
+			       enum hisi_oem_data_type data_type,
+			       int id, int64_t data, const char *text)
+{
+	switch (data_type) {
+	case hisi_oem_data_type_int:
+		sqlite3_bind_int(dec_tab->stmt_dec_record, id, data);
+		break;
+	case hisi_oem_data_type_int64:
+		sqlite3_bind_int64(dec_tab->stmt_dec_record, id, data);
+		break;
+	case hisi_oem_data_type_text:
+		sqlite3_bind_text(dec_tab->stmt_dec_record, id, text, -1, NULL);
+		break;
+	default:
+		break;
+	}
+}
+
+static int step_vendor_data_tab(struct ras_ns_dec_tab *dec_tab, char *name)
+{
+	int rc;
+
+	rc = sqlite3_step(dec_tab->stmt_dec_record);
+	if (rc != SQLITE_OK && rc != SQLITE_DONE)
+		log(TERM, LOG_ERR,
+		    "Failed to do %s step on sqlite: error = %d\n", name, rc);
+
+	rc = sqlite3_reset(dec_tab->stmt_dec_record);
+	if (rc != SQLITE_OK && rc != SQLITE_DONE)
+		log(TERM, LOG_ERR,
+		    "Failed to reset %s on sqlite: error = %d\n", name, rc);
+
+	rc = sqlite3_clear_bindings(dec_tab->stmt_dec_record);
+	if (rc != SQLITE_OK && rc != SQLITE_DONE)
+		log(TERM, LOG_ERR,
+		    "Failed to clear bindings %s on sqlite: error = %d\n",
+		    name, rc);
+
+	return rc;
+}
+#else
+static void record_vendor_data(struct ras_ns_dec_tab *dec_tab,
+			       enum hisi_oem_data_type data_type,
+			       int id, int64_t data, const char *text)
+{ }
+
+static int step_vendor_data_tab(struct ras_ns_dec_tab *dec_tab, char *name)
+{
+	return 0;
+}
+#endif
+
+/* error data decoding functions */
+static int decode_hip08_oem_type1_error(struct ras_events *ras,
+					struct ras_ns_dec_tab *dec_tab,
+					struct trace_seq *s, const void *error)
+{
+	const struct hisi_oem_type1_err_sec *err = error;
+	char buf[1024];
+	char *p = buf;
+
+	if (err->val_bits == 0) {
+		trace_seq_printf(s, "%s: no valid error information\n",
+				 __func__);
+		return -1;
+	}
+
+#ifdef HAVE_SQLITE3
+	if (!dec_tab->stmt_dec_record) {
+		if (ras_mc_add_vendor_table(ras, &dec_tab->stmt_dec_record,
+					    &hip08_oem_type1_event_tab)
+			!= SQLITE_OK) {
+			trace_seq_printf(s,
+					"create sql hip08_oem_type1_event_tab fail\n");
+			return -1;
+		}
+	}
+#endif
+
+	p += sprintf(p, "[ ");
+	p += sprintf(p, "Table version=%d ", err->version);
+	record_vendor_data(dec_tab, hisi_oem_data_type_int,
+			   hip08_oem_type1_field_version, err->version, NULL);
+
+	if (err->val_bits & HISI_OEM_VALID_SOC_ID) {
+		p += sprintf(p, "SOC ID=%d ", err->soc_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_soc_id,
+				   err->soc_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_SOCKET_ID) {
+		p += sprintf(p, "socket ID=%d ", err->socket_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_socket_id,
+				   err->socket_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_NIMBUS_ID) {
+		p += sprintf(p, "nimbus ID=%d ", err->nimbus_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_nimbus_id,
+				   err->nimbus_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_MODULE_ID) {
+		p += sprintf(p, "module=%s-",
+			     oem_type1_module_name(err->module_id));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_oem_type1_field_module_id,
+				   0, oem_type1_module_name(err->module_id));
+		if (err->val_bits & HISI_OEM_VALID_SUB_MODULE_ID) {
+			p += sprintf(p, "%d ", err->sub_module_id);
+			record_vendor_data(dec_tab, hisi_oem_data_type_int,
+					   hip08_oem_type1_field_sub_module_id,
+					   err->sub_module_id, NULL);
+		}
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_ERR_SEVERITY) {
+		p += sprintf(p, "error severity=%s ",
+			     err_severity(err->err_severity));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_oem_type1_field_err_sev,
+				   0, err_severity(err->err_severity));
+	}
+
+	p += sprintf(p, "]");
+	trace_seq_printf(s, "\nHISI HIP08: OEM Type-1 Error\n");
+	trace_seq_printf(s, "%s\n", buf);
+
+	trace_seq_printf(s, "Reg Dump:\n");
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_MISC_0) {
+		trace_seq_printf(s, "ERR_MISC0=0x%x\n", err->err_misc_0);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_err_misc_0,
+				   err->err_misc_0, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_MISC_1) {
+		trace_seq_printf(s, "ERR_MISC1=0x%x\n", err->err_misc_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_err_misc_1,
+				   err->err_misc_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_MISC_2) {
+		trace_seq_printf(s, "ERR_MISC2=0x%x\n", err->err_misc_2);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_err_misc_2,
+				   err->err_misc_2, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_MISC_3) {
+		trace_seq_printf(s, "ERR_MISC3=0x%x\n", err->err_misc_3);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_err_misc_3,
+				   err->err_misc_3, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_MISC_4) {
+		trace_seq_printf(s, "ERR_MISC4=0x%x\n", err->err_misc_4);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type1_field_err_misc_4,
+				   err->err_misc_4, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE1_VALID_ERR_ADDR) {
+		trace_seq_printf(s, "ERR_ADDR=0x%p\n", (void *)err->err_addr);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int64,
+				   hip08_oem_type1_field_err_addr,
+				   err->err_addr, NULL);
+	}
+
+	step_vendor_data_tab(dec_tab, "hip08_oem_type1_event_tab");
+
+	return 0;
+}
+
+struct ras_ns_dec_tab hip08_ns_oem_tab[] = {
+	{
+		.sec_type = "1f8161e155d641e6bd107afd1dc5f7c5",
+		.decode = decode_hip08_oem_type1_error,
+	},
+	{ /* sentinel */ }
+};
+
+__attribute__((constructor))
+static void hip08_init(void)
+{
+	register_ns_dec_tab(hip08_ns_oem_tab);
+}
diff --git a/ras-non-standard-handler.c b/ras-non-standard-handler.c
index 392bb27..4eda80b 100644
--- a/ras-non-standard-handler.c
+++ b/ras-non-standard-handler.c
@@ -166,7 +166,8 @@ int ras_non_standard_event_handler(struct trace_seq *s,
 		for (i = 0; dec_tab[i].decode; i++) {
 			if (uuid_le_cmp(ev.sec_type,
 					dec_tab[i].sec_type) == 0) {
-				dec_tab[i].decode(s, ev.error);
+				dec_tab[i].decode(ras, &dec_tab[i],
+						  s, ev.error);
 				dec_done = true;
 				break;
 			}
diff --git a/ras-non-standard-handler.h b/ras-non-standard-handler.h
index b2c9743..a7e48a3 100644
--- a/ras-non-standard-handler.h
+++ b/ras-non-standard-handler.h
@@ -22,7 +22,12 @@
 
 typedef struct ras_ns_dec_tab {
 	const char *sec_type;
-	int (*decode)(struct trace_seq *s, const void *err);
+	int (*decode)(struct ras_events *ras, struct ras_ns_dec_tab *dec_tab,
+		      struct trace_seq *s, const void *err);
+#ifdef HAVE_SQLITE3
+#include <sqlite3.h>
+	sqlite3_stmt *stmt_dec_record;
+#endif
 } *p_ns_dec_tab;
 
 int ras_non_standard_event_handler(struct trace_seq *s,
diff --git a/ras-record.c b/ras-record.c
index 4c8b55b..b212607 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -38,17 +38,6 @@
 
 #define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
 
-struct db_fields {
-	char *name;
-	char *type;
-};
-
-struct db_table_descriptor {
-	char			*name;
-	const struct db_fields	*fields;
-	size_t			num_fields;
-};
-
 /*
  * Table and functions to handle ras:mc_event
  */
@@ -511,7 +500,7 @@ static int ras_mc_create_table(struct sqlite3_priv *priv,
 {
 	const struct db_fields *field;
 	char sql[1024], *p = sql, *end = sql + sizeof(sql);
-	int i,rc;
+	int i, rc;
 
 	p += snprintf(p, end - p, "CREATE TABLE IF NOT EXISTS %s (",
 		      db_tab->name);
@@ -538,6 +527,23 @@ static int ras_mc_create_table(struct sqlite3_priv *priv,
 	return rc;
 }
 
+int ras_mc_add_vendor_table(struct ras_events *ras,
+			    sqlite3_stmt **stmt,
+			    const struct db_table_descriptor *db_tab)
+{
+	int rc;
+	struct sqlite3_priv *priv = ras->db_priv;
+
+	if (!priv)
+		return -1;
+
+	rc = ras_mc_create_table(priv, db_tab);
+	if (rc == SQLITE_OK)
+		rc = ras_mc_prepare_stmt(priv, stmt, db_tab);
+
+	return rc;
+}
+
 int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
 {
 	int rc;
diff --git a/ras-record.h b/ras-record.h
index 2183167..432a571 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -119,7 +119,20 @@ struct sqlite3_priv {
 #endif
 };
 
+struct db_fields {
+	char *name;
+	char *type;
+};
+
+struct db_table_descriptor {
+	char                    *name;
+	const struct db_fields  *fields;
+	size_t                  num_fields;
+};
+
 int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras);
+int ras_mc_add_vendor_table(struct ras_events *ras, sqlite3_stmt **stmt,
+			    const struct db_table_descriptor *db_tab);
 int ras_store_mc_event(struct ras_events *ras, struct ras_mc_event *ev);
 int ras_store_aer_event(struct ras_events *ras, struct ras_aer_event *ev);
 int ras_store_mce_record(struct ras_events *ras, struct mce_event *ev);
-- 
1.9.1

[PATCH 5/6] rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM format2

[Shiju Jose]

This patch adds logging the HiSilicon HIP08 H/W errors reported
in the non-standard OEM format2.
These errors are from the H/W modules SMMU, HHA, HLLC, PA and DDRC.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 non-standard-hisi_hip08.c | 300 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 300 insertions(+)

diff --git a/non-standard-hisi_hip08.c b/non-standard-hisi_hip08.c
index 240e832..6fe5cbe 100644
--- a/non-standard-hisi_hip08.c
+++ b/non-standard-hisi_hip08.c
@@ -43,6 +43,20 @@
 #define HISI_OEM_TYPE1_VALID_ERR_MISC_4	BIT(10)
 #define HISI_OEM_TYPE1_VALID_ERR_ADDR	BIT(11)
 
+/* HISI OEM format2 error definitions */
+#define HISI_OEM_MODULE_ID_SMMU	0
+#define HISI_OEM_MODULE_ID_HHA	1
+#define HISI_OEM_MODULE_ID_HLLC	2
+#define HISI_OEM_MODULE_ID_PA	3
+#define HISI_OEM_MODULE_ID_DDRC	4
+
+#define HISI_OEM_TYPE2_VALID_ERR_FR	BIT(6)
+#define HISI_OEM_TYPE2_VALID_ERR_CTRL	BIT(7)
+#define HISI_OEM_TYPE2_VALID_ERR_STATUS	BIT(8)
+#define HISI_OEM_TYPE2_VALID_ERR_ADDR	BIT(9)
+#define HISI_OEM_TYPE2_VALID_ERR_MISC_0	BIT(10)
+#define HISI_OEM_TYPE2_VALID_ERR_MISC_1	BIT(11)
+
 struct hisi_oem_type1_err_sec {
 	uint32_t   val_bits;
 	uint8_t    version;
@@ -61,6 +75,30 @@ struct hisi_oem_type1_err_sec {
 	uint64_t   err_addr;
 };
 
+struct hisi_oem_type2_err_sec {
+	uint32_t   val_bits;
+	uint8_t    version;
+	uint8_t    soc_id;
+	uint8_t    socket_id;
+	uint8_t    nimbus_id;
+	uint8_t    module_id;
+	uint8_t    sub_module_id;
+	uint8_t    err_severity;
+	uint8_t    reserv;
+	uint32_t   err_fr_0;
+	uint32_t   err_fr_1;
+	uint32_t   err_ctrl_0;
+	uint32_t   err_ctrl_1;
+	uint32_t   err_status_0;
+	uint32_t   err_status_1;
+	uint32_t   err_addr_0;
+	uint32_t   err_addr_1;
+	uint32_t   err_misc0_0;
+	uint32_t   err_misc0_1;
+	uint32_t   err_misc1_0;
+	uint32_t   err_misc1_1;
+};
+
 enum hisi_oem_data_type {
 	hisi_oem_data_type_int,
 	hisi_oem_data_type_int64,
@@ -84,6 +122,29 @@ enum {
 	hip08_oem_type1_field_err_addr,
 };
 
+enum {
+	hip08_oem_type2_field_id,
+	hip08_oem_type2_field_version,
+	hip08_oem_type2_field_soc_id,
+	hip08_oem_type2_field_socket_id,
+	hip08_oem_type2_field_nimbus_id,
+	hip08_oem_type2_field_module_id,
+	hip08_oem_type2_field_sub_module_id,
+	hip08_oem_type2_field_err_sev,
+	hip08_oem_type2_field_err_fr_0,
+	hip08_oem_type2_field_err_fr_1,
+	hip08_oem_type2_field_err_ctrl_0,
+	hip08_oem_type2_field_err_ctrl_1,
+	hip08_oem_type2_field_err_status_0,
+	hip08_oem_type2_field_err_status_1,
+	hip08_oem_type2_field_err_addr_0,
+	hip08_oem_type2_field_err_addr_1,
+	hip08_oem_type2_field_err_misc0_0,
+	hip08_oem_type2_field_err_misc0_1,
+	hip08_oem_type2_field_err_misc1_0,
+	hip08_oem_type2_field_err_misc1_1,
+};
+
 /* helper functions */
 static char *err_severity(uint8_t err_sev)
 {
@@ -113,6 +174,62 @@ static char *oem_type1_module_name(uint8_t module_id)
 	return "unknown";
 }
 
+static char *oem_type2_module_name(uint8_t module_id)
+{
+	switch (module_id) {
+	case HISI_OEM_MODULE_ID_SMMU: return "SMMU";
+	case HISI_OEM_MODULE_ID_HHA: return "HHA";
+	case HISI_OEM_MODULE_ID_HLLC: return "HLLC";
+	case HISI_OEM_MODULE_ID_PA: return "PA";
+	case HISI_OEM_MODULE_ID_DDRC: return "DDRC";
+	}
+	return "unknown module";
+}
+
+static char *oem_type2_sub_module_id(char *p, uint8_t module_id,
+				     uint8_t sub_module_id)
+{
+	switch (module_id) {
+	case HISI_OEM_MODULE_ID_SMMU:
+	case HISI_OEM_MODULE_ID_HLLC:
+	case HISI_OEM_MODULE_ID_PA:
+		p += sprintf(p, "%d ", sub_module_id);
+		break;
+
+	case HISI_OEM_MODULE_ID_HHA:
+		if (sub_module_id == 0)
+			p += sprintf(p, "TA HHA0 ");
+		else if (sub_module_id == 1)
+			p += sprintf(p, "TA HHA1 ");
+		else if (sub_module_id == 2)
+			p += sprintf(p, "TB HHA0 ");
+		else if (sub_module_id == 3)
+			p += sprintf(p, "TB HHA1 ");
+		break;
+
+	case HISI_OEM_MODULE_ID_DDRC:
+		if (sub_module_id == 0)
+			p += sprintf(p, "TA DDRC0 ");
+		else if (sub_module_id == 1)
+			p += sprintf(p, "TA DDRC1 ");
+		else if (sub_module_id == 2)
+			p += sprintf(p, "TA DDRC2 ");
+		else if (sub_module_id == 3)
+			p += sprintf(p, "TA DDRC3 ");
+		else if (sub_module_id == 4)
+			p += sprintf(p, "TB DDRC0 ");
+		else if (sub_module_id == 5)
+			p += sprintf(p, "TB DDRC1 ");
+		else if (sub_module_id == 6)
+			p += sprintf(p, "TB DDRC2 ");
+		else if (sub_module_id == 7)
+			p += sprintf(p, "TB DDRC3 ");
+		break;
+	}
+
+	return p;
+}
+
 #ifdef HAVE_SQLITE3
 static const struct db_fields hip08_oem_type1_event_fields[] = {
 	{ .name = "id",			.type = "INTEGER PRIMARY KEY" },
@@ -137,6 +254,35 @@ static const struct db_table_descriptor hip08_oem_type1_event_tab = {
 	.num_fields = ARRAY_SIZE(hip08_oem_type1_event_fields),
 };
 
+static const struct db_fields hip08_oem_type2_event_fields[] = {
+	{ .name = "id",                 .type = "INTEGER PRIMARY KEY" },
+	{ .name = "version",            .type = "INTEGER" },
+	{ .name = "soc_id",             .type = "INTEGER" },
+	{ .name = "socket_id",          .type = "INTEGER" },
+	{ .name = "nimbus_id",          .type = "INTEGER" },
+	{ .name = "module_id",          .type = "TEXT" },
+	{ .name = "sub_module_id",      .type = "INTEGER" },
+	{ .name = "err_severity",       .type = "TEXT" },
+	{ .name = "err_fr_0",		.type = "INTEGER" },
+	{ .name = "err_fr_1",		.type = "INTEGER" },
+	{ .name = "err_ctrl_0",		.type = "INTEGER" },
+	{ .name = "err_ctrl_1",		.type = "INTEGER" },
+	{ .name = "err_status_0",	.type = "INTEGER" },
+	{ .name = "err_status_1",	.type = "INTEGER" },
+	{ .name = "err_addr_0",         .type = "INTEGER" },
+	{ .name = "err_addr_1",         .type = "INTEGER" },
+	{ .name = "err_misc0_0",	.type = "INTEGER" },
+	{ .name = "err_misc0_1",	.type = "INTEGER" },
+	{ .name = "err_misc1_0",	.type = "INTEGER" },
+	{ .name = "err_misc1_1",	.type = "INTEGER" },
+};
+
+static const struct db_table_descriptor hip08_oem_type2_event_tab = {
+	.name = "hip08_oem_type2_event",
+	.fields = hip08_oem_type2_event_fields,
+	.num_fields = ARRAY_SIZE(hip08_oem_type2_event_fields),
+};
+
 static void record_vendor_data(struct ras_ns_dec_tab *dec_tab,
 			       enum hisi_oem_data_type data_type,
 			       int id, int64_t data, const char *text)
@@ -317,11 +463,165 @@ static int decode_hip08_oem_type1_error(struct ras_events *ras,
 	return 0;
 }
 
+static int decode_hip08_oem_type2_error(struct ras_events *ras,
+					struct ras_ns_dec_tab *dec_tab,
+					struct trace_seq *s, const void *error)
+{
+	const struct hisi_oem_type2_err_sec *err = error;
+	char buf[1024];
+	char *p = buf;
+
+	if (err->val_bits == 0) {
+		trace_seq_printf(s, "%s: no valid error information\n",
+				 __func__);
+		return -1;
+	}
+
+#ifdef HAVE_SQLITE3
+	if (!dec_tab->stmt_dec_record) {
+		if (ras_mc_add_vendor_table(ras, &dec_tab->stmt_dec_record,
+			&hip08_oem_type2_event_tab) != SQLITE_OK) {
+			trace_seq_printf(s,
+				"create sql hip08_oem_type2_event_tab fail\n");
+			return -1;
+		}
+	}
+#endif
+	p += sprintf(p, "[ ");
+	p += sprintf(p, "Table version=%d ", err->version);
+	record_vendor_data(dec_tab, hisi_oem_data_type_int,
+			   hip08_oem_type2_field_version,
+			   err->version, NULL);
+	if (err->val_bits & HISI_OEM_VALID_SOC_ID) {
+		p += sprintf(p, "SOC ID=%d ", err->soc_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_soc_id,
+				   err->soc_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_SOCKET_ID) {
+		p += sprintf(p, "socket ID=%d ", err->socket_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_socket_id,
+				   err->socket_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_NIMBUS_ID) {
+		p += sprintf(p, "nimbus ID=%d ", err->nimbus_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_nimbus_id,
+				   err->nimbus_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_MODULE_ID) {
+		p += sprintf(p, "module=%s ",
+			     oem_type2_module_name(err->module_id));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_oem_type2_field_module_id,
+				   0, oem_type2_module_name(err->module_id));
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_SUB_MODULE_ID) {
+		p =  oem_type2_sub_module_id(p, err->module_id,
+					     err->sub_module_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_sub_module_id,
+				   err->sub_module_id, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_VALID_ERR_SEVERITY) {
+		p += sprintf(p, "error severity=%s ",
+			     err_severity(err->err_severity));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_oem_type2_field_err_sev,
+				   0, err_severity(err->err_severity));
+	}
+
+	p += sprintf(p, "]");
+	trace_seq_printf(s, "\nHISI HIP08: OEM Type-2 Error\n");
+	trace_seq_printf(s, "%s\n", buf);
+
+	trace_seq_printf(s, "Reg Dump:\n");
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_FR) {
+		trace_seq_printf(s, "ERR_FR_0=0x%x\n", err->err_fr_0);
+		trace_seq_printf(s, "ERR_FR_1=0x%x\n", err->err_fr_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_fr_0,
+				   err->err_fr_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_fr_1,
+				   err->err_fr_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_CTRL) {
+		trace_seq_printf(s, "ERR_CTRL_0=0x%x\n", err->err_ctrl_0);
+		trace_seq_printf(s, "ERR_CTRL_1=0x%x\n", err->err_ctrl_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_ctrl_0,
+				   err->err_ctrl_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_ctrl_1,
+				   err->err_ctrl_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_STATUS) {
+		trace_seq_printf(s, "ERR_STATUS_0=0x%x\n", err->err_status_0);
+		trace_seq_printf(s, "ERR_STATUS_1=0x%x\n", err->err_status_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_status_0,
+				   err->err_status_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_status_1,
+				   err->err_status_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_ADDR) {
+		trace_seq_printf(s, "ERR_ADDR_0=0x%x\n", err->err_addr_0);
+		trace_seq_printf(s, "ERR_ADDR_1=0x%x\n", err->err_addr_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_addr_0,
+				   err->err_addr_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_addr_1,
+				   err->err_addr_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_MISC_0) {
+		trace_seq_printf(s, "ERR_MISC0_0=0x%x\n", err->err_misc0_0);
+		trace_seq_printf(s, "ERR_MISC0_1=0x%x\n", err->err_misc0_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_misc0_0,
+				   err->err_misc0_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_misc0_1,
+				   err->err_misc0_1, NULL);
+	}
+
+	if (err->val_bits & HISI_OEM_TYPE2_VALID_ERR_MISC_1) {
+		trace_seq_printf(s, "ERR_MISC1_0=0x%x\n", err->err_misc1_0);
+		trace_seq_printf(s, "ERR_MISC1_1=0x%x\n", err->err_misc1_1);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_misc1_0,
+				   err->err_misc1_0, NULL);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_oem_type2_field_err_misc1_1,
+				   err->err_misc1_1, NULL);
+	}
+
+	step_vendor_data_tab(dec_tab, "hip08_oem_type2_event_tab");
+
+	return 0;
+}
+
 struct ras_ns_dec_tab hip08_ns_oem_tab[] = {
 	{
 		.sec_type = "1f8161e155d641e6bd107afd1dc5f7c5",
 		.decode = decode_hip08_oem_type1_error,
 	},
+	{
+		.sec_type = "45534ea6ce2341158535e07ab3aef91d",
+		.decode = decode_hip08_oem_type2_error,
+	},
 	{ /* sentinel */ }
 };
 
-- 
1.9.1

[PATCH 6/6] rasdaemon:add logging HiSilicon HIP08 PCIe local errors

[Shiju Jose]

This patch adds logging for the HiSilicon HIP08 PCIe local errors.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 non-standard-hisi_hip08.c | 223 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 223 insertions(+)

diff --git a/non-standard-hisi_hip08.c b/non-standard-hisi_hip08.c
index 6fe5cbe..ae543d6 100644
--- a/non-standard-hisi_hip08.c
+++ b/non-standard-hisi_hip08.c
@@ -57,6 +57,24 @@
 #define HISI_OEM_TYPE2_VALID_ERR_MISC_0	BIT(10)
 #define HISI_OEM_TYPE2_VALID_ERR_MISC_1	BIT(11)
 
+/* HISI PCIe Local error definitions */
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
 struct hisi_oem_type1_err_sec {
 	uint32_t   val_bits;
 	uint8_t    version;
@@ -99,6 +117,21 @@ struct hisi_oem_type2_err_sec {
 	uint32_t   err_misc1_1;
 };
 
+struct hisi_pcie_local_err_sec {
+	uint64_t   val_bits;
+	uint8_t    version;
+	uint8_t    soc_id;
+	uint8_t    socket_id;
+	uint8_t    nimbus_id;
+	uint8_t    sub_module_id;
+	uint8_t    core_id;
+	uint8_t    port_id;
+	uint8_t    err_severity;
+	uint16_t   err_type;
+	uint8_t    reserv[2];
+	uint32_t   err_misc[33];
+};
+
 enum hisi_oem_data_type {
 	hisi_oem_data_type_int,
 	hisi_oem_data_type_int64,
@@ -145,6 +178,20 @@ enum {
 	hip08_oem_type2_field_err_misc1_1,
 };
 
+enum {
+	hip08_pcie_local_field_id,
+	hip08_pcie_local_field_version,
+	hip08_pcie_local_field_soc_id,
+	hip08_pcie_local_field_socket_id,
+	hip08_pcie_local_field_nimbus_id,
+	hip08_pcie_local_field_sub_module_id,
+	hip08_pcie_local_field_core_id,
+	hip08_pcie_local_field_port_id,
+	hip08_pcie_local_field_err_sev,
+	hip08_pcie_local_field_err_type,
+	hip08_pcie_local_field_err_misc,
+};
+
 /* helper functions */
 static char *err_severity(uint8_t err_sev)
 {
@@ -230,6 +277,18 @@ static char *oem_type2_sub_module_id(char *p, uint8_t module_id,
 	return p;
 }
 
+static char *pcie_local_sub_module_name(uint8_t id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+	return "unknown";
+}
+
 #ifdef HAVE_SQLITE3
 static const struct db_fields hip08_oem_type1_event_fields[] = {
 	{ .name = "id",			.type = "INTEGER PRIMARY KEY" },
@@ -283,6 +342,58 @@ static const struct db_table_descriptor hip08_oem_type2_event_tab = {
 	.num_fields = ARRAY_SIZE(hip08_oem_type2_event_fields),
 };
 
+static const struct db_fields hip08_pcie_local_event_fields[] = {
+	{ .name = "id",                 .type = "INTEGER PRIMARY KEY" },
+	{ .name = "version",            .type = "INTEGER" },
+	{ .name = "soc_id",             .type = "INTEGER" },
+	{ .name = "socket_id",          .type = "INTEGER" },
+	{ .name = "nimbus_id",          .type = "INTEGER" },
+	{ .name = "sub_module_id",      .type = "TEXT" },
+	{ .name = "core_id",		.type = "INTEGER" },
+	{ .name = "port_id",		.type = "INTEGER" },
+	{ .name = "err_severity",       .type = "TEXT" },
+	{ .name = "err_type",		.type = "INTEGER" },
+	{ .name = "err_misc0",		.type = "INTEGER" },
+	{ .name = "err_misc1",		.type = "INTEGER" },
+	{ .name = "err_misc2",		.type = "INTEGER" },
+	{ .name = "err_misc3",		.type = "INTEGER" },
+	{ .name = "err_misc4",		.type = "INTEGER" },
+	{ .name = "err_misc5",		.type = "INTEGER" },
+	{ .name = "err_misc6",		.type = "INTEGER" },
+	{ .name = "err_misc7",		.type = "INTEGER" },
+	{ .name = "err_misc8",		.type = "INTEGER" },
+	{ .name = "err_misc9",		.type = "INTEGER" },
+	{ .name = "err_misc10",		.type = "INTEGER" },
+	{ .name = "err_misc11",		.type = "INTEGER" },
+	{ .name = "err_misc12",		.type = "INTEGER" },
+	{ .name = "err_misc13",		.type = "INTEGER" },
+	{ .name = "err_misc14",		.type = "INTEGER" },
+	{ .name = "err_misc15",		.type = "INTEGER" },
+	{ .name = "err_misc16",		.type = "INTEGER" },
+	{ .name = "err_misc17",		.type = "INTEGER" },
+	{ .name = "err_misc18",		.type = "INTEGER" },
+	{ .name = "err_misc19",		.type = "INTEGER" },
+	{ .name = "err_misc20",		.type = "INTEGER" },
+	{ .name = "err_misc21",		.type = "INTEGER" },
+	{ .name = "err_misc22",		.type = "INTEGER" },
+	{ .name = "err_misc23",		.type = "INTEGER" },
+	{ .name = "err_misc24",		.type = "INTEGER" },
+	{ .name = "err_misc25",		.type = "INTEGER" },
+	{ .name = "err_misc26",		.type = "INTEGER" },
+	{ .name = "err_misc27",		.type = "INTEGER" },
+	{ .name = "err_misc28",		.type = "INTEGER" },
+	{ .name = "err_misc29",		.type = "INTEGER" },
+	{ .name = "err_misc30",		.type = "INTEGER" },
+	{ .name = "err_misc31",		.type = "INTEGER" },
+	{ .name = "err_misc32",		.type = "INTEGER" },
+};
+
+static const struct db_table_descriptor hip08_pcie_local_event_tab = {
+	.name = "hip08_pcie_local_event",
+	.fields = hip08_pcie_local_event_fields,
+	.num_fields = ARRAY_SIZE(hip08_pcie_local_event_fields),
+};
+
 static void record_vendor_data(struct ras_ns_dec_tab *dec_tab,
 			       enum hisi_oem_data_type data_type,
 			       int id, int64_t data, const char *text)
@@ -613,6 +724,114 @@ static int decode_hip08_oem_type2_error(struct ras_events *ras,
 	return 0;
 }
 
+static int decode_hip08_pcie_local_error(struct ras_events *ras,
+					 struct ras_ns_dec_tab *dec_tab,
+					 struct trace_seq *s, const void *error)
+{
+	const struct hisi_pcie_local_err_sec *err = error;
+	char buf[1024];
+	char *p = buf;
+	uint32_t i;
+
+	if (err->val_bits == 0) {
+		trace_seq_printf(s, "%s: no valid error information\n",
+				 __func__);
+		return -1;
+	}
+
+#ifdef HAVE_SQLITE3
+	if (!dec_tab->stmt_dec_record) {
+		if (ras_mc_add_vendor_table(ras, &dec_tab->stmt_dec_record,
+				&hip08_pcie_local_event_tab) != SQLITE_OK) {
+			trace_seq_printf(s,
+				"create sql hip08_pcie_local_event_tab fail\n");
+			return -1;
+		}
+	}
+#endif
+	p += sprintf(p, "[ ");
+	p += sprintf(p, "Table version=%d ", err->version);
+	record_vendor_data(dec_tab, hisi_oem_data_type_int,
+			   hip08_pcie_local_field_version,
+			   err->version, NULL);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID) {
+		p += sprintf(p, "SOC ID=%d ", err->soc_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_soc_id,
+				   err->soc_id, NULL);
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID) {
+		p += sprintf(p, "socket ID=%d ", err->socket_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_socket_id,
+				   err->socket_id, NULL);
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID) {
+		p += sprintf(p, "nimbus ID=%d ", err->nimbus_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_nimbus_id,
+				   err->nimbus_id, NULL);
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID) {
+		p += sprintf(p, "sub module=%s ",
+			     pcie_local_sub_module_name(err->sub_module_id));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_pcie_local_field_sub_module_id,
+				   0, pcie_local_sub_module_name(err->sub_module_id));
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID) {
+		p += sprintf(p, "core ID=core%d ", err->core_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_core_id,
+				   err->core_id, NULL);
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID) {
+		p += sprintf(p, "port ID=port%d ", err->port_id);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_port_id,
+				   err->port_id, NULL);
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY) {
+		p += sprintf(p, "error severity=%s ",
+			     err_severity(err->err_severity));
+		record_vendor_data(dec_tab, hisi_oem_data_type_text,
+				   hip08_pcie_local_field_err_sev,
+				   0, err_severity(err->err_severity));
+	}
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE) {
+		p += sprintf(p, "error type=0x%x ", err->err_type);
+		record_vendor_data(dec_tab, hisi_oem_data_type_int,
+				   hip08_pcie_local_field_err_type,
+				   err->err_type, NULL);
+	}
+	p += sprintf(p, "]");
+
+	trace_seq_printf(s, "\nHISI HIP08: PCIe local error\n");
+	trace_seq_printf(s, "%s\n", buf);
+
+	trace_seq_printf(s, "Reg Dump:\n");
+	for (i = 0; i < 33; i++) {
+		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i)) {
+			trace_seq_printf(s, "ERR_MISC_%d=0x%x\n", i,
+					 err->err_misc[i]);
+			record_vendor_data(dec_tab, hisi_oem_data_type_int,
+					   (hip08_pcie_local_field_err_misc + i),
+					   err->err_misc[i], NULL);
+		}
+	}
+
+	step_vendor_data_tab(dec_tab, "hip08_pcie_local_event_tab");
+
+	return 0;
+}
+
 struct ras_ns_dec_tab hip08_ns_oem_tab[] = {
 	{
 		.sec_type = "1f8161e155d641e6bd107afd1dc5f7c5",
@@ -622,6 +841,10 @@ struct ras_ns_dec_tab hip08_ns_oem_tab[] = {
 		.sec_type = "45534ea6ce2341158535e07ab3aef91d",
 		.decode = decode_hip08_oem_type2_error,
 	},
+	{
+		.sec_type = "b2889fc9e7d74f9da867af42e98be772",
+		.decode = decode_hip08_pcie_local_error,
+	},
 	{ /* sentinel */ }
 };
 
-- 
1.9.1

Re: [PATCH 0/6] rasdaemon:add logging of HiSilicon HIP08 non-standard H/W errors and changes in the error decoding code

[Mauro Carvalho Chehab]

Em Mon, 17 Jun 2019 15:28:46 +0100
Shiju Jose <shiju.jose@huawei.com> escreveu:

> This patch set add few changes in the non-standard error decoding code and
> logging for the HiSilicon HIP08 non-standard H/W errors.
> 
> Shiju Jose (6):
>   rasdaemon:print non-standard error data if not decoded
>   rasdaemon: rearrange HiSilicon HIP07 decoding function table
>   rasdaemon: update iteration logic for the non-standard error decoding
>     functions
>   rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM
>     format1
>   rasdaemon:add logging HiSilicon HIP08 H/W errors reported in the OEM
>     format2
>   rasdaemon:add logging HiSilicon HIP08 PCIe local errors
> 
>  Makefile.am                |   2 +-
>  non-standard-hisi_hip07.c  |  36 +-
>  non-standard-hisi_hip08.c  | 855 +++++++++++++++++++++++++++++++++++++++++++++
>  ras-non-standard-handler.c |  36 +-
>  ras-non-standard-handler.h |   8 +-
>  ras-record.c               |  30 +-
>  ras-record.h               |  13 +
>  7 files changed, 932 insertions(+), 48 deletions(-)
>  create mode 100644 non-standard-hisi_hip08.c
> 

Applied, thanks!


Thanks,
Mauro

[PATCH RFC 0/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently kernel does not support reporting the vendor specific HW errors,
in the non-standard format, to the vendor drivers for the recovery.

This patch set add this support and also move the existing handler
functions for the standard errors to the new callback method.
Also the CCIX RAS patches could be move to the proposed callback method.
https://www.spinics.net/lists/linux-edac/msg10508.html
https://patchwork.kernel.org/patch/10979491/

Shiju Jose (4):
  ACPI: APEI: Add support to notify the vendor specific HW errors
  ACPI: APEI: Add ghes_handle_memory_failure to the new notification
    method
  ACPI: APEI: Add ghes_handle_aer to the new notification method
  ACPI: APEI: Add log_arm_hw_error to the new notification method

 drivers/acpi/apei/ghes.c | 170 +++++++++++++++++++++++++++++++++++++++++------
 drivers/ras/ras.c        |   5 +-
 include/acpi/ghes.h      |  47 +++++++++++++
 include/linux/ras.h      |   7 +-
 4 files changed, 205 insertions(+), 24 deletions(-)

-- 
1.9.1

[PATCH RFC 1/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor specific HW errors, in the non-standard format,
are not reported to the vendor drivers for the recovery.

This patch adds support to notify the vendor specific HW errors to the
registered kernel drivers.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 118 +++++++++++++++++++++++++++++++++++++++++++++--
 include/acpi/ghes.h      |  47 +++++++++++++++++++
 2 files changed, 160 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index a66e00f..374d197 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -477,6 +477,77 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+struct ghes_error_notify {
+	struct list_head list;
+	struct rcu_head	rcu_head;
+	guid_t sec_type; /* guid of the error record */
+	error_handle handle; /* error handler function */
+	void *data; /* handler driver's private data if any */
+};
+
+/* List to store the registered error handling functions */
+static DEFINE_MUTEX(ghes_error_notify_mutex);
+static LIST_HEAD(ghes_error_notify_list);
+static refcount_t ghes_ref_count;
+
+/**
+ * ghes_error_notify_register - register an error handling function
+ * for the hw errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @handle: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_error_notify_register(guid_t sec_type, error_handle handle, void *data)
+{
+	struct ghes_error_notify *err_notify;
+
+	mutex_lock(&ghes_error_notify_mutex);
+	err_notify = kzalloc(sizeof(*err_notify), GFP_KERNEL);
+	if (!err_notify)
+		return -ENOMEM;
+
+	err_notify->handle = handle;
+	guid_copy(&err_notify->sec_type, &sec_type);
+	err_notify->data = data;
+	list_add_rcu(&err_notify->list, &ghes_error_notify_list);
+	mutex_unlock(&ghes_error_notify_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ghes_error_notify_register);
+
+/**
+ * ghes_error_notify_unregister - unregister an error handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @handle: pointer to the error handling function.
+ *
+ * return none.
+ */
+void ghes_error_notify_unregister(guid_t sec_type, error_handle handle)
+{
+	struct ghes_error_notify *err_notify;
+	bool found = 0;
+
+	mutex_lock(&ghes_error_notify_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(err_notify, &ghes_error_notify_list, list) {
+		if (guid_equal(&err_notify->sec_type, &sec_type) &&
+		    err_notify->handle == handle) {
+			list_del_rcu(&err_notify->list);
+			found = 1;
+			break;
+		}
+	}
+	rcu_read_unlock();
+	synchronize_rcu();
+	mutex_unlock(&ghes_error_notify_mutex);
+	if (found)
+		kfree(err_notify);
+}
+EXPORT_SYMBOL_GPL(ghes_error_notify_unregister);
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -485,6 +556,8 @@ static void ghes_do_proc(struct ghes *ghes,
 	guid_t *sec_type;
 	guid_t *fru_id = &NULL_UUID_LE;
 	char *fru_text = "";
+	bool is_notify = 0;
+	struct ghes_error_notify *err_notify;
 
 	sev = ghes_severity(estatus->error_severity);
 	apei_estatus_for_each_section(estatus, gdata) {
@@ -512,11 +585,29 @@ static void ghes_do_proc(struct ghes *ghes,
 
 			log_arm_hw_error(err);
 		} else {
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
+			rcu_read_lock();
+			list_for_each_entry_rcu(err_notify,
+						&ghes_error_notify_list, list) {
+				if (guid_equal(&err_notify->sec_type,
+					       sec_type)) {
+					/* The notification is called in the
+					 * interrupt context, thus the handler
+					 * functions should be take care of it.
+					 */
+					err_notify->handle(gdata, sev,
+							   err_notify->data);
+					is_notify = 1;
+				}
+			}
+			rcu_read_unlock();
+
+			if (!is_notify) {
+				void *err = acpi_hest_get_payload(gdata);
+
+				log_non_standard_event(sec_type, fru_id,
+						       fru_text, sec_sev, err,
+						       gdata->error_data_length);
+			}
 		}
 	}
 }
@@ -1217,6 +1308,11 @@ static int ghes_probe(struct platform_device *ghes_dev)
 
 	ghes_edac_register(ghes, &ghes_dev->dev);
 
+	if (!refcount_read(&ghes_ref_count))
+		refcount_set(&ghes_ref_count, 1);
+	else
+		refcount_inc(&ghes_ref_count);
+
 	/* Handle any pending errors right away */
 	spin_lock_irqsave(&ghes_notify_lock_irq, flags);
 	ghes_proc(ghes);
@@ -1237,6 +1333,7 @@ static int ghes_remove(struct platform_device *ghes_dev)
 	int rc;
 	struct ghes *ghes;
 	struct acpi_hest_generic *generic;
+	struct ghes_error_notify *err_notify, *tmp;
 
 	ghes = platform_get_drvdata(ghes_dev);
 	generic = ghes->generic;
@@ -1279,6 +1376,17 @@ static int ghes_remove(struct platform_device *ghes_dev)
 
 	ghes_fini(ghes);
 
+	if (refcount_dec_and_test(&ghes_ref_count) &&
+	    !list_empty(&ghes_error_notify_list)) {
+		mutex_lock(&ghes_error_notify_mutex);
+		list_for_each_entry_safe(err_notify, tmp,
+					 &ghes_error_notify_list, list) {
+			list_del_rcu(&err_notify->list);
+			kfree_rcu(err_notify, rcu_head);
+		}
+		mutex_unlock(&ghes_error_notify_mutex);
+	}
+
 	ghes_edac_unregister(ghes);
 
 	kfree(ghes);
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cdd..d480537 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,53 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+/**
+ * error_handle - error handling function for the hw errors.
+ * This handle function is called in the interrupt context.
+ * @gdata: acpi_hest_generic_data.
+ * @sev: error severity of the entire error event defined in the
+ * ACPI spec table generic error status block.
+ * @data: handler driver's private data.
+ *
+ * return : none.
+ */
+typedef void (*error_handle)(struct acpi_hest_generic_data *gdata, int sev,
+			     void *data);
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_error_notify_register - register an error handling function
+ * for the hw errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @handle: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_error_notify_register(guid_t sec_type, error_handle handle,
+			       void *data);
+
+/**
+ * ghes_error_notify_unregister - unregister an error handling function
+ * for the hw errors.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @handle: pointer to the error handling function.
+ *
+ * return none.
+ */
+void ghes_error_notify_unregister(guid_t sec_type, error_handle handle);
+
+#else
+int ghes_error_notify_register(guid_t sec_type, error_handle handle, void *data)
+{
+	return -ENODEV;
+}
+
+void ghes_error_notify_unregister(guid_t sec_type, error_handle handle)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
1.9.1

Re: [PATCH RFC 1/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi,

On 12/08/2019 11:11, Shiju Jose wrote:
> Presently the vendor specific HW errors, in the non-standard format,
> are not reported to the vendor drivers for the recovery.
> 
> This patch adds support to notify the vendor specific HW errors to the
> registered kernel drivers.

> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index a66e00f..374d197 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -477,6 +477,77 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
>  #endif
>  }
>  
> +struct ghes_error_notify {
> +	struct list_head list;> +	struct rcu_head	rcu_head;
> +	guid_t sec_type; /* guid of the error record */

> +	error_handle handle; /* error handler function */

ghes_error_handler_t error_handler; ?


> +	void *data; /* handler driver's private data if any */
> +};
> +
> +/* List to store the registered error handling functions */
> +static DEFINE_MUTEX(ghes_error_notify_mutex);
> +static LIST_HEAD(ghes_error_notify_list);

> +static refcount_t ghes_ref_count;

I don't think this refcount is needed.


> +/**
> + * ghes_error_notify_register - register an error handling function
> + * for the hw errors.
> + * @sec_type: sec_type of the corresponding CPER to be notified.
> + * @handle: pointer to the error handling function.
> + * @data: handler driver's private data.
> + *
> + * return 0 : SUCCESS, non-zero : FAIL
> + */
> +int ghes_error_notify_register(guid_t sec_type, error_handle handle, void *data)
> +{
> +	struct ghes_error_notify *err_notify;
> +
> +	mutex_lock(&ghes_error_notify_mutex);
> +	err_notify = kzalloc(sizeof(*err_notify), GFP_KERNEL);
> +	if (!err_notify)
> +		return -ENOMEM;

Leaving the mutex locked.
You may as well allocate the memory before taking the lock.


> +
> +	err_notify->handle = handle;
> +	guid_copy(&err_notify->sec_type, &sec_type);
> +	err_notify->data = data;
> +	list_add_rcu(&err_notify->list, &ghes_error_notify_list);
> +	mutex_unlock(&ghes_error_notify_mutex);
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(ghes_error_notify_register);

Could we leave exporting this to modules until there is a user?


> +/**
> + * ghes_error_notify_unregister - unregister an error handling function.
> + * @sec_type: sec_type of the corresponding CPER.
> + * @handle: pointer to the error handling function.
> + *
> + * return none.
> + */
> +void ghes_error_notify_unregister(guid_t sec_type, error_handle handle)

Why do we need the handle(r) a second time? Surely there can only be one callback for a
given guid.


> +{
> +	struct ghes_error_notify *err_notify;
> +	bool found = 0;
> +
> +	mutex_lock(&ghes_error_notify_mutex);
> +	rcu_read_lock();
> +	list_for_each_entry_rcu(err_notify, &ghes_error_notify_list, list) {
> +		if (guid_equal(&err_notify->sec_type, &sec_type) &&
> +		    err_notify->handle == handle) {
> +			list_del_rcu(&err_notify->list);
> +			found = 1;
> +			break;
> +		}
> +	}
> +	rcu_read_unlock();

> +	synchronize_rcu();

Is this for the kfree()? Please keep them together so its obvious what its for.
Putting it outside the mutex will also save any contended waiter some time.


> +	mutex_unlock(&ghes_error_notify_mutex);
> +	if (found)
> +		kfree(err_notify);
> +}
> +EXPORT_SYMBOL_GPL(ghes_error_notify_unregister);
> +

>  static void ghes_do_proc(struct ghes *ghes,
>  			 const struct acpi_hest_generic_status *estatus)
>  {> @@ -512,11 +585,29 @@ static void ghes_do_proc(struct ghes *ghes,
>  
>  			log_arm_hw_error(err);
>  		} else {
> -			void *err = acpi_hest_get_payload(gdata);
> -
> -			log_non_standard_event(sec_type, fru_id, fru_text,
> -					       sec_sev, err,
> -					       gdata->error_data_length);

> +			rcu_read_lock();
> +			list_for_each_entry_rcu(err_notify,
> +						&ghes_error_notify_list, list) {
> +				if (guid_equal(&err_notify->sec_type,
> +					       sec_type)) {

> +					/* The notification is called in the
> +					 * interrupt context, thus the handler
> +					 * functions should be take care of it.
> +					 */

I read this as "the handler will be called", which doesn't seem to be a useful comment.


> +					err_notify->handle(gdata, sev,
> +							   err_notify->data);
> +					is_notify = 1;

					break;

> +				}
> +			}
> +			rcu_read_unlock();

> +			if (!is_notify) {

if (!found) Seems more natural.


> +				void *err = acpi_hest_get_payload(gdata);
> +
> +				log_non_standard_event(sec_type, fru_id,
> +						       fru_text, sec_sev, err,
> +						       gdata->error_data_length);
> +			}

This is tricky to read as its so bunched up. Please pull it out into a separate function.
ghes_handle_non_standard_event() ?


Because you skip log_non_standard_event(), rasdaemon will no longer see these in
user-space. For any kernel consumer of these, we need to know we aren't breaking the
user-space component.


>  		}
>  	}
>  }
> @@ -1217,6 +1308,11 @@ static int ghes_probe(struct platform_device *ghes_dev)
>  
>  	ghes_edac_register(ghes, &ghes_dev->dev);
>  
> +	if (!refcount_read(&ghes_ref_count))
> +		refcount_set(&ghes_ref_count, 1);

What stops this from racing with itself if two ghes platform devices are probed at the
same time?

If the refcount needs initialising, please do it in ghes_init()....

> +	else
> +		refcount_inc(&ghes_ref_count);

.. but I don't think this refcount is needed.


>  	/* Handle any pending errors right away */
>  	spin_lock_irqsave(&ghes_notify_lock_irq, flags);
>  	ghes_proc(ghes);

> @@ -1279,6 +1376,17 @@ static int ghes_remove(struct platform_device *ghes_dev)
>  
>  	ghes_fini(ghes);
>  
> +	if (refcount_dec_and_test(&ghes_ref_count) &&
> +	    !list_empty(&ghes_error_notify_list)) {
> +		mutex_lock(&ghes_error_notify_mutex);> +		list_for_each_entry_safe(err_notify, tmp,
> +					 &ghes_error_notify_list, list) {
> +			list_del_rcu(&err_notify->list);
> +			kfree_rcu(err_notify, rcu_head);
> +		}
> +		mutex_unlock(&ghes_error_notify_mutex);
> +	}

... If someone unregisters, and re-registers all the GHES platform devices, the last one
out flushes the vendor-specific error handlers away. Then we re-probe the devices again,
but this time the vendor-specific error handlers don't work.

As you have an add/remove API for drivers, its up to drivers to cleanup when they are
removed. The comings and goings of GHES platform devices isn't relevant.


>  	ghes_edac_unregister(ghes);
>  
>  	kfree(ghes);
> diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
> index e3f1cdd..d480537 100644
> --- a/include/acpi/ghes.h
> +++ b/include/acpi/ghes.h
> @@ -50,6 +50,53 @@ enum {
>  	GHES_SEV_PANIC = 0x3,
>  };
>  
> +/**
> + * error_handle - error handling function for the hw errors.

Fatal errors get dealt with earlier, so drivers will never see them.
| error handling function for non-fatal hardware errors.


> + * This handle function is called in the interrupt context.

As this overrides ghes's logging of the error, we should mention:
| The handler is responsible for any logging of the error.


> + * @gdata: acpi_hest_generic_data.
> + * @sev: error severity of the entire error event defined in the
> + * ACPI spec table generic error status block.
> + * @data: handler driver's private data.
> + *
> + * return : none.
> + */
> +typedef void (*error_handle)(struct acpi_hest_generic_data *gdata, int sev,
> +			     void *data);


Thanks,

James

RE: [PATCH RFC 1/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi James,

>-----Original Message-----
>From: James Morse [mailto:james.morse@arm.com]
>Sent: 21 August 2019 18:24
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-edac@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; lenb@kernel.org;
>tony.luck@intel.com; bp@alien8.de; baicar@os.amperecomputing.com;
>Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>
>Subject: Re: [PATCH RFC 1/4] ACPI: APEI: Add support to notify the vendor
>specific HW errors
>
>Hi,
>
>On 12/08/2019 11:11, Shiju Jose wrote:
>> Presently the vendor specific HW errors, in the non-standard format,
>> are not reported to the vendor drivers for the recovery.
>>
>> This patch adds support to notify the vendor specific HW errors to the
>> registered kernel drivers.
>
>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c index
>> a66e00f..374d197 100644
>> --- a/drivers/acpi/apei/ghes.c
>> +++ b/drivers/acpi/apei/ghes.c
>> @@ -477,6 +477,77 @@ static void ghes_handle_aer(struct
>> acpi_hest_generic_data *gdata)  #endif  }
>>
>> +struct ghes_error_notify {
>> +	struct list_head list;> +	struct rcu_head	rcu_head;
>> +	guid_t sec_type; /* guid of the error record */
>
>> +	error_handle handle; /* error handler function */
>
>ghes_error_handler_t error_handler; ?
Sure.

>
>
>> +	void *data; /* handler driver's private data if any */ };
>> +
>> +/* List to store the registered error handling functions */ static
>> +DEFINE_MUTEX(ghes_error_notify_mutex);
>> +static LIST_HEAD(ghes_error_notify_list);
>
>> +static refcount_t ghes_ref_count;
>
>I don't think this refcount is needed.
refcount was added to register standard error handlers with this notification method one time when
multiple ghes platform devices are probed.
 
>
>
>> +/**
>> + * ghes_error_notify_register - register an error handling function
>> + * for the hw errors.
>> + * @sec_type: sec_type of the corresponding CPER to be notified.
>> + * @handle: pointer to the error handling function.
>> + * @data: handler driver's private data.
>> + *
>> + * return 0 : SUCCESS, non-zero : FAIL  */ int
>> +ghes_error_notify_register(guid_t sec_type, error_handle handle, void
>> +*data) {
>> +	struct ghes_error_notify *err_notify;
>> +
>> +	mutex_lock(&ghes_error_notify_mutex);
>> +	err_notify = kzalloc(sizeof(*err_notify), GFP_KERNEL);
>> +	if (!err_notify)
>> +		return -ENOMEM;
>
>Leaving the mutex locked.
>You may as well allocate the memory before taking the lock.
Good spot. I will fix.

>
>
>> +
>> +	err_notify->handle = handle;
>> +	guid_copy(&err_notify->sec_type, &sec_type);
>> +	err_notify->data = data;
>> +	list_add_rcu(&err_notify->list, &ghes_error_notify_list);
>> +	mutex_unlock(&ghes_error_notify_mutex);
>> +
>> +	return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(ghes_error_notify_register);
>
>Could we leave exporting this to modules until there is a user?
>
>
>> +/**
>> + * ghes_error_notify_unregister - unregister an error handling function.
>> + * @sec_type: sec_type of the corresponding CPER.
>> + * @handle: pointer to the error handling function.
>> + *
>> + * return none.
>> + */
>> +void ghes_error_notify_unregister(guid_t sec_type, error_handle
>> +handle)
>
>Why do we need the handle(r) a second time? Surely there can only be one
>callback for a given guid.
There is a possibility of sharing the guid between drivers if the non-standard error section format is common
for more than one devices if the error data to be reported is in the same format.
 
>
>
>> +{
>> +	struct ghes_error_notify *err_notify;
>> +	bool found = 0;
>> +
>> +	mutex_lock(&ghes_error_notify_mutex);
>> +	rcu_read_lock();
>> +	list_for_each_entry_rcu(err_notify, &ghes_error_notify_list, list) {
>> +		if (guid_equal(&err_notify->sec_type, &sec_type) &&
>> +		    err_notify->handle == handle) {
>> +			list_del_rcu(&err_notify->list);
>> +			found = 1;
>> +			break;
>> +		}
>> +	}
>> +	rcu_read_unlock();
>
>> +	synchronize_rcu();
>
>Is this for the kfree()? Please keep them together so its obvious what its for.
>Putting it outside the mutex will also save any contended waiter some time.
Yes. I will move synchronize_rcu () just before kfree. 
 
>
>
>> +	mutex_unlock(&ghes_error_notify_mutex);
>> +	if (found)
>> +		kfree(err_notify);
>> +}
>> +EXPORT_SYMBOL_GPL(ghes_error_notify_unregister);
>> +
>
>>  static void ghes_do_proc(struct ghes *ghes,
>>  			 const struct acpi_hest_generic_status *estatus)  {>
>@@ -512,11
>> +585,29 @@ static void ghes_do_proc(struct ghes *ghes,
>>
>>  			log_arm_hw_error(err);
>>  		} else {
>> -			void *err = acpi_hest_get_payload(gdata);
>> -
>> -			log_non_standard_event(sec_type, fru_id, fru_text,
>> -					       sec_sev, err,
>> -					       gdata->error_data_length);
>
>> +			rcu_read_lock();
>> +			list_for_each_entry_rcu(err_notify,
>> +						&ghes_error_notify_list, list) {
>> +				if (guid_equal(&err_notify->sec_type,
>> +					       sec_type)) {
>
>> +					/* The notification is called in the
>> +					 * interrupt context, thus the handler
>> +					 * functions should be take care of it.
>> +					 */
>
>I read this as "the handler will be called", which doesn't seem to be a useful
>comment.
Ok. I will correct the comment.
>
>
>> +					err_notify->handle(gdata, sev,
>> +							   err_notify->data);
>> +					is_notify = 1;
>
>					break;
>
>> +				}
>> +			}
>> +			rcu_read_unlock();
>
>> +			if (!is_notify) {
>
>if (!found) Seems more natural.
Ok. I will change to "is_notify"  to "found".

>
>
>> +				void *err = acpi_hest_get_payload(gdata);
>> +
>> +				log_non_standard_event(sec_type, fru_id,
>> +						       fru_text, sec_sev, err,
>> +						       gdata->error_data_length);
>> +			}
>
>This is tricky to read as its so bunched up. Please pull it out into a separate
>function.
>ghes_handle_non_standard_event() ?
Ok. I will add to new ghes_handle_non_standard_event() function.

>
>
>Because you skip log_non_standard_event(), rasdaemon will no longer see
>these in user-space. For any kernel consumer of these, we need to know we
>aren't breaking the user-space component.
>
>
>>  		}
>>  	}
>>  }
>> @@ -1217,6 +1308,11 @@ static int ghes_probe(struct platform_device
>> *ghes_dev)
>>
>>  	ghes_edac_register(ghes, &ghes_dev->dev);
>>
>> +	if (!refcount_read(&ghes_ref_count))
>> +		refcount_set(&ghes_ref_count, 1);
>
>What stops this from racing with itself if two ghes platform devices are probed
>at the same time?
yes. It is an issue.
>
>If the refcount needs initialising, please do it in ghes_init()....
refcount was added to register the standard error handlers to the notification
method only for the first time when the ghes device probed multiple times.
I will check is it possible to avoid using refcount by moving the above registration
of standard error handlers to the ghes_init().
>
>> +	else
>> +		refcount_inc(&ghes_ref_count);
>
>.. but I don't think this refcount is needed.
>
>
>>  	/* Handle any pending errors right away */
>>  	spin_lock_irqsave(&ghes_notify_lock_irq, flags);
>>  	ghes_proc(ghes);
>
>> @@ -1279,6 +1376,17 @@ static int ghes_remove(struct platform_device
>> *ghes_dev)
>>
>>  	ghes_fini(ghes);
>>
>> +	if (refcount_dec_and_test(&ghes_ref_count) &&
>> +	    !list_empty(&ghes_error_notify_list)) {
>> +		mutex_lock(&ghes_error_notify_mutex);> +
>	list_for_each_entry_safe(err_notify, tmp,
>> +					 &ghes_error_notify_list, list) {
>> +			list_del_rcu(&err_notify->list);
>> +			kfree_rcu(err_notify, rcu_head);
>> +		}
>> +		mutex_unlock(&ghes_error_notify_mutex);
>> +	}
>
>... If someone unregisters, and re-registers all the GHES platform devices, the
>last one out flushes the vendor-specific error handlers away. Then we re-probe
>the devices again, but this time the vendor-specific error handlers don't work.
>
>As you have an add/remove API for drivers, its up to drivers to cleanup when
>they are removed. The comings and goings of GHES platform devices isn't
>relevant.
Ok. Got it. I will either keep the unregister for the standard error handlers only
if the standard error handlers can be part of the notification method or remove completely
if the registration can be done in the ghes_init().    
>
>
>>  	ghes_edac_unregister(ghes);
>>
>>  	kfree(ghes);
>> diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h index
>> e3f1cdd..d480537 100644
>> --- a/include/acpi/ghes.h
>> +++ b/include/acpi/ghes.h
>> @@ -50,6 +50,53 @@ enum {
>>  	GHES_SEV_PANIC = 0x3,
>>  };
>>
>> +/**
>> + * error_handle - error handling function for the hw errors.
>
>Fatal errors get dealt with earlier, so drivers will never see them.
>| error handling function for non-fatal hardware errors.
Ok. I will change the comment as recoverable HW errors.
>
>
>> + * This handle function is called in the interrupt context.
>
>As this overrides ghes's logging of the error, we should mention:
>| The handler is responsible for any logging of the error.
Ok. I will add in the comment.
>
>
>> + * @gdata: acpi_hest_generic_data.
>> + * @sev: error severity of the entire error event defined in the
>> + * ACPI spec table generic error status block.
>> + * @data: handler driver's private data.
>> + *
>> + * return : none.
>> + */
>> +typedef void (*error_handle)(struct acpi_hest_generic_data *gdata, int sev,
>> +			     void *data);
>
>
>Thanks,
>
>James
Thanks,
Shiju

[PATCH RFC 2/4] ACPI: APEI: Add ghes_handle_memory_failure to the new notification method

[Shiju Jose]

This patch adds ghes_handle_memory_failure to the new error
notification method.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 51 ++++++++++++++++++++++++++++++++++--------------
 1 file changed, 36 insertions(+), 15 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 374d197..4400d56 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -401,14 +401,18 @@ static void ghes_clear_estatus(struct ghes *ghes,
 		ghes_ack_error(ghes->generic_v2);
 }
 
-static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int sev)
+static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
+				       int sev, void *data)
 {
-#ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
+	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
+	int sec_sev = ghes_severity(gdata->error_severity);
 	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);
 
+	ghes_edac_report_mem_error(sev, mem_err);
+	arch_apei_report_mem_error(sev, mem_err);
+
+#ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
 	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
 		return;
 
@@ -569,15 +573,7 @@ static void ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
-		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
-			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
-
-			ghes_edac_report_mem_error(sev, mem_err);
-
-			arch_apei_report_mem_error(sev, mem_err);
-			ghes_handle_memory_failure(gdata, sev);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
+		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
@@ -1190,11 +1186,25 @@ static int apei_sdei_unregister_ghes(struct ghes *ghes)
 	return sdei_unregister_ghes(ghes);
 }
 
+struct ghes_err_handler_tab {
+	guid_t sec_type;
+	error_handle handle;
+};
+
+static struct ghes_err_handler_tab handler_tab[] = {
+	{
+		.sec_type = CPER_SEC_PLATFORM_MEM,
+		.handle = ghes_handle_memory_failure,
+	},
+	{ /* sentinel */ }
+};
+
 static int ghes_probe(struct platform_device *ghes_dev)
 {
 	struct acpi_hest_generic *generic;
 	struct ghes *ghes = NULL;
 	unsigned long flags;
+	int i;
 
 	int rc = -EINVAL;
 
@@ -1308,9 +1318,20 @@ static int ghes_probe(struct platform_device *ghes_dev)
 
 	ghes_edac_register(ghes, &ghes_dev->dev);
 
-	if (!refcount_read(&ghes_ref_count))
+	if (!refcount_read(&ghes_ref_count)) {
 		refcount_set(&ghes_ref_count, 1);
-	else
+		/* register handler functions for the standard errors.
+		 * This may be done from the corresponding drivers.
+		 */
+		for (i = 0; handler_tab[i].handle; i++) {
+			if (ghes_error_notify_register(handler_tab[i].sec_type,
+						handler_tab[i].handle, NULL)) {
+				ghes_edac_unregister(ghes);
+				platform_set_drvdata(ghes_dev, NULL);
+				goto err;
+			}
+		}
+	} else
 		refcount_inc(&ghes_ref_count);
 
 	/* Handle any pending errors right away */
-- 
1.9.1

Re: [PATCH RFC 2/4] ACPI: APEI: Add ghes_handle_memory_failure to the new notification method

[James Morse]

Hi,

On 12/08/2019 11:11, Shiju Jose wrote:
> This patch adds ghes_handle_memory_failure to the new error
> notification method.

The commit message doesn't answer the question: why?

The existing code works. This just looks like additional churn.
Given a user, I think the vendor specific example is useful. I don't think making this
thing more pluggable is a good idea.


Thanks,

James

RE: [PATCH RFC 2/4] ACPI: APEI: Add ghes_handle_memory_failure to the new notification method

[Shiju Jose]

Hi James,

>-----Original Message-----
>From: James Morse [mailto:james.morse@arm.com]
>Sent: 21 August 2019 18:23
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-edac@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; lenb@kernel.org;
>tony.luck@intel.com; bp@alien8.de; baicar@os.amperecomputing.com;
>Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>
>Subject: Re: [PATCH RFC 2/4] ACPI: APEI: Add ghes_handle_memory_failure to
>the new notification method
>
>Hi,
>
>On 12/08/2019 11:11, Shiju Jose wrote:
>> This patch adds ghes_handle_memory_failure to the new error
>> notification method.
>
>The commit message doesn't answer the question: why?
>
>The existing code works. This just looks like additional churn.
>Given a user, I think the vendor specific example is useful. I don't think making
>this thing more pluggable is a good idea.
This was intended to replace the  number of if(guid_equal(...)) else if(guid_equal(...)) checks in the ghes_do_proc() , which would grow when new UEFI defined error sections would be added in the future.
>
>
>Thanks,
>
>James

Thanks,
Shiju

[PATCH RFC 3/4] ACPI: APEI: Add ghes_handle_aer to the new notification method

[Shiju Jose]

This patch adds ghes_handle_aer to the new error notification method.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 4400d56..ffc309c 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -450,7 +450,8 @@ static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata,
  * GHES_SEV_PANIC does not make it to this handling since the kernel must
  *     panic.
  */
-static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
+static void ghes_handle_aer(struct acpi_hest_generic_data *gdata,
+			    int sev, void *data)
 {
 #ifdef CONFIG_ACPI_APEI_PCIEAER
 	struct cper_sec_pcie *pcie_err = acpi_hest_get_payload(gdata);
@@ -573,10 +574,7 @@ static void ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
-		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
-			ghes_handle_aer(gdata);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
+		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);
@@ -1196,6 +1194,10 @@ struct ghes_err_handler_tab {
 		.sec_type = CPER_SEC_PLATFORM_MEM,
 		.handle = ghes_handle_memory_failure,
 	},
+	{
+		.sec_type = CPER_SEC_PCIE,
+		.handle = ghes_handle_aer,
+	},
 	{ /* sentinel */ }
 };
 
-- 
1.9.1

[PATCH RFC 4/4] ACPI: APEI: Add log_arm_hw_error to the new notification method

[Shiju Jose]

This patch adds log_arm_hw_error to the new error notification
method.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 47 ++++++++++++++++++++++-------------------------
 drivers/ras/ras.c        |  5 ++++-
 include/linux/ras.h      |  7 +++++--
 3 files changed, 31 insertions(+), 28 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index ffc309c..013fea0 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -574,34 +574,27 @@ static void ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
-		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);
-		} else {
-			rcu_read_lock();
-			list_for_each_entry_rcu(err_notify,
-						&ghes_error_notify_list, list) {
-				if (guid_equal(&err_notify->sec_type,
-					       sec_type)) {
-					/* The notification is called in the
-					 * interrupt context, thus the handler
-					 * functions should be take care of it.
-					 */
-					err_notify->handle(gdata, sev,
-							   err_notify->data);
-					is_notify = 1;
-				}
+		rcu_read_lock();
+		list_for_each_entry_rcu(err_notify, &ghes_error_notify_list,
+					list) {
+			if (guid_equal(&err_notify->sec_type, sec_type)) {
+				/* The notification is called in the
+				 * interrupt context, thus the handler
+				 * functions should be take care of it.
+				 */
+				err_notify->handle(gdata, sev,
+						   err_notify->data);
+				is_notify = 1;
 			}
-			rcu_read_unlock();
+		}
+		rcu_read_unlock();
 
-			if (!is_notify) {
-				void *err = acpi_hest_get_payload(gdata);
+		if (!is_notify) {
+			void *err = acpi_hest_get_payload(gdata);
 
-				log_non_standard_event(sec_type, fru_id,
-						       fru_text, sec_sev, err,
-						       gdata->error_data_length);
-			}
+			log_non_standard_event(sec_type, fru_id,
+					       fru_text, sec_sev, err,
+					       gdata->error_data_length);
 		}
 	}
 }
@@ -1198,6 +1191,10 @@ struct ghes_err_handler_tab {
 		.sec_type = CPER_SEC_PCIE,
 		.handle = ghes_handle_aer,
 	},
+	{
+		.sec_type = CPER_SEC_PROC_ARM,
+		.handle = log_arm_hw_error,
+	},
 	{ /* sentinel */ }
 };
 
diff --git a/drivers/ras/ras.c b/drivers/ras/ras.c
index 95540ea..7ec3eeb 100644
--- a/drivers/ras/ras.c
+++ b/drivers/ras/ras.c
@@ -21,8 +21,11 @@ void log_non_standard_event(const guid_t *sec_type, const guid_t *fru_id,
 	trace_non_standard_event(sec_type, fru_id, fru_text, sev, err, len);
 }
 
-void log_arm_hw_error(struct cper_sec_proc_arm *err)
+void log_arm_hw_error(struct acpi_hest_generic_data *gdata,
+		      int sev, void *data)
 {
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+
 	trace_arm_event(err);
 }
 
diff --git a/include/linux/ras.h b/include/linux/ras.h
index 7c3debb..05b662d 100644
--- a/include/linux/ras.h
+++ b/include/linux/ras.h
@@ -5,6 +5,7 @@
 #include <asm/errno.h>
 #include <linux/uuid.h>
 #include <linux/cper.h>
+#include <acpi/ghes.h>
 
 #ifdef CONFIG_DEBUG_FS
 int ras_userspace_consumers(void);
@@ -29,7 +30,8 @@ static inline void __init cec_init(void)	{ }
 void log_non_standard_event(const guid_t *sec_type,
 			    const guid_t *fru_id, const char *fru_text,
 			    const u8 sev, const u8 *err, const u32 len);
-void log_arm_hw_error(struct cper_sec_proc_arm *err);
+void log_arm_hw_error(struct acpi_hest_generic_data *gdata,
+		      int sev, void *data);
 #else
 static inline void
 log_non_standard_event(const guid_t *sec_type,
@@ -37,7 +39,8 @@ void log_non_standard_event(const guid_t *sec_type,
 		       const u8 sev, const u8 *err, const u32 len)
 { return; }
 static inline void
-log_arm_hw_error(struct cper_sec_proc_arm *err) { return; }
+log_arm_hw_error(struct acpi_hest_generic_data *gdata,
+		 int sev, void *data) { return; }
 #endif
 
 #endif /* __RAS_H__ */
-- 
1.9.1

Re: [PATCH RFC 0/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi,

On 12/08/2019 11:11, Shiju Jose wrote:
> Presently kernel does not support reporting the vendor specific HW errors,
> in the non-standard format, to the vendor drivers for the recovery.

'non standard' here is probably a little jarring to the casual reader. You're referring to
the UEFI spec's "N.2.3 Non-standard Section Body", which refers to any section type
published somewhere other than the UEFI spec.

These still have to have a GUID to identify them, so they still have the same section
header format.


> This patch set add this support and also move the existing handler
> functions for the standard errors to the new callback method.

Could you give an example of where this would be useful? You're adding an API with no
caller to justify its existence.


GUIDs should only belong to one driver.

I don't think we should call drivers for something described as a fatal error. (which is
the case with what you have here)


> Also the CCIX RAS patches could be move to the proposed callback method.

Presumably for any vendor-specific stuff?


Thanks,

James

RE: [PATCH RFC 0/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi James, 

Thanks for the feedback.

>-----Original Message-----
>From: linux-acpi-owner@vger.kernel.org [mailto:linux-acpi-
>owner@vger.kernel.org] On Behalf Of James Morse
>Sent: 21 August 2019 18:23
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-edac@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; lenb@kernel.org;
>tony.luck@intel.com; bp@alien8.de; baicar@os.amperecomputing.com;
>Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>
>Subject: Re: [PATCH RFC 0/4] ACPI: APEI: Add support to notify the vendor
>specific HW errors
>
>Hi,
>
>On 12/08/2019 11:11, Shiju Jose wrote:
>> Presently kernel does not support reporting the vendor specific HW
>> errors, in the non-standard format, to the vendor drivers for the recovery.
>
>'non standard' here is probably a little jarring to the casual reader. You're
>referring to the UEFI spec's "N.2.3 Non-standard Section Body", which refers to
>any section type published somewhere other than the UEFI spec.
OK. I will change it.  
>
>These still have to have a GUID to identify them, so they still have the same
>section header format.
Yes. 
 
>
>
>> This patch set add this support and also move the existing handler
>> functions for the standard errors to the new callback method.
>
>Could you give an example of where this would be useful? You're adding an API
>with no caller to justify its existence.
One such example is handling the local errors occurred in a device controller, such as PCIe.

>
>
>GUIDs should only belong to one driver.
UEFI spec's N.2.3 Non-standard Section Body mentioned,  "The type (e.g. format) of a non-standard section is identified by the GUID populated in the Section Descriptor's Section Type field." 
There is a possibility to define common non-standard error section format which will be used for more than one driver if the error data to be reported is in the same format. Then can the same GUID belong to multiple drivers?

>
>I don't think we should call drivers for something described as a fatal error.
>(which is the case with what you have here)
The notification is intended only for the recoverable errors as the ghes_proc() call panic for the fatal errors in the early stage.

>
>
>> Also the CCIX RAS patches could be move to the proposed callback method.
>
>Presumably for any vendor-specific stuff?
This information was related to the proposal to replace the  number of if(guid_equal(...)) else if(guid_equal(...)) checks in the ghes_do_proc() for the existing UEFI spec defined error sections(such as PCIe,  Memory, ARM HW error) by registering the corresponding handler functions to the proposed notification method. The same apply to the CCIX error sections and any other error sections defined by the UEFI spec in the future.  

>
>
>Thanks,
>
>James

Thanks,
Shiju

Re: [PATCH RFC 0/4] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi Shiju,

On 22/08/2019 17:56, Shiju Jose wrote:
> James Morse wrote:
>> On 12/08/2019 11:11, Shiju Jose wrote:
>>> Presently kernel does not support reporting the vendor specific HW
>>> errors, in the non-standard format, to the vendor drivers for the recovery.
>>
>> 'non standard' here is probably a little jarring to the casual reader. You're
>> referring to the UEFI spec's "N.2.3 Non-standard Section Body", which refers to
>> any section type published somewhere other than the UEFI spec.

>>> This patch set add this support and also move the existing handler
>>> functions for the standard errors to the new callback method.
>>
>> Could you give an example of where this would be useful? You're adding an API
>> with no caller to justify its existence.

> One such example is handling the local errors occurred in a device controller, such as PCIe.

Could we have the example in the form of patches? (sorry, I wasn't clear)

I don't think its realistic that a PCIe device driver would want to know about errors on
other devices in the system. (SAS-HBA meet the GPU).

PCIe's has AER for handling errors that (may have) occurred on a PCIe link, and this has
its own CPER records.


>> GUIDs should only belong to one driver.

> UEFI spec's N.2.3 Non-standard Section Body mentioned,  "The type (e.g. format) of a
> non-standard section is identified by the GUID populated in the Section Descriptor's
> Section Type field." 
> There is a possibility to define common non-standard error section format

I agree the GUID describes the format of the error record,


> which will
> be used for more than one driver if the error data to be reported is in the same format.
> Then can the same GUID belong to multiple drivers?

... but here we disagree.

CPER has a component/block-diagram view of the system. It describes a Memory error or an
error with a PCIe endpoint. An error record affects one component.

If you wanted to describe an error caused by a failed transaction between a PCIe device
and memory, you would need two of these records, and its guesswork as to what happened
between them.

But the PCIe device has no business poking around in the memory error. Even if it did APEI
would be the wrong place to do this as its not the only caller of memory_failure().


>>> Also the CCIX RAS patches could be move to the proposed callback method.
>>
>> Presumably for any vendor-specific stuff?

> This information was related to the proposal to replace the  number of if(guid_equal(...)) else
> if(guid_equal(...)) checks in the ghes_do_proc() for the existing UEFI spec defined error 
> sections(such as PCIe,  Memory, ARM HW error)

'the standard ones'

> by registering the corresponding handler functions to the proposed notification method.

I really don't like this. Registering a handler for 'memory corruption' would require
walking a list of dynamically allocated pointers. Can there be more than one entry? Can
random drivers block memory_failure() while they allocate more memory to send packets over
USB? What if it loops?

For the standard error sources the kernel needs to run 'the' handler as quickly as
possible, with a minimum of code/memory-access in the meantime. It already takes too long.


Thanks,

James


> The same apply to the CCIX error sections and any other
> error sections defined by the UEFI spec in the future.  

[PATCH 0/7] rasdaemon: add fixes, database closure and signal handling

[Shiju Jose]

This patch set add
1. fixes for some memory leaks and file closure.
2. closure for the sqlite3 database.
3. signal handling for the cleanup.

Shiju Jose (7):
  rasdaemon: fix cleanup issues in
    ras-events.c:read_ras_event_all_cpus()
  rasdaemon: fix memory leak in ras-events.c:handle_ras_events()
  rasdaemon: fix missing fclose in
    ras-events.c:select_tracing_timestamp()
  rasdaemon: fix memory leak in ras-events.c:add_event_handler()
  rasdaemon: delete multiple definitions of ARRAY_SIZE
  rasdaemon: add closure and cleanups for the database
  rasdaemon: add signal handling for the cleanup

 ras-diskerror-handler.c    |   2 -
 ras-events.c               |  88 +++++++++++++++++++++++++++-----
 ras-mce-handler.h          |   3 --
 ras-non-standard-handler.c |  16 ++++++
 ras-non-standard-handler.h |   6 ++-
 ras-record.c               | 123 +++++++++++++++++++++++++++++++++++++++++++--
 ras-record.h               |   5 ++
 7 files changed, 222 insertions(+), 21 deletions(-)

-- 
2.1.4

[PATCH 1/7] rasdaemon: fix cleanup issues in ras-events.c:read_ras_event_all_cpus()

[Shiju Jose]

This patch fix memory leaks and close the open files if the
open_trace() or read(fds[i].fd, page, pdata[i].ras->page_size)
function calls fail.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c | 23 ++++++++++++++++-------
 1 file changed, 16 insertions(+), 7 deletions(-)

diff --git a/ras-events.c b/ras-events.c
index 3cdac19..d1773b1 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -353,6 +353,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 	struct pollfd fds[n_cpus];
 	int warnonce[n_cpus];
 	char pipe_raw[PATH_MAX];
+	int legacy_kernel = 0;
 #if 0
 	int need_sleep = 0;
 #endif
@@ -372,6 +373,9 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 		return -ENOMEM;
 	}
 
+	for (i = 0; i < n_cpus; i++)
+		fds[i].fd = -1;
+
 	for (i = 0; i < n_cpus; i++) {
 		fds[i].events = POLLIN;
 
@@ -382,9 +386,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 		fds[i].fd = open_trace(pdata[0].ras, pipe_raw, O_RDONLY);
 		if (fds[i].fd < 0) {
 			log(TERM, LOG_ERR, "Can't open trace_pipe_raw\n");
-			kbuffer_free(kbuf);
-			free(page);
-			return -EINVAL;
+			goto error;
 		}
 	}
 
@@ -416,7 +418,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 			size = read(fds[i].fd, page, pdata[i].ras->page_size);
 			if (size < 0) {
 				log(TERM, LOG_WARNING, "read\n");
-				return -1;
+				goto error;
 			} else if (size > 0) {
 				kbuffer_load_subbuffer(kbuf, page);
 
@@ -441,6 +443,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 		 */
 		if (count_nready == n_cpus) {
 			/* Should only happen with legacy kernels */
+			legacy_kernel = 1;
 			break;
 		}
 #endif
@@ -449,12 +452,18 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 	/* poll() is not supported. We need to fallback to the old way */
 	log(TERM, LOG_INFO,
 	    "Old kernel detected. Stop listening and fall back to pthread way.\n");
+error:
 	kbuffer_free(kbuf);
 	free(page);
-	for (i = 0; i < n_cpus; i++)
-		close(fds[i].fd);
+	for (i = 0; i < n_cpus; i++) {
+		if (fds[i].fd > 0)
+			close(fds[i].fd);
+	}
 
-	return -255;
+	if (legacy_kernel)
+		return -255;
+	else
+		return -1;
 }
 
 static int read_ras_event(int fd,
-- 
2.1.4

[PATCH 2/7] rasdaemon: fix memory leak in ras-events.c:handle_ras_events()

[Shiju Jose]

This patch fix memory leak in handle_ras_events()
when failed to trace all supported RAS events.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ras-events.c b/ras-events.c
index d1773b1..d543251 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -846,7 +846,8 @@ int handle_ras_events(int record_events)
 	if (!num_events) {
 		log(ALL, LOG_INFO,
 		    "Failed to trace all supported RAS events. Aborting.\n");
-		return EINVAL;
+		rc = -EINVAL;
+		goto err;
 	}
 
 	data = calloc(sizeof(*data), cpus);
-- 
2.1.4

[PATCH 3/7] rasdaemon: fix missing fclose in ras-events.c:select_tracing_timestamp()

[Shiju Jose]

This patch adds fix for missing fclose() in select_tracing_timestamp()
when return fail if can't parse /proc/uptime.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ras-events.c b/ras-events.c
index d543251..fc6b288 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -600,12 +600,12 @@ static int select_tracing_timestamp(struct ras_events *ras)
 		return 0;
 	}
 	rc = fscanf(fp, "%zu.%u ", &uptime, &j1);
+	fclose(fp);
 	if (rc <= 0) {
 		log(TERM, LOG_ERR, "Can't parse /proc/uptime!\n");
 		return -1;
 	}
 	now = time(NULL);
-	fclose(fp);
 
 	ras->use_uptime = 1;
 	ras->uptime_diff = now - uptime;
-- 
2.1.4

[PATCH 4/7] rasdaemon: fix memory leak in ras-events.c:add_event_handler()

[Shiju Jose]

This patch rearranges the free(page) call to prevent the
memory leak when __toggle_ras_mc_event() fail.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ras-events.c b/ras-events.c
index fc6b288..f912dae 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -688,6 +688,7 @@ static int add_event_handler(struct ras_events *ras, struct pevent *pevent,
 
 	/* Enable RAS events */
 	rc = __toggle_ras_mc_event(ras, group, event, 1);
+	free(page);
 	if (rc < 0) {
 		log(TERM, LOG_ERR, "Can't enable %s:%s tracing\n",
 		    group, event);
@@ -697,7 +698,6 @@ static int add_event_handler(struct ras_events *ras, struct pevent *pevent,
 
 	log(ALL, LOG_INFO, "Enabled event %s:%s\n", group, event);
 
-	free(page);
 	return 0;
 }
 
-- 
2.1.4

[PATCH 5/7] rasdaemon: delete multiple definitions of ARRAY_SIZE

[Shiju Jose]

This patch deletes multiple definitions of ARRAY_SIZE and
move the definition to a common file.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-diskerror-handler.c | 2 --
 ras-mce-handler.h       | 3 ---
 ras-record.c            | 3 ---
 ras-record.h            | 2 ++
 4 files changed, 2 insertions(+), 8 deletions(-)

diff --git a/ras-diskerror-handler.c b/ras-diskerror-handler.c
index 271dfac..68c0c77 100644
--- a/ras-diskerror-handler.c
+++ b/ras-diskerror-handler.c
@@ -50,8 +50,6 @@ static const struct {
 	{ -EIO,       "I/O error" },
 };
 
-#define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
-
 static const char *get_blk_error(int err)
 {
 	int i;
diff --git a/ras-mce-handler.h b/ras-mce-handler.h
index 94395eb..4d615b4 100644
--- a/ras-mce-handler.h
+++ b/ras-mce-handler.h
@@ -24,9 +24,6 @@
 #include "ras-events.h"
 #include "libtrace/event-parse.h"
 
-
-#define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
-
 enum cputype {
 	CPU_GENERIC,
 	CPU_P6OLD,
diff --git a/ras-record.c b/ras-record.c
index ae5d359..8f1c550 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -35,9 +35,6 @@
 
 #define SQLITE_RAS_DB RASSTATEDIR "/" RAS_DB_FNAME
 
-
-#define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
-
 /*
  * Table and functions to handle ras:mc_event
  */
diff --git a/ras-record.h b/ras-record.h
index 5311c67..c9af5ae 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -23,6 +23,8 @@
 #include <stdint.h>
 #include "config.h"
 
+#define ARRAY_SIZE(x) (sizeof(x)/sizeof(*(x)))
+
 extern long user_hz;
 
 struct ras_events *ras;
-- 
2.1.4

[PATCH 6/7] rasdaemon: add closure and cleanups for the database

[Shiju Jose]

This patch adds closure and cleanups for the sqlite3 database.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c               |  14 +++++-
 ras-non-standard-handler.c |  16 ++++++
 ras-non-standard-handler.h |   6 ++-
 ras-record.c               | 120 +++++++++++++++++++++++++++++++++++++++++++++
 ras-record.h               |   3 ++
 5 files changed, 157 insertions(+), 2 deletions(-)

diff --git a/ras-events.c b/ras-events.c
index f912dae..d155caa 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -418,7 +418,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 			size = read(fds[i].fd, page, pdata[i].ras->page_size);
 			if (size < 0) {
 				log(TERM, LOG_WARNING, "read\n");
-				goto error;
+				goto cleanup;
 			} else if (size > 0) {
 				kbuffer_load_subbuffer(kbuf, page);
 
@@ -452,6 +452,13 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 	/* poll() is not supported. We need to fallback to the old way */
 	log(TERM, LOG_INFO,
 	    "Old kernel detected. Stop listening and fall back to pthread way.\n");
+
+cleanup:
+	if (pdata[0].ras->record_events) {
+		unregister_ns_dec_tab();
+		ras_mc_event_closedb(pdata[0].cpu, pdata[0].ras);
+	}
+
 error:
 	kbuffer_free(kbuf);
 	free(page);
@@ -540,6 +547,11 @@ static void *handle_ras_events_cpu(void *priv)
 
 	read_ras_event(fd, pdata, kbuf, page);
 
+	if (pdata->ras->record_events) {
+		unregister_ns_dec_tab();
+		ras_mc_event_closedb(pdata->cpu, pdata->ras);
+	}
+
 	close(fd);
 	kbuffer_free(kbuf);
 	free(page);
diff --git a/ras-non-standard-handler.c b/ras-non-standard-handler.c
index 1b5d67a..d92fd42 100644
--- a/ras-non-standard-handler.c
+++ b/ras-non-standard-handler.c
@@ -41,6 +41,22 @@ int register_ns_dec_tab(const p_ns_dec_tab tab)
 void unregister_ns_dec_tab(void)
 {
 	if (ns_dec_tab) {
+#ifdef HAVE_SQLITE3
+		p_ns_dec_tab dec_tab;
+		int i, count;
+
+		for (count = 0; count < dec_tab_count; count++) {
+			dec_tab = ns_dec_tab[count];
+			for (i = 0; dec_tab[i].decode; i++) {
+				if (dec_tab[i].stmt_dec_record) {
+					ras_mc_finalize_vendor_table(
+						dec_tab[i].stmt_dec_record);
+					dec_tab[i].stmt_dec_record = NULL;
+				}
+			}
+		}
+#endif
+
 		free(ns_dec_tab);
 		ns_dec_tab = NULL;
 		dec_tab_count = 0;
diff --git a/ras-non-standard-handler.h b/ras-non-standard-handler.h
index fd9dd92..2b9bf40 100644
--- a/ras-non-standard-handler.h
+++ b/ras-non-standard-handler.h
@@ -36,8 +36,12 @@ int ras_non_standard_event_handler(struct trace_seq *s,
 
 void print_le_hex(struct trace_seq *s, const uint8_t *buf, int index);
 
+#ifdef HAVE_NON_STANDARD
 int register_ns_dec_tab(const p_ns_dec_tab tab);
-
 void unregister_ns_dec_tab(void);
+#else
+static inline int register_ns_dec_tab(const p_ns_dec_tab tab) { return 0; };
+static inline void unregister_ns_dec_tab(void) { return; };
+#endif
 
 #endif
diff --git a/ras-record.c b/ras-record.c
index 8f1c550..c6cf61a 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -595,6 +595,18 @@ int ras_mc_add_vendor_table(struct ras_events *ras,
 	return rc;
 }
 
+int ras_mc_finalize_vendor_table(sqlite3_stmt *stmt)
+{
+	int rc;
+
+	rc = sqlite3_finalize(stmt);
+	if (rc != SQLITE_OK)
+		log(TERM, LOG_ERR,
+		    "Failed to finalize sqlite: error = %d\n", rc);
+
+	return rc;
+}
+
 int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
 {
 	int rc;
@@ -692,3 +704,111 @@ int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras)
 		ras->db_priv = priv;
 	return 0;
 }
+
+int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras)
+{
+	int rc;
+	sqlite3 *db;
+	struct sqlite3_priv *priv = ras->db_priv;
+
+	printf("Calling %s()\n", __func__);
+
+	if (!priv)
+		return -1;
+
+	db = priv->db;
+	if (!db)
+		return -1;
+
+	if (priv->stmt_mc_event) {
+		rc = sqlite3_finalize(priv->stmt_mc_event);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize mc_event sqlite: error = %d\n",
+			    cpu, rc);
+	}
+
+#ifdef HAVE_AER
+	if (priv->stmt_aer_event) {
+		rc = sqlite3_finalize(priv->stmt_aer_event);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize aer_event sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+#ifdef HAVE_EXTLOG
+	if (priv->stmt_extlog_record) {
+		rc = sqlite3_finalize(priv->stmt_extlog_record);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize extlog_record sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+
+#ifdef HAVE_MCE
+	if (priv->stmt_mce_record) {
+		rc = sqlite3_finalize(priv->stmt_mce_record);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize mce_record sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+#ifdef HAVE_NON_STANDARD
+	if (priv->stmt_non_standard_record) {
+		rc = sqlite3_finalize(priv->stmt_non_standard_record);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize non_standard_record sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+#ifdef HAVE_ARM
+	if (priv->stmt_arm_record) {
+		rc = sqlite3_finalize(priv->stmt_arm_record);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize arm_record sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+#ifdef HAVE_DEVLINK
+	if (priv->stmt_devlink_event) {
+		rc = sqlite3_finalize(priv->stmt_devlink_event);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize devlink_event sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+#ifdef HAVE_DISKERROR
+	if (priv->stmt_diskerror_event) {
+		rc = sqlite3_finalize(priv->stmt_diskerror_event);
+		if (rc != SQLITE_OK)
+			log(TERM, LOG_ERR,
+			    "cpu %u: Failed to finalize diskerror_event sqlite: error = %d\n",
+			    cpu, rc);
+	}
+#endif
+
+	rc = sqlite3_close_v2(db);
+	if (rc != SQLITE_OK)
+		log(TERM, LOG_ERR,
+		    "cpu %u: Failed to close sqlite: error = %d\n", cpu, rc);
+
+	rc = sqlite3_shutdown();
+	if (rc != SQLITE_OK)
+		log(TERM, LOG_ERR,
+		    "cpu %u: Failed to shutdown sqlite: error = %d\n", cpu, rc);
+	free(priv);
+
+	return 0;
+}
diff --git a/ras-record.h b/ras-record.h
index c9af5ae..a67b193 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -147,8 +147,10 @@ struct db_table_descriptor {
 };
 
 int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras);
+int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras);
 int ras_mc_add_vendor_table(struct ras_events *ras, sqlite3_stmt **stmt,
 			    const struct db_table_descriptor *db_tab);
+int ras_mc_finalize_vendor_table(sqlite3_stmt *stmt);
 int ras_store_mc_event(struct ras_events *ras, struct ras_mc_event *ev);
 int ras_store_aer_event(struct ras_events *ras, struct ras_aer_event *ev);
 int ras_store_mce_record(struct ras_events *ras, struct mce_event *ev);
@@ -160,6 +162,7 @@ int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev
 
 #else
 static inline int ras_mc_event_opendb(unsigned cpu, struct ras_events *ras) { return 0; };
+static inline int ras_mc_event_closedb(unsigned int cpu, struct ras_events *ras) { return 0; };
 static inline int ras_store_mc_event(struct ras_events *ras, struct ras_mc_event *ev) { return 0; };
 static inline int ras_store_aer_event(struct ras_events *ras, struct ras_aer_event *ev) { return 0; };
 static inline int ras_store_mce_record(struct ras_events *ras, struct mce_event *ev) { return 0; };
-- 
2.1.4

[PATCH 7/7] rasdaemon: add signal handling for the cleanup

[Shiju Jose]

Presently rasdaemon would not free allocated memory and
would not do other cleanup when the rasdaemon closed
with ctrl+c or kill etc.
This patch adds handling of the signals SIGINT, SIGTERM, SIGHUP
and SIGQUIT and do necessary clean ups when receive the
specified signals.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-events.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 46 insertions(+), 4 deletions(-)

diff --git a/ras-events.c b/ras-events.c
index d155caa..511c93d 100644
--- a/ras-events.c
+++ b/ras-events.c
@@ -25,6 +25,8 @@
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/poll.h>
+#include <signal.h>
+#include <sys/signalfd.h>
 #include "libtrace/kbuffer.h"
 #include "libtrace/event-parse.h"
 #include "ras-mc-handler.h"
@@ -350,7 +352,9 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 	int ready, i, count_nready;
 	struct kbuffer *kbuf;
 	void *page;
-	struct pollfd fds[n_cpus];
+	struct pollfd fds[n_cpus + 1];
+	struct signalfd_siginfo fdsiginfo;
+	sigset_t mask;
 	int warnonce[n_cpus];
 	char pipe_raw[PATH_MAX];
 	int legacy_kernel = 0;
@@ -373,7 +377,7 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 		return -ENOMEM;
 	}
 
-	for (i = 0; i < n_cpus; i++)
+	for (i = 0; i < (n_cpus + 1); i++)
 		fds[i].fd = -1;
 
 	for (i = 0; i < n_cpus; i++) {
@@ -390,15 +394,51 @@ static int read_ras_event_all_cpus(struct pthread_data *pdata,
 		}
 	}
 
+	sigemptyset(&mask);
+	sigaddset(&mask, SIGINT);
+	sigaddset(&mask, SIGTERM);
+	sigaddset(&mask, SIGHUP);
+	sigaddset(&mask, SIGQUIT);
+	if (sigprocmask(SIG_BLOCK, &mask, NULL) == -1)
+		log(TERM, LOG_WARNING, "sigprocmask\n");
+	fds[n_cpus].events = POLLIN;
+	fds[n_cpus].fd = signalfd(-1, &mask, 0);
+	if (fds[n_cpus].fd < 0) {
+		log(TERM, LOG_WARNING, "signalfd\n");
+		goto error;
+	}
+
 	log(TERM, LOG_INFO, "Listening to events for cpus 0 to %d\n", n_cpus - 1);
 	if (pdata[0].ras->record_events)
 		ras_mc_event_opendb(pdata[0].cpu, pdata[0].ras);
 
 	do {
-		ready = poll(fds, n_cpus, -1);
+		ready = poll(fds, (n_cpus + 1), -1);
 		if (ready < 0) {
 			log(TERM, LOG_WARNING, "poll\n");
 		}
+
+		/* check for the signal */
+		if (fds[n_cpus].revents & POLLIN) {
+			size = read(fds[n_cpus].fd, &fdsiginfo,
+				    sizeof(struct signalfd_siginfo));
+			if (size != sizeof(struct signalfd_siginfo))
+				log(TERM, LOG_WARNING, "signalfd read\n");
+
+			if (fdsiginfo.ssi_signo == SIGINT ||
+			    fdsiginfo.ssi_signo == SIGTERM ||
+			    fdsiginfo.ssi_signo == SIGHUP ||
+			    fdsiginfo.ssi_signo == SIGQUIT) {
+				log(TERM, LOG_INFO, "Recevied signal=%d\n",
+				    fdsiginfo.ssi_signo);
+				goto  cleanup;
+			} else {
+				log(TERM, LOG_INFO,
+				    "Received unexpected signal=%d\n",
+				    fdsiginfo.ssi_signo);
+			}
+		}
+
 		count_nready = 0;
 		for (i = 0; i < n_cpus; i++) {
 			if (fds[i].revents & POLLERR) {
@@ -462,7 +502,9 @@ cleanup:
 error:
 	kbuffer_free(kbuf);
 	free(page);
-	for (i = 0; i < n_cpus; i++) {
+	sigprocmask(SIG_UNBLOCK, &mask, NULL);
+
+	for (i = 0; i < (n_cpus + 1); i++) {
 		if (fds[i].fd > 0)
 			close(fds[i].fd);
 	}
-- 
2.1.4

RE: [PATCH 0/7] rasdaemon: add fixes, database closure and signal handling

[Shiju Jose]

Hi Mauro,

Can you please review this patch set?

Thanks,
Shiju

>-----Original Message-----
>From: Shiju Jose
>Sent: 16 October 2019 17:34
>To: mchehab@kernel.org; linux-edac@vger.kernel.org
>Cc: Linuxarm <linuxarm@huawei.com>; Shiju Jose <shiju.jose@huawei.com>
>Subject: [PATCH 0/7] rasdaemon: add fixes, database closure and signal
>handling
>
>This patch set add
>1. fixes for some memory leaks and file closure.
>2. closure for the sqlite3 database.
>3. signal handling for the cleanup.
>
>Shiju Jose (7):
>  rasdaemon: fix cleanup issues in
>    ras-events.c:read_ras_event_all_cpus()
>  rasdaemon: fix memory leak in ras-events.c:handle_ras_events()
>  rasdaemon: fix missing fclose in
>    ras-events.c:select_tracing_timestamp()
>  rasdaemon: fix memory leak in ras-events.c:add_event_handler()
>  rasdaemon: delete multiple definitions of ARRAY_SIZE
>  rasdaemon: add closure and cleanups for the database
>  rasdaemon: add signal handling for the cleanup
>
> ras-diskerror-handler.c    |   2 -
> ras-events.c               |  88 +++++++++++++++++++++++++++-----
> ras-mce-handler.h          |   3 --
> ras-non-standard-handler.c |  16 ++++++
> ras-non-standard-handler.h |   6 ++-
> ras-record.c               | 123
>+++++++++++++++++++++++++++++++++++++++++++--
> ras-record.h               |   5 ++
> 7 files changed, 222 insertions(+), 21 deletions(-)
>
>--
>2.1.4
>

Re: [PATCH 0/7] rasdaemon: add fixes, database closure and signal handling

[Mauro Carvalho Chehab]

Em Wed, 13 Nov 2019 16:38:45 +0000
Shiju Jose <shiju.jose@huawei.com> escreveu:

> Hi Mauro,
> 
> Can you please review this patch set?

Reviewed both rasdaemon patchsets and applied.

Thanks!
Mauro

> 
> Thanks,
> Shiju
> 
> >-----Original Message-----
> >From: Shiju Jose
> >Sent: 16 October 2019 17:34
> >To: mchehab@kernel.org; linux-edac@vger.kernel.org
> >Cc: Linuxarm <linuxarm@huawei.com>; Shiju Jose <shiju.jose@huawei.com>
> >Subject: [PATCH 0/7] rasdaemon: add fixes, database closure and signal
> >handling
> >
> >This patch set add
> >1. fixes for some memory leaks and file closure.
> >2. closure for the sqlite3 database.
> >3. signal handling for the cleanup.
> >
> >Shiju Jose (7):
> >  rasdaemon: fix cleanup issues in
> >    ras-events.c:read_ras_event_all_cpus()
> >  rasdaemon: fix memory leak in ras-events.c:handle_ras_events()
> >  rasdaemon: fix missing fclose in
> >    ras-events.c:select_tracing_timestamp()
> >  rasdaemon: fix memory leak in ras-events.c:add_event_handler()
> >  rasdaemon: delete multiple definitions of ARRAY_SIZE
> >  rasdaemon: add closure and cleanups for the database
> >  rasdaemon: add signal handling for the cleanup
> >
> > ras-diskerror-handler.c    |   2 -
> > ras-events.c               |  88 +++++++++++++++++++++++++++-----
> > ras-mce-handler.h          |   3 --
> > ras-non-standard-handler.c |  16 ++++++
> > ras-non-standard-handler.h |   6 ++-
> > ras-record.c               | 123
> >+++++++++++++++++++++++++++++++++++++++++++--
> > ras-record.h               |   5 ++
> > 7 files changed, 222 insertions(+), 21 deletions(-)
> >
> >--
> >2.1.4
> >
> 




Cheers,
Mauro

[PATCH rasdaemon 0/2] rasdaemon: add fix for the sql table

[Shiju Jose]

rasdaemon fail to prepare the sql table when new fields are added on top
of the existing sql table present in the system.

This patch set add solution for this issue and a patch for adding missing
information to the aer_event table.

Shiju Jose (2):
  rasdaemon: fix for the ras-record.c:ras_mc_prepare_stmt() failure when
    new fields added to the sql table
  rasdaemon: store PCIe dev name and TLP header for the aer event

 ras-aer-handler.c | 21 +++++++++++-
 ras-record.c      | 99 +++++++++++++++++++++++++++++++++++++++++++++++++++----
 ras-record.h      |  2 ++
 3 files changed, 115 insertions(+), 7 deletions(-)

-- 
1.9.1

[PATCH rasdaemon 1/2] rasdaemon: fix for the ras-record.c:ras_mc_prepare_stmt() failure when new fields added to the sql table

[Shiju Jose]

rasdaemon fails in the ras_mc_prepare_stmt() function when new fields are
added to the table's db_fields on top of the existing sql table in the
system.

This patch adds solution for this issue.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-record.c | 93 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 89 insertions(+), 4 deletions(-)

diff --git a/ras-record.c b/ras-record.c
index c6cf61a..ca58b22 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -499,10 +499,9 @@ int ras_store_diskerror_event(struct ras_events *ras, struct diskerror_event *ev
 /*
  * Generic code
  */
-
-static int ras_mc_prepare_stmt(struct sqlite3_priv *priv,
-			       sqlite3_stmt **stmt,
-			       const struct db_table_descriptor *db_tab)
+static int __ras_mc_prepare_stmt(struct sqlite3_priv *priv,
+				 sqlite3_stmt **stmt,
+				 const struct db_table_descriptor *db_tab)
 
 {
 	int i, rc;
@@ -578,6 +577,92 @@ static int ras_mc_create_table(struct sqlite3_priv *priv,
 	return rc;
 }
 
+static int ras_mc_alter_table(struct sqlite3_priv *priv,
+			      sqlite3_stmt **stmt,
+			      const struct db_table_descriptor *db_tab)
+{
+	char sql[1024], *p = sql, *end = sql + sizeof(sql);
+	const struct db_fields *field;
+	int col_count;
+	int i, j, rc, found;
+
+	snprintf(p, end - p, "SELECT * FROM %s", db_tab->name);
+	rc = sqlite3_prepare_v2(priv->db, sql, -1, stmt, NULL);
+	if (rc != SQLITE_OK) {
+		log(TERM, LOG_ERR,
+		    "Failed to query fields from the table %s on %s: error = %d\n",
+		    db_tab->name, SQLITE_RAS_DB, rc);
+		return rc;
+	}
+
+	col_count = sqlite3_column_count(*stmt);
+	for (i = 0; i < db_tab->num_fields; i++) {
+		field = &db_tab->fields[i];
+		found = 0;
+		for (j = 0; j < col_count; j++) {
+			if (!strcmp(field->name,
+			    sqlite3_column_name(*stmt, j))) {
+				found = 1;
+				break;
+			}
+		}
+
+		if (!found) {
+			/* add new field */
+			p += snprintf(p, end - p, "ALTER TABLE %s ADD ",
+				      db_tab->name);
+			p += snprintf(p, end - p,
+				      "%s %s", field->name, field->type);
+#ifdef DEBUG_SQL
+			log(TERM, LOG_INFO, "SQL: %s\n", sql);
+#endif
+			rc = sqlite3_exec(priv->db, sql, NULL, NULL, NULL);
+			if (rc != SQLITE_OK) {
+				log(TERM, LOG_ERR,
+				    "Failed to add new field %s to the table %s on %s: error = %d\n",
+				    field->name, db_tab->name,
+				    SQLITE_RAS_DB, rc);
+				return rc;
+			}
+			p = sql;
+			memset(sql, 0, sizeof(sql));
+		}
+	}
+
+	return rc;
+}
+
+static int ras_mc_prepare_stmt(struct sqlite3_priv *priv,
+			       sqlite3_stmt **stmt,
+			       const struct db_table_descriptor *db_tab)
+{
+	int rc;
+
+	rc = __ras_mc_prepare_stmt(priv, stmt, db_tab);
+	if (rc != SQLITE_OK) {
+		log(TERM, LOG_ERR,
+		    "Failed to prepare insert db at table %s (db %s): error = %s\n",
+		    db_tab->name, SQLITE_RAS_DB, sqlite3_errmsg(priv->db));
+
+		log(TERM, LOG_INFO, "Trying to alter db at table %s (db %s)\n",
+		    db_tab->name, SQLITE_RAS_DB);
+
+		rc = ras_mc_alter_table(priv, stmt, db_tab);
+		if (rc != SQLITE_OK && rc != SQLITE_DONE) {
+			log(TERM, LOG_ERR,
+			    "Failed to alter db at table %s (db %s): error = %s\n",
+			    db_tab->name, SQLITE_RAS_DB,
+			    sqlite3_errmsg(priv->db));
+			stmt = NULL;
+			return rc;
+		}
+
+		rc = __ras_mc_prepare_stmt(priv, stmt, db_tab);
+	}
+
+	return rc;
+}
+
 int ras_mc_add_vendor_table(struct ras_events *ras,
 			    sqlite3_stmt **stmt,
 			    const struct db_table_descriptor *db_tab)
-- 
1.9.1

[PATCH rasdaemon 2/2] rasdaemon: store PCIe dev name and TLP header for the aer event

[Shiju Jose]

This patch adds logging and recording of the PCIe dev name and the
TLP header for the aer event.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 ras-aer-handler.c | 21 ++++++++++++++++++++-
 ras-record.c      |  6 ++++--
 ras-record.h      |  2 ++
 3 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/ras-aer-handler.c b/ras-aer-handler.c
index 664f7b4..8ddd439 100644
--- a/ras-aer-handler.c
+++ b/ras-aer-handler.c
@@ -52,6 +52,8 @@ static const char *aer_uncor_errors[32] = {
 	[20] = "Unsupported Request",
 };
 
+#define BUF_LEN	1024
+
 int ras_aer_event_handler(struct trace_seq *s,
 			 struct pevent_record *record,
 			 struct event_format *event, void *context)
@@ -59,11 +61,12 @@ int ras_aer_event_handler(struct trace_seq *s,
 	int len;
 	unsigned long long severity_val;
 	unsigned long long status_val;
+	unsigned long long val;
 	struct ras_events *ras = context;
 	time_t now;
 	struct tm *tm;
 	struct ras_aer_event ev;
-	char buf[1024];
+	char buf[BUF_LEN];
 
 	/*
 	 * Newer kernels (3.10-rc1 or upper) provide an uptime clock.
@@ -89,6 +92,7 @@ int ras_aer_event_handler(struct trace_seq *s,
 					   record, &len, 1);
 	if (!ev.dev_name)
 		return -1;
+	trace_seq_printf(s, "%s ", ev.dev_name);
 
 	if (pevent_get_field_val(s,  event, "status", record, &status_val, 1) < 0)
 		return -1;
@@ -104,6 +108,21 @@ int ras_aer_event_handler(struct trace_seq *s,
 	else
 		bitfield_msg(buf, sizeof(buf), aer_uncor_errors, 32, 0, 0, status_val);
 	ev.msg = buf;
+
+	if (pevent_get_field_val(s, event, "tlp_header_valid",
+				record, &val, 1) < 0)
+		return -1;
+
+	ev.tlp_header_valid = val;
+	if (ev.tlp_header_valid) {
+		ev.tlp_header = pevent_get_field_raw(s, event, "tlp_header",
+						     record, &len, 1);
+		snprintf((buf + strlen(ev.msg)), BUF_LEN - strlen(ev.msg),
+			 " TLP Header: %08x %08x %08x %08x",
+			 ev.tlp_header[0], ev.tlp_header[1],
+			 ev.tlp_header[2], ev.tlp_header[3]);
+	}
+
 	trace_seq_printf(s, "%s ", ev.msg);
 
 	/* Use hw_event_aer_err_type switch between different severity_val */
diff --git a/ras-record.c b/ras-record.c
index ca58b22..318bace 100644
--- a/ras-record.c
+++ b/ras-record.c
@@ -106,6 +106,7 @@ int ras_store_mc_event(struct ras_events *ras, struct ras_mc_event *ev)
 static const struct db_fields aer_event_fields[] = {
 		{ .name="id",			.type="INTEGER PRIMARY KEY" },
 		{ .name="timestamp",		.type="TEXT" },
+		{ .name="dev_name",		.type="TEXT" },
 		{ .name="err_type",		.type="TEXT" },
 		{ .name="err_msg",		.type="TEXT" },
 };
@@ -126,8 +127,9 @@ int ras_store_aer_event(struct ras_events *ras, struct ras_aer_event *ev)
 	log(TERM, LOG_INFO, "aer_event store: %p\n", priv->stmt_aer_event);
 
 	sqlite3_bind_text(priv->stmt_aer_event,  1, ev->timestamp, -1, NULL);
-	sqlite3_bind_text(priv->stmt_aer_event,  2, ev->error_type, -1, NULL);
-	sqlite3_bind_text(priv->stmt_aer_event,  3, ev->msg, -1, NULL);
+	sqlite3_bind_text(priv->stmt_aer_event,  2, ev->dev_name, -1, NULL);
+	sqlite3_bind_text(priv->stmt_aer_event,  3, ev->error_type, -1, NULL);
+	sqlite3_bind_text(priv->stmt_aer_event,  4, ev->msg, -1, NULL);
 
 	rc = sqlite3_step(priv->stmt_aer_event);
 	if (rc != SQLITE_OK && rc != SQLITE_DONE)
diff --git a/ras-record.h b/ras-record.h
index 440669d..cc217a9 100644
--- a/ras-record.h
+++ b/ras-record.h
@@ -43,6 +43,8 @@ struct ras_aer_event {
 	char timestamp[64];
 	const char *error_type;
 	const char *dev_name;
+	uint8_t tlp_header_valid;
+	uint32_t *tlp_header;
 	const char *msg;
 };
 
-- 
1.9.1

[RFC PATCH 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor drivers are unable to do the recovery for the vendor
specific HW errors, reported to the APEI driver in the vendor defined sections,
because APEI driver does not support reporting the same to the vendor drivers.

This patch set
1. add an interface to the APEI driver to enable the vendor
drivers to register the event handling functions for the corresponding
vendor specific HW errors.

2. add driver to handle HiSilicon hip08 PCIe controller's errors
   which is an application of the above interface.

Changes from the previous version
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an example of the above interface.

Shiju Jose (1):
  ACPI: APEI: Add support to notify the vendor specific HW errors

Yicong Yang (1):
  PCI:hip08:Add driver to handle HiSilicon hip08 PCIe controller's
    errors

 drivers/acpi/apei/ghes.c                       | 110 ++++++++-
 drivers/pci/controller/Kconfig                 |   8 +
 drivers/pci/controller/Makefile                |   1 +
 drivers/pci/controller/pcie-hisi-hip08-error.c | 323 +++++++++++++++++++++++++
 include/acpi/ghes.h                            |  49 ++++
 5 files changed, 486 insertions(+), 5 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-hip08-error.c

-- 
1.9.1

[RFC PATCH 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently APEI does not support reporting the vendor specific
HW errors, received in the vendor defined table entries, to the
vendor drivers for any recovery.

This patch adds the support to register and unregister the
error handling function for the vendor specific HW errors and
notify the registered kernel driver.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 110 ++++++++++++++++++++++++++++++++++++++++++++---
 include/acpi/ghes.h      |  49 +++++++++++++++++++++
 2 files changed, 154 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 8906c80..3ba43b0 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -490,6 +490,103 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+struct ghes_event_notify {
+	struct list_head list;
+	struct rcu_head	rcu_head;
+	guid_t sec_type; /* guid of the error record */
+	ghes_event_handler_t event_handler; /* event handler function */
+	void *data; /* handler driver's private data if any */
+};
+
+/* List to store the registered event handling functions */
+static DEFINE_MUTEX(ghes_event_notify_mutex);
+static LIST_HEAD(ghes_event_handler_list);
+
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data)
+{
+	struct ghes_event_notify *event_notify;
+
+	event_notify = kzalloc(sizeof(*event_notify), GFP_KERNEL);
+	if (!event_notify)
+		return -ENOMEM;
+
+	event_notify->event_handler = event_handler;
+	guid_copy(&event_notify->sec_type, &sec_type);
+	event_notify->data = data;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	list_add_rcu(&event_notify->list, &ghes_event_handler_list);
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_handler);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ */
+void ghes_unregister_event_handler(guid_t sec_type)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, &sec_type)) {
+			list_del_rcu(&event_notify->list);
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	if (!found) {
+		pr_err("Tried to unregister a GHES event handler that has not been registered\n");
+		return;
+	}
+
+	synchronize_rcu();
+	kfree(event_notify);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);
+
+static int ghes_handle_non_standard_event(guid_t *sec_type,
+	struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, sec_type)) {
+			event_notify->event_handler(gdata, sev,
+						    event_notify->data);
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+
+	return found;
+}
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -525,11 +622,14 @@ static void ghes_do_proc(struct ghes *ghes,
 
 			log_arm_hw_error(err);
 		} else {
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
+			if (!ghes_handle_non_standard_event(sec_type, gdata,
+							    sev)) {
+				void *err = acpi_hest_get_payload(gdata);
+
+				log_non_standard_event(sec_type, fru_id,
+						       fru_text, sec_sev, err,
+						       gdata->error_data_length);
+			}
 		}
 	}
 }
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cdd..2564860 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,55 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+/**
+ * typedef ghes_event_handler_t - event handling function
+ * for the non-fatal HW errors.
+ *
+ * @gdata: acpi_hest_generic_data.
+ * @sev: error severity of the entire error event defined in the
+ *       ACPI spec table generic error status block.
+ * @data: handler driver's private data.
+ *
+ * The error handling function is responsible for logging error and
+ * this function would be called in the interrupt context.
+ */
+typedef void (*ghes_event_handler_t)(struct acpi_hest_generic_data *gdata,
+				     int sev, void *data);
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the event handling function.
+ * @data: handler driver's private data.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ */
+void ghes_unregister_event_handler(guid_t sec_type);
+
+#else
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data)
+{
+	return -ENODEV;
+}
+
+void ghes_unregister_event_handler(guid_t sec_type)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
1.9.1

Re: [RFC PATCH 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 1978 bytes --]

Hi Shiju,

[FYI, it's a private test report for your RFC patch.]
[auto build test ERROR on pm/linux-next]
[also build test ERROR on pci/next linux/master linus/master v5.5-rc6 next-20200117]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200116-203108
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next
config: i386-randconfig-f001-20200118 (attached as .config)
compiler: gcc-7 (Debian 7.5.0-3) 7.5.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   ld: drivers/firmware/efi/cper.o: in function `ghes_register_event_handler':
>> include/acpi/ghes.h:93: multiple definition of `ghes_register_event_handler'; drivers/acpi/apei/hest.o:include/acpi/ghes.h:93: first defined here
   ld: drivers/firmware/efi/cper.o: in function `ghes_unregister_event_handler':
>> include/acpi/ghes.h:98: multiple definition of `ghes_unregister_event_handler'; drivers/acpi/apei/hest.o:include/acpi/ghes.h:98: first defined here

vim +93 include/acpi/ghes.h

    88	
    89	#else
    90	int ghes_register_event_handler(guid_t sec_type,
    91					ghes_event_handler_t event_handler,
    92					void *data)
  > 93	{
    94		return -ENODEV;
    95	}
    96	
    97	void ghes_unregister_event_handler(guid_t sec_type)
  > 98	{
    99	}
   100	#endif
   101	

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org Intel Corporation

[-- Attachment #2: config.gz --]
[-- Type: application/gzip, Size: 38339 bytes --]

[RFC PATCH 2/2] PCI:hip08:Add driver to handle HiSilicon hip08 PCIe controller's errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The hip08 error handle driver logs and reports PCIe controller's
recoverable errors.
Perform root port reset and restore link status for the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

RFC: The appropriate location for this driver may be discussed.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/pci/controller/Kconfig                 |   8 +
 drivers/pci/controller/Makefile                |   1 +
 drivers/pci/controller/pcie-hisi-hip08-error.c | 323 +++++++++++++++++++++++++
 3 files changed, 332 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-hip08-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index c77069c..0ee99b8 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_HIP08_ERR_HANDLER
+	depends on ARM64 || COMPILE_TEST
+	depends on (ACPI && PCI_QUIRKS)
+	bool "HiSilicon hip08 Soc PCIe local error handling driver"
+	help
+	  Say Y here if you want PCIe error handling support
+	  for the PCIe local(controller) errors on HiSilicon hip08 SoC
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 3d4f597..ac9852f 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
 obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
+obj-$(CONFIG_PCIE_HISI_HIP08_ERR_HANDLER) += pcie-hisi-hip08-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-hip08-error.c b/drivers/pci/controller/pcie-hisi-hip08-error.c
new file mode 100644
index 0000000..6f5d002
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-hip08-error.c
@@ -0,0 +1,323 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * PCIe driver for handling PCIe local errors occurred on
+ * HiSilicon hip08 PCIe controller.
+ *
+ * Copyright (c) 2018-2019 Hisilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/bitfield.h>
+#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/list.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/resource.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#include "../pci.h"
+
+#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe Local error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+guid_t hip08_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
+				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       ((v >> 1) + (core << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        ((v % 8) << 1)
+#define HISI_PCIE_ROOT_BUSNR(v)          ((v) ? 0x80 : 0)
+
+struct hisi_pcie_local_err_data {
+	uint64_t   val_bits;
+	uint8_t    version;
+	uint8_t    soc_id;
+	uint8_t    socket_id;
+	uint8_t    nimbus_id;
+	uint8_t    sub_module_id;
+	uint8_t    core_id;
+	uint8_t    port_id;
+	uint8_t    err_severity;
+	uint16_t   err_type;
+	uint8_t    reserv[2];
+	uint32_t   err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct pcie_err_info {
+	struct hisi_pcie_local_err_data err_data;
+	struct platform_device *pdev;
+};
+
+static char *pcie_local_sub_module_name(uint8_t id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *err_severity(uint8_t err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static struct pci_dev *hisi_hip08_pcie_get_rp(u32 chip_id, u32 port_id)
+{
+	u32 devfn = PCI_DEVFN(port_id, 0);
+	u32 busnr = HISI_PCIE_ROOT_BUSNR(chip_id);
+
+	return pci_get_domain_bus_and_slot(0, busnr, devfn);
+}
+
+static int hisi_hip08_pcie_port_acpi_reset(struct platform_device *pdev,
+					u32 chip_id, u32 port_id)
+{
+	struct device *dev = &(pdev->dev);
+	struct acpi_object_list arg_list;
+	union acpi_object arg[3];
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+	/* Call the ACPI handle to reset root port  */
+	if (ACPI_HANDLE(dev)) {
+		unsigned long long data = 0;
+		acpi_status s;
+
+		s = acpi_evaluate_integer(ACPI_HANDLE(dev),
+				"RST", &arg_list, &data);
+
+		if (ACPI_FAILURE(s)) {
+			dev_err(dev, "No Reset method\n");
+			return -EIO;
+		}
+
+		if (data) {
+			dev_err(dev, "Failed to Reset\n");
+			return -EIO;
+		}
+
+	} else {
+		dev_err(dev, "No Reset method\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int hisi_hip08_pcie_port_reset(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	struct pci_dev *pdev;
+	struct pci_bus *root_bus;
+
+	pdev = hisi_hip08_pcie_get_rp(chip_id, port_id);
+	if (!pdev) {
+		dev_info(&(dev->dev), "Fail to get root port device\n");
+		return -ENODEV;
+	}
+	root_bus = pdev->bus;
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+
+	if (hisi_hip08_pcie_port_acpi_reset(dev, chip_id, port_id))
+		return -EIO;
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void pcie_local_error_handle(const struct hisi_pcie_local_err_data *err,
+				    struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &(pdev->dev);
+	uint32_t i;
+	int rc;
+
+	if (err->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      pcie_local_sub_module_name(err->sub_module_id));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      err_severity(err->err_severity));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI HIP08: PCIe local error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe local errors */
+	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_hip08_pcie_port_reset(pdev, err->socket_id,
+				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hip08 pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static DEFINE_KFIFO(pcie_err_recover_ring, struct pcie_err_info,
+		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
+static DEFINE_SPINLOCK(pcie_err_recover_ring_lock);
+
+static void pcie_local_err_recover_work_func(struct work_struct *work)
+{
+	struct pcie_err_info pcie_err_entry;
+
+	while (kfifo_get(&pcie_err_recover_ring, &pcie_err_entry)) {
+		pcie_local_error_handle(&pcie_err_entry.err_data,
+					pcie_err_entry.pdev);
+	}
+}
+
+static DECLARE_WORK(pcie_err_recover_work, pcie_local_err_recover_work_func);
+
+
+static void hip08_pcie_local_error_handle(struct acpi_hest_generic_data *gdata,
+					  int sev, void *data)
+{
+	const struct hisi_pcie_local_err_data *err_data =
+					acpi_hest_get_payload(gdata);
+	struct pcie_err_info err_info;
+	struct platform_device *pdev = data;
+	struct device *dev = &(pdev->dev);
+
+	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
+	err_info.pdev = pdev;
+
+	if (kfifo_in_spinlocked(&pcie_err_recover_ring, &err_info, 1,
+				&pcie_err_recover_ring_lock))
+		schedule_work(&pcie_err_recover_work);
+	else
+		dev_warn(dev, "Buffer overflow when recovering PCIe local error\n");
+}
+
+static int hisi_hip08_pcie_err_handler_probe(struct platform_device *pdev)
+{
+	int ret = 0;
+
+	if (ghes_register_event_handler(hip08_pcie_sec_type,
+					hip08_pcie_local_error_handle,
+					pdev)) {
+		dev_err(&(pdev->dev), "%s : ghes_register_event_handler fail\n",
+			__func__);
+		return ret;
+}
+
+	return 0;
+}
+
+static int hisi_hip08_pcie_err_handler_remove(struct platform_device *pdev)
+{
+	ghes_unregister_event_handler(hip08_pcie_sec_type);
+
+	return 0;
+}
+
+static const struct acpi_device_id hip08_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_hip08_pcie_err_handler_driver = {
+	.driver = {
+		.name	= "hisi-hip08-pcie-err-handler",
+		.acpi_match_table = hip08_pcie_acpi_match,
+	},
+	.probe		= hisi_hip08_pcie_err_handler_probe,
+	.remove		= hisi_hip08_pcie_err_handler_remove,
+};
+module_platform_driver(hisi_hip08_pcie_err_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP08 PCIe controller error handling driver");
+MODULE_LICENSE("GPL v2");
+
-- 
1.9.1

Re: [RFC PATCH 2/2] PCI:hip08:Add driver to handle HiSilicon hip08 PCIe controller's errors

[Bjorn Helgaas]

Follow convention for subject line.

On Wed, Jan 15, 2020 at 11:01:40AM +0000, Shiju Jose wrote:
> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> The hip08 error handle driver logs and reports PCIe controller's
> recoverable errors.
> Perform root port reset and restore link status for the recovery.
> 
> Following are some of the PCIe controller's recoverable errors
> 1. completion transmission timeout error.
> 2. CRS retry counter over the threshold error.
> 3. ECC 2 bit errors
> 4. AXI bresponse/rresponse errors etc.
> 
> RFC: The appropriate location for this driver may be discussed.
> 
> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
> ---
>  drivers/pci/controller/Kconfig                 |   8 +
>  drivers/pci/controller/Makefile                |   1 +
>  drivers/pci/controller/pcie-hisi-hip08-error.c | 323 +++++++++++++++++++++++++

Seems like this driver and its Kconfig should be near
drivers/pci/controller/dwc/pcie-hisi.c.

>  3 files changed, 332 insertions(+)
>  create mode 100644 drivers/pci/controller/pcie-hisi-hip08-error.c
> 
> diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
> index c77069c..0ee99b8 100644
> --- a/drivers/pci/controller/Kconfig
> +++ b/drivers/pci/controller/Kconfig
> @@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
>  	  The Hyper-V PCI Interface is a helper driver allows other drivers to
>  	  have a common interface with the Hyper-V PCI frontend driver.
>  
> +config PCIE_HISI_HIP08_ERR_HANDLER

Config symbol is too long, maybe CONFIG_PCI_HISI_ERR or similar (to be
parallel with existing CONFIG_PCI_HISI).  Both should probably be
"CONFIG_PCIE", not "CONFIG_PCI".  I can't remember why CONFIG_PCI_HISI
is that way.

> +	depends on ARM64 || COMPILE_TEST
> +	depends on (ACPI && PCI_QUIRKS)
> +	bool "HiSilicon hip08 Soc PCIe local error handling driver"
> +	help
> +	  Say Y here if you want PCIe error handling support
> +	  for the PCIe local(controller) errors on HiSilicon hip08 SoC
> +
>  source "drivers/pci/controller/dwc/Kconfig"
>  source "drivers/pci/controller/cadence/Kconfig"
>  endmenu
> diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
> index 3d4f597..ac9852f 100644
> --- a/drivers/pci/controller/Makefile
> +++ b/drivers/pci/controller/Makefile
> @@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
>  obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
>  obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
>  obj-$(CONFIG_VMD) += vmd.o
> +obj-$(CONFIG_PCIE_HISI_HIP08_ERR_HANDLER) += pcie-hisi-hip08-error.o
>  # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
>  obj-y				+= dwc/
>  
> diff --git a/drivers/pci/controller/pcie-hisi-hip08-error.c b/drivers/pci/controller/pcie-hisi-hip08-error.c
> new file mode 100644
> index 0000000..6f5d002
> --- /dev/null
> +++ b/drivers/pci/controller/pcie-hisi-hip08-error.c
> @@ -0,0 +1,323 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * PCIe driver for handling PCIe local errors occurred on
> + * HiSilicon hip08 PCIe controller.

"PCIe" occurs too many times in this sentence.  Strictly speaking this
is not a "PCIe driver"; it's a driver for an ACPI device that reports
hip08-related errors.  Hopefully we don't need a separate driver for
every hip* device, so maybe the "hip08" name is too specific.

> + * Copyright (c) 2018-2019 Hisilicon Limited.

Capitalize "HiSilicon" consistently.  Also "hip08"; previous practice
in drivers/pci is "Hip05", "Hip06", so use that unless HiSilicon
itself does it differently.

> +#include <linux/acpi.h>
> +#include <acpi/ghes.h>
> +#include <linux/bitfield.h>
> +#include <linux/bitops.h>
> +#include <linux/delay.h>
> +#include <linux/irq.h>
> +#include <linux/irqdomain.h>
> +#include <linux/list.h>
> +#include <linux/pci.h>
> +#include <linux/platform_device.h>
> +#include <linux/resource.h>
> +#include <linux/kfifo.h>
> +#include <linux/spinlock.h>
> +
> +#include "../pci.h"
> +
> +#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
> +#define	HISI_PCIE_ERR_INFO_SIZE	1024
> +
> +/* HISI PCIe Local error definitions */
> +#define HISI_PCIE_ERR_MISC_REGS	33
> +
> +#define HISI_PCIE_SUB_MODULE_ID_AP	0
> +#define HISI_PCIE_SUB_MODULE_ID_TL	1
> +#define HISI_PCIE_SUB_MODULE_ID_MAC	2
> +#define HISI_PCIE_SUB_MODULE_ID_DL	3
> +#define HISI_PCIE_SUB_MODULE_ID_SDI	4
> +
> +#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
> +#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
> +#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
> +#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
> +#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
> +#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
> +#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
> +#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
> +#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
> +#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
> +
> +#define HISI_ERR_SEV_RECOVERABLE	0
> +#define HISI_ERR_SEV_FATAL		1
> +#define HISI_ERR_SEV_CORRECTED		2
> +#define HISI_ERR_SEV_NONE		3
> +
> +guid_t hip08_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
> +				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
> +
> +#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
> +#define HISI_PCIE_PORT_ID(core, v)       ((v >> 1) + (core << 3))
> +#define HISI_PCIE_CORE_PORT_ID(v)        ((v % 8) << 1)
> +#define HISI_PCIE_ROOT_BUSNR(v)          ((v) ? 0x80 : 0)

Is the root bus number really hard-wired in the chip?  You're saying
the only possible root bus numbers are 0x80 and 0x00?  Typically this
bus number is programmable.

Why parens around "v" (sometimes) but not others and "core"?

> +struct hisi_pcie_local_err_data {
> +	uint64_t   val_bits;
> +	uint8_t    version;
> +	uint8_t    soc_id;
> +	uint8_t    socket_id;
> +	uint8_t    nimbus_id;
> +	uint8_t    sub_module_id;
> +	uint8_t    core_id;
> +	uint8_t    port_id;
> +	uint8_t    err_severity;
> +	uint16_t   err_type;
> +	uint8_t    reserv[2];
> +	uint32_t   err_misc[HISI_PCIE_ERR_MISC_REGS];

Use u64, u8, u32 throughout instead.

> +};
> +
> +struct pcie_err_info {
> +	struct hisi_pcie_local_err_data err_data;
> +	struct platform_device *pdev;
> +};
> +
> +static char *pcie_local_sub_module_name(uint8_t id)
> +{
> +	switch (id) {
> +	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
> +	}
> +
> +	return "unknown";
> +}
> +
> +static char *err_severity(uint8_t err_sev)
> +{
> +	switch (err_sev) {
> +	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
> +	case HISI_ERR_SEV_FATAL: return "fatal";
> +	case HISI_ERR_SEV_CORRECTED: return "corrected";
> +	case HISI_ERR_SEV_NONE: return "none";
> +	}
> +
> +	return "unknown";
> +}
> +
> +static struct pci_dev *hisi_hip08_pcie_get_rp(u32 chip_id, u32 port_id)
> +{
> +	u32 devfn = PCI_DEVFN(port_id, 0);
> +	u32 busnr = HISI_PCIE_ROOT_BUSNR(chip_id);
> +
> +	return pci_get_domain_bus_and_slot(0, busnr, devfn);
> +}
> +
> +static int hisi_hip08_pcie_port_acpi_reset(struct platform_device *pdev,
> +					u32 chip_id, u32 port_id)
> +{
> +	struct device *dev = &(pdev->dev);

Unnecessary parens.  More occurrences below.

> +	struct acpi_object_list arg_list;
> +	union acpi_object arg[3];
> +
> +	arg[0].type = ACPI_TYPE_INTEGER;
> +	arg[0].integer.value = chip_id;
> +	arg[1].type = ACPI_TYPE_INTEGER;
> +	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
> +	arg[2].type = ACPI_TYPE_INTEGER;
> +	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
> +
> +	arg_list.count = 3;
> +	arg_list.pointer = arg;
> +	/* Call the ACPI handle to reset root port  */

s/root port  /root port /

> +	if (ACPI_HANDLE(dev)) {

Restructure this to return early for error and unindent the following,
e.g.,

  acpi_handle handle = ACPI_HANDLE(dev);

  if (!handle) {
    dev_err(...);
    return -EINVAL;
  }

  arg[0].type = ACPI_TYPE_INTEGER;
  ...
  s = acpi_evaluate_integer(handle, ...);

> +		unsigned long long data = 0;
> +		acpi_status s;
> +
> +		s = acpi_evaluate_integer(ACPI_HANDLE(dev),
> +				"RST", &arg_list, &data);
> +
> +		if (ACPI_FAILURE(s)) {
> +			dev_err(dev, "No Reset method\n");
> +			return -EIO;
> +		}
> +
> +		if (data) {
> +			dev_err(dev, "Failed to Reset\n");
> +			return -EIO;
> +		}
> +
> +	} else {
> +		dev_err(dev, "No Reset method\n");
> +		return -EINVAL;
> +	}
> +
> +	return 0;
> +}
> +
> +static int hisi_hip08_pcie_port_reset(struct platform_device *dev,
> +				      u32 chip_id, u32 port_id)
> +{
> +	struct pci_dev *pdev;
> +	struct pci_bus *root_bus;
> +
> +	pdev = hisi_hip08_pcie_get_rp(chip_id, port_id);
> +	if (!pdev) {
> +		dev_info(&(dev->dev), "Fail to get root port device\n");
> +		return -ENODEV;
> +	}
> +	root_bus = pdev->bus;
> +
> +	pci_stop_and_remove_bus_device_locked(pdev);
> +
> +	if (hisi_hip08_pcie_port_acpi_reset(dev, chip_id, port_id))
> +		return -EIO;
> +	ssleep(1UL);

Please include a comment that cites the spec section that requires
this sleep.

> +
> +	/* add root port and downstream devices */
> +	pci_lock_rescan_remove();
> +	pci_rescan_bus(root_bus);
> +	pci_unlock_rescan_remove();
> +
> +	return 0;
> +}
> +
> +static void pcie_local_error_handle(const struct hisi_pcie_local_err_data *err,
> +				    struct platform_device *pdev)
> +{
> +	char buf[HISI_PCIE_ERR_INFO_SIZE];
> +	char *p = buf, *end = buf + sizeof(buf);
> +	struct device *dev = &(pdev->dev);
> +	uint32_t i;
> +	int rc;
> +
> +	if (err->val_bits == 0) {
> +		dev_warn(dev, "%s: no valid error information\n", __func__);
> +		return;
> +	}
> +
> +	/* Logging */
> +	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
> +		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
> +		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
> +		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
> +		p += snprintf(p, end - p, "sub module=%s ",
> +			      pcie_local_sub_module_name(err->sub_module_id));
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
> +		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
> +		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
> +		p += snprintf(p, end - p, "error severity=%s ",
> +			      err_severity(err->err_severity));
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
> +		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
> +
> +	p += snprintf(p, end - p, "]\n");
> +	dev_info(dev, "\nHISI HIP08: PCIe local error\n");
> +	dev_info(dev, "%s\n", buf);
> +
> +	dev_info(dev, "Reg Dump:\n");
> +	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
> +		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
> +			dev_info(dev,
> +				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
> +	}
> +
> +	/* Recovery for the PCIe local errors */
> +	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
> +		/* try reset PCI port for the error recovery */
> +		rc = hisi_hip08_pcie_port_reset(pdev, err->socket_id,
> +				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
> +		if (rc) {
> +			dev_info(dev, "fail to do hip08 pcie port reset\n");
> +			return;
> +		}
> +	}
> +}
> +
> +static DEFINE_KFIFO(pcie_err_recover_ring, struct pcie_err_info,
> +		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
> +static DEFINE_SPINLOCK(pcie_err_recover_ring_lock);
> +
> +static void pcie_local_err_recover_work_func(struct work_struct *work)
> +{
> +	struct pcie_err_info pcie_err_entry;
> +
> +	while (kfifo_get(&pcie_err_recover_ring, &pcie_err_entry)) {
> +		pcie_local_error_handle(&pcie_err_entry.err_data,
> +					pcie_err_entry.pdev);
> +	}
> +}
> +
> +static DECLARE_WORK(pcie_err_recover_work, pcie_local_err_recover_work_func);
> +
> +
> +static void hip08_pcie_local_error_handle(struct acpi_hest_generic_data *gdata,
> +					  int sev, void *data)
> +{
> +	const struct hisi_pcie_local_err_data *err_data =
> +					acpi_hest_get_payload(gdata);
> +	struct pcie_err_info err_info;
> +	struct platform_device *pdev = data;
> +	struct device *dev = &(pdev->dev);
> +
> +	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
> +	err_info.pdev = pdev;
> +
> +	if (kfifo_in_spinlocked(&pcie_err_recover_ring, &err_info, 1,
> +				&pcie_err_recover_ring_lock))
> +		schedule_work(&pcie_err_recover_work);
> +	else
> +		dev_warn(dev, "Buffer overflow when recovering PCIe local error\n");

I'd call this a "queue full" warning or similar.  "Buffer overflow"
suggests that we wrote past the end of a buffer and corrupted some
memory, but that's not the case here.

> +}
> +
> +static int hisi_hip08_pcie_err_handler_probe(struct platform_device *pdev)
> +{
> +	int ret = 0;

Pointless local variable; maybe you meant to return failure if
ghes_register_event_handler() fails?  Don't initialize unless it's
necessary.

> +
> +	if (ghes_register_event_handler(hip08_pcie_sec_type,
> +					hip08_pcie_local_error_handle,
> +					pdev)) {
> +		dev_err(&(pdev->dev), "%s : ghes_register_event_handler fail\n",
> +			__func__);
> +		return ret;
> +}

Indentation error.

> +
> +	return 0;
> +}
> +
> +static int hisi_hip08_pcie_err_handler_remove(struct platform_device *pdev)
> +{
> +	ghes_unregister_event_handler(hip08_pcie_sec_type);
> +
> +	return 0;
> +}
> +
> +static const struct acpi_device_id hip08_pcie_acpi_match[] = {
> +	{ "HISI0361", 0 },
> +	{ }
> +};
> +
> +static struct platform_driver hisi_hip08_pcie_err_handler_driver = {
> +	.driver = {
> +		.name	= "hisi-hip08-pcie-err-handler",
> +		.acpi_match_table = hip08_pcie_acpi_match,
> +	},
> +	.probe		= hisi_hip08_pcie_err_handler_probe,
> +	.remove		= hisi_hip08_pcie_err_handler_remove,
> +};
> +module_platform_driver(hisi_hip08_pcie_err_handler_driver);
> +
> +MODULE_DESCRIPTION("HiSilicon HIP08 PCIe controller error handling driver");
> +MODULE_LICENSE("GPL v2");
> +
> -- 
> 1.9.1
> 
> 

RE: [RFC PATCH 2/2] PCI:hip08:Add driver to handle HiSilicon hip08 PCIe controller's errors

[Shiju Jose]

Hi Bjorn,

Thanks for reviewing the patch.
Please find reply inline.

>-----Original Message-----
>From: Bjorn Helgaas [mailto:helgaas@kernel.org]
>Sent: 15 January 2020 14:14
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; lenb@kernel.org; bp@alien8.de;
>james.morse@arm.com; tony.luck@intel.com; gregkh@linuxfoundation.org;
>zhangliguang@linux.alibaba.com; tglx@linutronix.de; Linuxarm
><linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [RFC PATCH 2/2] PCI:hip08:Add driver to handle HiSilicon hip08
>PCIe controller's errors
>
>Follow convention for subject line.

Ok. We will ix it.

>
>On Wed, Jan 15, 2020 at 11:01:40AM +0000, Shiju Jose wrote:
>> From: Yicong Yang <yangyicong@hisilicon.com>
>>
>> The hip08 error handle driver logs and reports PCIe controller's
>> recoverable errors.
>> Perform root port reset and restore link status for the recovery.
>>
>> Following are some of the PCIe controller's recoverable errors 1.
>> completion transmission timeout error.
>> 2. CRS retry counter over the threshold error.
>> 3. ECC 2 bit errors
>> 4. AXI bresponse/rresponse errors etc.
>>
>> RFC: The appropriate location for this driver may be discussed.
>>
>> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
>> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
>> ---
>>  drivers/pci/controller/Kconfig                 |   8 +
>>  drivers/pci/controller/Makefile                |   1 +
>>  drivers/pci/controller/pcie-hisi-hip08-error.c | 323
>> +++++++++++++++++++++++++
>
>Seems like this driver and its Kconfig should be near
>drivers/pci/controller/dwc/pcie-hisi.c.

pcie-hisi.c was for our old hip* devices.
Our hip08 PCIe controller doesn't use DWC ip.
Thus the driver to be in /drivers/pci/controller/ ?

>
>>  3 files changed, 332 insertions(+)
>>  create mode 100644 drivers/pci/controller/pcie-hisi-hip08-error.c
>>
>> diff --git a/drivers/pci/controller/Kconfig
>> b/drivers/pci/controller/Kconfig index c77069c..0ee99b8 100644
>> --- a/drivers/pci/controller/Kconfig
>> +++ b/drivers/pci/controller/Kconfig
>> @@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
>>  	  The Hyper-V PCI Interface is a helper driver allows other drivers to
>>  	  have a common interface with the Hyper-V PCI frontend driver.
>>
>> +config PCIE_HISI_HIP08_ERR_HANDLER
>
>Config symbol is too long, maybe CONFIG_PCI_HISI_ERR or similar (to be
>parallel with existing CONFIG_PCI_HISI).  Both should probably be
>"CONFIG_PCIE", not "CONFIG_PCI".  I can't remember why CONFIG_PCI_HISI is
>that way.

Ok. We will change it to CONFIG_PCI_HISI_ERR.

>
>> +	depends on ARM64 || COMPILE_TEST
>> +	depends on (ACPI && PCI_QUIRKS)
>> +	bool "HiSilicon hip08 Soc PCIe local error handling driver"
>> +	help
>> +	  Say Y here if you want PCIe error handling support
>> +	  for the PCIe local(controller) errors on HiSilicon hip08 SoC
>> +
>>  source "drivers/pci/controller/dwc/Kconfig"
>>  source "drivers/pci/controller/cadence/Kconfig"
>>  endmenu
>> diff --git a/drivers/pci/controller/Makefile
>> b/drivers/pci/controller/Makefile index 3d4f597..ac9852f 100644
>> --- a/drivers/pci/controller/Makefile
>> +++ b/drivers/pci/controller/Makefile
>> @@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
>>  obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
>>  obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
>>  obj-$(CONFIG_VMD) += vmd.o
>> +obj-$(CONFIG_PCIE_HISI_HIP08_ERR_HANDLER) += pcie-hisi-hip08-error.o
>>  # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
>>  obj-y				+= dwc/
>>
>> diff --git a/drivers/pci/controller/pcie-hisi-hip08-error.c
>> b/drivers/pci/controller/pcie-hisi-hip08-error.c
>> new file mode 100644
>> index 0000000..6f5d002
>> --- /dev/null
>> +++ b/drivers/pci/controller/pcie-hisi-hip08-error.c
>> @@ -0,0 +1,323 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * PCIe driver for handling PCIe local errors occurred on
>> + * HiSilicon hip08 PCIe controller.
>
>"PCIe" occurs too many times in this sentence.  Strictly speaking this is not a
>"PCIe driver"; it's a driver for an ACPI device that reports hip08-related errors.

Ok. Will remove it.

>Hopefully we don't need a separate driver for every hip* device, so maybe the
>"hip08" name is too specific.

Currently we are using this single driver for all hip08 device. so maybe this is better:
yes, a separate driver is unnecessary. We'll remove "hip08" name and
make it more generic for hip* device.

>
>> + * Copyright (c) 2018-2019 Hisilicon Limited.
>
>Capitalize "HiSilicon" consistently.  Also "hip08"; previous practice in
>drivers/pci is "Hip05", "Hip06", so use that unless HiSilicon itself does it
>differently.

Ok. We will fix.

>
>> +#include <linux/acpi.h>
>> +#include <acpi/ghes.h>
>> +#include <linux/bitfield.h>
>> +#include <linux/bitops.h>
>> +#include <linux/delay.h>
>> +#include <linux/irq.h>
>> +#include <linux/irqdomain.h>
>> +#include <linux/list.h>
>> +#include <linux/pci.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/resource.h>
>> +#include <linux/kfifo.h>
>> +#include <linux/spinlock.h>
>> +
>> +#include "../pci.h"
>> +
>> +#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
>> +#define	HISI_PCIE_ERR_INFO_SIZE	1024
>> +
>> +/* HISI PCIe Local error definitions */
>> +#define HISI_PCIE_ERR_MISC_REGS	33
>> +
>> +#define HISI_PCIE_SUB_MODULE_ID_AP	0
>> +#define HISI_PCIE_SUB_MODULE_ID_TL	1
>> +#define HISI_PCIE_SUB_MODULE_ID_MAC	2
>> +#define HISI_PCIE_SUB_MODULE_ID_DL	3
>> +#define HISI_PCIE_SUB_MODULE_ID_SDI	4
>> +
>> +#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
>> +#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
>> +#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
>> +#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
>> +#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
>> +#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
>> +#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
>> +#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
>> +#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
>> +#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
>> +
>> +#define HISI_ERR_SEV_RECOVERABLE	0
>> +#define HISI_ERR_SEV_FATAL		1
>> +#define HISI_ERR_SEV_CORRECTED		2
>> +#define HISI_ERR_SEV_NONE		3
>> +
>> +guid_t hip08_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
>0xA8, 0x67,
>> +				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
>> +
>> +#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
>> +#define HISI_PCIE_PORT_ID(core, v)       ((v >> 1) + (core << 3))
>> +#define HISI_PCIE_CORE_PORT_ID(v)        ((v % 8) << 1)
>> +#define HISI_PCIE_ROOT_BUSNR(v)          ((v) ? 0x80 : 0)
>
>Is the root bus number really hard-wired in the chip?  You're saying the only
>possible root bus numbers are 0x80 and 0x00?  Typically this bus number is
>programmable.

We will fix it and remove the macro. We'll get the root bus number from acpi instead.

>
>Why parens around "v" (sometimes) but not others and "core"?

We will correct it.

>
>> +struct hisi_pcie_local_err_data {
>> +	uint64_t   val_bits;
>> +	uint8_t    version;
>> +	uint8_t    soc_id;
>> +	uint8_t    socket_id;
>> +	uint8_t    nimbus_id;
>> +	uint8_t    sub_module_id;
>> +	uint8_t    core_id;
>> +	uint8_t    port_id;
>> +	uint8_t    err_severity;
>> +	uint16_t   err_type;
>> +	uint8_t    reserv[2];
>> +	uint32_t   err_misc[HISI_PCIE_ERR_MISC_REGS];
>
>Use u64, u8, u32 throughout instead.

We will change it.

>
>> +};
>> +
>> +struct pcie_err_info {
>> +	struct hisi_pcie_local_err_data err_data;
>> +	struct platform_device *pdev;
>> +};
>> +
>> +static char *pcie_local_sub_module_name(uint8_t id) {
>> +	switch (id) {
>> +	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
>> +	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
>> +	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
>> +	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
>> +	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
>> +	}
>> +
>> +	return "unknown";
>> +}
>> +
>> +static char *err_severity(uint8_t err_sev) {
>> +	switch (err_sev) {
>> +	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
>> +	case HISI_ERR_SEV_FATAL: return "fatal";
>> +	case HISI_ERR_SEV_CORRECTED: return "corrected";
>> +	case HISI_ERR_SEV_NONE: return "none";
>> +	}
>> +
>> +	return "unknown";
>> +}
>> +
>> +static struct pci_dev *hisi_hip08_pcie_get_rp(u32 chip_id, u32
>> +port_id) {
>> +	u32 devfn = PCI_DEVFN(port_id, 0);
>> +	u32 busnr = HISI_PCIE_ROOT_BUSNR(chip_id);
>> +
>> +	return pci_get_domain_bus_and_slot(0, busnr, devfn); }
>> +
>> +static int hisi_hip08_pcie_port_acpi_reset(struct platform_device *pdev,
>> +					u32 chip_id, u32 port_id)
>> +{
>> +	struct device *dev = &(pdev->dev);
>
>Unnecessary parens.  More occurrences below.

We will change it.

>
>> +	struct acpi_object_list arg_list;
>> +	union acpi_object arg[3];
>> +
>> +	arg[0].type = ACPI_TYPE_INTEGER;
>> +	arg[0].integer.value = chip_id;
>> +	arg[1].type = ACPI_TYPE_INTEGER;
>> +	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
>> +	arg[2].type = ACPI_TYPE_INTEGER;
>> +	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
>> +
>> +	arg_list.count = 3;
>> +	arg_list.pointer = arg;
>> +	/* Call the ACPI handle to reset root port  */
>
>s/root port  /root port /
>
>> +	if (ACPI_HANDLE(dev)) {
>
>Restructure this to return early for error and unindent the following, e.g.,

Ok. We will change it.

>
>  acpi_handle handle = ACPI_HANDLE(dev);
>
>  if (!handle) {
>    dev_err(...);
>    return -EINVAL;
>  }
>
>  arg[0].type = ACPI_TYPE_INTEGER;
>  ...
>  s = acpi_evaluate_integer(handle, ...);
>
>> +		unsigned long long data = 0;
>> +		acpi_status s;
>> +
>> +		s = acpi_evaluate_integer(ACPI_HANDLE(dev),
>> +				"RST", &arg_list, &data);
>> +
>> +		if (ACPI_FAILURE(s)) {
>> +			dev_err(dev, "No Reset method\n");
>> +			return -EIO;
>> +		}
>> +
>> +		if (data) {
>> +			dev_err(dev, "Failed to Reset\n");
>> +			return -EIO;
>> +		}
>> +
>> +	} else {
>> +		dev_err(dev, "No Reset method\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	return 0;
>> +}
>> +
>> +static int hisi_hip08_pcie_port_reset(struct platform_device *dev,
>> +				      u32 chip_id, u32 port_id)
>> +{
>> +	struct pci_dev *pdev;
>> +	struct pci_bus *root_bus;
>> +
>> +	pdev = hisi_hip08_pcie_get_rp(chip_id, port_id);
>> +	if (!pdev) {
>> +		dev_info(&(dev->dev), "Fail to get root port device\n");
>> +		return -ENODEV;
>> +	}
>> +	root_bus = pdev->bus;
>> +
>> +	pci_stop_and_remove_bus_device_locked(pdev);
>> +
>> +	if (hisi_hip08_pcie_port_acpi_reset(dev, chip_id, port_id))
>> +		return -EIO;
>> +	ssleep(1UL);
>
>Please include a comment that cites the spec section that requires this sleep.

we'll add a comment. We use a 1s delay here as does in pci_reset_secondary_bus().
It's necessary for re-initialization of subordinate devices.

>
>> +
>> +	/* add root port and downstream devices */
>> +	pci_lock_rescan_remove();
>> +	pci_rescan_bus(root_bus);
>> +	pci_unlock_rescan_remove();
>> +
>> +	return 0;
>> +}
>> +
>> +static void pcie_local_error_handle(const struct hisi_pcie_local_err_data
>*err,
>> +				    struct platform_device *pdev) {
>> +	char buf[HISI_PCIE_ERR_INFO_SIZE];
>> +	char *p = buf, *end = buf + sizeof(buf);
>> +	struct device *dev = &(pdev->dev);
>> +	uint32_t i;
>> +	int rc;
>> +
>> +	if (err->val_bits == 0) {
>> +		dev_warn(dev, "%s: no valid error information\n", __func__);
>> +		return;
>> +	}
>> +
>> +	/* Logging */
>> +	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
>> +		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
>> +		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
>> +		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
>> +		p += snprintf(p, end - p, "sub module=%s ",
>> +			      pcie_local_sub_module_name(err-
>>sub_module_id));
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
>> +		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
>> +		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
>> +		p += snprintf(p, end - p, "error severity=%s ",
>> +			      err_severity(err->err_severity));
>> +
>> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
>> +		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
>> +
>> +	p += snprintf(p, end - p, "]\n");
>> +	dev_info(dev, "\nHISI HIP08: PCIe local error\n");
>> +	dev_info(dev, "%s\n", buf);
>> +
>> +	dev_info(dev, "Reg Dump:\n");
>> +	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
>> +		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
>> +			dev_info(dev,
>> +				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
>> +	}
>> +
>> +	/* Recovery for the PCIe local errors */
>> +	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
>> +		/* try reset PCI port for the error recovery */
>> +		rc = hisi_hip08_pcie_port_reset(pdev, err->socket_id,
>> +				HISI_PCIE_PORT_ID(err->core_id, err-
>>port_id));
>> +		if (rc) {
>> +			dev_info(dev, "fail to do hip08 pcie port reset\n");
>> +			return;
>> +		}
>> +	}
>> +}
>> +
>> +static DEFINE_KFIFO(pcie_err_recover_ring, struct pcie_err_info,
>> +		    HISI_PCIE_ERR_RECOVER_RING_SIZE); static
>> +DEFINE_SPINLOCK(pcie_err_recover_ring_lock);
>> +
>> +static void pcie_local_err_recover_work_func(struct work_struct
>> +*work) {
>> +	struct pcie_err_info pcie_err_entry;
>> +
>> +	while (kfifo_get(&pcie_err_recover_ring, &pcie_err_entry)) {
>> +		pcie_local_error_handle(&pcie_err_entry.err_data,
>> +					pcie_err_entry.pdev);
>> +	}
>> +}
>> +
>> +static DECLARE_WORK(pcie_err_recover_work,
>> +pcie_local_err_recover_work_func);
>> +
>> +
>> +static void hip08_pcie_local_error_handle(struct acpi_hest_generic_data
>*gdata,
>> +					  int sev, void *data)
>> +{
>> +	const struct hisi_pcie_local_err_data *err_data =
>> +					acpi_hest_get_payload(gdata);
>> +	struct pcie_err_info err_info;
>> +	struct platform_device *pdev = data;
>> +	struct device *dev = &(pdev->dev);
>> +
>> +	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
>> +	err_info.pdev = pdev;
>> +
>> +	if (kfifo_in_spinlocked(&pcie_err_recover_ring, &err_info, 1,
>> +				&pcie_err_recover_ring_lock))
>> +		schedule_work(&pcie_err_recover_work);
>> +	else
>> +		dev_warn(dev, "Buffer overflow when recovering PCIe local
>> +error\n");
>
>I'd call this a "queue full" warning or similar.  "Buffer overflow"
>suggests that we wrote past the end of a buffer and corrupted some memory,
>but that's not the case here.

We will change it to "queue full".

>
>> +}
>> +
>> +static int hisi_hip08_pcie_err_handler_probe(struct platform_device
>> +*pdev) {
>> +	int ret = 0;
>
>Pointless local variable; maybe you meant to return failure if
>ghes_register_event_handler() fails?  Don't initialize unless it's necessary.

We will fix it. We  need to return error value on ghes_register_event_handler  failure.

>
>> +
>> +	if (ghes_register_event_handler(hip08_pcie_sec_type,
>> +					hip08_pcie_local_error_handle,
>> +					pdev)) {
>> +		dev_err(&(pdev->dev), "%s : ghes_register_event_handler
>fail\n",
>> +			__func__);
>> +		return ret;
>> +}
>
>Indentation error.

Ok. we will correct it.

>
>> +
>> +	return 0;
>> +}
>> +
>> +static int hisi_hip08_pcie_err_handler_remove(struct platform_device
>> +*pdev) {
>> +	ghes_unregister_event_handler(hip08_pcie_sec_type);
>> +
>> +	return 0;
>> +}
>> +
>> +static const struct acpi_device_id hip08_pcie_acpi_match[] = {
>> +	{ "HISI0361", 0 },
>> +	{ }
>> +};
>> +
>> +static struct platform_driver hisi_hip08_pcie_err_handler_driver = {
>> +	.driver = {
>> +		.name	= "hisi-hip08-pcie-err-handler",
>> +		.acpi_match_table = hip08_pcie_acpi_match,
>> +	},
>> +	.probe		= hisi_hip08_pcie_err_handler_probe,
>> +	.remove		= hisi_hip08_pcie_err_handler_remove,
>> +};
>> +module_platform_driver(hisi_hip08_pcie_err_handler_driver);
>> +
>> +MODULE_DESCRIPTION("HiSilicon HIP08 PCIe controller error handling
>> +driver"); MODULE_LICENSE("GPL v2");
>> +
>> --
>> 1.9.1
>>
>>

Thanks,
Shiju

[PATCH v2 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor drivers are unable to do the recovery for the vendor specific
recoverable HW errors, reported to the APEI driver in the vendor defined sections,
because APEI driver does not support reporting the same to the vendor drivers.

This patch set
1. add an interface to the APEI driver to enable the vendor
drivers to register the event handling functions for the corresponding
vendor specific HW errors and report the error to the vendor driver.

2. add driver to handle HiSilicon hip08 PCIe controller's errors
   which is an example application of the above APEI interface.

Changes:

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (1):
  ACPI: APEI: Add support to notify the vendor specific HW errors

Yicong Yang (1):
  PCI: hip: Add handling of HiSilicon hip PCIe controller's errors

 drivers/acpi/apei/ghes.c                 | 116 ++++++++++-
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
 include/acpi/ghes.h                      |  56 ++++++
 5 files changed, 512 insertions(+), 5 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
1.9.1

[PATCH v2 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently APEI does not support reporting the vendor specific
recoverable HW errors, received in the vendor defined table entries, to the
vendor drivers for the recovery.

This patch adds the support to register and unregister the
error handling function for the vendor specific HW errors and
to notify the registered kernel driver.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 116 +++++++++++++++++++++++++++++++++++++++++++++--
 include/acpi/ghes.h      |  56 +++++++++++++++++++++++
 2 files changed, 167 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 103acbb..69e18d7 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -490,6 +490,109 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+struct ghes_event_notify {
+	struct list_head list;
+	struct rcu_head	rcu_head;
+	guid_t sec_type; /* guid of the error record */
+	ghes_event_handler_t event_handler; /* event handler function */
+	void *data; /* handler driver's private data if any */
+};
+
+/* List to store the registered event handling functions */
+static DEFINE_MUTEX(ghes_event_notify_mutex);
+static LIST_HEAD(ghes_event_handler_list);
+
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data)
+{
+	struct ghes_event_notify *event_notify;
+
+	event_notify = kzalloc(sizeof(*event_notify), GFP_KERNEL);
+	if (!event_notify)
+		return -ENOMEM;
+
+	event_notify->event_handler = event_handler;
+	guid_copy(&event_notify->sec_type, &sec_type);
+	event_notify->data = data;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	list_add_rcu(&event_notify->list, &ghes_event_handler_list);
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_handler);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, &sec_type)) {
+			if (data != event_notify->data)
+				continue;
+			list_del_rcu(&event_notify->list);
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	if (!found) {
+		pr_err("Tried to unregister a GHES event handler that has not been registered\n");
+		return;
+	}
+
+	synchronize_rcu();
+	kfree(event_notify);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);
+
+static int ghes_handle_non_standard_event(guid_t *sec_type,
+	struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+	int ret;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, sec_type)) {
+			ret = event_notify->event_handler(gdata, sev,
+						    event_notify->data);
+			if (!ret)
+				continue;
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+
+	return found;
+}
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
 
 			log_arm_hw_error(err);
 		} else {
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
+			if (!ghes_handle_non_standard_event(sec_type, gdata,
+							    sev)) {
+				void *err = acpi_hest_get_payload(gdata);
+
+				log_non_standard_event(sec_type, fru_id,
+						       fru_text, sec_sev, err,
+						       gdata->error_data_length);
+			}
 		}
 	}
 }
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cdd..e3387cf 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,62 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+enum {
+	GHES_EVENT_NONE	= 0x0,
+	GHES_EVENT_HANDLED	= 0x1,
+};
+
+/**
+ * typedef ghes_event_handler_t - event handling function
+ * for the non-fatal HW errors.
+ *
+ * @gdata: acpi_hest_generic_data.
+ * @sev: error severity of the entire error event defined in the
+ *       ACPI spec table generic error status block.
+ * @data: handler driver's private data.
+ *
+ * Return : GHES_EVENT_NONE - event not handled, GHES_EVENT_HANDLED - handled.
+ *
+ * The error handling function is responsible for logging error and
+ * this function would be called in the interrupt context.
+ */
+typedef int (*ghes_event_handler_t)(struct acpi_hest_generic_data *gdata,
+				    int sev, void *data);
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the event handling function.
+ * @data: handler driver's private data.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_typei, void *data);
+#else
+static inline int ghes_register_event_handler(guid_t sec_type,
+					ghes_event_handler_t event_handler,
+					void *data)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
1.9.1

[PATCH v2 2/2] PCI: hip: Add handling of HiSilicon hip PCIe controller's errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The error handling driver logs and reports hip PCIe controller's
recoverable errors.
Perform root port reset and restore link status for the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
 3 files changed, 345 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
 3 files changed, 345 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index c77069c..769fce7 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on (ACPI && PCI_QUIRKS)
+	bool "HiSilicon hip PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon hip SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 3d4f597..2d1565f 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
 obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 0000000..27520ad
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,336 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller's errors on
+ * HiSilicon hip SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/bitfield.h>
+#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/list.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/resource.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#include "../pci.h"
+
+#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller's error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
+				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_err_data {
+	u64   val_bits;
+	u8    version;
+	u8    soc_id;
+	u8    socket_id;
+	u8    nimbus_id;
+	u8    sub_module_id;
+	u8    core_id;
+	u8    port_id;
+	u8    err_severity;
+	u16   err_type;
+	u8    reserv[2];
+	u32   err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_err_info {
+	struct hisi_pcie_err_data err_data;
+	struct platform_device *pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_err_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+					u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	union acpi_object arg[3];
+	unsigned long long data = 0;
+	struct device *dev = &pdev->dev;
+	struct acpi_object_list arg_list;
+	acpi_handle handle = ACPI_HANDLE(dev);
+
+	if (!handle) {
+		dev_err(dev, "No Reset method\n");
+		return -EINVAL;
+	}
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	/* Call the ACPI handle to reset root port  */
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No Reset method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	u32 busnr, devfn;
+	struct pci_dev *pdev;
+	struct pci_bus *root_bus;
+
+	devfn = PCI_DEVFN(port_id, 0);
+	if (device_property_read_u32(&dev->dev, "busnr", &busnr))
+		goto failed;
+
+	pdev = pci_get_domain_bus_and_slot(0, busnr, devfn);
+	if (!pdev)
+		goto failed;
+
+	root_bus = pdev->bus;
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/**
+	 * In pci_reset_secondary_bus(), using 1s delay before subordinates
+	 * devices to be re-initialized. Use the same delay here to ensure
+	 * we can get all the devices after root port reset.
+	 **/
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+
+failed:
+	dev_info(&(dev->dev), "Fail to get root port device\n");
+	return -ENODEV;
+}
+
+static void hisi_pcie_handle_one_error(const struct hisi_pcie_err_data *err,
+				    struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (err->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(err->sub_module_id));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_err_severity(err->err_severity));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : hip : PCIe controller's error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller's errors */
+	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, err->socket_id,
+				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static DEFINE_KFIFO(hisi_pcie_err_recover_ring, struct hisi_pcie_err_info,
+		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
+static DEFINE_SPINLOCK(hisi_pcie_err_recover_ring_lock);
+
+static void hisi_pcie_err_recover_work_func(struct work_struct *work)
+{
+	struct hisi_pcie_err_info pcie_err_entry;
+
+	while (kfifo_get(&hisi_pcie_err_recover_ring, &pcie_err_entry)) {
+		hisi_pcie_handle_one_error(&pcie_err_entry.err_data,
+					pcie_err_entry.pdev);
+	}
+}
+
+static DECLARE_WORK(hisi_pcie_err_recover_work,
+		    hisi_pcie_err_recover_work_func);
+
+static int hisi_pcie_error_handle(struct acpi_hest_generic_data *gdata,
+				  int sev, void *data)
+{
+	const struct hisi_pcie_err_data *err_data =
+					acpi_hest_get_payload(gdata);
+	struct hisi_pcie_err_info err_info;
+	struct platform_device *pdev = data;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return GHES_EVENT_NONE;
+
+	if (err_data->socket_id != socket)
+		return GHES_EVENT_NONE;
+
+	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
+	err_info.pdev = pdev;
+
+	if (kfifo_in_spinlocked(&hisi_pcie_err_recover_ring, &err_info, 1,
+				&hisi_pcie_err_recover_ring_lock))
+		schedule_work(&hisi_pcie_err_recover_work);
+	else
+		dev_warn(dev, "queue full when recovering PCIe controller's error\n");
+
+	return GHES_EVENT_HANDLED;
+}
+
+static int hisi_pcie_err_handler_probe(struct platform_device *pdev)
+{
+	int ret;
+
+	ret = ghes_register_event_handler(hisi_pcie_sec_type,
+					  hisi_pcie_error_handle, pdev);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_handler fail\n",
+			__func__);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_err_handler_remove(struct platform_device *pdev)
+{
+	ghes_unregister_event_handler(hisi_pcie_sec_type, pdev);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_err_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-err-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_err_handler_probe,
+	.remove		= hisi_pcie_err_handler_remove,
+};
+module_platform_driver(hisi_pcie_err_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon hip PCIe controller's error handling driver");
+MODULE_LICENSE("GPL v2");
+
-- 
1.9.1

Re: [PATCH v2 2/2] PCI: hip: Add handling of HiSilicon hip PCIe controller's errors

[Bjorn Helgaas]

On Fri, Jan 24, 2020 at 12:39:38PM +0000, Shiju Jose wrote:
> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> The error handling driver logs and reports hip PCIe controller's
> recoverable errors.
> Perform root port reset and restore link status for the recovery.

If the preceding is two paragraphs, there should be a blank line
between them.  If it's a single paragraph, it should be rewrapped to
use the entire line width.

> Following are some of the PCIe controller's recoverable errors
> 1. completion transmission timeout error.
> 2. CRS retry counter over the threshold error.
> 3. ECC 2 bit errors
> 4. AXI bresponse/rresponse errors etc.
> 
> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
> --
>  drivers/pci/controller/Kconfig           |   8 +
>  drivers/pci/controller/Makefile          |   1 +
>  drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
>  3 files changed, 345 insertions(+)
>  create mode 100644 drivers/pci/controller/pcie-hisi-error.c
> ---
>  drivers/pci/controller/Kconfig           |   8 +
>  drivers/pci/controller/Makefile          |   1 +
>  drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
>  3 files changed, 345 insertions(+)
>  create mode 100644 drivers/pci/controller/pcie-hisi-error.c
> 
> diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
> index c77069c..769fce7 100644
> --- a/drivers/pci/controller/Kconfig
> +++ b/drivers/pci/controller/Kconfig
> @@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
>  	  The Hyper-V PCI Interface is a helper driver allows other drivers to
>  	  have a common interface with the Hyper-V PCI frontend driver.
>  
> +config PCIE_HISI_ERR
> +	depends on ARM64 || COMPILE_TEST
> +	depends on (ACPI && PCI_QUIRKS)

Why does this depend on PCI_QUIRKS?  If it's needed, please mention
the reason somewhere (maybe in the commit log, since there's not
really a good way to do it in Kconfig itself).

> +	bool "HiSilicon hip PCIe controller error handling driver"
> +	help
> +	  Say Y here if you want error handling support
> +	  for the PCIe controller's errors on HiSilicon hip SoCs

"hip" above refers to the hardware device and should be capitalized
(two occurrences above and at least two below).

>  source "drivers/pci/controller/dwc/Kconfig"
>  source "drivers/pci/controller/cadence/Kconfig"
>  endmenu
> diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
> index 3d4f597..2d1565f 100644
> --- a/drivers/pci/controller/Makefile
> +++ b/drivers/pci/controller/Makefile
> @@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
>  obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
>  obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
>  obj-$(CONFIG_VMD) += vmd.o
> +obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
>  # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
>  obj-y				+= dwc/
>  
> diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
> new file mode 100644
> index 0000000..27520ad
> --- /dev/null
> +++ b/drivers/pci/controller/pcie-hisi-error.c
> @@ -0,0 +1,336 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Driver for handling the PCIe controller's errors on
> + * HiSilicon hip SoCs.
> + *
> + * Copyright (c) 2018-2019 HiSilicon Limited.
> + */
> +
> +#include <linux/acpi.h>
> +#include <acpi/ghes.h>
> +#include <linux/bitfield.h>
> +#include <linux/bitops.h>
> +#include <linux/delay.h>
> +#include <linux/irq.h>
> +#include <linux/irqdomain.h>
> +#include <linux/list.h>
> +#include <linux/pci.h>
> +#include <linux/platform_device.h>
> +#include <linux/resource.h>
> +#include <linux/kfifo.h>
> +#include <linux/spinlock.h>

Why do you need bitfield.h, bitops.h, irq.h, irqdomain.h, list.h,
resource.h?  You *do* need bits.h, which is included by at least
bitops.h.  But if you don't need bitops.h itself, include bits.h
directly instead.

> +
> +#include "../pci.h"
> +
> +#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
> +#define	HISI_PCIE_ERR_INFO_SIZE	1024
> +
> +/* HISI PCIe controller's error definitions */
> +#define HISI_PCIE_ERR_MISC_REGS	33
> +
> +#define HISI_PCIE_SUB_MODULE_ID_AP	0
> +#define HISI_PCIE_SUB_MODULE_ID_TL	1
> +#define HISI_PCIE_SUB_MODULE_ID_MAC	2
> +#define HISI_PCIE_SUB_MODULE_ID_DL	3
> +#define HISI_PCIE_SUB_MODULE_ID_SDI	4
> +
> +#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
> +#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
> +#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
> +#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
> +#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
> +#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
> +#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
> +#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
> +#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
> +#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
> +
> +#define HISI_ERR_SEV_RECOVERABLE	0
> +#define HISI_ERR_SEV_FATAL		1
> +#define HISI_ERR_SEV_CORRECTED		2
> +#define HISI_ERR_SEV_NONE		3
> +
> +guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
> +				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
> +
> +#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
> +#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
> +#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
> +
> +struct hisi_pcie_err_data {
> +	u64   val_bits;
> +	u8    version;
> +	u8    soc_id;
> +	u8    socket_id;
> +	u8    nimbus_id;
> +	u8    sub_module_id;
> +	u8    core_id;
> +	u8    port_id;
> +	u8    err_severity;
> +	u16   err_type;
> +	u8    reserv[2];
> +	u32   err_misc[HISI_PCIE_ERR_MISC_REGS];
> +};
> +
> +struct hisi_pcie_err_info {
> +	struct hisi_pcie_err_data err_data;
> +	struct platform_device *pdev;
> +};
> +
> +static char *hisi_pcie_sub_module_name(u8 id)
> +{
> +	switch (id) {
> +	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
> +	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
> +	}
> +
> +	return "unknown";
> +}
> +
> +static char *hisi_pcie_err_severity(u8 err_sev)
> +{
> +	switch (err_sev) {
> +	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
> +	case HISI_ERR_SEV_FATAL: return "fatal";
> +	case HISI_ERR_SEV_CORRECTED: return "corrected";
> +	case HISI_ERR_SEV_NONE: return "none";
> +	}
> +
> +	return "unknown";
> +}
> +
> +static int hisi_pcie_port_reset(struct platform_device *pdev,
> +					u32 chip_id, u32 port_id)
> +{
> +	acpi_status s;
> +	union acpi_object arg[3];
> +	unsigned long long data = 0;
> +	struct device *dev = &pdev->dev;
> +	struct acpi_object_list arg_list;
> +	acpi_handle handle = ACPI_HANDLE(dev);

Order these as:

  struct device *dev = &pdev->dev;
  acpi_handle handle = ACPI_HANDLE(dev);
  union acpi_object arg[3];
  ...

so they're in order of use.

> +
> +	if (!handle) {
> +		dev_err(dev, "No Reset method\n");

Technically it's not the RST *method* that's missing; it's the entire
object that would *enclose* the RST method.  I guess that would be the
HISI0361 device?  Is it even possible for that to not exist, since
this driver binds to that device?

> +		return -EINVAL;
> +	}
> +
> +	arg[0].type = ACPI_TYPE_INTEGER;
> +	arg[0].integer.value = chip_id;
> +	arg[1].type = ACPI_TYPE_INTEGER;
> +	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
> +	arg[2].type = ACPI_TYPE_INTEGER;
> +	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
> +
> +	arg_list.count = 3;
> +	arg_list.pointer = arg;
> +
> +	/* Call the ACPI handle to reset root port  */

s/root port  /root port / (mentioned last time as well)
(Remove the extra space after "root port")

This isn't actually "calling the ACPI handle"; the handle is a
reference to the HISI0361 device, which *contains* the RST method.
You could just drop the comment altogether since it's pretty obvious
what's going on.

> +	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
> +	if (ACPI_FAILURE(s)) {
> +		dev_err(dev, "No Reset method\n");

I think it'd be better to use the exact method name ("RST") in the
message.

> +		return -EIO;
> +	}
> +
> +	if (data) {
> +		dev_err(dev, "Failed to Reset\n");
> +		return -EIO;
> +	}
> +
> +	return 0;
> +}
> +
> +static int hisi_pcie_port_do_recovery(struct platform_device *dev,
> +				      u32 chip_id, u32 port_id)
> +{
> +	u32 busnr, devfn;
> +	struct pci_dev *pdev;
> +	struct pci_bus *root_bus;
> +
> +	devfn = PCI_DEVFN(port_id, 0);
> +	if (device_property_read_u32(&dev->dev, "busnr", &busnr))
> +		goto failed;
> +
> +	pdev = pci_get_domain_bus_and_slot(0, busnr, devfn);
> +	if (!pdev)
> +		goto failed;
> +
> +	root_bus = pdev->bus;
> +
> +	pci_stop_and_remove_bus_device_locked(pdev);
> +	pci_dev_put(pdev);
> +
> +	if (hisi_pcie_port_reset(dev, chip_id, port_id))
> +		return -EIO;
> +
> +	/**
> +	 * In pci_reset_secondary_bus(), using 1s delay before subordinates
> +	 * devices to be re-initialized. Use the same delay here to ensure
> +	 * we can get all the devices after root port reset.

Use normal comment style (no "**").  Include the specific spec section
reference, e.g., PCIe r5.0, sec XXX.

> +	 **/
> +	ssleep(1UL);
> +
> +	/* add root port and downstream devices */
> +	pci_lock_rescan_remove();
> +	pci_rescan_bus(root_bus);
> +	pci_unlock_rescan_remove();
> +
> +	return 0;
> +
> +failed:
> +	dev_info(&(dev->dev), "Fail to get root port device\n");

For debugging purposes, include the address of the root port you tried
to find, e.g., ssss:bb:dd.f from segment, busnr, and devfn.

Since you're using ACPI, the segment number should not have to be
hard-coded; it should come from the host bridge's _SEG method.  I
don't know how you *find* that.  Is the HISI0361 device in the scope
of the PNP0A08 device, so you could just search up the chain?  You
need some way to associate this with the host bridge device.

> +	return -ENODEV;
> +}
> +
> +static void hisi_pcie_handle_one_error(const struct hisi_pcie_err_data *err,
> +				    struct platform_device *pdev)
> +{
> +	char buf[HISI_PCIE_ERR_INFO_SIZE];
> +	char *p = buf, *end = buf + sizeof(buf);
> +	struct device *dev = &pdev->dev;
> +	u32 i;
> +	int rc;
> +
> +	if (err->val_bits == 0) {
> +		dev_warn(dev, "%s: no valid error information\n", __func__);
> +		return;
> +	}
> +
> +	/* Logging */
> +	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
> +		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
> +		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
> +		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
> +		p += snprintf(p, end - p, "sub module=%s ",
> +			      hisi_pcie_sub_module_name(err->sub_module_id));
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
> +		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
> +		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
> +		p += snprintf(p, end - p, "error severity=%s ",
> +			      hisi_pcie_err_severity(err->err_severity));
> +
> +	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
> +		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
> +
> +	p += snprintf(p, end - p, "]\n");
> +	dev_info(dev, "\nHISI : hip : PCIe controller's error\n");
> +	dev_info(dev, "%s\n", buf);
> +
> +	dev_info(dev, "Reg Dump:\n");
> +	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
> +		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
> +			dev_info(dev,
> +				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
> +	}
> +
> +	/* Recovery for the PCIe controller's errors */
> +	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
> +		/* try reset PCI port for the error recovery */
> +		rc = hisi_pcie_port_do_recovery(pdev, err->socket_id,
> +				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
> +		if (rc) {
> +			dev_info(dev, "fail to do hisi pcie port reset\n");
> +			return;
> +		}
> +	}
> +}
> +
> +static DEFINE_KFIFO(hisi_pcie_err_recover_ring, struct hisi_pcie_err_info,
> +		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
> +static DEFINE_SPINLOCK(hisi_pcie_err_recover_ring_lock);
> +
> +static void hisi_pcie_err_recover_work_func(struct work_struct *work)
> +{
> +	struct hisi_pcie_err_info pcie_err_entry;
> +
> +	while (kfifo_get(&hisi_pcie_err_recover_ring, &pcie_err_entry)) {
> +		hisi_pcie_handle_one_error(&pcie_err_entry.err_data,
> +					pcie_err_entry.pdev);
> +	}
> +}
> +
> +static DECLARE_WORK(hisi_pcie_err_recover_work,
> +		    hisi_pcie_err_recover_work_func);
> +
> +static int hisi_pcie_error_handle(struct acpi_hest_generic_data *gdata,
> +				  int sev, void *data)
> +{
> +	const struct hisi_pcie_err_data *err_data =
> +					acpi_hest_get_payload(gdata);
> +	struct hisi_pcie_err_info err_info;
> +	struct platform_device *pdev = data;
> +	struct device *dev = &pdev->dev;
> +	u8 socket;
> +
> +	if (device_property_read_u8(dev, "socket", &socket))
> +		return GHES_EVENT_NONE;
> +
> +	if (err_data->socket_id != socket)
> +		return GHES_EVENT_NONE;
> +
> +	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
> +	err_info.pdev = pdev;
> +
> +	if (kfifo_in_spinlocked(&hisi_pcie_err_recover_ring, &err_info, 1,
> +				&hisi_pcie_err_recover_ring_lock))
> +		schedule_work(&hisi_pcie_err_recover_work);
> +	else
> +		dev_warn(dev, "queue full when recovering PCIe controller's error\n");
> +
> +	return GHES_EVENT_HANDLED;
> +}
> +
> +static int hisi_pcie_err_handler_probe(struct platform_device *pdev)
> +{
> +	int ret;
> +
> +	ret = ghes_register_event_handler(hisi_pcie_sec_type,
> +					  hisi_pcie_error_handle, pdev);
> +	if (ret) {
> +		dev_err(&pdev->dev, "%s : ghes_register_event_handler fail\n",
> +			__func__);
> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int hisi_pcie_err_handler_remove(struct platform_device *pdev)
> +{
> +	ghes_unregister_event_handler(hisi_pcie_sec_type, pdev);
> +
> +	return 0;
> +}
> +
> +static const struct acpi_device_id hisi_pcie_acpi_match[] = {
> +	{ "HISI0361", 0 },
> +	{ }
> +};
> +
> +static struct platform_driver hisi_pcie_err_handler_driver = {
> +	.driver = {
> +		.name	= "hisi-pcie-err-handler",
> +		.acpi_match_table = hisi_pcie_acpi_match,
> +	},
> +	.probe		= hisi_pcie_err_handler_probe,
> +	.remove		= hisi_pcie_err_handler_remove,
> +};
> +module_platform_driver(hisi_pcie_err_handler_driver);
> +
> +MODULE_DESCRIPTION("HiSilicon hip PCIe controller's error handling driver");
> +MODULE_LICENSE("GPL v2");
> +
> -- 
> 1.9.1
> 
> 

Re: [PATCH v2 2/2] PCI: hip: Add handling of HiSilicon hip PCIe controller's errors

[kbuild test robot]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on pm/linux-next]
[also build test WARNING on linux/master linus/master v5.5-rc7]
[cannot apply to pci/next next-20200121]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200125-171952
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next
reproduce:
        # apt-get install sparse
        # sparse version: v0.6.1-153-g47b6dfef-dirty
        make ARCH=x86_64 allmodconfig
        make C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__'

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)

>> drivers/pci/controller/pcie-hisi-error.c:53:8: sparse: sparse: symbol 'hisi_pcie_sec_type' was not declared. Should it be static?

Please review and possibly fold the followup patch.

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org Intel Corporation

Re: [PATCH v2 2/2] PCI: hip: Add handling of HiSilicon hip PCIe controller's errors

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 1397 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on pm/linux-next]
[also build test WARNING on linux/master linus/master v5.5-rc7]
[cannot apply to pci/next next-20200121]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200125-171952
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next
reproduce:
        # apt-get install sparse
        # sparse version: v0.6.1-153-g47b6dfef-dirty
        make ARCH=x86_64 allmodconfig
        make C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__'

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)

>> drivers/pci/controller/pcie-hisi-error.c:53:8: sparse: sparse: symbol 'hisi_pcie_sec_type' was not declared. Should it be static?

Please review and possibly fold the followup patch.

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org Intel Corporation

[RFC PATCH] PCI: hip: hisi_pcie_sec_type can be static

[kbuild test robot]

Fixes: 9bf6eb234cd2 ("PCI: hip: Add handling of HiSilicon hip PCIe controller's errors")
Signed-off-by: kbuild test robot <lkp@intel.com>
---
 pcie-hisi-error.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
index 7c669043a4e0e..76c37abaaf968 100644
--- a/drivers/pci/controller/pcie-hisi-error.c
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -50,7 +50,7 @@
 #define HISI_ERR_SEV_CORRECTED		2
 #define HISI_ERR_SEV_NONE		3
 
-guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
 				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
 
 #define HISI_PCIE_CORE_ID(v)             ((v) >> 3)

[RFC PATCH] PCI: hip: hisi_pcie_sec_type can be static

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 825 bytes --]


Fixes: 9bf6eb234cd2 ("PCI: hip: Add handling of HiSilicon hip PCIe controller's errors")
Signed-off-by: kbuild test robot <lkp@intel.com>
---
 pcie-hisi-error.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
index 7c669043a4e0e..76c37abaaf968 100644
--- a/drivers/pci/controller/pcie-hisi-error.c
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -50,7 +50,7 @@
 #define HISI_ERR_SEV_CORRECTED		2
 #define HISI_ERR_SEV_NONE		3
 
-guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D, 0xA8, 0x67,
 				       0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
 
 #define HISI_PCIE_CORE_ID(v)             ((v) >> 3)

[PATCH v3 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor drivers are unable to do the recovery for the
vendor specific recoverable HW errors, reported to the APEI driver
in the vendor defined sections, because APEI driver does not support
reporting the same to the vendor drivers.

This patch set
1. add an interface to the APEI driver to enable the vendor
drivers to register the event handling functions for the corresponding
vendor specific HW errors and report the error to the vendor driver.

2. add driver to handle HiSilicon hip08 PCIe controller's errors
   which is an example application of the above APEI interface.

Changes:

V3:
1. Fix the comments from Bjorn Helgaas.

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (1):
  ACPI: APEI: Add support to notify the vendor specific HW errors

Yicong Yang (1):
  PCI: HIP: Add handling of HiSilicon HIP PCIe controller's errors

 drivers/acpi/apei/ghes.c                 | 116 ++++++++++-
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 334 +++++++++++++++++++++++++++++++
 include/acpi/ghes.h                      |  56 ++++++
 5 files changed, 510 insertions(+), 5 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
1.9.1

[PATCH v3 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently APEI does not support reporting the vendor specific
HW errors, received in the vendor defined table entries, to the
vendor drivers for any recovery.

This patch adds the support to register and unregister the
error handling function for the vendor specific HW errors and
notify the registered kernel driver.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 116 +++++++++++++++++++++++++++++++++++++++++++++--
 include/acpi/ghes.h      |  56 +++++++++++++++++++++++
 2 files changed, 167 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 103acbb..69e18d7 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -490,6 +490,109 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+struct ghes_event_notify {
+	struct list_head list;
+	struct rcu_head	rcu_head;
+	guid_t sec_type; /* guid of the error record */
+	ghes_event_handler_t event_handler; /* event handler function */
+	void *data; /* handler driver's private data if any */
+};
+
+/* List to store the registered event handling functions */
+static DEFINE_MUTEX(ghes_event_notify_mutex);
+static LIST_HEAD(ghes_event_handler_list);
+
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data)
+{
+	struct ghes_event_notify *event_notify;
+
+	event_notify = kzalloc(sizeof(*event_notify), GFP_KERNEL);
+	if (!event_notify)
+		return -ENOMEM;
+
+	event_notify->event_handler = event_handler;
+	guid_copy(&event_notify->sec_type, &sec_type);
+	event_notify->data = data;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	list_add_rcu(&event_notify->list, &ghes_event_handler_list);
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_handler);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, &sec_type)) {
+			if (data != event_notify->data)
+				continue;
+			list_del_rcu(&event_notify->list);
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	if (!found) {
+		pr_err("Tried to unregister a GHES event handler that has not been registered\n");
+		return;
+	}
+
+	synchronize_rcu();
+	kfree(event_notify);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);
+
+static int ghes_handle_non_standard_event(guid_t *sec_type,
+	struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+	int ret;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, sec_type)) {
+			ret = event_notify->event_handler(gdata, sev,
+						    event_notify->data);
+			if (!ret)
+				continue;
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+
+	return found;
+}
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
 
 			log_arm_hw_error(err);
 		} else {
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
+			if (!ghes_handle_non_standard_event(sec_type, gdata,
+							    sev)) {
+				void *err = acpi_hest_get_payload(gdata);
+
+				log_non_standard_event(sec_type, fru_id,
+						       fru_text, sec_sev, err,
+						       gdata->error_data_length);
+			}
 		}
 	}
 }
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cdd..e3387cf 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,62 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+enum {
+	GHES_EVENT_NONE	= 0x0,
+	GHES_EVENT_HANDLED	= 0x1,
+};
+
+/**
+ * typedef ghes_event_handler_t - event handling function
+ * for the non-fatal HW errors.
+ *
+ * @gdata: acpi_hest_generic_data.
+ * @sev: error severity of the entire error event defined in the
+ *       ACPI spec table generic error status block.
+ * @data: handler driver's private data.
+ *
+ * Return : GHES_EVENT_NONE - event not handled, GHES_EVENT_HANDLED - handled.
+ *
+ * The error handling function is responsible for logging error and
+ * this function would be called in the interrupt context.
+ */
+typedef int (*ghes_event_handler_t)(struct acpi_hest_generic_data *gdata,
+				    int sev, void *data);
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the event handling function.
+ * @data: handler driver's private data.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_typei, void *data);
+#else
+static inline int ghes_register_event_handler(guid_t sec_type,
+					ghes_event_handler_t event_handler,
+					void *data)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
1.9.1

[PATCH v3 2/2] PCI: HIP: Add handling of HiSilicon HIP PCIe controller's errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The HiSilicon HIP PCIe controller is capable of handling errors
on root port and perform port reset separately at each root port.

This patch add error handling driver for HIP PCIe controller to log
and report recoverable errors. Perform root port reset and restore
link status after the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
drivers/pci/controller/Kconfig           |   8 +
drivers/pci/controller/Makefile          |   1 +
drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
3 files changed, 345 insertions(+)
create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 334 +++++++++++++++++++++++++++++++
 3 files changed, 343 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index c77069c..5dad1ca 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on ACPI
+	bool "HiSilicon HIP PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon HIP SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 3d4f597..2d1565f 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
 obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 0000000..5b33a63
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,334 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller's errors on
+ * HiSilicon HIP SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#include "../pci.h"
+
+#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller's error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
+			0xA8, 0x67, 0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_err_data {
+	u64   val_bits;
+	u8    version;
+	u8    soc_id;
+	u8    socket_id;
+	u8    nimbus_id;
+	u8    sub_module_id;
+	u8    core_id;
+	u8    port_id;
+	u8    err_severity;
+	u16   err_type;
+	u8    reserv[2];
+	u32   err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_err_info {
+	struct hisi_pcie_err_data err_data;
+	struct platform_device *pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_err_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+					u32 chip_id, u32 port_id)
+{
+	struct device *dev = &pdev->dev;
+	acpi_handle handle = ACPI_HANDLE(dev);
+	union acpi_object arg[3];
+	struct acpi_object_list arg_list;
+	acpi_status s;
+	unsigned long long data = 0;
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	/* Call the ACPI handle to reset root port */
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No RST method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	struct device *device = &dev->dev;
+	acpi_handle root_handle = ACPI_HANDLE(device);
+	struct acpi_pci_root *pci_root;
+	struct pci_bus *root_bus;
+	struct pci_dev *pdev;
+	u32 domain, busnr, devfn;
+
+	s = acpi_get_parent(root_handle, &root_handle);
+	if (ACPI_FAILURE(s))
+		return -ENODEV;
+	pci_root = acpi_pci_find_root(root_handle);
+	if (!pci_root)
+		return -ENODEV;
+	root_bus = pci_root->bus;
+	domain = pci_root->segment;
+
+	busnr = root_bus->number;
+	devfn = PCI_DEVFN(port_id, 0);
+	pdev = pci_get_domain_bus_and_slot(domain, busnr, devfn);
+	if (!pdev) {
+		dev_info(device, "Fail to get root port %04x:%02x:%02x.%d device\n",
+			 domain, busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
+		return -ENODEV;
+	}
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/*
+	 * The initialization time of subordinate devices after
+	 * hot reset is no more than 1s, which is required by
+	 * the PCI spec v5.0 sec 6.6.1. The time will shorten
+	 * if Readiness Notifications mechanisms are used. But
+	 * wait 1s here to adapt any conditions.
+	 */
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void hisi_pcie_handle_one_error(const struct hisi_pcie_err_data *err,
+				    struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (err->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(err->sub_module_id));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_err_severity(err->err_severity));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller's errors */
+	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, err->socket_id,
+				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static DEFINE_KFIFO(hisi_pcie_err_recover_ring, struct hisi_pcie_err_info,
+		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
+static DEFINE_SPINLOCK(hisi_pcie_err_recover_ring_lock);
+
+static void hisi_pcie_err_recover_work_func(struct work_struct *work)
+{
+	struct hisi_pcie_err_info pcie_err_entry;
+
+	while (kfifo_get(&hisi_pcie_err_recover_ring, &pcie_err_entry)) {
+		hisi_pcie_handle_one_error(&pcie_err_entry.err_data,
+					pcie_err_entry.pdev);
+	}
+}
+
+static DECLARE_WORK(hisi_pcie_err_recover_work,
+		    hisi_pcie_err_recover_work_func);
+
+static int hisi_pcie_error_handle(struct acpi_hest_generic_data *gdata,
+				  int sev, void *data)
+{
+	const struct hisi_pcie_err_data *err_data =
+					acpi_hest_get_payload(gdata);
+	struct hisi_pcie_err_info err_info;
+	struct platform_device *pdev = data;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return GHES_EVENT_NONE;
+
+	if (err_data->socket_id != socket)
+		return GHES_EVENT_NONE;
+
+	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
+	err_info.pdev = pdev;
+
+	if (kfifo_in_spinlocked(&hisi_pcie_err_recover_ring, &err_info, 1,
+				&hisi_pcie_err_recover_ring_lock))
+		schedule_work(&hisi_pcie_err_recover_work);
+	else
+		dev_warn(dev, "queue full when recovering PCIe controller's error\n");
+
+	return GHES_EVENT_HANDLED;
+}
+
+static int hisi_pcie_err_handler_probe(struct platform_device *pdev)
+{
+	int ret;
+
+	ret = ghes_register_event_handler(hisi_pcie_sec_type,
+					  hisi_pcie_error_handle, pdev);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_handler fail\n",
+			__func__);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_err_handler_remove(struct platform_device *pdev)
+{
+	ghes_unregister_event_handler(hisi_pcie_sec_type, pdev);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_err_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-err-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_err_handler_probe,
+	.remove		= hisi_pcie_err_handler_remove,
+};
+module_platform_driver(hisi_pcie_err_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP PCIe controller's error handling driver");
+MODULE_LICENSE("GPL v2");
-- 
1.9.1

Re: [PATCH v3 2/2] PCI: HIP: Add handling of HiSilicon HIP PCIe controller's errors

[Dan Carpenter]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200204-073736
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>

smatch warnings:
drivers/pci/controller/pcie-hisi-error.c:234 hisi_pcie_handle_one_error() warn: should '((((1))) << (9 + i))' be a 64 bit type?

# https://github.com/0day-ci/linux/commit/71688ac6d222c137b66a707f8a6fdf28b48e1942
git remote add linux-review https://github.com/0day-ci/linux
git remote update linux-review
git checkout 71688ac6d222c137b66a707f8a6fdf28b48e1942
vim +234 drivers/pci/controller/pcie-hisi-error.c

71688ac6d222c1 Yicong Yang 2020-02-03  228  	p += snprintf(p, end - p, "]\n");
71688ac6d222c1 Yicong Yang 2020-02-03  229  	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
71688ac6d222c1 Yicong Yang 2020-02-03  230  	dev_info(dev, "%s\n", buf);
71688ac6d222c1 Yicong Yang 2020-02-03  231  
71688ac6d222c1 Yicong Yang 2020-02-03  232  	dev_info(dev, "Reg Dump:\n");
71688ac6d222c1 Yicong Yang 2020-02-03  233  	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
71688ac6d222c1 Yicong Yang 2020-02-03 @234  		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
                                                                            ^^^
This should be BIT_ULL() because it goes up to 9 + 32.

71688ac6d222c1 Yicong Yang 2020-02-03  235  			dev_info(dev,
71688ac6d222c1 Yicong Yang 2020-02-03  236  				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
71688ac6d222c1 Yicong Yang 2020-02-03  237  	}
71688ac6d222c1 Yicong Yang 2020-02-03  238  

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org Intel Corporation

Re: [PATCH v3 2/2] PCI: HIP: Add handling of HiSilicon HIP PCIe controller's errors

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 2025 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200204-073736
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>

smatch warnings:
drivers/pci/controller/pcie-hisi-error.c:234 hisi_pcie_handle_one_error() warn: should '((((1))) << (9 + i))' be a 64 bit type?

# https://github.com/0day-ci/linux/commit/71688ac6d222c137b66a707f8a6fdf28b48e1942
git remote add linux-review https://github.com/0day-ci/linux
git remote update linux-review
git checkout 71688ac6d222c137b66a707f8a6fdf28b48e1942
vim +234 drivers/pci/controller/pcie-hisi-error.c

71688ac6d222c1 Yicong Yang 2020-02-03  228  	p += snprintf(p, end - p, "]\n");
71688ac6d222c1 Yicong Yang 2020-02-03  229  	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
71688ac6d222c1 Yicong Yang 2020-02-03  230  	dev_info(dev, "%s\n", buf);
71688ac6d222c1 Yicong Yang 2020-02-03  231  
71688ac6d222c1 Yicong Yang 2020-02-03  232  	dev_info(dev, "Reg Dump:\n");
71688ac6d222c1 Yicong Yang 2020-02-03  233  	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
71688ac6d222c1 Yicong Yang 2020-02-03 @234  		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
                                                                            ^^^
This should be BIT_ULL() because it goes up to 9 + 32.

71688ac6d222c1 Yicong Yang 2020-02-03  235  			dev_info(dev,
71688ac6d222c1 Yicong Yang 2020-02-03  236  				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
71688ac6d222c1 Yicong Yang 2020-02-03  237  	}
71688ac6d222c1 Yicong Yang 2020-02-03  238  

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org Intel Corporation

Re: [PATCH v3 2/2] PCI: HIP: Add handling of HiSilicon HIP PCIe controller's errors

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 2025 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-the-vendor-specific-HW-errors/20200204-073736
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>

smatch warnings:
drivers/pci/controller/pcie-hisi-error.c:234 hisi_pcie_handle_one_error() warn: should '((((1))) << (9 + i))' be a 64 bit type?

# https://github.com/0day-ci/linux/commit/71688ac6d222c137b66a707f8a6fdf28b48e1942
git remote add linux-review https://github.com/0day-ci/linux
git remote update linux-review
git checkout 71688ac6d222c137b66a707f8a6fdf28b48e1942
vim +234 drivers/pci/controller/pcie-hisi-error.c

71688ac6d222c1 Yicong Yang 2020-02-03  228  	p += snprintf(p, end - p, "]\n");
71688ac6d222c1 Yicong Yang 2020-02-03  229  	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
71688ac6d222c1 Yicong Yang 2020-02-03  230  	dev_info(dev, "%s\n", buf);
71688ac6d222c1 Yicong Yang 2020-02-03  231  
71688ac6d222c1 Yicong Yang 2020-02-03  232  	dev_info(dev, "Reg Dump:\n");
71688ac6d222c1 Yicong Yang 2020-02-03  233  	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
71688ac6d222c1 Yicong Yang 2020-02-03 @234  		if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
                                                                            ^^^
This should be BIT_ULL() because it goes up to 9 + 32.

71688ac6d222c1 Yicong Yang 2020-02-03  235  			dev_info(dev,
71688ac6d222c1 Yicong Yang 2020-02-03  236  				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
71688ac6d222c1 Yicong Yang 2020-02-03  237  	}
71688ac6d222c1 Yicong Yang 2020-02-03  238  

---
0-DAY kernel test infrastructure                 Open Source Technology Center
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org Intel Corporation

[PATCH v4 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor drivers are unable to do the recovery for the
vendor specific recoverable HW errors, reported to the APEI driver
in the vendor defined sections, because APEI driver does not support
reporting the same to the vendor drivers.

This patch set
1. add an interface to the APEI driver to enable the vendor
drivers to register the event handling functions for the corresponding
vendor specific HW errors and report the error to the vendor driver.

2. add driver to handle HiSilicon hip08 PCIe controller's errors
   which is an example application of the above APEI interface.

Changes:

V4:
1. Fix for the smatch warning in the PCIe error driver:
   warn: should '((((1))) << (9 + i))' be a 64 bit type?
   if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
	^^^ This should be BIT_ULL() because it goes up to 9 + 32.

V3:
1. Fix the comments from Bjorn Helgaas.

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (1):
  ACPI: APEI: Add support to notify the vendor specific HW errors

Yicong Yang (1):
  PCI: HIP: Add handling of HiSilicon HIP PCIe controller errors

 drivers/acpi/apei/ghes.c                 | 116 ++++++++++-
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 334 +++++++++++++++++++++++++++++++
 include/acpi/ghes.h                      |  56 ++++++
 5 files changed, 510 insertions(+), 5 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
1.9.1

[PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently APEI does not support reporting the vendor specific
HW errors, received in the vendor defined table entries, to the
vendor drivers for any recovery.

This patch adds the support to register and unregister the
error handling function for the vendor specific HW errors and
notify the registered kernel driver.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 116 +++++++++++++++++++++++++++++++++++++++++++++--
 include/acpi/ghes.h      |  56 +++++++++++++++++++++++
 2 files changed, 167 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 103acbb..69e18d7 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -490,6 +490,109 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+struct ghes_event_notify {
+	struct list_head list;
+	struct rcu_head	rcu_head;
+	guid_t sec_type; /* guid of the error record */
+	ghes_event_handler_t event_handler; /* event handler function */
+	void *data; /* handler driver's private data if any */
+};
+
+/* List to store the registered event handling functions */
+static DEFINE_MUTEX(ghes_event_notify_mutex);
+static LIST_HEAD(ghes_event_handler_list);
+
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the error handling function.
+ * @data: handler driver's private data.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data)
+{
+	struct ghes_event_notify *event_notify;
+
+	event_notify = kzalloc(sizeof(*event_notify), GFP_KERNEL);
+	if (!event_notify)
+		return -ENOMEM;
+
+	event_notify->event_handler = event_handler;
+	guid_copy(&event_notify->sec_type, &sec_type);
+	event_notify->data = data;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	list_add_rcu(&event_notify->list, &ghes_event_handler_list);
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_handler);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+
+	mutex_lock(&ghes_event_notify_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, &sec_type)) {
+			if (data != event_notify->data)
+				continue;
+			list_del_rcu(&event_notify->list);
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+	mutex_unlock(&ghes_event_notify_mutex);
+
+	if (!found) {
+		pr_err("Tried to unregister a GHES event handler that has not been registered\n");
+		return;
+	}
+
+	synchronize_rcu();
+	kfree(event_notify);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);
+
+static int ghes_handle_non_standard_event(guid_t *sec_type,
+	struct acpi_hest_generic_data *gdata, int sev)
+{
+	struct ghes_event_notify *event_notify;
+	bool found = false;
+	int ret;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(event_notify,
+				&ghes_event_handler_list, list) {
+		if (guid_equal(&event_notify->sec_type, sec_type)) {
+			ret = event_notify->event_handler(gdata, sev,
+						    event_notify->data);
+			if (!ret)
+				continue;
+			found = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+
+	return found;
+}
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
 
 			log_arm_hw_error(err);
 		} else {
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
+			if (!ghes_handle_non_standard_event(sec_type, gdata,
+							    sev)) {
+				void *err = acpi_hest_get_payload(gdata);
+
+				log_non_standard_event(sec_type, fru_id,
+						       fru_text, sec_sev, err,
+						       gdata->error_data_length);
+			}
 		}
 	}
 }
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cdd..e3387cf 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,62 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+enum {
+	GHES_EVENT_NONE	= 0x0,
+	GHES_EVENT_HANDLED	= 0x1,
+};
+
+/**
+ * typedef ghes_event_handler_t - event handling function
+ * for the non-fatal HW errors.
+ *
+ * @gdata: acpi_hest_generic_data.
+ * @sev: error severity of the entire error event defined in the
+ *       ACPI spec table generic error status block.
+ * @data: handler driver's private data.
+ *
+ * Return : GHES_EVENT_NONE - event not handled, GHES_EVENT_HANDLED - handled.
+ *
+ * The error handling function is responsible for logging error and
+ * this function would be called in the interrupt context.
+ */
+typedef int (*ghes_event_handler_t)(struct acpi_hest_generic_data *gdata,
+				    int sev, void *data);
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_handler - register an event handling
+ * function for the non-fatal HW errors.
+ * @sec_type: sec_type of the corresponding CPER to be notified.
+ * @event_handler: pointer to the event handling function.
+ * @data: handler driver's private data.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_handler(guid_t sec_type,
+				ghes_event_handler_t event_handler,
+				void *data);
+
+/**
+ * ghes_unregister_event_handler - unregister the previously
+ * registered event handling function.
+ * @sec_type: sec_type of the corresponding CPER.
+ * @data: driver specific data to distinguish devices.
+ */
+void ghes_unregister_event_handler(guid_t sec_typei, void *data);
+#else
+static inline int ghes_register_event_handler(guid_t sec_type,
+					ghes_event_handler_t event_handler,
+					void *data)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_handler(guid_t sec_type, void *data)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
1.9.1

Re: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi Shiju,

On 07/02/2020 10:31, Shiju Jose wrote:
> Presently APEI does not support reporting the vendor specific
> HW errors, received in the vendor defined table entries, to the
> vendor drivers for any recovery.
> 
> This patch adds the support to register and unregister the
> error handling function for the vendor specific HW errors and
> notify the registered kernel driver.

Is it possible to use the kernel's existing atomic_notifier_chain_register() API for this?

The one thing that can't be done in the same way is the GUID filtering in ghes.c. Each
driver would need to check if the call matched a GUID they knew about, and return
NOTIFY_DONE if they "don't care".

I think this patch would be a lot smaller if it was tweaked to be able to use the existing
API. If there is a reason not to use it, it would be good to know what it is.


> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index 103acbb..69e18d7 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -490,6 +490,109 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)

> +/**
> + * ghes_unregister_event_handler - unregister the previously
> + * registered event handling function.
> + * @sec_type: sec_type of the corresponding CPER.
> + * @data: driver specific data to distinguish devices.
> + */
> +void ghes_unregister_event_handler(guid_t sec_type, void *data)
> +{
> +	struct ghes_event_notify *event_notify;
> +	bool found = false;
> +
> +	mutex_lock(&ghes_event_notify_mutex);
> +	rcu_read_lock();
> +	list_for_each_entry_rcu(event_notify,
> +				&ghes_event_handler_list, list) {
> +		if (guid_equal(&event_notify->sec_type, &sec_type)) {

> +			if (data != event_notify->data)

It looks like you need multiple drivers to handle the same GUID because of multiple root
ports. Can't the handler lookup the right device?


> +				continue;
> +			list_del_rcu(&event_notify->list);
> +			found = true;
> +			break;
> +		}
> +	}
> +	rcu_read_unlock();
> +	mutex_unlock(&ghes_event_notify_mutex);
> +
> +	if (!found) {
> +		pr_err("Tried to unregister a GHES event handler that has not been registered\n");
> +		return;
> +	}
> +
> +	synchronize_rcu();
> +	kfree(event_notify);
> +}
> +EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);

> @@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
>  
>  			log_arm_hw_error(err);
>  		} else {
> -			void *err = acpi_hest_get_payload(gdata);
> -
> -			log_non_standard_event(sec_type, fru_id, fru_text,
> -					       sec_sev, err,
> -					       gdata->error_data_length);
> +			if (!ghes_handle_non_standard_event(sec_type, gdata,
> +							    sev)) {
> +				void *err = acpi_hest_get_payload(gdata);
> +
> +				log_non_standard_event(sec_type, fru_id,
> +						       fru_text, sec_sev, err,
> +						       gdata->error_data_length);
> +			}

So, a side effect of the kernel handling these is they no longer get logged out of trace
points?

I guess the driver the claims this logs some more accurate information. Are there expected
to be any user-space programs doing something useful with B2889FC9... today?


Thanks,

James

RE: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi James,

Thanks for reviewing the code.

>-----Original Message-----
>From: linux-pci-owner@vger.kernel.org [mailto:linux-pci-
>owner@vger.kernel.org] On Behalf Of James Morse
>Sent: 11 March 2020 17:30
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; bp@alien8.de; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor
>specific HW errors
>
>Hi Shiju,
>
>On 07/02/2020 10:31, Shiju Jose wrote:
>> Presently APEI does not support reporting the vendor specific HW
>> errors, received in the vendor defined table entries, to the vendor
>> drivers for any recovery.
>>
>> This patch adds the support to register and unregister the error
>> handling function for the vendor specific HW errors and notify the
>> registered kernel driver.
>
>Is it possible to use the kernel's existing atomic_notifier_chain_register() API for
>this?
>
>The one thing that can't be done in the same way is the GUID filtering in ghes.c.
>Each driver would need to check if the call matched a GUID they knew about,
>and return NOTIFY_DONE if they "don't care".
>
>I think this patch would be a lot smaller if it was tweaked to be able to use the
>existing API. If there is a reason not to use it, it would be good to know what it
>is.
I think when using atomic_notifier_chain_register we have following limitations,
1. All the registered error handlers would get called, though an error is not related to those handlers.    
    Also this may lead to mishandling of the error information if a handler does not
    implement GUID checking etc.
2. atomic_notifier_chain_register (notifier_chain_register) looks like does not support 
    pass the handler's private data during the registration which supposed to 
    passed later in the call back function *notifier_fn_t(... ,void *data) to the handler.
3. Also got difficulty in passing the ghes error data(acpi_hest_generic_data), GUID
    for the error received to the handler through the notifier_chain  callback interface. 
    
Sorry if I did not  understood your suggestion correctly.
 
>
>
>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c index
>> 103acbb..69e18d7 100644
>> --- a/drivers/acpi/apei/ghes.c
>> +++ b/drivers/acpi/apei/ghes.c
>> @@ -490,6 +490,109 @@ static void ghes_handle_aer(struct
>> acpi_hest_generic_data *gdata)
>
>> +/**
>> + * ghes_unregister_event_handler - unregister the previously
>> + * registered event handling function.
>> + * @sec_type: sec_type of the corresponding CPER.
>> + * @data: driver specific data to distinguish devices.
>> + */
>> +void ghes_unregister_event_handler(guid_t sec_type, void *data) {
>> +	struct ghes_event_notify *event_notify;
>> +	bool found = false;
>> +
>> +	mutex_lock(&ghes_event_notify_mutex);
>> +	rcu_read_lock();
>> +	list_for_each_entry_rcu(event_notify,
>> +				&ghes_event_handler_list, list) {
>> +		if (guid_equal(&event_notify->sec_type, &sec_type)) {
>
>> +			if (data != event_notify->data)
>
>It looks like you need multiple drivers to handle the same GUID because of
>multiple root ports. Can't the handler lookup the right device?
This check was because GUID is shared among multiple devices with one driver as seen
in the B2889FC9 driver (pcie-hisi-error.c). 
  
>
>
>> +				continue;
>> +			list_del_rcu(&event_notify->list);
>> +			found = true;
>> +			break;
>> +		}
>> +	}
>> +	rcu_read_unlock();
>> +	mutex_unlock(&ghes_event_notify_mutex);
>> +
>> +	if (!found) {
>> +		pr_err("Tried to unregister a GHES event handler that has not
>been registered\n");
>> +		return;
>> +	}
>> +
>> +	synchronize_rcu();
>> +	kfree(event_notify);
>> +}
>> +EXPORT_SYMBOL_GPL(ghes_unregister_event_handler);
>
>> @@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
>>
>>  			log_arm_hw_error(err);
>>  		} else {
>> -			void *err = acpi_hest_get_payload(gdata);
>> -
>> -			log_non_standard_event(sec_type, fru_id, fru_text,
>> -					       sec_sev, err,
>> -					       gdata->error_data_length);
>> +			if (!ghes_handle_non_standard_event(sec_type, gdata,
>> +							    sev)) {
>> +				void *err = acpi_hest_get_payload(gdata);
>> +
>> +				log_non_standard_event(sec_type, fru_id,
>> +						       fru_text, sec_sev, err,
>> +						       gdata->error_data_length);
>> +			}
>
>So, a side effect of the kernel handling these is they no longer get logged out of
>trace points?
>
>I guess the driver the claims this logs some more accurate information. Are
>there expected to be any user-space programs doing something useful with
>B2889FC9... today?
The B2889FC9 driver does not expect any corresponding user space programs. 
The driver mainly for the error recovery and basic error decoding and logging.
Previously we added the error logging for the B2889FC9 in the rasdaemon.
>
>
>Thanks,
>
>James

Thanks,
Shiju

Re: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi Shiju,

On 3/12/20 12:10 PM, Shiju Jose wrote:
>> On 07/02/2020 10:31, Shiju Jose wrote:
>>> Presently APEI does not support reporting the vendor specific HW
>>> errors, received in the vendor defined table entries, to the vendor
>>> drivers for any recovery.
>>>
>>> This patch adds the support to register and unregister the error
>>> handling function for the vendor specific HW errors and notify the
>>> registered kernel driver.
>>
>> Is it possible to use the kernel's existing atomic_notifier_chain_register() API for
>> this?
>>
>> The one thing that can't be done in the same way is the GUID filtering in ghes.c.
>> Each driver would need to check if the call matched a GUID they knew about,
>> and return NOTIFY_DONE if they "don't care".
>>
>> I think this patch would be a lot smaller if it was tweaked to be able to use the
>> existing API. If there is a reason not to use it, it would be good to know what it
>> is.

> I think when using atomic_notifier_chain_register we have following limitations,
> 1. All the registered error handlers would get called, though an error is not related to those handlers.    

The notifier chain provides NOTIFY_STOP_MASK, so that one of the callers
can say the work is done. We only expect a handful of these, so I don't
think there is going to be a scalability problem.


>     Also this may lead to mishandling of the error information if a handler does not
>     implement GUID checking etc.

Which would be a bug we can fix.
There is no point worrying about bugs in out of tree code.


> 2. atomic_notifier_chain_register (notifier_chain_register) looks like does not support 
>     pass the handler's private data during the registration which supposed to 
>     passed later in the call back function *notifier_fn_t(... ,void *data) to the handler.

The callback is provided with the struct notifier_block. A bit of
container_of() magic will give you whatever structure you embedded it in!


> 3. Also got difficulty in passing the ghes error data(acpi_hest_generic_data), GUID
>     for the error received to the handler through the notifier_chain  callback interface. 

Here you've lost me. Because you need to pass more than one thing? Can't
we have a struct for that?

But, isn't it all in struct acpi_hest_generic_data already? That is
where the guid and severity come from.


>>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c index
>>> 103acbb..69e18d7 100644
>>> --- a/drivers/acpi/apei/ghes.c
>>> +++ b/drivers/acpi/apei/ghes.c
>>> @@ -490,6 +490,109 @@ static void ghes_handle_aer(struct
>>> acpi_hest_generic_data *gdata)
>>
>>> +/**
>>> + * ghes_unregister_event_handler - unregister the previously
>>> + * registered event handling function.
>>> + * @sec_type: sec_type of the corresponding CPER.
>>> + * @data: driver specific data to distinguish devices.
>>> + */
>>> +void ghes_unregister_event_handler(guid_t sec_type, void *data) {
>>> +	struct ghes_event_notify *event_notify;
>>> +	bool found = false;
>>> +
>>> +	mutex_lock(&ghes_event_notify_mutex);
>>> +	rcu_read_lock();
>>> +	list_for_each_entry_rcu(event_notify,
>>> +				&ghes_event_handler_list, list) {
>>> +		if (guid_equal(&event_notify->sec_type, &sec_type)) {
>>
>>> +			if (data != event_notify->data)
>>
>> It looks like you need multiple drivers to handle the same GUID because of
>> multiple root ports. Can't the handler lookup the right device?

> This check was because GUID is shared among multiple devices with one driver as seen
> in the B2889FC9 driver (pcie-hisi-error.c). 

(we should stop calling it by its guid ... does it have a name?!)


This must be some kind of error collector for a bus right?

I agree we may need to have multiple drivers register to handle vendor
events, but it looks like you are registering the same handler multiple
times, with different private structures.

Can't it find the affected device from the error description?


>>> @@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
>>>
>>>  			log_arm_hw_error(err);
>>>  		} else {
>>> -			void *err = acpi_hest_get_payload(gdata);
>>> -
>>> -			log_non_standard_event(sec_type, fru_id, fru_text,
>>> -					       sec_sev, err,
>>> -					       gdata->error_data_length);
>>> +			if (!ghes_handle_non_standard_event(sec_type, gdata,
>>> +							    sev)) {
>>> +				void *err = acpi_hest_get_payload(gdata);
>>> +
>>> +				log_non_standard_event(sec_type, fru_id,
>>> +						       fru_text, sec_sev, err,
>>> +						       gdata->error_data_length);
>>> +			}
>>
>> So, a side effect of the kernel handling these is they no longer get logged out of
>> trace points?
>>
>> I guess the driver the claims this logs some more accurate information. Are
>> there expected to be any user-space programs doing something useful with
>> B2889FC9... today?

> The B2889FC9 driver does not expect any corresponding user space programs. 
> The driver mainly for the error recovery and basic error decoding and logging.

> Previously we added the error logging for the B2889FC9 in the rasdaemon.

So this series would break the error logging in rasdaemon.

User-space would need to be upgraded to receive the trace information
from the specific driver instead. (how does it know?!)

Could we log_non_standard_event() unconditionally, maybe adding a field
to indicate that a driver claimed it, so there may be more data
somewhere else...


Thanks,

James

RE: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi James,

>-----Original Message-----
>From: James Morse [mailto:james.morse@arm.com]
>Sent: 13 March 2020 15:17
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; bp@alien8.de; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v4 1/2] ACPI: APEI: Add support to notify the vendor
>specific HW errors
>
>Hi Shiju,
>
>On 3/12/20 12:10 PM, Shiju Jose wrote:
>>> On 07/02/2020 10:31, Shiju Jose wrote:
>>>> Presently APEI does not support reporting the vendor specific HW
>>>> errors, received in the vendor defined table entries, to the vendor
>>>> drivers for any recovery.
>>>>
>>>> This patch adds the support to register and unregister the error
>>>> handling function for the vendor specific HW errors and notify the
>>>> registered kernel driver.
>>>
>>> Is it possible to use the kernel's existing
>>> atomic_notifier_chain_register() API for this?
>>>
>>> The one thing that can't be done in the same way is the GUID filtering in
>ghes.c.
>>> Each driver would need to check if the call matched a GUID they knew
>>> about, and return NOTIFY_DONE if they "don't care".
>>>
>>> I think this patch would be a lot smaller if it was tweaked to be
>>> able to use the existing API. If there is a reason not to use it, it
>>> would be good to know what it is.
>
>> I think when using atomic_notifier_chain_register we have following
>limitations,
>> 1. All the registered error handlers would get called, though an error is not
>related to those handlers.
>
>The notifier chain provides NOTIFY_STOP_MASK, so that one of the callers can
>say the work is done. We only expect a handful of these, so I don't think there is
>going to be a scalability problem.
Ok. I will check the error reporting by using atomic_notifier_chain and test.

>
>
>>     Also this may lead to mishandling of the error information if a handler does
>not
>>     implement GUID checking etc.
>
>Which would be a bug we can fix.
>There is no point worrying about bugs in out of tree code.
Ok.

>
>
>> 2. atomic_notifier_chain_register (notifier_chain_register) looks like does not
>support
>>     pass the handler's private data during the registration which supposed to
>>     passed later in the call back function *notifier_fn_t(... ,void *data) to the
>handler.
>
>The callback is provided with the struct notifier_block. A bit of
>container_of() magic will give you whatever structure you embedded it in!
Ok. I will check this.
 
>
>
>> 3. Also got difficulty in passing the ghes error data(acpi_hest_generic_data),
>GUID
>>     for the error received to the handler through the notifier_chain  callback
>interface.
>
>Here you've lost me. Because you need to pass more than one thing? Can't we
>have a struct for that?
>
>But, isn't it all in struct acpi_hest_generic_data already? That is where the guid
>and severity come from.
Ok.  right. 
 
>
>
>>>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
>>>> index
>>>> 103acbb..69e18d7 100644
>>>> --- a/drivers/acpi/apei/ghes.c
>>>> +++ b/drivers/acpi/apei/ghes.c
>>>> @@ -490,6 +490,109 @@ static void ghes_handle_aer(struct
>>>> acpi_hest_generic_data *gdata)
>>>
>>>> +/**
>>>> + * ghes_unregister_event_handler - unregister the previously
>>>> + * registered event handling function.
>>>> + * @sec_type: sec_type of the corresponding CPER.
>>>> + * @data: driver specific data to distinguish devices.
>>>> + */
>>>> +void ghes_unregister_event_handler(guid_t sec_type, void *data) {
>>>> +	struct ghes_event_notify *event_notify;
>>>> +	bool found = false;
>>>> +
>>>> +	mutex_lock(&ghes_event_notify_mutex);
>>>> +	rcu_read_lock();
>>>> +	list_for_each_entry_rcu(event_notify,
>>>> +				&ghes_event_handler_list, list) {
>>>> +		if (guid_equal(&event_notify->sec_type, &sec_type)) {
>>>
>>>> +			if (data != event_notify->data)
>>>
>>> It looks like you need multiple drivers to handle the same GUID
>>> because of multiple root ports. Can't the handler lookup the right device?
>
>> This check was because GUID is shared among multiple devices with one
>> driver as seen in the B2889FC9 driver (pcie-hisi-error.c).
>
>(we should stop calling it by its guid ... does it have a name?!)
>
>
>This must be some kind of error collector for a bus right?
>
>I agree we may need to have multiple drivers register to handle vendor events,
>but it looks like you are registering the same handler multiple times, with
>different private structures.
>
>Can't it find the affected device from the error description?
Yes. We already have the code in the PCIe error handling driver to identify the right device
from the error information.

>
>
>>>> @@ -525,11 +628,14 @@ static void ghes_do_proc(struct ghes *ghes,
>>>>
>>>>  			log_arm_hw_error(err);
>>>>  		} else {
>>>> -			void *err = acpi_hest_get_payload(gdata);
>>>> -
>>>> -			log_non_standard_event(sec_type, fru_id, fru_text,
>>>> -					       sec_sev, err,
>>>> -					       gdata->error_data_length);
>>>> +			if (!ghes_handle_non_standard_event(sec_type, gdata,
>>>> +							    sev)) {
>>>> +				void *err = acpi_hest_get_payload(gdata);
>>>> +
>>>> +				log_non_standard_event(sec_type, fru_id,
>>>> +						       fru_text, sec_sev, err,
>>>> +						       gdata->error_data_length);
>>>> +			}
>>>
>>> So, a side effect of the kernel handling these is they no longer get
>>> logged out of trace points?
>>>
>>> I guess the driver the claims this logs some more accurate
>>> information. Are there expected to be any user-space programs doing
>>> something useful with B2889FC9... today?
>
>> The B2889FC9 driver does not expect any corresponding user space
>programs.
>> The driver mainly for the error recovery and basic error decoding and logging.
>
>> Previously we added the error logging for the B2889FC9 in the rasdaemon.
>
>So this series would break the error logging in rasdaemon.
It does not affect the logging information to the user for the HiSilicon PCIe controller errors
because the level of logging information is the same both in the rasdaemon and in the
newly adding HiSilicon PCIe controller error handling driver.
>
>User-space would need to be upgraded to receive the trace information from
>the specific driver instead. (how does it know?!)
>
>Could we log_non_standard_event() unconditionally, maybe adding a field to
>indicate that a driver claimed it, so there may be more data somewhere else...
sure, I will check the possibility of adding the field to indicate driver claimed it and
calling log_non_standard_event() always.
>
>
>Thanks,
>
>James

Thanks,
Shiju

[PATCH v4 2/2] PCI: HIP: Add handling of HiSilicon HIP PCIe controller errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The HiSilicon HIP PCIe controller is capable of handling errors
on root port and perform port reset separately at each root port.

This patch add error handling driver for HIP PCIe controller to log
and report recoverable errors. Perform root port reset and restore
link status after the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

Also fix the following Smatch warning:
warn: should '((((1))) << (9 + i))' be a 64 bit type?
if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
     ^^^ This should be BIT_ULL() because it goes up to 9 + 32.
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
drivers/pci/controller/Kconfig           |   8 +
drivers/pci/controller/Makefile          |   1 +
drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
3 files changed, 345 insertions(+)
create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 334 +++++++++++++++++++++++++++++++
 3 files changed, 343 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index c77069c..5dad1ca 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -260,6 +260,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on ACPI
+	bool "HiSilicon HIP PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon HIP SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 3d4f597..2d1565f 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -28,6 +28,7 @@ obj-$(CONFIG_PCIE_MEDIATEK) += pcie-mediatek.o
 obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 0000000..7867612
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,334 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller errors on
+ * HiSilicon HIP SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#include "../pci.h"
+
+#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
+			0xA8, 0x67, 0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_err_data {
+	u64   val_bits;
+	u8    version;
+	u8    soc_id;
+	u8    socket_id;
+	u8    nimbus_id;
+	u8    sub_module_id;
+	u8    core_id;
+	u8    port_id;
+	u8    err_severity;
+	u16   err_type;
+	u8    reserv[2];
+	u32   err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_err_info {
+	struct hisi_pcie_err_data err_data;
+	struct platform_device *pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_err_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+					u32 chip_id, u32 port_id)
+{
+	struct device *dev = &pdev->dev;
+	acpi_handle handle = ACPI_HANDLE(dev);
+	union acpi_object arg[3];
+	struct acpi_object_list arg_list;
+	acpi_status s;
+	unsigned long long data = 0;
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	/* Call the ACPI handle to reset root port */
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No RST method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	struct device *device = &dev->dev;
+	acpi_handle root_handle = ACPI_HANDLE(device);
+	struct acpi_pci_root *pci_root;
+	struct pci_bus *root_bus;
+	struct pci_dev *pdev;
+	u32 domain, busnr, devfn;
+
+	s = acpi_get_parent(root_handle, &root_handle);
+	if (ACPI_FAILURE(s))
+		return -ENODEV;
+	pci_root = acpi_pci_find_root(root_handle);
+	if (!pci_root)
+		return -ENODEV;
+	root_bus = pci_root->bus;
+	domain = pci_root->segment;
+
+	busnr = root_bus->number;
+	devfn = PCI_DEVFN(port_id, 0);
+	pdev = pci_get_domain_bus_and_slot(domain, busnr, devfn);
+	if (!pdev) {
+		dev_info(device, "Fail to get root port %04x:%02x:%02x.%d device\n",
+			 domain, busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
+		return -ENODEV;
+	}
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/*
+	 * The initialization time of subordinate devices after
+	 * hot reset is no more than 1s, which is required by
+	 * the PCI spec v5.0 sec 6.6.1. The time will shorten
+	 * if Readiness Notifications mechanisms are used. But
+	 * wait 1s here to adapt any conditions.
+	 */
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void hisi_pcie_handle_one_error(const struct hisi_pcie_err_data *err,
+				    struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (err->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(err->sub_module_id));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_err_severity(err->err_severity));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (err->val_bits & BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller errors */
+	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, err->socket_id,
+				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static DEFINE_KFIFO(hisi_pcie_err_recover_ring, struct hisi_pcie_err_info,
+		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
+static DEFINE_SPINLOCK(hisi_pcie_err_recover_ring_lock);
+
+static void hisi_pcie_err_recover_work_func(struct work_struct *work)
+{
+	struct hisi_pcie_err_info pcie_err_entry;
+
+	while (kfifo_get(&hisi_pcie_err_recover_ring, &pcie_err_entry)) {
+		hisi_pcie_handle_one_error(&pcie_err_entry.err_data,
+					pcie_err_entry.pdev);
+	}
+}
+
+static DECLARE_WORK(hisi_pcie_err_recover_work,
+		    hisi_pcie_err_recover_work_func);
+
+static int hisi_pcie_error_handle(struct acpi_hest_generic_data *gdata,
+				  int sev, void *data)
+{
+	const struct hisi_pcie_err_data *err_data =
+					acpi_hest_get_payload(gdata);
+	struct hisi_pcie_err_info err_info;
+	struct platform_device *pdev = data;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return GHES_EVENT_NONE;
+
+	if (err_data->socket_id != socket)
+		return GHES_EVENT_NONE;
+
+	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
+	err_info.pdev = pdev;
+
+	if (kfifo_in_spinlocked(&hisi_pcie_err_recover_ring, &err_info, 1,
+				&hisi_pcie_err_recover_ring_lock))
+		schedule_work(&hisi_pcie_err_recover_work);
+	else
+		dev_warn(dev, "queue full when recovering PCIe controller error\n");
+
+	return GHES_EVENT_HANDLED;
+}
+
+static int hisi_pcie_err_handler_probe(struct platform_device *pdev)
+{
+	int ret;
+
+	ret = ghes_register_event_handler(hisi_pcie_sec_type,
+					  hisi_pcie_error_handle, pdev);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_handler fail\n",
+			__func__);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_err_handler_remove(struct platform_device *pdev)
+{
+	ghes_unregister_event_handler(hisi_pcie_sec_type, pdev);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_err_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-err-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_err_handler_probe,
+	.remove		= hisi_pcie_err_handler_remove,
+};
+module_platform_driver(hisi_pcie_err_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP PCIe controller error handling driver");
+MODULE_LICENSE("GPL v2");
-- 
1.9.1

RE: [PATCH v4 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi All,

Gentle reminder on this patch set.

Thanks,
Shiju

>-----Original Message-----
>From: linux-acpi-owner@vger.kernel.org [mailto:linux-acpi-
>owner@vger.kernel.org] On Behalf Of Shiju Jose
>Sent: 07 February 2020 10:32
>To: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; bp@alien8.de; james.morse@arm.com; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de
>Cc: Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>; Shiju Jose <shiju.jose@huawei.com>
>Subject: [PATCH v4 0/2] ACPI: APEI: Add support to notify the vendor specific
>HW errors
>
>Presently the vendor drivers are unable to do the recovery for the vendor
>specific recoverable HW errors, reported to the APEI driver in the vendor
>defined sections, because APEI driver does not support reporting the same to
>the vendor drivers.
>
>This patch set
>1. add an interface to the APEI driver to enable the vendor drivers to register
>the event handling functions for the corresponding vendor specific HW errors
>and report the error to the vendor driver.
>
>2. add driver to handle HiSilicon hip08 PCIe controller's errors
>   which is an example application of the above APEI interface.
>
>Changes:
>
>V4:
>1. Fix for the smatch warning in the PCIe error driver:
>   warn: should '((((1))) << (9 + i))' be a 64 bit type?
>   if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
>	^^^ This should be BIT_ULL() because it goes up to 9 + 32.
>
>V3:
>1. Fix the comments from Bjorn Helgaas.
>
>V2:
>1. Changes in the HiSilicon PCIe controller's error handling driver
>   for the comments from Bjorn Helgaas.
>
>2. Changes in the APEI interface to support reporting the vendor error
>   for module with multiple devices, but use the same section type.
>   In the error handler will use socket id/sub module id etc to distinguish
>   the device.
>
>V1:
>1. Fix comments from James Morse.
>
>2. add driver to handle HiSilicon hip08 PCIe controller's errors,
>   which is an application of the above interface.
>
>Shiju Jose (1):
>  ACPI: APEI: Add support to notify the vendor specific HW errors
>
>Yicong Yang (1):
>  PCI: HIP: Add handling of HiSilicon HIP PCIe controller errors
>
> drivers/acpi/apei/ghes.c                 | 116 ++++++++++-
> drivers/pci/controller/Kconfig           |   8 +
> drivers/pci/controller/Makefile          |   1 +
> drivers/pci/controller/pcie-hisi-error.c | 334
>+++++++++++++++++++++++++++++++
> include/acpi/ghes.h                      |  56 ++++++
> 5 files changed, 510 insertions(+), 5 deletions(-)  create mode 100644
>drivers/pci/controller/pcie-hisi-error.c
>
>--
>1.9.1
>

Re: [PATCH v4 0/2] ACPI: APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi Shiju,

On 09/03/2020 09:23, Shiju Jose wrote:
> Gentle reminder on this patch set.

Your cover-letter has:
| X-Mailer: git-send-email 2.19.2.windows.1
| In-Reply-To: <Shiju Jose>
| References: <Shiju Jose>

Which causes my mail client to thread this with year-old mail ... hence I've only just
seen this. Other people may have the same problem.
If you're feeding these headers into git-send-email, it expects the value from the
original message's 'Message-Id'... but you don't want this for a cover letter!


Thanks,

James

[PATCH v6 0/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently the vendor drivers are unable to do the recovery for the
vendor specific recoverable HW errors, reported to the APEI driver
in the vendor defined sections, because APEI driver does not support
reporting the same to the vendor drivers.

This patch set
1. add an interface to the APEI driver to enable the vendor
drivers to register the event handling functions for the corresponding
vendor specific HW errors and report the error to the vendor driver.

2. add driver to handle HiSilicon hip08 PCIe controller's errors
   which is an example application of the above APEI interface.

Changes:

V6:
1. Fix few changes in the patch subject line suggested by Bjorn Helgaas.

V5:
1. Fix comments from James Morse.
1.1 Changed the notification method to use the atomic_notifier_chain.
1.2 Add the error handled status for the user space.  

V4:
1. Fix for the smatch warning in the PCIe error driver:
   warn: should '((((1))) << (9 + i))' be a 64 bit type?
   if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
	^^^ This should be BIT_ULL() because it goes up to 9 + 32.

V3:
1. Fix the comments from Bjorn Helgaas.

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (1):
  ACPI / APEI: Add support to notify the vendor specific HW errors

Yicong Yang (1):
  PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

 drivers/acpi/apei/ghes.c                 |  35 ++-
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 357 +++++++++++++++++++++++
 drivers/ras/ras.c                        |   5 +-
 include/acpi/ghes.h                      |  28 ++
 include/linux/ras.h                      |   6 +-
 include/ras/ras_event.h                  |   7 +-
 8 files changed, 440 insertions(+), 7 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
2.17.1

[PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Presently APEI does not support reporting the vendor specific
HW errors, received in the vendor defined table entries, to the
vendor drivers for any recovery.

This patch adds the support to register and unregister the
error handling function for the vendor specific HW errors and
notify the registered kernel driver.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 35 ++++++++++++++++++++++++++++++++++-
 drivers/ras/ras.c        |  5 +++--
 include/acpi/ghes.h      | 28 ++++++++++++++++++++++++++++
 include/linux/ras.h      |  6 ++++--
 include/ras/ras_event.h  |  7 +++++--
 5 files changed, 74 insertions(+), 7 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 24c9642e8fc7..d83f0b1aad0d 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -490,6 +490,32 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+static ATOMIC_NOTIFIER_HEAD(ghes_event_notify_list);
+
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return atomic_notifier_chain_register(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_notifier);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+	atomic_notifier_chain_unregister(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_notifier);
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
 			log_arm_hw_error(err);
 		} else {
 			void *err = acpi_hest_get_payload(gdata);
+			u8 error_handled = false;
+			int ret;
+
+			ret = atomic_notifier_call_chain(&ghes_event_notify_list, 0, gdata);
+			if (ret & NOTIFY_OK)
+				error_handled = true;
 
 			log_non_standard_event(sec_type, fru_id, fru_text,
 					       sec_sev, err,
-					       gdata->error_data_length);
+					       gdata->error_data_length,
+					       error_handled);
 		}
 	}
 }
diff --git a/drivers/ras/ras.c b/drivers/ras/ras.c
index 95540ea8dd9d..0ed784a8466e 100644
--- a/drivers/ras/ras.c
+++ b/drivers/ras/ras.c
@@ -16,9 +16,10 @@
 
 void log_non_standard_event(const guid_t *sec_type, const guid_t *fru_id,
 			    const char *fru_text, const u8 sev, const u8 *err,
-			    const u32 len)
+			    const u32 len, const u8 error_handled)
 {
-	trace_non_standard_event(sec_type, fru_id, fru_text, sev, err, len);
+	trace_non_standard_event(sec_type, fru_id, fru_text, sev,
+				 err, len, error_handled);
 }
 
 void log_arm_hw_error(struct cper_sec_proc_arm *err)
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cddb4ac8..a3dd82069069 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,34 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_notifier(struct notifier_block *nb);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb);
+#else
+static inline int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
diff --git a/include/linux/ras.h b/include/linux/ras.h
index 7c3debb47c87..6ed3c67ab905 100644
--- a/include/linux/ras.h
+++ b/include/linux/ras.h
@@ -28,13 +28,15 @@ static inline int cec_add_elem(u64 pfn)		{ return -ENODEV; }
 #ifdef CONFIG_RAS
 void log_non_standard_event(const guid_t *sec_type,
 			    const guid_t *fru_id, const char *fru_text,
-			    const u8 sev, const u8 *err, const u32 len);
+			    const u8 sev, const u8 *err, const u32 len,
+			    const u8 error_handled);
 void log_arm_hw_error(struct cper_sec_proc_arm *err);
 #else
 static inline void
 log_non_standard_event(const guid_t *sec_type,
 		       const guid_t *fru_id, const char *fru_text,
-		       const u8 sev, const u8 *err, const u32 len)
+		       const u8 sev, const u8 *err, const u32 len,
+		       const u8 error_handled);
 { return; }
 static inline void
 log_arm_hw_error(struct cper_sec_proc_arm *err) { return; }
diff --git a/include/ras/ras_event.h b/include/ras/ras_event.h
index 36c5c5e38c1d..38fd05d82d8e 100644
--- a/include/ras/ras_event.h
+++ b/include/ras/ras_event.h
@@ -223,9 +223,10 @@ TRACE_EVENT(non_standard_event,
 		 const char *fru_text,
 		 const u8 sev,
 		 const u8 *err,
-		 const u32 len),
+		 const u32 len,
+		 const u8 error_handled),
 
-	TP_ARGS(sec_type, fru_id, fru_text, sev, err, len),
+	TP_ARGS(sec_type, fru_id, fru_text, sev, err, len, error_handled),
 
 	TP_STRUCT__entry(
 		__array(char, sec_type, UUID_SIZE)
@@ -234,6 +235,7 @@ TRACE_EVENT(non_standard_event,
 		__field(u8, sev)
 		__field(u32, len)
 		__dynamic_array(u8, buf, len)
+		__field(u8, error_handled)
 	),
 
 	TP_fast_assign(
@@ -243,6 +245,7 @@ TRACE_EVENT(non_standard_event,
 		__entry->sev = sev;
 		__entry->len = len;
 		memcpy(__get_dynamic_array(buf), err, len);
+		__entry->error_handled = error_handled;
 	),
 
 	TP_printk("severity: %d; sec type:%pU; FRU: %pU %s; data len:%d; raw data:%s",
-- 
2.17.1

Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Borislav Petkov]

On Wed, Mar 25, 2020 at 04:42:22PM +0000, Shiju Jose wrote:
> Presently APEI does not support reporting the vendor specific
> HW errors, received in the vendor defined table entries, to the
> vendor drivers for any recovery.
> 
> This patch adds the support to register and unregister the

Avoid having "This patch" or "This commit" in the commit message. It is
tautologically useless.

Also, do

$ git grep 'This patch' Documentation/process

for more details.

> error handling function for the vendor specific HW errors and
> notify the registered kernel driver.
> 
> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
> ---
>  drivers/acpi/apei/ghes.c | 35 ++++++++++++++++++++++++++++++++++-
>  drivers/ras/ras.c        |  5 +++--
>  include/acpi/ghes.h      | 28 ++++++++++++++++++++++++++++
>  include/linux/ras.h      |  6 ++++--
>  include/ras/ras_event.h  |  7 +++++--
>  5 files changed, 74 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index 24c9642e8fc7..d83f0b1aad0d 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -490,6 +490,32 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
>  #endif
>  }
>  
> +static ATOMIC_NOTIFIER_HEAD(ghes_event_notify_list);
> +
> +/**
> + * ghes_register_event_notifier - register an event notifier
> + * for the non-fatal HW errors.
> + * @nb: pointer to the notifier_block structure of the event handler.
> + *
> + * return 0 : SUCCESS, non-zero : FAIL
> + */
> +int ghes_register_event_notifier(struct notifier_block *nb)
> +{
> +	return atomic_notifier_chain_register(&ghes_event_notify_list, nb);
> +}
> +EXPORT_SYMBOL_GPL(ghes_register_event_notifier);
> +
> +/**
> + * ghes_unregister_event_notifier - unregister the previously
> + * registered event notifier.
> + * @nb: pointer to the notifier_block structure of the event handler.
> + */
> +void ghes_unregister_event_notifier(struct notifier_block *nb)
> +{
> +	atomic_notifier_chain_unregister(&ghes_event_notify_list, nb);
> +}
> +EXPORT_SYMBOL_GPL(ghes_unregister_event_notifier);
> +
>  static void ghes_do_proc(struct ghes *ghes,
>  			 const struct acpi_hest_generic_status *estatus)
>  {
> @@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
>  			log_arm_hw_error(err);
>  		} else {
>  			void *err = acpi_hest_get_payload(gdata);
> +			u8 error_handled = false;
> +			int ret;
> +
> +			ret = atomic_notifier_call_chain(&ghes_event_notify_list, 0, gdata);

Well, this is a notifier with standard name for a non-standard event.
Not optimal.

Why does only this event need a notifier? Because your driver is
interested in only those events?

> +			if (ret & NOTIFY_OK)
> +				error_handled = true;
>  
>  			log_non_standard_event(sec_type, fru_id, fru_text,
>  					       sec_sev, err,
> -					       gdata->error_data_length);
> +					       gdata->error_data_length,
> +					       error_handled);

What's that error_handled thing for? That's just silly.

Your notifier returns NOTIFY_STOP when it has queued the error. If you
don't want to log it, just test == NOTIFY_STOP and do not log it then.

Then your notifier callback is queuing the error into a kfifo for
whatever reason and then scheduling a workqueue to handle it in user
context...

So I'm thinking that it would be better if you:

* make that kfifo generic and part of ghes.c and queue all types of
error records into it in ghes_do_proc() - not just the non-standard
ones.

* then, when you're done queuing, you kick a workqueue.

* that workqueue runs a normal, blocking notifier to which drivers
register.

Your driver can register to that notifier too and do the normal handling
then and not have this ad-hoc, semi-generic, semi-vendor-specific thing.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi Borislav,

Thanks for reviewing the patches.

>-----Original Message-----
>From: linux-acpi-owner@vger.kernel.org [mailto:linux-acpi-
>owner@vger.kernel.org] On Behalf Of Borislav Petkov
>Sent: 27 March 2020 18:22
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; james.morse@arm.com; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>On Wed, Mar 25, 2020 at 04:42:22PM +0000, Shiju Jose wrote:
>> Presently APEI does not support reporting the vendor specific HW
>> errors, received in the vendor defined table entries, to the vendor
>> drivers for any recovery.
>>
>> This patch adds the support to register and unregister the
>
>Avoid having "This patch" or "This commit" in the commit message. It is
>tautologically useless.
>
Sure.

>Also, do
>
>$ git grep 'This patch' Documentation/process
>
>for more details.
Sure.

>
>> error handling function for the vendor specific HW errors and notify
>> the registered kernel driver.
>>
>> Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
>> ---
>>  drivers/acpi/apei/ghes.c | 35 ++++++++++++++++++++++++++++++++++-
>>  drivers/ras/ras.c        |  5 +++--
>>  include/acpi/ghes.h      | 28 ++++++++++++++++++++++++++++
>>  include/linux/ras.h      |  6 ++++--
>>  include/ras/ras_event.h  |  7 +++++--
>>  5 files changed, 74 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c index
>> 24c9642e8fc7..d83f0b1aad0d 100644
>> --- a/drivers/acpi/apei/ghes.c
>> +++ b/drivers/acpi/apei/ghes.c
>> @@ -490,6 +490,32 @@ static void ghes_handle_aer(struct
>> acpi_hest_generic_data *gdata)  #endif  }
>>
>> +static ATOMIC_NOTIFIER_HEAD(ghes_event_notify_list);
>> +
>> +/**
>> + * ghes_register_event_notifier - register an event notifier
>> + * for the non-fatal HW errors.
>> + * @nb: pointer to the notifier_block structure of the event handler.
>> + *
>> + * return 0 : SUCCESS, non-zero : FAIL  */ int
>> +ghes_register_event_notifier(struct notifier_block *nb) {
>> +	return atomic_notifier_chain_register(&ghes_event_notify_list, nb);
>> +} EXPORT_SYMBOL_GPL(ghes_register_event_notifier);
>> +
>> +/**
>> + * ghes_unregister_event_notifier - unregister the previously
>> + * registered event notifier.
>> + * @nb: pointer to the notifier_block structure of the event handler.
>> + */
>> +void ghes_unregister_event_notifier(struct notifier_block *nb)
>> +{
>> +	atomic_notifier_chain_unregister(&ghes_event_notify_list, nb);
>> +}
>> +EXPORT_SYMBOL_GPL(ghes_unregister_event_notifier);
>> +
>>  static void ghes_do_proc(struct ghes *ghes,
>>  			 const struct acpi_hest_generic_status *estatus)
>>  {
>> @@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
>>  			log_arm_hw_error(err);
>>  		} else {
>>  			void *err = acpi_hest_get_payload(gdata);
>> +			u8 error_handled = false;
>> +			int ret;
>> +
>> +			ret =
>atomic_notifier_call_chain(&ghes_event_notify_list, 0, gdata);
>
>Well, this is a notifier with standard name for a non-standard event.
>Not optimal.
Ok.

>
>Why does only this event need a notifier? Because your driver is
>interested in only those events?
The error events for the PCIe controller can be reported to the kernel in the vendor defined format
[as per the"N.2.3 Non-standard Section Body" of the UEFI spec]. 
Thus these events require a notifier from APEI to the corresponding kernel driver. 

>
>> +			if (ret & NOTIFY_OK)
>> +				error_handled = true;
>>
>>  			log_non_standard_event(sec_type, fru_id, fru_text,
>>  					       sec_sev, err,
>> -					       gdata->error_data_length);
>> +					       gdata->error_data_length,
>> +					       error_handled);
>
>What's that error_handled thing for? That's just silly.
This field added based on the input from James Morse on v4 patch to enable the user space application(rasdaemon)
do the decoding and logging of the any extra error information shared by the corresponding  kernel driver to the user space.

>
>Your notifier returns NOTIFY_STOP when it has queued the error. If you
>don't want to log it, just test == NOTIFY_STOP and do not log it then.
sure.
   
>
>Then your notifier callback is queuing the error into a kfifo for
>whatever reason and then scheduling a workqueue to handle it in user
>context...
>
>So I'm thinking that it would be better if you:
>
>* make that kfifo generic and part of ghes.c and queue all types of
>error records into it in ghes_do_proc() - not just the non-standard
>ones.
>
>* then, when you're done queuing, you kick a workqueue.
>
>* that workqueue runs a normal, blocking notifier to which drivers
>register.
Sure. I will test this method and update.
Can you please confirm you want all the existing standard errors(memory, ARM, PCIE) in the ghes_do_proc ()
to be reported through the blocking notifier?

>
>Your driver can register to that notifier too and do the normal handling
>then and not have this ad-hoc, semi-generic, semi-vendor-specific thing.
>
>Thx.
>
>--
>Regards/Gruss,
>    Boris.
>
>https://people.kernel.org/tglx/notes-about-netiquette

Thanks,
Shiju

Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Borislav Petkov]

On Mon, Mar 30, 2020 at 10:14:20AM +0000, Shiju Jose wrote:
> This field added based on the input from James Morse on v4 patch to
> enable the user space application(rasdaemon) do the decoding and
> logging of the any extra error information shared by the corresponding
> kernel driver to the user space.

How is your error reporting supposed to work?

Your driver is printing error information in dmesg and, at the same
time, you want to report errors with the rasdaemon.

Currently, the kernel does not report any error info if there's a user
agent like rasdaemon registered so you need to think about what exactly
you're trying to achieve here wrt to error handling. Port resetting,
printing error info, etc. Always ask yourself, what can the user do with
the information you're printing. And so on...

> Can you please confirm you want all the existing standard
> errors(memory, ARM, PCIE) in the ghes_do_proc () to be reported
> through the blocking notifier?

Yes, I would very much prefer to have a generic solution instead of
vendor-specific stuff left and right.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi Boris,

>-----Original Message-----
>From: Borislav Petkov [mailto:bp@alien8.de]
>Sent: 30 March 2020 11:34
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; james.morse@arm.com; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>On Mon, Mar 30, 2020 at 10:14:20AM +0000, Shiju Jose wrote:
>> This field added based on the input from James Morse on v4 patch to
>> enable the user space application(rasdaemon) do the decoding and
>> logging of the any extra error information shared by the corresponding
>> kernel driver to the user space.
>
>How is your error reporting supposed to work?
>
>Your driver is printing error information in dmesg and, at the same time, you
>want to report errors with the rasdaemon.
>
>Currently, the kernel does not report any error info if there's a user agent like
>rasdaemon registered so you need to think about what exactly you're trying
>to achieve here wrt to error handling. Port resetting, printing error info, etc.
>Always ask yourself, what can the user do with the information you're
>printing. And so on...
The error_handled field added on the generic basis for the non-standard errors.
rasdaemon supports adding decoding of the vendor-specific error data, printing and 
storing the decoded vendor error information to the sql database. 
The idea was the  error handled field  will help the decoding part of the rasdaemon to do the
appropriate steps for logging the vendor error information depending on whether a corresponding kernel driver
has handled the error or not.  
However I think the same can be achieved by adding an error handling status field to the vendor-specific data, which
the kernel  driver will set after handling the error and corresponding vendor-specific code in the rasdaemon will use it 
while logging the vendor error data.
>
>> Can you please confirm you want all the existing standard
>> errors(memory, ARM, PCIE) in the ghes_do_proc () to be reported
>> through the blocking notifier?
>
>Yes, I would very much prefer to have a generic solution instead of vendor-
>specific stuff left and right.
Sure.

>
>Thx.
>
>--
>Regards/Gruss,
>    Boris.
>
>https://people.kernel.org/tglx/notes-about-netiquette

Thanks,
Shiju

Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Borislav Petkov]

On Mon, Mar 30, 2020 at 11:55:35AM +0000, Shiju Jose wrote:
> The idea was the error handled field will help the decoding part of
> the rasdaemon to do the appropriate steps for logging the vendor error
> information depending on whether a corresponding kernel driver has
> handled the error or not.

What's the difference for rasdaemon whether the error has been handled
or not?

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi Boris,

>-----Original Message-----
>From: Borislav Petkov [mailto:bp@alien8.de]
>Sent: 30 March 2020 14:43
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; james.morse@arm.com; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>On Mon, Mar 30, 2020 at 11:55:35AM +0000, Shiju Jose wrote:
>> The idea was the error handled field will help the decoding part of
>> the rasdaemon to do the appropriate steps for logging the vendor error
>> information depending on whether a corresponding kernel driver has
>> handled the error or not.
>
>What's the difference for rasdaemon whether the error has been handled or
>not?
Following are some of the examples of the usage of error handled status
in the vendor specific code of the rasdaemon,
1. rasdaemon need not to print the vendor error data reported by the firmware if the 
    kernel driver already print those information. In this case rasdaemon will only need to store
    the decoded vendor error data to the SQL database.  
2. If the vendor kernel driver want to report extra error information through
    the vendor specific data (though presently we do not have any such use case) for the rasdamon to log. 
    I think the error handled status useful to indicate that the kernel driver has filled the extra information and
    rasdaemon to decode and log them after extra data specific validity check.
      
>
>--
>Regards/Gruss,
>    Boris.
>
>https://people.kernel.org/tglx/notes-about-netiquette

Thanks,
Shiju

Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Borislav Petkov]

On Mon, Mar 30, 2020 at 03:44:29PM +0000, Shiju Jose wrote:
> 1. rasdaemon need not to print the vendor error data reported by the firmware if the 
>     kernel driver already print those information. In this case rasdaemon will only need to store
>     the decoded vendor error data to the SQL database.

Well, there's a problem with this:

rasdaemon printing != kernel driver printing

Because printing in dmesg would need people to go grep dmesg.

Printing through rasdaemon or any userspace agent, OTOH, is a lot more
flexible wrt analyzing and collecting those error records. Especially
if you are a data center admin and you want to collect all your error
records: grepping dmesg simply doesn't scale versus all the rasdaemon
agents reporting to a centrallized location.

> 2. If the vendor kernel driver want to report extra error information through
>     the vendor specific data (though presently we do not have any such use case) for the rasdamon to log. 
>     I think the error handled status useful to indicate that the kernel driver has filled the extra information and
>     rasdaemon to decode and log them after extra data specific validity check.

The kernel driver can report that extra information without the kernel
saying that the error was handled.

So I still see no sense for the kernel to tell userspace explicitly that
it handled the error. There might be a valid reason, though, of which I
cannot think of right now.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi Boris,

>-----Original Message-----
>From: Borislav Petkov [mailto:bp@alien8.de]
>Sent: 31 March 2020 10:09
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; james.morse@arm.com; tony.luck@intel.com;
>gregkh@linuxfoundation.org; zhangliguang@linux.alibaba.com;
>tglx@linutronix.de; Linuxarm <linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>On Mon, Mar 30, 2020 at 03:44:29PM +0000, Shiju Jose wrote:
>> 1. rasdaemon need not to print the vendor error data reported by the
>firmware if the
>>     kernel driver already print those information. In this case rasdaemon will
>only need to store
>>     the decoded vendor error data to the SQL database.
>
>Well, there's a problem with this:
>
>rasdaemon printing != kernel driver printing
>
>Because printing in dmesg would need people to go grep dmesg.
>
>Printing through rasdaemon or any userspace agent, OTOH, is a lot more
>flexible wrt analyzing and collecting those error records. Especially if you are a
>data center admin and you want to collect all your error
>records: grepping dmesg simply doesn't scale versus all the rasdaemon
>agents reporting to a centrallized location.
Ok.
I posted V7 of this series.  
"[v7 PATCH 0/6] ACPI / APEI: Add support to notify non-fatal HW errors"

>
>> 2. If the vendor kernel driver want to report extra error information
>through
>>     the vendor specific data (though presently we do not have any such use
>case) for the rasdamon to log.
>>     I think the error handled status useful to indicate that the kernel driver
>has filled the extra information and
>>     rasdaemon to decode and log them after extra data specific validity
>check.
>
>The kernel driver can report that extra information without the kernel saying
>that the error was handled.
>
>So I still see no sense for the kernel to tell userspace explicitly that it handled
>the error. There might be a valid reason, though, of which I cannot think of
>right now.
Ok.

>
>Thx.
>
>--
>Regards/Gruss,
>    Boris.
>
>https://people.kernel.org/tglx/notes-about-netiquette

Thanks,
Shiju

Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[James Morse]

Hi Boris, Shiju,

Sorry for not spotting this reply earlier: Its in-reply to v1, so gets buried.

On 27/03/2020 18:22, Borislav Petkov wrote:
> On Wed, Mar 25, 2020 at 04:42:22PM +0000, Shiju Jose wrote:
>> Presently APEI does not support reporting the vendor specific
>> HW errors, received in the vendor defined table entries, to the
>> vendor drivers for any recovery.
>>
>> This patch adds the support to register and unregister the
> 
> Avoid having "This patch" or "This commit" in the commit message. It is
> tautologically useless.
> 
> Also, do
> 
> $ git grep 'This patch' Documentation/process
> 
> for more details.
> 
>> error handling function for the vendor specific HW errors and
>> notify the registered kernel driver.

>> @@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
>>  			log_arm_hw_error(err);
>>  		} else {
>>  			void *err = acpi_hest_get_payload(gdata);
>> +			u8 error_handled = false;
>> +			int ret;
>> +
>> +			ret = atomic_notifier_call_chain(&ghes_event_notify_list, 0, gdata);
> 
> Well, this is a notifier with standard name for a non-standard event.
> Not optimal.
> 
> Why does only this event need a notifier? Because your driver is
> interested in only those events?

Its the 'else' catch-all for stuff drivers/acpi/apei  doesn't know to handle.

In this case its because its a vendor specific GUID that only the vendor driver knows how
to parse.


>> +			if (ret & NOTIFY_OK)
>> +				error_handled = true;
>>  
>>  			log_non_standard_event(sec_type, fru_id, fru_text,
>>  					       sec_sev, err,
>> -					       gdata->error_data_length);
>> +					       gdata->error_data_length,
>> +					       error_handled);
> 
> What's that error_handled thing for? That's just silly.
> 
> Your notifier returns NOTIFY_STOP when it has queued the error. If you
> don't want to log it, just test == NOTIFY_STOP and do not log it then.

My thinking for this being needed was so user-space consumers of those tracepoints keep
working. Otherwise you upgrade, get this feature, and your user-space counters stop working.

You'd need to know this error source was now managed by an in-kernel driver, which may
report the errors somewhere else...


> Then your notifier callback is queuing the error into a kfifo for
> whatever reason and then scheduling a workqueue to handle it in user
> context...
> 
> So I'm thinking that it would be better if you:
> 
> * make that kfifo generic and part of ghes.c and queue all types of
> error records into it in ghes_do_proc() - not just the non-standard
> ones.

Move the drop to process context into ghes.c? This should result in less code.

I asked for this hooking to only be for the 'catch all' don't-know case so that we don't
get drivers trying to hook and handle memory errors. (if we ever wanted that, it should be
from part of memory_failure() so it catches all the ways of reporting memory-failure)
32bit arm has prior in this area.


> * then, when you're done queuing, you kick a workqueue.
> 
> * that workqueue runs a normal, blocking notifier to which drivers
> register.
> 
> Your driver can register to that notifier too and do the normal handling
> then and not have this ad-hoc, semi-generic, semi-vendor-specific thing.

As long as we don't walk a list of things that might handle a memory-error, and have some
random driver try and NOTIFY_STOP it....

aer_recover_queue() would be replaced by this. memory_failure_queue() has one additional
caller in drivers/ras/cec.c.


Thanks,

James

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi James,

>-----Original Message-----
>From: linux-pci-owner@vger.kernel.org [mailto:linux-pci-
>owner@vger.kernel.org] On Behalf Of James Morse
>Sent: 08 April 2020 11:03
>To: Borislav Petkov <bp@alien8.de>; Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; tony.luck@intel.com; gregkh@linuxfoundation.org;
>zhangliguang@linux.alibaba.com; tglx@linutronix.de; Linuxarm
><linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>Hi Boris, Shiju,
>
>Sorry for not spotting this reply earlier: Its in-reply to v1, so gets buried.
I will resend the v7 patch solving this issue.
I guess the remaining  questions here are for Boris. May be can we discuss
your comments with V7 patch, which I will send?

>
>On 27/03/2020 18:22, Borislav Petkov wrote:
>> On Wed, Mar 25, 2020 at 04:42:22PM +0000, Shiju Jose wrote:
>>> Presently APEI does not support reporting the vendor specific HW
>>> errors, received in the vendor defined table entries, to the vendor
>>> drivers for any recovery.
>>>
>>> This patch adds the support to register and unregister the
>>
>> Avoid having "This patch" or "This commit" in the commit message. It
>> is tautologically useless.
>>
>> Also, do
>>
>> $ git grep 'This patch' Documentation/process
>>
>> for more details.
>>
>>> error handling function for the vendor specific HW errors and notify
>>> the registered kernel driver.
>
>>> @@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
>>>  			log_arm_hw_error(err);
>>>  		} else {
>>>  			void *err = acpi_hest_get_payload(gdata);
>>> +			u8 error_handled = false;
>>> +			int ret;
>>> +
>>> +			ret =
>atomic_notifier_call_chain(&ghes_event_notify_list, 0,
>>> +gdata);
>>
>> Well, this is a notifier with standard name for a non-standard event.
>> Not optimal.
>>
>> Why does only this event need a notifier? Because your driver is
>> interested in only those events?
>
>Its the 'else' catch-all for stuff drivers/acpi/apei  doesn't know to handle.
>
>In this case its because its a vendor specific GUID that only the vendor driver
>knows how to parse.
>
>
>>> +			if (ret & NOTIFY_OK)
>>> +				error_handled = true;
>>>
>>>  			log_non_standard_event(sec_type, fru_id, fru_text,
>>>  					       sec_sev, err,
>>> -					       gdata->error_data_length);
>>> +					       gdata->error_data_length,
>>> +					       error_handled);
>>
>> What's that error_handled thing for? That's just silly.
>>
>> Your notifier returns NOTIFY_STOP when it has queued the error. If you
>> don't want to log it, just test == NOTIFY_STOP and do not log it then.
>
>My thinking for this being needed was so user-space consumers of those
>tracepoints keep working. Otherwise you upgrade, get this feature, and your
>user-space counters stop working.
>
>You'd need to know this error source was now managed by an in-kernel
>driver, which may report the errors somewhere else...
>
>
>> Then your notifier callback is queuing the error into a kfifo for
>> whatever reason and then scheduling a workqueue to handle it in user
>> context...
>>
>> So I'm thinking that it would be better if you:
>>
>> * make that kfifo generic and part of ghes.c and queue all types of
>> error records into it in ghes_do_proc() - not just the non-standard
>> ones.
>
>Move the drop to process context into ghes.c? This should result in less code.
>
>I asked for this hooking to only be for the 'catch all' don't-know case so that
>we don't get drivers trying to hook and handle memory errors. (if we ever
>wanted that, it should be from part of memory_failure() so it catches all the
>ways of reporting memory-failure) 32bit arm has prior in this area.
>
>
>> * then, when you're done queuing, you kick a workqueue.
>>
>> * that workqueue runs a normal, blocking notifier to which drivers
>> register.
>>
>> Your driver can register to that notifier too and do the normal
>> handling then and not have this ad-hoc, semi-generic, semi-vendor-specific
>thing.
>
>As long as we don't walk a list of things that might handle a memory-error,
>and have some random driver try and NOTIFY_STOP it....
>
>aer_recover_queue() would be replaced by this. memory_failure_queue() has
>one additional caller in drivers/ras/cec.c.
>
>
>Thanks,
>
>James
Thanks,
Shiju

RE: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Shiju Jose]

Hi Boris, Hi James,

>-----Original Message-----
>From: linux-pci-owner@vger.kernel.org [mailto:linux-pci-
>owner@vger.kernel.org] On Behalf Of James Morse
>Sent: 08 April 2020 11:03
>To: Borislav Petkov <bp@alien8.de>; Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; helgaas@kernel.org;
>lenb@kernel.org; tony.luck@intel.com; gregkh@linuxfoundation.org;
>zhangliguang@linux.alibaba.com; tglx@linutronix.de; Linuxarm
><linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [PATCH v6 1/2] ACPI / APEI: Add support to notify the vendor
>specific HW errors
>
>Hi Boris, Shiju,
>
>Sorry for not spotting this reply earlier: Its in-reply to v1, so gets buried.
>
>On 27/03/2020 18:22, Borislav Petkov wrote:
>> On Wed, Mar 25, 2020 at 04:42:22PM +0000, Shiju Jose wrote:
>>> Presently APEI does not support reporting the vendor specific HW
>>> errors, received in the vendor defined table entries, to the vendor
>>> drivers for any recovery.
>>>
>>> This patch adds the support to register and unregister the
>>
>> Avoid having "This patch" or "This commit" in the commit message. It
>> is tautologically useless.
>>
>> Also, do
>>
>> $ git grep 'This patch' Documentation/process
>>
>> for more details.
>>
>>> error handling function for the vendor specific HW errors and notify
>>> the registered kernel driver.
>
>>> @@ -526,10 +552,17 @@ static void ghes_do_proc(struct ghes *ghes,
>>>  			log_arm_hw_error(err);
>>>  		} else {
>>>  			void *err = acpi_hest_get_payload(gdata);
>>> +			u8 error_handled = false;
>>> +			int ret;
>>> +
>>> +			ret =
>atomic_notifier_call_chain(&ghes_event_notify_list, 0,
>>> +gdata);
>>
>> Well, this is a notifier with standard name for a non-standard event.
>> Not optimal.
>>
>> Why does only this event need a notifier? Because your driver is
>> interested in only those events?
>
>Its the 'else' catch-all for stuff drivers/acpi/apei  doesn't know to handle.
>
>In this case its because its a vendor specific GUID that only the vendor driver
>knows how to parse.
>
>
>>> +			if (ret & NOTIFY_OK)
>>> +				error_handled = true;
>>>
>>>  			log_non_standard_event(sec_type, fru_id, fru_text,
>>>  					       sec_sev, err,
>>> -					       gdata->error_data_length);
>>> +					       gdata->error_data_length,
>>> +					       error_handled);
>>
>> What's that error_handled thing for? That's just silly.
>>
>> Your notifier returns NOTIFY_STOP when it has queued the error. If you
>> don't want to log it, just test == NOTIFY_STOP and do not log it then.
>
>My thinking for this being needed was so user-space consumers of those
>tracepoints keep working. Otherwise you upgrade, get this feature, and your
>user-space counters stop working.
>
>You'd need to know this error source was now managed by an in-kernel
>driver, which may report the errors somewhere else...
>
>
>> Then your notifier callback is queuing the error into a kfifo for
>> whatever reason and then scheduling a workqueue to handle it in user
>> context...
>>
>> So I'm thinking that it would be better if you:
>>
>> * make that kfifo generic and part of ghes.c and queue all types of
>> error records into it in ghes_do_proc() - not just the non-standard
>> ones.
>
>Move the drop to process context into ghes.c? This should result in less code.
>
>I asked for this hooking to only be for the 'catch all' don't-know case so that
>we don't get drivers trying to hook and handle memory errors. (if we ever
>wanted that, it should be from part of memory_failure() so it catches all the
>ways of reporting memory-failure) 32bit arm has prior in this area.
>
>
>> * then, when you're done queuing, you kick a workqueue.
>>
>> * that workqueue runs a normal, blocking notifier to which drivers
>> register.
>>
>> Your driver can register to that notifier too and do the normal
>> handling then and not have this ad-hoc, semi-generic, semi-vendor-specific
>thing.
>
>As long as we don't walk a list of things that might handle a memory-error,
>and have some random driver try and NOTIFY_STOP it....
>
>aer_recover_queue() would be replaced by this. memory_failure_queue() has
>one additional caller in drivers/ras/cec.c.

Can you suggest whether the standard errors can report through the 
notifier (kfifo + blocking notifier), [which implemented  in V7 patch], or not 
so that we can proceed with the changes to notify the vendor specific errors?

>
>
>Thanks,
>
>James

Thanks,
Shiju

[PATCH v6 2/2] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The HiSilicon HIP PCIe controller is capable of handling errors
on root port and perform port reset separately at each root port.

This patch add error handling driver for HIP PCIe controller to log
and report recoverable errors. Perform root port reset and restore
link status after the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

Also fix the following Smatch warning:
warn: should '((((1))) << (9 + i))' be a 64 bit type?
if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
     ^^^ This should be BIT_ULL() because it goes up to 9 + 32.
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
drivers/pci/controller/Kconfig           |   8 +
drivers/pci/controller/Makefile          |   1 +
drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
3 files changed, 345 insertions(+)
create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 357 +++++++++++++++++++++++
 3 files changed, 366 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index 20bf00f587bd..8bc6111480c8 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -268,6 +268,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on ACPI
+	bool "HiSilicon HIP PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon HIP SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 01b2502a5323..94f37b3d9929 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -29,6 +29,7 @@ obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
 obj-$(CONFIG_PCIE_BRCMSTB) += pcie-brcmstb.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 000000000000..73304512af92
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,357 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller errors on
+ * HiSilicon HIP SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#include "../pci.h"
+
+#define HISI_PCIE_ERR_RECOVER_RING_SIZE           16
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
+			0xA8, 0x67, 0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_err_data {
+	u64   val_bits;
+	u8    version;
+	u8    soc_id;
+	u8    socket_id;
+	u8    nimbus_id;
+	u8    sub_module_id;
+	u8    core_id;
+	u8    port_id;
+	u8    err_severity;
+	u16   err_type;
+	u8    reserv[2];
+	u32   err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_err_info {
+	struct hisi_pcie_err_data err_data;
+	struct platform_device *pdev;
+};
+
+struct hisi_pcie_err_private {
+	struct notifier_block nb;
+	struct platform_device *pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_err_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+					u32 chip_id, u32 port_id)
+{
+	struct device *dev = &pdev->dev;
+	acpi_handle handle = ACPI_HANDLE(dev);
+	union acpi_object arg[3];
+	struct acpi_object_list arg_list;
+	acpi_status s;
+	unsigned long long data = 0;
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	/* Call the ACPI handle to reset root port */
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No RST method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	struct device *device = &dev->dev;
+	acpi_handle root_handle = ACPI_HANDLE(device);
+	struct acpi_pci_root *pci_root;
+	struct pci_bus *root_bus;
+	struct pci_dev *pdev;
+	u32 domain, busnr, devfn;
+
+	s = acpi_get_parent(root_handle, &root_handle);
+	if (ACPI_FAILURE(s))
+		return -ENODEV;
+	pci_root = acpi_pci_find_root(root_handle);
+	if (!pci_root)
+		return -ENODEV;
+	root_bus = pci_root->bus;
+	domain = pci_root->segment;
+
+	busnr = root_bus->number;
+	devfn = PCI_DEVFN(port_id, 0);
+	pdev = pci_get_domain_bus_and_slot(domain, busnr, devfn);
+	if (!pdev) {
+		dev_warn(device, "Fail to get root port %04x:%02x:%02x.%d device\n",
+			 domain, busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
+		return -ENODEV;
+	}
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/*
+	 * The initialization time of subordinate devices after
+	 * hot reset is no more than 1s, which is required by
+	 * the PCI spec v5.0 sec 6.6.1. The time will shorten
+	 * if Readiness Notifications mechanisms are used. But
+	 * wait 1s here to adapt any conditions.
+	 */
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void hisi_pcie_handle_one_error(const struct hisi_pcie_err_data *err,
+				    struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (err->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", err->version);
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", err->soc_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", err->socket_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", err->nimbus_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(err->sub_module_id));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", err->core_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", err->port_id);
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_err_severity(err->err_severity));
+
+	if (err->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", err->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (err->val_bits & BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, err->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller errors */
+	if (err->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, err->socket_id,
+				HISI_PCIE_PORT_ID(err->core_id, err->port_id));
+		if (rc) {
+			dev_warn(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static DEFINE_KFIFO(hisi_pcie_err_recover_ring, struct hisi_pcie_err_info,
+		    HISI_PCIE_ERR_RECOVER_RING_SIZE);
+static DEFINE_SPINLOCK(hisi_pcie_err_recover_ring_lock);
+
+static void hisi_pcie_err_recover_work_func(struct work_struct *work)
+{
+	struct hisi_pcie_err_info pcie_err_entry;
+
+	while (kfifo_get(&hisi_pcie_err_recover_ring, &pcie_err_entry)) {
+		hisi_pcie_handle_one_error(&pcie_err_entry.err_data,
+					pcie_err_entry.pdev);
+	}
+}
+
+static DECLARE_WORK(hisi_pcie_err_recover_work,
+		    hisi_pcie_err_recover_work_func);
+
+
+static int hisi_pcie_error_notify(struct notifier_block *nb,
+				  unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	const struct hisi_pcie_err_data *err_data =
+				acpi_hest_get_payload(gdata);
+	struct hisi_pcie_err_info err_info;
+	struct hisi_pcie_err_private *priv =
+			container_of(nb, struct hisi_pcie_err_private, nb);
+	struct platform_device *pdev = priv->pdev;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return NOTIFY_DONE;
+
+	if (!guid_equal((guid_t *)gdata->section_type, &hisi_pcie_sec_type) ||
+	    err_data->socket_id != socket)
+		return NOTIFY_DONE;
+
+	memcpy(&err_info.err_data, err_data, sizeof(*err_data));
+	err_info.pdev = pdev;
+
+	if (kfifo_in_spinlocked(&hisi_pcie_err_recover_ring, &err_info, 1,
+				&hisi_pcie_err_recover_ring_lock))
+		schedule_work(&hisi_pcie_err_recover_work);
+	else
+		dev_warn(dev, "queue full when recovering PCIe controller error\n");
+
+	return NOTIFY_STOP;
+}
+
+static int hisi_pcie_err_handler_probe(struct platform_device *pdev)
+{
+	struct hisi_pcie_err_private *priv;
+	int ret;
+
+	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->nb.notifier_call = hisi_pcie_error_notify;
+	priv->pdev = pdev;
+	ret = ghes_register_event_notifier(&priv->nb);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_notifier fail\n",
+			__func__);
+		return ret;
+	}
+
+	platform_set_drvdata(pdev, priv);
+
+	return 0;
+}
+
+static int hisi_pcie_err_handler_remove(struct platform_device *pdev)
+{
+	struct hisi_pcie_err_private *priv = platform_get_drvdata(pdev);
+
+	if (priv)
+		ghes_unregister_event_notifier(&priv->nb);
+
+	kfree(priv);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_err_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-err-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_err_handler_probe,
+	.remove		= hisi_pcie_err_handler_remove,
+};
+module_platform_driver(hisi_pcie_err_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP PCIe controller error handling driver");
+MODULE_LICENSE("GPL v2");
-- 
2.17.1

Re: [PATCH v6 0/2] ACPI / APEI: Add support to notify the vendor specific HW errors

[Bjorn Helgaas]

On Wed, Mar 25, 2020 at 04:42:21PM +0000, Shiju Jose wrote:
> Presently the vendor drivers are unable to do the recovery for the
> vendor specific recoverable HW errors, reported to the APEI driver
> in the vendor defined sections, because APEI driver does not support
> reporting the same to the vendor drivers.
> 
> This patch set
> 1. add an interface to the APEI driver to enable the vendor
> drivers to register the event handling functions for the corresponding
> vendor specific HW errors and report the error to the vendor driver.
> 
> 2. add driver to handle HiSilicon hip08 PCIe controller's errors
>    which is an example application of the above APEI interface.
> 
> Changes:
> 
> V6:
> 1. Fix few changes in the patch subject line suggested by Bjorn Helgaas.

I think it will save everybody a little work if you can wait a day or
two so you can address more comments at once.  You posted v6 only
about three hours after v5, which isn't enough time for people to
respond to v5.

I'm not going to even look at v6 because it doesn't address some of my
v5 comments.  Please wait a few days before v7 to see if Rafael has
any thoughts on where the error driver should live.

Bjorn

[v7 PATCH 0/6] ACPI / APEI: Add support to notify non-fatal HW errors

[Shiju Jose]

Add common interface for queuing up the non-fatal HW errors and notify
the registered kernel drivers.The interface supports drivers to register
to receive the callback for the non-fatal HW errors, including the vendor
specific HW errors, for the recovery and supports handling the non-fatal
errors in the bottom half.

Patch set
1. add the new interface to the APEI driver for the non-fatal HW
   error notification.
2. change the existing error handling for the standard errors
   to use the above notification interface.
2. add driver to handle HiSilicon hip PCIe controller's errors.

Changes:

V7:
1. Add changes in the APEI driver suggested by Borislav Petkov, for
   queuing up all the non-fatal HW errors to the work queue and
   notify the registered kernel drivers from the bottom half using
   blocking notifier, common interface for both standard and
   vendor-spcific errors.
2. Fix for further feedbacks in v5 HIP PCIe error handler driver
   by Bjorn Helgaas.

V6:
1. Fix few changes in the patch subject line suggested by Bjorn Helgaas.

V5:
1. Fix comments from James Morse.
1.1 Changed the notification method to use the atomic_notifier_chain.
1.2 Add the error handled status for the user space.  

V4:
1. Fix for the following smatch warning in the PCIe error driver,
   reported by kbuild test robot<lkp@intel.com>:
   warn: should '((((1))) << (9 + i))' be a 64 bit type?
   if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
	^^^ This should be BIT_ULL() because it goes up to 9 + 32.

V3:
1. Fix the comments from Bjorn Helgaas.

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (5):
  ACPI / APEI: Add support to queuing up the non-fatal HW errors and
    notify
  ACPI / APEI: Add callback for memory errors to the GHES notifier
  ACPI / APEI: Add callback for AER to the GHES notifier
  ACPI / APEI: Add callback for ARM HW errors to the GHES notifier
  ACPI / APEI: Add callback for non-standard HW errors to the GHES
    notifier

Yicong Yang (1):
  PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

 drivers/acpi/apei/ghes.c                 | 268 ++++++++++++++++---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 323 +++++++++++++++++++++++
 include/acpi/ghes.h                      |  28 ++
 5 files changed, 595 insertions(+), 33 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
2.17.1

[v7 PATCH 1/6] ACPI / APEI: Add support to queuing up the non-fatal HW errors and notify

[Shiju Jose]

Add queuing up the non-fatal HW errors to the work queue and interface to
notify the registered kernel drivers for the error recovery.
The interface enables drivers to register the callback functions to
receive the error notification, including the vendor-specific HW errors
and supports handling the non-fatal HW errors in the bottom half.

Suggested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 127 +++++++++++++++++++++++++++++++++++++++
 include/acpi/ghes.h      |  28 +++++++++
 2 files changed, 155 insertions(+)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 24c9642e8fc7..5c0ab5422311 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -33,6 +33,7 @@
 #include <linux/irq_work.h>
 #include <linux/llist.h>
 #include <linux/genalloc.h>
+#include <linux/kfifo.h>
 #include <linux/pci.h>
 #include <linux/pfn.h>
 #include <linux/aer.h>
@@ -63,6 +64,11 @@
 #define GHES_ESTATUS_CACHES_SIZE	4
 
 #define GHES_ESTATUS_IN_CACHE_MAX_NSEC	10000000000ULL
+
+#define GHES_EVENT_RING_SIZE	256
+#define GHES_GDATA_POOL_MIN_ALLOC_ORDER	3
+#define GHES_GDATA_POOL_MIN_SIZE	65536
+
 /* Prevent too many caches are allocated because of RCU */
 #define GHES_ESTATUS_CACHE_ALLOCED_MAX	(GHES_ESTATUS_CACHES_SIZE * 3 / 2)
 
@@ -122,6 +128,19 @@ static DEFINE_MUTEX(ghes_list_mutex);
  */
 static DEFINE_SPINLOCK(ghes_notify_lock_irq);
 
+struct ghes_event_entry {
+	struct acpi_hest_generic_data *gdata;
+	int error_severity;
+};
+
+static DEFINE_KFIFO(ghes_event_ring, struct ghes_event_entry,
+		    GHES_EVENT_RING_SIZE);
+
+static DEFINE_SPINLOCK(ghes_event_ring_lock);
+
+static struct gen_pool *ghes_gdata_pool;
+static unsigned long ghes_gdata_pool_size_request;
+
 static struct gen_pool *ghes_estatus_pool;
 static unsigned long ghes_estatus_pool_size_request;
 
@@ -188,6 +207,40 @@ int ghes_estatus_pool_init(int num_ghes)
 	return -ENOMEM;
 }
 
+int ghes_gdata_pool_init(void)
+{
+	unsigned long addr, len;
+	int rc;
+
+	ghes_gdata_pool = gen_pool_create(GHES_GDATA_POOL_MIN_ALLOC_ORDER, -1);
+	if (!ghes_gdata_pool)
+		return -ENOMEM;
+
+	if (ghes_gdata_pool_size_request < GHES_GDATA_POOL_MIN_SIZE)
+		ghes_gdata_pool_size_request = GHES_GDATA_POOL_MIN_SIZE;
+
+	len = ghes_gdata_pool_size_request;
+	addr = (unsigned long)vmalloc(PAGE_ALIGN(len));
+	if (!addr)
+		goto err_pool_alloc;
+
+	vmalloc_sync_mappings();
+
+	rc = gen_pool_add(ghes_gdata_pool, addr, PAGE_ALIGN(len), -1);
+	if (rc)
+		goto err_pool_add;
+
+	return 0;
+
+err_pool_add:
+	vfree((void *)addr);
+
+err_pool_alloc:
+	gen_pool_destroy(ghes_gdata_pool);
+
+	return -ENOMEM;
+}
+
 static int map_gen_v2(struct ghes *ghes)
 {
 	return apei_map_generic_address(&ghes->generic_v2->read_ack_register);
@@ -247,6 +300,10 @@ static struct ghes *ghes_new(struct acpi_hest_generic *generic)
 		goto err_unmap_status_addr;
 	}
 
+	ghes_gdata_pool_size_request += generic->records_to_preallocate *
+					generic->max_sections_per_record *
+					generic->max_raw_data_length;
+
 	return ghes;
 
 err_unmap_status_addr:
@@ -490,6 +547,49 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
+
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return blocking_notifier_chain_register(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_notifier);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+	blocking_notifier_chain_unregister(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_notifier);
+
+static void ghes_event_work_func(struct work_struct *work)
+{
+	struct ghes_event_entry entry;
+	u32 len;
+
+	while (kfifo_get(&ghes_event_ring, &entry)) {
+		blocking_notifier_call_chain(&ghes_event_notify_list,
+					     entry.error_severity,
+					     entry.gdata);
+		len = acpi_hest_get_size(entry.gdata) +
+				acpi_hest_get_error_length(entry.gdata);
+		gen_pool_free(ghes_gdata_pool, (unsigned long)entry.gdata, len);
+	}
+}
+
+static DECLARE_WORK(ghes_event_work, ghes_event_work_func);
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -498,6 +598,8 @@ static void ghes_do_proc(struct ghes *ghes,
 	guid_t *sec_type;
 	const guid_t *fru_id = &guid_null;
 	char *fru_text = "";
+	struct ghes_event_entry event_entry;
+	u32 len;
 
 	sev = ghes_severity(estatus->error_severity);
 	apei_estatus_for_each_section(estatus, gdata) {
@@ -509,6 +611,25 @@ static void ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
+		len = acpi_hest_get_record_size(gdata);
+		event_entry.gdata = (void *)gen_pool_alloc(ghes_gdata_pool,
+							   len);
+		if (!event_entry.gdata) {
+			pr_warn(GHES_PFX "ghes gdata pool alloc fail\n");
+			break;
+		}
+
+		memcpy(event_entry.gdata, gdata, len);
+		event_entry.error_severity = sev;
+
+		if (kfifo_in_spinlocked(&ghes_event_ring, &event_entry, 1,
+					&ghes_event_ring_lock))
+			schedule_work(&ghes_event_work);
+		else {
+			pr_warn(GHES_PFX "ghes event queue full\n");
+			break;
+		}
+
 		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
 			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
@@ -1346,6 +1467,12 @@ static int __init ghes_init(void)
 	else
 		pr_info(GHES_PFX "Failed to enable APEI firmware first mode.\n");
 
+	rc = ghes_gdata_pool_init();
+	if (rc) {
+		pr_warn(GHES_PFX "ghes_gdata_pool_init failed\n");
+		goto err;
+	}
+
 	return 0;
 err:
 	return rc;
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cddb4ac8..a3dd82069069 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,34 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_notifier(struct notifier_block *nb);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb);
+#else
+static inline int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
2.17.1

Re: [v7 PATCH 1/6] ACPI / APEI: Add support to queuing up the non-fatal HW errors and notify

[kbuild test robot]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v5.6]
[also build test WARNING on next-20200408]
[cannot apply to pm/linux-next pci/next linus/master linux/master]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-non-fatal-HW-errors/20200408-014447
base:    7111951b8d4973bda27ff663f2cf18b663d15b48
reproduce:
        # apt-get install sparse
        # sparse version: v0.6.1-188-g79f7ac98-dirty
        make ARCH=x86_64 allmodconfig
        make C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__'

If you fix the issue, kindly add following tag as appropriate
Reported-by: kbuild test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)

>> drivers/acpi/apei/ghes.c:210:5: sparse: sparse: symbol 'ghes_gdata_pool_init' was not declared. Should it be static?
   drivers/acpi/apei/ghes.c:715:25: sparse: sparse: incompatible types in comparison expression (different address spaces):
   drivers/acpi/apei/ghes.c:715:25: sparse:    struct ghes_estatus_cache [noderef] <asn:4> *
   drivers/acpi/apei/ghes.c:715:25: sparse:    struct ghes_estatus_cache *
   drivers/acpi/apei/ghes.c:795:25: sparse: sparse: incompatible types in comparison expression (different address spaces):
   drivers/acpi/apei/ghes.c:795:25: sparse:    struct ghes_estatus_cache [noderef] <asn:4> *
   drivers/acpi/apei/ghes.c:795:25: sparse:    struct ghes_estatus_cache *

Please review and possibly fold the followup patch.

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [v7 PATCH 1/6] ACPI / APEI: Add support to queuing up the non-fatal HW errors and notify

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 1889 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v5.6]
[also build test WARNING on next-20200408]
[cannot apply to pm/linux-next pci/next linus/master linux/master]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-non-fatal-HW-errors/20200408-014447
base:    7111951b8d4973bda27ff663f2cf18b663d15b48
reproduce:
        # apt-get install sparse
        # sparse version: v0.6.1-188-g79f7ac98-dirty
        make ARCH=x86_64 allmodconfig
        make C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__'

If you fix the issue, kindly add following tag as appropriate
Reported-by: kbuild test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)

>> drivers/acpi/apei/ghes.c:210:5: sparse: sparse: symbol 'ghes_gdata_pool_init' was not declared. Should it be static?
   drivers/acpi/apei/ghes.c:715:25: sparse: sparse: incompatible types in comparison expression (different address spaces):
   drivers/acpi/apei/ghes.c:715:25: sparse:    struct ghes_estatus_cache [noderef] <asn:4> *
   drivers/acpi/apei/ghes.c:715:25: sparse:    struct ghes_estatus_cache *
   drivers/acpi/apei/ghes.c:795:25: sparse: sparse: incompatible types in comparison expression (different address spaces):
   drivers/acpi/apei/ghes.c:795:25: sparse:    struct ghes_estatus_cache [noderef] <asn:4> *
   drivers/acpi/apei/ghes.c:795:25: sparse:    struct ghes_estatus_cache *

Please review and possibly fold the followup patch.

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

[RFC PATCH] ACPI / APEI: ghes_gdata_pool_init() can be static

[kbuild test robot]

Signed-off-by: kbuild test robot <lkp@intel.com>
---
 ghes.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 5c0ab54223118..6d698867468d1 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -207,7 +207,7 @@ int ghes_estatus_pool_init(int num_ghes)
 	return -ENOMEM;
 }
 
-int ghes_gdata_pool_init(void)
+static int ghes_gdata_pool_init(void)
 {
 	unsigned long addr, len;
 	int rc;

[RFC PATCH] ACPI / APEI: ghes_gdata_pool_init() can be static

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 501 bytes --]


Signed-off-by: kbuild test robot <lkp@intel.com>
---
 ghes.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 5c0ab54223118..6d698867468d1 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -207,7 +207,7 @@ int ghes_estatus_pool_init(int num_ghes)
 	return -ENOMEM;
 }
 
-int ghes_gdata_pool_init(void)
+static int ghes_gdata_pool_init(void)
 {
 	unsigned long addr, len;
 	int rc;

[v7 PATCH 2/6] ACPI / APEI: Add callback for memory errors to the GHES notifier

[Shiju Jose]

Add callback function for handling the memory errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 55 ++++++++++++++++++++++++++++++----------
 1 file changed, 42 insertions(+), 13 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 5c0ab5422311..053c4a2ed96c 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -471,23 +471,33 @@ static void ghes_clear_estatus(struct ghes *ghes,
 		ghes_ack_error(ghes->generic_v2);
 }
 
-static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int sev)
+static int ghes_handle_memory_failure(struct notifier_block *nb,
+				      unsigned long event, void *data)
 {
 #ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
 	unsigned long pfn;
 	int flags = -1;
+	int sev = event;
+	struct acpi_hest_generic_data *gdata = data;
 	int sec_sev = ghes_severity(gdata->error_severity);
 	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PLATFORM_MEM))
+		return NOTIFY_DONE;
+
+	ghes_edac_report_mem_error(sev, mem_err);
+
+	arch_apei_report_mem_error(sev, mem_err);
+
 	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return;
+		return NOTIFY_STOP;
 
 	pfn = mem_err->physical_addr >> PAGE_SHIFT;
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		mem_err->physical_addr);
-		return;
+		return NOTIFY_STOP;
 	}
 
 	/* iff following two events can be handled properly by now */
@@ -500,6 +510,7 @@ static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int
 	if (flags != -1)
 		memory_failure_queue(pfn, flags);
 #endif
+	return NOTIFY_STOP;
 }
 
 /*
@@ -547,6 +558,22 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+static struct notifier_block ghes_notifier_mem_error = {
+	.notifier_call = ghes_handle_memory_failure,
+};
+
+struct ghes_error_handler_list {
+	const char *name;
+	struct notifier_block *nb;
+};
+
+static const struct ghes_error_handler_list ghes_error_handler_list[] = {
+	{
+		.name = "ghes_notifier_mem_error",
+		.nb = &ghes_notifier_mem_error,
+	},
+};
+
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
 
 /**
@@ -630,15 +657,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			break;
 		}
 
-		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
-			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
-
-			ghes_edac_report_mem_error(sev, mem_err);
-
-			arch_apei_report_mem_error(sev, mem_err);
-			ghes_handle_memory_failure(gdata, sev);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
+		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
@@ -1431,7 +1450,7 @@ static struct platform_driver ghes_platform_driver = {
 
 static int __init ghes_init(void)
 {
-	int rc;
+	int rc, i;
 
 	if (acpi_disabled)
 		return -ENODEV;
@@ -1473,6 +1492,16 @@ static int __init ghes_init(void)
 		goto err;
 	}
 
+	for (i = 0; i < ARRAY_SIZE(ghes_error_handler_list); i++) {
+		const struct ghes_error_handler_list *list =
+						&ghes_error_handler_list[i];
+		rc = ghes_register_event_notifier(list->nb);
+		if (rc) {
+			pr_warn(GHES_PFX "fail to register %s\n", list->name);
+			goto err;
+		}
+	}
+
 	return 0;
 err:
 	return rc;
-- 
2.17.1

[v7 PATCH 3/6] ACPI / APEI: Add callback for AER to the GHES notifier

[Shiju Jose]

Add callback function for handling the AER to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 053c4a2ed96c..67ef1742fc93 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -527,11 +527,16 @@ static int ghes_handle_memory_failure(struct notifier_block *nb,
  * GHES_SEV_PANIC does not make it to this handling since the kernel must
  *     panic.
  */
-static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
+static int ghes_handle_aer(struct notifier_block *nb, unsigned long event,
+			   void *data)
 {
 #ifdef CONFIG_ACPI_APEI_PCIEAER
+	struct acpi_hest_generic_data *gdata = data;
 	struct cper_sec_pcie *pcie_err = acpi_hest_get_payload(gdata);
 
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PCIE))
+		return NOTIFY_DONE;
+
 	if (pcie_err->validation_bits & CPER_PCIE_VALID_DEVICE_ID &&
 	    pcie_err->validation_bits & CPER_PCIE_VALID_AER_INFO) {
 		unsigned int devfn;
@@ -556,12 +561,17 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 				  pcie_err->aer_info);
 	}
 #endif
+	return NOTIFY_STOP;
 }
 
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
 
+static struct notifier_block ghes_notifier_aer = {
+	.notifier_call = ghes_handle_aer,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -572,6 +582,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_mem_error",
 		.nb = &ghes_notifier_mem_error,
 	},
+	{
+		.name = "ghes_notifier_aer",
+		.nb = &ghes_notifier_aer,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -657,10 +671,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			break;
 		}
 
-		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
-			ghes_handle_aer(gdata);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
+		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);
-- 
2.17.1

[v7 PATCH 4/6] ACPI / APEI: Add callback for ARM HW errors to the GHES notifier

[Shiju Jose]

Add callback function for handling the ARM HW errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 29 +++++++++++++++++++++++------
 1 file changed, 23 insertions(+), 6 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 67ef1742fc93..3b89c7621a0d 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -564,6 +564,20 @@ static int ghes_handle_aer(struct notifier_block *nb, unsigned long event,
 	return NOTIFY_STOP;
 }
 
+static int ghes_handle_arm_hw_error(struct notifier_block *nb,
+				    unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PROC_ARM))
+		return NOTIFY_DONE;
+
+	log_arm_hw_error(err);
+
+	return NOTIFY_STOP;
+}
+
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
@@ -572,6 +586,10 @@ static struct notifier_block ghes_notifier_aer = {
 	.notifier_call = ghes_handle_aer,
 };
 
+static struct notifier_block ghes_notifier_arm_hw_error = {
+	.notifier_call = ghes_handle_arm_hw_error,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -586,6 +604,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_aer",
 		.nb = &ghes_notifier_aer,
 	},
+	{
+		.name = "ghes_notifier_arm_hw_error",
+		.nb = &ghes_notifier_arm_hw_error,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -670,12 +692,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			pr_warn(GHES_PFX "ghes event queue full\n");
 			break;
 		}
-
-		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);
-		} else {
+		{
 			void *err = acpi_hest_get_payload(gdata);
 
 			log_non_standard_event(sec_type, fru_id, fru_text,
-- 
2.17.1

[v7 PATCH 5/6] ACPI / APEI: Add callback for non-standard HW errors to the GHES notifier

[Shiju Jose]

Add callback function for the non-standard HW errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 56 ++++++++++++++++++++++++++--------------
 1 file changed, 37 insertions(+), 19 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 3b89c7621a0d..0c27ea8ea943 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -578,6 +578,34 @@ static int ghes_handle_arm_hw_error(struct notifier_block *nb,
 	return NOTIFY_STOP;
 }
 
+static int ghes_handle_non_standard_event(struct notifier_block *nb,
+					  unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	void *err = acpi_hest_get_payload(gdata);
+	int sec_sev;
+	guid_t *sec_type;
+	const guid_t *fru_id = &guid_null;
+	char *fru_text = "";
+
+	if (guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PLATFORM_MEM) ||
+	    guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PCIE) ||
+	    guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PROC_ARM))
+		return NOTIFY_DONE;
+
+	sec_type = (guid_t *)gdata->section_type;
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
+		fru_id = (guid_t *)gdata->fru_id;
+
+	if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
+		fru_text = gdata->fru_text;
+
+	log_non_standard_event(sec_type, fru_id, fru_text, sec_sev,
+			       err, gdata->error_data_length);
+	return NOTIFY_OK;
+}
+
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
@@ -590,6 +618,10 @@ static struct notifier_block ghes_notifier_arm_hw_error = {
 	.notifier_call = ghes_handle_arm_hw_error,
 };
 
+static struct notifier_block ghes_notifier_non_standard_event = {
+	.notifier_call = ghes_handle_non_standard_event,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -608,6 +640,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_arm_hw_error",
 		.nb = &ghes_notifier_arm_hw_error,
 	},
+	{
+		.name = "ghes_notifier_non_standard_event",
+		.nb = &ghes_notifier_non_standard_event,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -656,24 +692,13 @@ static DECLARE_WORK(ghes_event_work, ghes_event_work_func);
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
-	int sev, sec_sev;
+	int sev;
 	struct acpi_hest_generic_data *gdata;
-	guid_t *sec_type;
-	const guid_t *fru_id = &guid_null;
-	char *fru_text = "";
 	struct ghes_event_entry event_entry;
 	u32 len;
 
 	sev = ghes_severity(estatus->error_severity);
 	apei_estatus_for_each_section(estatus, gdata) {
-		sec_type = (guid_t *)gdata->section_type;
-		sec_sev = ghes_severity(gdata->error_severity);
-		if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
-			fru_id = (guid_t *)gdata->fru_id;
-
-		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
-			fru_text = gdata->fru_text;
-
 		len = acpi_hest_get_record_size(gdata);
 		event_entry.gdata = (void *)gen_pool_alloc(ghes_gdata_pool,
 							   len);
@@ -692,13 +717,6 @@ static void ghes_do_proc(struct ghes *ghes,
 			pr_warn(GHES_PFX "ghes event queue full\n");
 			break;
 		}
-		{
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
-		}
 	}
 }
 
-- 
2.17.1

[v7 PATCH 6/6] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The HiSilicon HIP PCIe controller is capable of handling errors
on root port and perform port reset separately at each root port.

Add error handling driver for HIP PCIe controller to log
and report recoverable errors. Perform root port reset and restore
link status after the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

The driver placed in the drivers/pci/controller/ because the
HIP PCIe controller does not use DWC ip.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
drivers/pci/controller/Kconfig           |   8 +
drivers/pci/controller/Makefile          |   1 +
drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
3 files changed, 345 insertions(+)
create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 323 +++++++++++++++++++++++
 3 files changed, 332 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index 20bf00f587bd..8bc6111480c8 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -268,6 +268,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on ACPI
+	bool "HiSilicon HIP PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon HIP SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 01b2502a5323..94f37b3d9929 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -29,6 +29,7 @@ obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
 obj-$(CONFIG_PCIE_BRCMSTB) += pcie-brcmstb.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 000000000000..cc721070e07b
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,323 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller errors on
+ * HiSilicon HIP SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
+			0xA8, 0x67, 0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_error_data {
+	u64	val_bits;
+	u8	version;
+	u8	soc_id;
+	u8	socket_id;
+	u8	nimbus_id;
+	u8	sub_module_id;
+	u8	core_id;
+	u8	port_id;
+	u8	err_severity;
+	u16	err_type;
+	u8	reserv[2];
+	u32	err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_error_private {
+	struct notifier_block	nb;
+	struct platform_device	*pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_error_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+				u32 chip_id, u32 port_id)
+{
+	struct device *dev = &pdev->dev;
+	acpi_handle handle = ACPI_HANDLE(dev);
+	union acpi_object arg[3];
+	struct acpi_object_list arg_list;
+	acpi_status s;
+	unsigned long long data = 0;
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No RST method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	struct device *device = &dev->dev;
+	acpi_handle root_handle = ACPI_HANDLE(device);
+	struct acpi_pci_root *pci_root;
+	struct pci_bus *root_bus;
+	struct pci_dev *pdev;
+	u32 domain, busnr, devfn;
+
+	s = acpi_get_parent(root_handle, &root_handle);
+	if (ACPI_FAILURE(s))
+		return -ENODEV;
+	pci_root = acpi_pci_find_root(root_handle);
+	if (!pci_root)
+		return -ENODEV;
+	root_bus = pci_root->bus;
+	domain = pci_root->segment;
+
+	busnr = root_bus->number;
+	devfn = PCI_DEVFN(port_id, 0);
+	pdev = pci_get_domain_bus_and_slot(domain, busnr, devfn);
+	if (!pdev) {
+		dev_info(device, "Fail to get root port %04x:%02x:%02x.%d device\n",
+			 domain, busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
+		return -ENODEV;
+	}
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/*
+	 * The initialization time of subordinate devices after
+	 * hot reset is no more than 1s, which is required by
+	 * the PCI spec v5.0 sec 6.6.1. The time will shorten
+	 * if Readiness Notifications mechanisms are used. But
+	 * wait 1s here to adapt any conditions.
+	 */
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void hisi_pcie_handle_error(const struct hisi_pcie_error_data *error,
+				   struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (error->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", error->version);
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", error->soc_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", error->socket_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", error->nimbus_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(error->sub_module_id));
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", error->core_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", error->port_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_error_severity(error->err_severity));
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", error->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (error->val_bits &
+				BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, error->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller errors */
+	if (error->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, error->socket_id,
+			HISI_PCIE_PORT_ID(error->core_id, error->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static int hisi_pcie_notify_error(struct notifier_block *nb,
+				  unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	const struct hisi_pcie_error_data *error_data =
+				acpi_hest_get_payload(gdata);
+	struct hisi_pcie_error_private *priv =
+			container_of(nb, struct hisi_pcie_error_private, nb);
+	struct platform_device *pdev = priv->pdev;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return NOTIFY_DONE;
+
+	if (!guid_equal((guid_t *)gdata->section_type, &hisi_pcie_sec_type) ||
+	    error_data->socket_id != socket)
+		return NOTIFY_DONE;
+
+	hisi_pcie_handle_error(error_data, pdev);
+
+	return NOTIFY_OK;
+}
+
+static int hisi_pcie_error_handler_probe(struct platform_device *pdev)
+{
+	struct hisi_pcie_error_private *priv;
+	int ret;
+
+	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->nb.notifier_call = hisi_pcie_notify_error;
+	priv->pdev = pdev;
+	ret = ghes_register_event_notifier(&priv->nb);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_notifier fail\n",
+			__func__);
+		return ret;
+	}
+
+	platform_set_drvdata(pdev, priv);
+
+	return 0;
+}
+
+static int hisi_pcie_error_handler_remove(struct platform_device *pdev)
+{
+	struct hisi_pcie_error_private *priv = platform_get_drvdata(pdev);
+
+	if (priv)
+		ghes_unregister_event_notifier(&priv->nb);
+
+	kfree(priv);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_error_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-error-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_error_handler_probe,
+	.remove		= hisi_pcie_error_handler_remove,
+};
+module_platform_driver(hisi_pcie_error_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP PCIe controller error handling driver");
+MODULE_LICENSE("GPL v2");
-- 
2.17.1

Re: [v7 PATCH 6/6] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 4017 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v5.6]
[also build test WARNING on next-20200407]
[cannot apply to pm/linux-next pci/next linus/master linux/master]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-non-fatal-HW-errors/20200408-014447
base:    7111951b8d4973bda27ff663f2cf18b663d15b48
config: i386-allyesconfig (attached as .config)
compiler: gcc-7 (Ubuntu 7.5.0-6ubuntu2) 7.5.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kbuild test robot <lkp@intel.com>

All warnings (new ones prefixed by >>):

   drivers/pci/controller/pcie-hisi-error.c: In function 'hisi_pcie_handle_error':
>> drivers/pci/controller/pcie-hisi-error.c:246:1: warning: the frame size of 1124 bytes is larger than 1024 bytes [-Wframe-larger-than=]
    }
    ^

vim +246 drivers/pci/controller/pcie-hisi-error.c

   181	
   182	static void hisi_pcie_handle_error(const struct hisi_pcie_error_data *error,
   183					   struct platform_device *pdev)
   184	{
   185		char buf[HISI_PCIE_ERR_INFO_SIZE];
   186		char *p = buf, *end = buf + sizeof(buf);
   187		struct device *dev = &pdev->dev;
   188		u32 i;
   189		int rc;
   190	
   191		if (error->val_bits == 0) {
   192			dev_warn(dev, "%s: no valid error information\n", __func__);
   193			return;
   194		}
   195	
   196		/* Logging */
   197		p += snprintf(p, end - p, "[ Table version=%d ", error->version);
   198		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
   199			p += snprintf(p, end - p, "SOC ID=%d ", error->soc_id);
   200	
   201		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
   202			p += snprintf(p, end - p, "socket ID=%d ", error->socket_id);
   203	
   204		if (error->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
   205			p += snprintf(p, end - p, "nimbus ID=%d ", error->nimbus_id);
   206	
   207		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
   208			p += snprintf(p, end - p, "sub module=%s ",
   209				      hisi_pcie_sub_module_name(error->sub_module_id));
   210	
   211		if (error->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
   212			p += snprintf(p, end - p, "core ID=core%d ", error->core_id);
   213	
   214		if (error->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
   215			p += snprintf(p, end - p, "port ID=port%d ", error->port_id);
   216	
   217		if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
   218			p += snprintf(p, end - p, "error severity=%s ",
   219				      hisi_pcie_error_severity(error->err_severity));
   220	
   221		if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
   222			p += snprintf(p, end - p, "error type=0x%x ", error->err_type);
   223	
   224		p += snprintf(p, end - p, "]\n");
   225		dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
   226		dev_info(dev, "%s\n", buf);
   227	
   228		dev_info(dev, "Reg Dump:\n");
   229		for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
   230			if (error->val_bits &
   231					BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
   232				dev_info(dev,
   233					 "ERR_MISC_%d=0x%x\n", i, error->err_misc[i]);
   234		}
   235	
   236		/* Recovery for the PCIe controller errors */
   237		if (error->err_severity == HISI_ERR_SEV_RECOVERABLE) {
   238			/* try reset PCI port for the error recovery */
   239			rc = hisi_pcie_port_do_recovery(pdev, error->socket_id,
   240				HISI_PCIE_PORT_ID(error->core_id, error->port_id));
   241			if (rc) {
   242				dev_info(dev, "fail to do hisi pcie port reset\n");
   243				return;
   244			}
   245		}
 > 246	}
   247	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 71594 bytes --]

Re: [v7 PATCH 6/6] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 4121 bytes --]

Hi Shiju,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v5.6]
[also build test WARNING on next-20200407]
[cannot apply to pm/linux-next pci/next linus/master linux/master]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Shiju-Jose/ACPI-APEI-Add-support-to-notify-non-fatal-HW-errors/20200408-014447
base:    7111951b8d4973bda27ff663f2cf18b663d15b48
config: i386-allyesconfig (attached as .config)
compiler: gcc-7 (Ubuntu 7.5.0-6ubuntu2) 7.5.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kbuild test robot <lkp@intel.com>

All warnings (new ones prefixed by >>):

   drivers/pci/controller/pcie-hisi-error.c: In function 'hisi_pcie_handle_error':
>> drivers/pci/controller/pcie-hisi-error.c:246:1: warning: the frame size of 1124 bytes is larger than 1024 bytes [-Wframe-larger-than=]
    }
    ^

vim +246 drivers/pci/controller/pcie-hisi-error.c

   181	
   182	static void hisi_pcie_handle_error(const struct hisi_pcie_error_data *error,
   183					   struct platform_device *pdev)
   184	{
   185		char buf[HISI_PCIE_ERR_INFO_SIZE];
   186		char *p = buf, *end = buf + sizeof(buf);
   187		struct device *dev = &pdev->dev;
   188		u32 i;
   189		int rc;
   190	
   191		if (error->val_bits == 0) {
   192			dev_warn(dev, "%s: no valid error information\n", __func__);
   193			return;
   194		}
   195	
   196		/* Logging */
   197		p += snprintf(p, end - p, "[ Table version=%d ", error->version);
   198		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
   199			p += snprintf(p, end - p, "SOC ID=%d ", error->soc_id);
   200	
   201		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
   202			p += snprintf(p, end - p, "socket ID=%d ", error->socket_id);
   203	
   204		if (error->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
   205			p += snprintf(p, end - p, "nimbus ID=%d ", error->nimbus_id);
   206	
   207		if (error->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
   208			p += snprintf(p, end - p, "sub module=%s ",
   209				      hisi_pcie_sub_module_name(error->sub_module_id));
   210	
   211		if (error->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
   212			p += snprintf(p, end - p, "core ID=core%d ", error->core_id);
   213	
   214		if (error->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
   215			p += snprintf(p, end - p, "port ID=port%d ", error->port_id);
   216	
   217		if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
   218			p += snprintf(p, end - p, "error severity=%s ",
   219				      hisi_pcie_error_severity(error->err_severity));
   220	
   221		if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
   222			p += snprintf(p, end - p, "error type=0x%x ", error->err_type);
   223	
   224		p += snprintf(p, end - p, "]\n");
   225		dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
   226		dev_info(dev, "%s\n", buf);
   227	
   228		dev_info(dev, "Reg Dump:\n");
   229		for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
   230			if (error->val_bits &
   231					BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
   232				dev_info(dev,
   233					 "ERR_MISC_%d=0x%x\n", i, error->err_misc[i]);
   234		}
   235	
   236		/* Recovery for the PCIe controller errors */
   237		if (error->err_severity == HISI_ERR_SEV_RECOVERABLE) {
   238			/* try reset PCI port for the error recovery */
   239			rc = hisi_pcie_port_do_recovery(pdev, error->socket_id,
   240				HISI_PCIE_PORT_ID(error->core_id, error->port_id));
   241			if (rc) {
   242				dev_info(dev, "fail to do hisi pcie port reset\n");
   243				return;
   244			}
   245		}
 > 246	}
   247	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

[-- Attachment #2: config.gz --]
[-- Type: application/gzip, Size: 71594 bytes --]

[RESEND PATCH v7 0/6] ACPI / APEI: Add support to notify non-fatal HW errors

[Shiju Jose]

Add common interface for queuing up the non-fatal HW errors and notify
the registered kernel drivers.The interface supports drivers to register
to receive the callback for the non-fatal HW errors, including the vendor
specific HW errors, for the recovery and supports handling the non-fatal
errors in the process context.

Patch set
1. add the new interface to the APEI driver for the non-fatal HW
   error notification.
2. change the existing error handling for the standard errors
   to use the above notification interface.
2. add driver to handle HiSilicon hip PCIe controller's errors.

Changes:

V7:
1. Add changes in the APEI driver suggested by Borislav Petkov, for
   queuing up all the non-fatal HW errors to the work queue and
   notify the registered kernel drivers from the bottom half using
   blocking notifier, common interface for both standard and
   vendor-spcific errors.
2. Fix for further feedbacks in v5 HIP PCIe error handler driver
   by Bjorn Helgaas.

V6:
1. Fix few changes in the patch subject line suggested by Bjorn Helgaas.

V5:
1. Fix comments from James Morse.
1.1 Changed the notification method to use the atomic_notifier_chain.
1.2 Add the error handled status for the user space.  

V4:
1. Fix for the following smatch warning in the PCIe error driver,
   reported by kbuild test robot<lkp@intel.com>:
   warn: should '((((1))) << (9 + i))' be a 64 bit type?
   if (err->val_bits & BIT(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
	^^^ This should be BIT_ULL() because it goes up to 9 + 32.

V3:
1. Fix the comments from Bjorn Helgaas.

V2:
1. Changes in the HiSilicon PCIe controller's error handling driver
   for the comments from Bjorn Helgaas.
   
2. Changes in the APEI interface to support reporting the vendor error
   for module with multiple devices, but use the same section type.
   In the error handler will use socket id/sub module id etc to distinguish
   the device.

V1:  
1. Fix comments from James Morse.

2. add driver to handle HiSilicon hip08 PCIe controller's errors,
   which is an application of the above interface.

Shiju Jose (5):
  ACPI / APEI: Add support to queuing up the non-fatal HW errors and
    notify
  ACPI / APEI: Add callback for memory errors to the GHES notifier
  ACPI / APEI: Add callback for AER to the GHES notifier
  ACPI / APEI: Add callback for ARM HW errors to the GHES notifier
  ACPI / APEI: Add callback for non-standard HW errors to the GHES
    notifier

Yicong Yang (1):
  PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

 drivers/acpi/apei/ghes.c                 | 268 ++++++++++++++++---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 323 +++++++++++++++++++++++
 include/acpi/ghes.h                      |  28 ++
 5 files changed, 595 insertions(+), 33 deletions(-)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

-- 
2.17.1

[RESEND PATCH v7 1/6] ACPI / APEI: Add support to queuing up the non-fatal HW errors and notify

[Shiju Jose]

Add queuing up the non-fatal HW errors to the work queue and interface to
notify the registered kernel drivers for the error recovery.
The interface enables drivers to register the callback functions to
receive the error notification, including the vendor-specific HW errors
and supports handling the non-fatal HW errors in the process context.

Suggested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 127 +++++++++++++++++++++++++++++++++++++++
 include/acpi/ghes.h      |  28 +++++++++
 2 files changed, 155 insertions(+)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 24c9642e8fc7..5c0ab5422311 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -33,6 +33,7 @@
 #include <linux/irq_work.h>
 #include <linux/llist.h>
 #include <linux/genalloc.h>
+#include <linux/kfifo.h>
 #include <linux/pci.h>
 #include <linux/pfn.h>
 #include <linux/aer.h>
@@ -63,6 +64,11 @@
 #define GHES_ESTATUS_CACHES_SIZE	4
 
 #define GHES_ESTATUS_IN_CACHE_MAX_NSEC	10000000000ULL
+
+#define GHES_EVENT_RING_SIZE	256
+#define GHES_GDATA_POOL_MIN_ALLOC_ORDER	3
+#define GHES_GDATA_POOL_MIN_SIZE	65536
+
 /* Prevent too many caches are allocated because of RCU */
 #define GHES_ESTATUS_CACHE_ALLOCED_MAX	(GHES_ESTATUS_CACHES_SIZE * 3 / 2)
 
@@ -122,6 +128,19 @@ static DEFINE_MUTEX(ghes_list_mutex);
  */
 static DEFINE_SPINLOCK(ghes_notify_lock_irq);
 
+struct ghes_event_entry {
+	struct acpi_hest_generic_data *gdata;
+	int error_severity;
+};
+
+static DEFINE_KFIFO(ghes_event_ring, struct ghes_event_entry,
+		    GHES_EVENT_RING_SIZE);
+
+static DEFINE_SPINLOCK(ghes_event_ring_lock);
+
+static struct gen_pool *ghes_gdata_pool;
+static unsigned long ghes_gdata_pool_size_request;
+
 static struct gen_pool *ghes_estatus_pool;
 static unsigned long ghes_estatus_pool_size_request;
 
@@ -188,6 +207,40 @@ int ghes_estatus_pool_init(int num_ghes)
 	return -ENOMEM;
 }
 
+int ghes_gdata_pool_init(void)
+{
+	unsigned long addr, len;
+	int rc;
+
+	ghes_gdata_pool = gen_pool_create(GHES_GDATA_POOL_MIN_ALLOC_ORDER, -1);
+	if (!ghes_gdata_pool)
+		return -ENOMEM;
+
+	if (ghes_gdata_pool_size_request < GHES_GDATA_POOL_MIN_SIZE)
+		ghes_gdata_pool_size_request = GHES_GDATA_POOL_MIN_SIZE;
+
+	len = ghes_gdata_pool_size_request;
+	addr = (unsigned long)vmalloc(PAGE_ALIGN(len));
+	if (!addr)
+		goto err_pool_alloc;
+
+	vmalloc_sync_mappings();
+
+	rc = gen_pool_add(ghes_gdata_pool, addr, PAGE_ALIGN(len), -1);
+	if (rc)
+		goto err_pool_add;
+
+	return 0;
+
+err_pool_add:
+	vfree((void *)addr);
+
+err_pool_alloc:
+	gen_pool_destroy(ghes_gdata_pool);
+
+	return -ENOMEM;
+}
+
 static int map_gen_v2(struct ghes *ghes)
 {
 	return apei_map_generic_address(&ghes->generic_v2->read_ack_register);
@@ -247,6 +300,10 @@ static struct ghes *ghes_new(struct acpi_hest_generic *generic)
 		goto err_unmap_status_addr;
 	}
 
+	ghes_gdata_pool_size_request += generic->records_to_preallocate *
+					generic->max_sections_per_record *
+					generic->max_raw_data_length;
+
 	return ghes;
 
 err_unmap_status_addr:
@@ -490,6 +547,49 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
+
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ *
+ * return 0 : SUCCESS, non-zero : FAIL
+ */
+int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return blocking_notifier_chain_register(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_register_event_notifier);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event handler.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+	blocking_notifier_chain_unregister(&ghes_event_notify_list, nb);
+}
+EXPORT_SYMBOL_GPL(ghes_unregister_event_notifier);
+
+static void ghes_event_work_func(struct work_struct *work)
+{
+	struct ghes_event_entry entry;
+	u32 len;
+
+	while (kfifo_get(&ghes_event_ring, &entry)) {
+		blocking_notifier_call_chain(&ghes_event_notify_list,
+					     entry.error_severity,
+					     entry.gdata);
+		len = acpi_hest_get_size(entry.gdata) +
+				acpi_hest_get_error_length(entry.gdata);
+		gen_pool_free(ghes_gdata_pool, (unsigned long)entry.gdata, len);
+	}
+}
+
+static DECLARE_WORK(ghes_event_work, ghes_event_work_func);
+
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
@@ -498,6 +598,8 @@ static void ghes_do_proc(struct ghes *ghes,
 	guid_t *sec_type;
 	const guid_t *fru_id = &guid_null;
 	char *fru_text = "";
+	struct ghes_event_entry event_entry;
+	u32 len;
 
 	sev = ghes_severity(estatus->error_severity);
 	apei_estatus_for_each_section(estatus, gdata) {
@@ -509,6 +611,25 @@ static void ghes_do_proc(struct ghes *ghes,
 		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
 			fru_text = gdata->fru_text;
 
+		len = acpi_hest_get_record_size(gdata);
+		event_entry.gdata = (void *)gen_pool_alloc(ghes_gdata_pool,
+							   len);
+		if (!event_entry.gdata) {
+			pr_warn(GHES_PFX "ghes gdata pool alloc fail\n");
+			break;
+		}
+
+		memcpy(event_entry.gdata, gdata, len);
+		event_entry.error_severity = sev;
+
+		if (kfifo_in_spinlocked(&ghes_event_ring, &event_entry, 1,
+					&ghes_event_ring_lock))
+			schedule_work(&ghes_event_work);
+		else {
+			pr_warn(GHES_PFX "ghes event queue full\n");
+			break;
+		}
+
 		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
 			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
@@ -1346,6 +1467,12 @@ static int __init ghes_init(void)
 	else
 		pr_info(GHES_PFX "Failed to enable APEI firmware first mode.\n");
 
+	rc = ghes_gdata_pool_init();
+	if (rc) {
+		pr_warn(GHES_PFX "ghes_gdata_pool_init failed\n");
+		goto err;
+	}
+
 	return 0;
 err:
 	return rc;
diff --git a/include/acpi/ghes.h b/include/acpi/ghes.h
index e3f1cddb4ac8..a3dd82069069 100644
--- a/include/acpi/ghes.h
+++ b/include/acpi/ghes.h
@@ -50,6 +50,34 @@ enum {
 	GHES_SEV_PANIC = 0x3,
 };
 
+
+#ifdef CONFIG_ACPI_APEI_GHES
+/**
+ * ghes_register_event_notifier - register an event notifier
+ * for the non-fatal HW errors.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ *
+ * Return : 0 - SUCCESS, non-zero - FAIL.
+ */
+int ghes_register_event_notifier(struct notifier_block *nb);
+
+/**
+ * ghes_unregister_event_notifier - unregister the previously
+ * registered event notifier.
+ * @nb: pointer to the notifier_block structure of the event notifier.
+ */
+void ghes_unregister_event_notifier(struct notifier_block *nb);
+#else
+static inline int ghes_register_event_notifier(struct notifier_block *nb)
+{
+	return -ENODEV;
+}
+
+static inline void ghes_unregister_event_notifier(struct notifier_block *nb)
+{
+}
+#endif
+
 int ghes_estatus_pool_init(int num_ghes);
 
 /* From drivers/edac/ghes_edac.c */
-- 
2.17.1

Re: [RESEND PATCH v7 1/6] ACPI / APEI: Add support to queuing up the non-fatal HW errors and notify

[Dan Carpenter]

On Tue, Apr 21, 2020 at 02:21:31PM +0100, Shiju Jose wrote:
> @@ -1346,6 +1467,12 @@ static int __init ghes_init(void)
>  	else
>  		pr_info(GHES_PFX "Failed to enable APEI firmware first mode.\n");
>  
> +	rc = ghes_gdata_pool_init();
> +	if (rc) {
> +		pr_warn(GHES_PFX "ghes_gdata_pool_init failed\n");
> +		goto err;

There probably should be some clean up on this error path.

> +	}
> +
>  	return 0;
>  err:
>  	return rc;

regards,
dan carpenter

[RESEND PATCH v7 2/6] ACPI / APEI: Add callback for memory errors to the GHES notifier

[Shiju Jose]

Add callback function for handling the memory errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 55 ++++++++++++++++++++++++++++++----------
 1 file changed, 42 insertions(+), 13 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 5c0ab5422311..053c4a2ed96c 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -471,23 +471,33 @@ static void ghes_clear_estatus(struct ghes *ghes,
 		ghes_ack_error(ghes->generic_v2);
 }
 
-static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int sev)
+static int ghes_handle_memory_failure(struct notifier_block *nb,
+				      unsigned long event, void *data)
 {
 #ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
 	unsigned long pfn;
 	int flags = -1;
+	int sev = event;
+	struct acpi_hest_generic_data *gdata = data;
 	int sec_sev = ghes_severity(gdata->error_severity);
 	struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
 
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PLATFORM_MEM))
+		return NOTIFY_DONE;
+
+	ghes_edac_report_mem_error(sev, mem_err);
+
+	arch_apei_report_mem_error(sev, mem_err);
+
 	if (!(mem_err->validation_bits & CPER_MEM_VALID_PA))
-		return;
+		return NOTIFY_STOP;
 
 	pfn = mem_err->physical_addr >> PAGE_SHIFT;
 	if (!pfn_valid(pfn)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		mem_err->physical_addr);
-		return;
+		return NOTIFY_STOP;
 	}
 
 	/* iff following two events can be handled properly by now */
@@ -500,6 +510,7 @@ static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int
 	if (flags != -1)
 		memory_failure_queue(pfn, flags);
 #endif
+	return NOTIFY_STOP;
 }
 
 /*
@@ -547,6 +558,22 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 #endif
 }
 
+static struct notifier_block ghes_notifier_mem_error = {
+	.notifier_call = ghes_handle_memory_failure,
+};
+
+struct ghes_error_handler_list {
+	const char *name;
+	struct notifier_block *nb;
+};
+
+static const struct ghes_error_handler_list ghes_error_handler_list[] = {
+	{
+		.name = "ghes_notifier_mem_error",
+		.nb = &ghes_notifier_mem_error,
+	},
+};
+
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
 
 /**
@@ -630,15 +657,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			break;
 		}
 
-		if (guid_equal(sec_type, &CPER_SEC_PLATFORM_MEM)) {
-			struct cper_sec_mem_err *mem_err = acpi_hest_get_payload(gdata);
-
-			ghes_edac_report_mem_error(sev, mem_err);
-
-			arch_apei_report_mem_error(sev, mem_err);
-			ghes_handle_memory_failure(gdata, sev);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
+		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
 			ghes_handle_aer(gdata);
 		}
 		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
@@ -1431,7 +1450,7 @@ static struct platform_driver ghes_platform_driver = {
 
 static int __init ghes_init(void)
 {
-	int rc;
+	int rc, i;
 
 	if (acpi_disabled)
 		return -ENODEV;
@@ -1473,6 +1492,16 @@ static int __init ghes_init(void)
 		goto err;
 	}
 
+	for (i = 0; i < ARRAY_SIZE(ghes_error_handler_list); i++) {
+		const struct ghes_error_handler_list *list =
+						&ghes_error_handler_list[i];
+		rc = ghes_register_event_notifier(list->nb);
+		if (rc) {
+			pr_warn(GHES_PFX "fail to register %s\n", list->name);
+			goto err;
+		}
+	}
+
 	return 0;
 err:
 	return rc;
-- 
2.17.1

[RESEND PATCH v7 3/6] ACPI / APEI: Add callback for AER to the GHES notifier

[Shiju Jose]

Add callback function for handling the AER to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 053c4a2ed96c..67ef1742fc93 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -527,11 +527,16 @@ static int ghes_handle_memory_failure(struct notifier_block *nb,
  * GHES_SEV_PANIC does not make it to this handling since the kernel must
  *     panic.
  */
-static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
+static int ghes_handle_aer(struct notifier_block *nb, unsigned long event,
+			   void *data)
 {
 #ifdef CONFIG_ACPI_APEI_PCIEAER
+	struct acpi_hest_generic_data *gdata = data;
 	struct cper_sec_pcie *pcie_err = acpi_hest_get_payload(gdata);
 
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PCIE))
+		return NOTIFY_DONE;
+
 	if (pcie_err->validation_bits & CPER_PCIE_VALID_DEVICE_ID &&
 	    pcie_err->validation_bits & CPER_PCIE_VALID_AER_INFO) {
 		unsigned int devfn;
@@ -556,12 +561,17 @@ static void ghes_handle_aer(struct acpi_hest_generic_data *gdata)
 				  pcie_err->aer_info);
 	}
 #endif
+	return NOTIFY_STOP;
 }
 
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
 
+static struct notifier_block ghes_notifier_aer = {
+	.notifier_call = ghes_handle_aer,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -572,6 +582,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_mem_error",
 		.nb = &ghes_notifier_mem_error,
 	},
+	{
+		.name = "ghes_notifier_aer",
+		.nb = &ghes_notifier_aer,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -657,10 +671,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			break;
 		}
 
-		if (guid_equal(sec_type, &CPER_SEC_PCIE)) {
-			ghes_handle_aer(gdata);
-		}
-		else if (guid_equal(sec_type, &CPER_SEC_PROC_ARM)) {
+		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);
-- 
2.17.1

[RESEND PATCH v7 4/6] ACPI / APEI: Add callback for ARM HW errors to the GHES notifier

[Shiju Jose]

Add callback function for handling the ARM HW errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 29 +++++++++++++++++++++++------
 1 file changed, 23 insertions(+), 6 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 67ef1742fc93..3b89c7621a0d 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -564,6 +564,20 @@ static int ghes_handle_aer(struct notifier_block *nb, unsigned long event,
 	return NOTIFY_STOP;
 }
 
+static int ghes_handle_arm_hw_error(struct notifier_block *nb,
+				    unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	struct cper_sec_proc_arm *err = acpi_hest_get_payload(gdata);
+
+	if (!guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PROC_ARM))
+		return NOTIFY_DONE;
+
+	log_arm_hw_error(err);
+
+	return NOTIFY_STOP;
+}
+
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
@@ -572,6 +586,10 @@ static struct notifier_block ghes_notifier_aer = {
 	.notifier_call = ghes_handle_aer,
 };
 
+static struct notifier_block ghes_notifier_arm_hw_error = {
+	.notifier_call = ghes_handle_arm_hw_error,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -586,6 +604,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_aer",
 		.nb = &ghes_notifier_aer,
 	},
+	{
+		.name = "ghes_notifier_arm_hw_error",
+		.nb = &ghes_notifier_arm_hw_error,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -670,12 +692,7 @@ static void ghes_do_proc(struct ghes *ghes,
 			pr_warn(GHES_PFX "ghes event queue full\n");
 			break;
 		}
-
-		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);
-		} else {
+		{
 			void *err = acpi_hest_get_payload(gdata);
 
 			log_non_standard_event(sec_type, fru_id, fru_text,
-- 
2.17.1

Re: [RESEND PATCH v7 4/6] ACPI / APEI: Add callback for ARM HW errors to the GHES notifier

[Dan Carpenter]

On Tue, Apr 21, 2020 at 02:21:34PM +0100, Shiju Jose wrote:
>  static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
> @@ -670,12 +692,7 @@ static void ghes_do_proc(struct ghes *ghes,
>  			pr_warn(GHES_PFX "ghes event queue full\n");
>  			break;
>  		}
> -
> -		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);
> -		} else {
> +		{
>  			void *err = acpi_hest_get_payload(gdata);

This is ugly.  Just move the "void *err;" declaration to the top of the
function so we can delete this block.

>  
>  			log_non_standard_event(sec_type, fru_id, fru_text,
> -- 

regards,
dan carpenter

RE: [RESEND PATCH v7 4/6] ACPI / APEI: Add callback for ARM HW errors to the GHES notifier

[Shiju Jose]

Hi Dan,

>-----Original Message-----
>From: Dan Carpenter [mailto:dan.carpenter@oracle.com]
>Sent: 21 April 2020 15:14
>To: Shiju Jose <shiju.jose@huawei.com>
>Cc: linux-acpi@vger.kernel.org; linux-pci@vger.kernel.org; linux-
>kernel@vger.kernel.org; rjw@rjwysocki.net; bp@alien8.de;
>james.morse@arm.com; helgaas@kernel.org; lenb@kernel.org;
>tony.luck@intel.com; gregkh@linuxfoundation.org;
>zhangliguang@linux.alibaba.com; tglx@linutronix.de; Linuxarm
><linuxarm@huawei.com>; Jonathan Cameron
><jonathan.cameron@huawei.com>; tanxiaofei <tanxiaofei@huawei.com>;
>yangyicong <yangyicong@huawei.com>
>Subject: Re: [RESEND PATCH v7 4/6] ACPI / APEI: Add callback for ARM HW
>errors to the GHES notifier
>
>On Tue, Apr 21, 2020 at 02:21:34PM +0100, Shiju Jose wrote:
>>  static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
>> @@ -670,12 +692,7 @@ static void ghes_do_proc(struct ghes *ghes,
>>  			pr_warn(GHES_PFX "ghes event queue full\n");
>>  			break;
>>  		}
>> -
>> -		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);
>> -		} else {
>> +		{
>>  			void *err = acpi_hest_get_payload(gdata);
>
>This is ugly.  Just move the "void *err;" declaration to the top of the function
>so we can delete this block.
Ok. I will change. 
>
>>
>>  			log_non_standard_event(sec_type, fru_id, fru_text,
>> --
>
>regards,
>dan carpenter

Thanks,
Shiju

[RESEND PATCH v7 5/6] ACPI / APEI: Add callback for non-standard HW errors to the GHES notifier

[Shiju Jose]

Add callback function for the non-standard HW errors to the GHES notifier.

Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
---
 drivers/acpi/apei/ghes.c | 56 ++++++++++++++++++++++++++--------------
 1 file changed, 37 insertions(+), 19 deletions(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 3b89c7621a0d..0c27ea8ea943 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -578,6 +578,34 @@ static int ghes_handle_arm_hw_error(struct notifier_block *nb,
 	return NOTIFY_STOP;
 }
 
+static int ghes_handle_non_standard_event(struct notifier_block *nb,
+					  unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	void *err = acpi_hest_get_payload(gdata);
+	int sec_sev;
+	guid_t *sec_type;
+	const guid_t *fru_id = &guid_null;
+	char *fru_text = "";
+
+	if (guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PLATFORM_MEM) ||
+	    guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PCIE) ||
+	    guid_equal((guid_t *)gdata->section_type, &CPER_SEC_PROC_ARM))
+		return NOTIFY_DONE;
+
+	sec_type = (guid_t *)gdata->section_type;
+	sec_sev = ghes_severity(gdata->error_severity);
+	if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
+		fru_id = (guid_t *)gdata->fru_id;
+
+	if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
+		fru_text = gdata->fru_text;
+
+	log_non_standard_event(sec_type, fru_id, fru_text, sec_sev,
+			       err, gdata->error_data_length);
+	return NOTIFY_OK;
+}
+
 static struct notifier_block ghes_notifier_mem_error = {
 	.notifier_call = ghes_handle_memory_failure,
 };
@@ -590,6 +618,10 @@ static struct notifier_block ghes_notifier_arm_hw_error = {
 	.notifier_call = ghes_handle_arm_hw_error,
 };
 
+static struct notifier_block ghes_notifier_non_standard_event = {
+	.notifier_call = ghes_handle_non_standard_event,
+};
+
 struct ghes_error_handler_list {
 	const char *name;
 	struct notifier_block *nb;
@@ -608,6 +640,10 @@ static const struct ghes_error_handler_list ghes_error_handler_list[] = {
 		.name = "ghes_notifier_arm_hw_error",
 		.nb = &ghes_notifier_arm_hw_error,
 	},
+	{
+		.name = "ghes_notifier_non_standard_event",
+		.nb = &ghes_notifier_non_standard_event,
+	},
 };
 
 static BLOCKING_NOTIFIER_HEAD(ghes_event_notify_list);
@@ -656,24 +692,13 @@ static DECLARE_WORK(ghes_event_work, ghes_event_work_func);
 static void ghes_do_proc(struct ghes *ghes,
 			 const struct acpi_hest_generic_status *estatus)
 {
-	int sev, sec_sev;
+	int sev;
 	struct acpi_hest_generic_data *gdata;
-	guid_t *sec_type;
-	const guid_t *fru_id = &guid_null;
-	char *fru_text = "";
 	struct ghes_event_entry event_entry;
 	u32 len;
 
 	sev = ghes_severity(estatus->error_severity);
 	apei_estatus_for_each_section(estatus, gdata) {
-		sec_type = (guid_t *)gdata->section_type;
-		sec_sev = ghes_severity(gdata->error_severity);
-		if (gdata->validation_bits & CPER_SEC_VALID_FRU_ID)
-			fru_id = (guid_t *)gdata->fru_id;
-
-		if (gdata->validation_bits & CPER_SEC_VALID_FRU_TEXT)
-			fru_text = gdata->fru_text;
-
 		len = acpi_hest_get_record_size(gdata);
 		event_entry.gdata = (void *)gen_pool_alloc(ghes_gdata_pool,
 							   len);
@@ -692,13 +717,6 @@ static void ghes_do_proc(struct ghes *ghes,
 			pr_warn(GHES_PFX "ghes event queue full\n");
 			break;
 		}
-		{
-			void *err = acpi_hest_get_payload(gdata);
-
-			log_non_standard_event(sec_type, fru_id, fru_text,
-					       sec_sev, err,
-					       gdata->error_data_length);
-		}
 	}
 }
 
-- 
2.17.1

[RESEND PATCH v7 6/6] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[Shiju Jose]

From: Yicong Yang <yangyicong@hisilicon.com>

The HiSilicon HIP PCIe controller is capable of handling errors
on root port and perform port reset separately at each root port.

Add error handling driver for HIP PCIe controller to log
and report recoverable errors. Perform root port reset and restore
link status after the recovery.

Following are some of the PCIe controller's recoverable errors
1. completion transmission timeout error.
2. CRS retry counter over the threshold error.
3. ECC 2 bit errors
4. AXI bresponse/rresponse errors etc.

The driver placed in the drivers/pci/controller/ because the
HIP PCIe controller does not use DWC ip.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Shiju Jose <shiju.jose@huawei.com>
--
drivers/pci/controller/Kconfig           |   8 +
drivers/pci/controller/Makefile          |   1 +
drivers/pci/controller/pcie-hisi-error.c | 336 +++++++++++++++++++++++++++++++
3 files changed, 345 insertions(+)
create mode 100644 drivers/pci/controller/pcie-hisi-error.c
---
 drivers/pci/controller/Kconfig           |   8 +
 drivers/pci/controller/Makefile          |   1 +
 drivers/pci/controller/pcie-hisi-error.c | 323 +++++++++++++++++++++++
 3 files changed, 332 insertions(+)
 create mode 100644 drivers/pci/controller/pcie-hisi-error.c

diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
index 20bf00f587bd..8bc6111480c8 100644
--- a/drivers/pci/controller/Kconfig
+++ b/drivers/pci/controller/Kconfig
@@ -268,6 +268,14 @@ config PCI_HYPERV_INTERFACE
 	  The Hyper-V PCI Interface is a helper driver allows other drivers to
 	  have a common interface with the Hyper-V PCI frontend driver.
 
+config PCIE_HISI_ERR
+	depends on ARM64 || COMPILE_TEST
+	depends on ACPI
+	bool "HiSilicon HIP PCIe controller error handling driver"
+	help
+	  Say Y here if you want error handling support
+	  for the PCIe controller's errors on HiSilicon HIP SoCs
+
 source "drivers/pci/controller/dwc/Kconfig"
 source "drivers/pci/controller/cadence/Kconfig"
 endmenu
diff --git a/drivers/pci/controller/Makefile b/drivers/pci/controller/Makefile
index 01b2502a5323..94f37b3d9929 100644
--- a/drivers/pci/controller/Makefile
+++ b/drivers/pci/controller/Makefile
@@ -29,6 +29,7 @@ obj-$(CONFIG_PCIE_MOBIVEIL) += pcie-mobiveil.o
 obj-$(CONFIG_PCIE_TANGO_SMP8759) += pcie-tango.o
 obj-$(CONFIG_VMD) += vmd.o
 obj-$(CONFIG_PCIE_BRCMSTB) += pcie-brcmstb.o
+obj-$(CONFIG_PCIE_HISI_ERR) += pcie-hisi-error.o
 # pcie-hisi.o quirks are needed even without CONFIG_PCIE_DW
 obj-y				+= dwc/
 
diff --git a/drivers/pci/controller/pcie-hisi-error.c b/drivers/pci/controller/pcie-hisi-error.c
new file mode 100644
index 000000000000..cc721070e07b
--- /dev/null
+++ b/drivers/pci/controller/pcie-hisi-error.c
@@ -0,0 +1,323 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for handling the PCIe controller errors on
+ * HiSilicon HIP SoCs.
+ *
+ * Copyright (c) 2018-2019 HiSilicon Limited.
+ */
+
+#include <linux/acpi.h>
+#include <acpi/ghes.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/kfifo.h>
+#include <linux/spinlock.h>
+
+#define	HISI_PCIE_ERR_INFO_SIZE	1024
+
+/* HISI PCIe controller error definitions */
+#define HISI_PCIE_ERR_MISC_REGS	33
+
+#define HISI_PCIE_SUB_MODULE_ID_AP	0
+#define HISI_PCIE_SUB_MODULE_ID_TL	1
+#define HISI_PCIE_SUB_MODULE_ID_MAC	2
+#define HISI_PCIE_SUB_MODULE_ID_DL	3
+#define HISI_PCIE_SUB_MODULE_ID_SDI	4
+
+#define HISI_PCIE_LOCAL_VALID_VERSION		BIT(0)
+#define HISI_PCIE_LOCAL_VALID_SOC_ID		BIT(1)
+#define HISI_PCIE_LOCAL_VALID_SOCKET_ID		BIT(2)
+#define HISI_PCIE_LOCAL_VALID_NIMBUS_ID		BIT(3)
+#define HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID	BIT(4)
+#define HISI_PCIE_LOCAL_VALID_CORE_ID		BIT(5)
+#define HISI_PCIE_LOCAL_VALID_PORT_ID		BIT(6)
+#define HISI_PCIE_LOCAL_VALID_ERR_TYPE		BIT(7)
+#define HISI_PCIE_LOCAL_VALID_ERR_SEVERITY	BIT(8)
+#define HISI_PCIE_LOCAL_VALID_ERR_MISC		9
+
+#define HISI_ERR_SEV_RECOVERABLE	0
+#define HISI_ERR_SEV_FATAL		1
+#define HISI_ERR_SEV_CORRECTED		2
+#define HISI_ERR_SEV_NONE		3
+
+static guid_t hisi_pcie_sec_type = GUID_INIT(0xB2889FC9, 0xE7D7, 0x4F9D,
+			0xA8, 0x67, 0xAF, 0x42, 0xE9, 0x8B, 0xE7, 0x72);
+
+#define HISI_PCIE_CORE_ID(v)             ((v) >> 3)
+#define HISI_PCIE_PORT_ID(core, v)       (((v) >> 1) + ((core) << 3))
+#define HISI_PCIE_CORE_PORT_ID(v)        (((v) % 8) << 1)
+
+struct hisi_pcie_error_data {
+	u64	val_bits;
+	u8	version;
+	u8	soc_id;
+	u8	socket_id;
+	u8	nimbus_id;
+	u8	sub_module_id;
+	u8	core_id;
+	u8	port_id;
+	u8	err_severity;
+	u16	err_type;
+	u8	reserv[2];
+	u32	err_misc[HISI_PCIE_ERR_MISC_REGS];
+};
+
+struct hisi_pcie_error_private {
+	struct notifier_block	nb;
+	struct platform_device	*pdev;
+};
+
+static char *hisi_pcie_sub_module_name(u8 id)
+{
+	switch (id) {
+	case HISI_PCIE_SUB_MODULE_ID_AP: return "AP Layer";
+	case HISI_PCIE_SUB_MODULE_ID_TL: return "TL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_MAC: return "MAC Layer";
+	case HISI_PCIE_SUB_MODULE_ID_DL: return "DL Layer";
+	case HISI_PCIE_SUB_MODULE_ID_SDI: return "SDI Layer";
+	}
+
+	return "unknown";
+}
+
+static char *hisi_pcie_error_severity(u8 err_sev)
+{
+	switch (err_sev) {
+	case HISI_ERR_SEV_RECOVERABLE: return "recoverable";
+	case HISI_ERR_SEV_FATAL: return "fatal";
+	case HISI_ERR_SEV_CORRECTED: return "corrected";
+	case HISI_ERR_SEV_NONE: return "none";
+	}
+
+	return "unknown";
+}
+
+static int hisi_pcie_port_reset(struct platform_device *pdev,
+				u32 chip_id, u32 port_id)
+{
+	struct device *dev = &pdev->dev;
+	acpi_handle handle = ACPI_HANDLE(dev);
+	union acpi_object arg[3];
+	struct acpi_object_list arg_list;
+	acpi_status s;
+	unsigned long long data = 0;
+
+	arg[0].type = ACPI_TYPE_INTEGER;
+	arg[0].integer.value = chip_id;
+	arg[1].type = ACPI_TYPE_INTEGER;
+	arg[1].integer.value = HISI_PCIE_CORE_ID(port_id);
+	arg[2].type = ACPI_TYPE_INTEGER;
+	arg[2].integer.value = HISI_PCIE_CORE_PORT_ID(port_id);
+
+	arg_list.count = 3;
+	arg_list.pointer = arg;
+
+	s = acpi_evaluate_integer(handle, "RST", &arg_list, &data);
+	if (ACPI_FAILURE(s)) {
+		dev_err(dev, "No RST method\n");
+		return -EIO;
+	}
+
+	if (data) {
+		dev_err(dev, "Failed to Reset\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int hisi_pcie_port_do_recovery(struct platform_device *dev,
+				      u32 chip_id, u32 port_id)
+{
+	acpi_status s;
+	struct device *device = &dev->dev;
+	acpi_handle root_handle = ACPI_HANDLE(device);
+	struct acpi_pci_root *pci_root;
+	struct pci_bus *root_bus;
+	struct pci_dev *pdev;
+	u32 domain, busnr, devfn;
+
+	s = acpi_get_parent(root_handle, &root_handle);
+	if (ACPI_FAILURE(s))
+		return -ENODEV;
+	pci_root = acpi_pci_find_root(root_handle);
+	if (!pci_root)
+		return -ENODEV;
+	root_bus = pci_root->bus;
+	domain = pci_root->segment;
+
+	busnr = root_bus->number;
+	devfn = PCI_DEVFN(port_id, 0);
+	pdev = pci_get_domain_bus_and_slot(domain, busnr, devfn);
+	if (!pdev) {
+		dev_info(device, "Fail to get root port %04x:%02x:%02x.%d device\n",
+			 domain, busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
+		return -ENODEV;
+	}
+
+	pci_stop_and_remove_bus_device_locked(pdev);
+	pci_dev_put(pdev);
+
+	if (hisi_pcie_port_reset(dev, chip_id, port_id))
+		return -EIO;
+
+	/*
+	 * The initialization time of subordinate devices after
+	 * hot reset is no more than 1s, which is required by
+	 * the PCI spec v5.0 sec 6.6.1. The time will shorten
+	 * if Readiness Notifications mechanisms are used. But
+	 * wait 1s here to adapt any conditions.
+	 */
+	ssleep(1UL);
+
+	/* add root port and downstream devices */
+	pci_lock_rescan_remove();
+	pci_rescan_bus(root_bus);
+	pci_unlock_rescan_remove();
+
+	return 0;
+}
+
+static void hisi_pcie_handle_error(const struct hisi_pcie_error_data *error,
+				   struct platform_device *pdev)
+{
+	char buf[HISI_PCIE_ERR_INFO_SIZE];
+	char *p = buf, *end = buf + sizeof(buf);
+	struct device *dev = &pdev->dev;
+	u32 i;
+	int rc;
+
+	if (error->val_bits == 0) {
+		dev_warn(dev, "%s: no valid error information\n", __func__);
+		return;
+	}
+
+	/* Logging */
+	p += snprintf(p, end - p, "[ Table version=%d ", error->version);
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOC_ID)
+		p += snprintf(p, end - p, "SOC ID=%d ", error->soc_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SOCKET_ID)
+		p += snprintf(p, end - p, "socket ID=%d ", error->socket_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_NIMBUS_ID)
+		p += snprintf(p, end - p, "nimbus ID=%d ", error->nimbus_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_SUB_MODULE_ID)
+		p += snprintf(p, end - p, "sub module=%s ",
+			      hisi_pcie_sub_module_name(error->sub_module_id));
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_CORE_ID)
+		p += snprintf(p, end - p, "core ID=core%d ", error->core_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_PORT_ID)
+		p += snprintf(p, end - p, "port ID=port%d ", error->port_id);
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_SEVERITY)
+		p += snprintf(p, end - p, "error severity=%s ",
+			      hisi_pcie_error_severity(error->err_severity));
+
+	if (error->val_bits & HISI_PCIE_LOCAL_VALID_ERR_TYPE)
+		p += snprintf(p, end - p, "error type=0x%x ", error->err_type);
+
+	p += snprintf(p, end - p, "]\n");
+	dev_info(dev, "\nHISI : HIP : PCIe controller error\n");
+	dev_info(dev, "%s\n", buf);
+
+	dev_info(dev, "Reg Dump:\n");
+	for (i = 0; i < HISI_PCIE_ERR_MISC_REGS; i++) {
+		if (error->val_bits &
+				BIT_ULL(HISI_PCIE_LOCAL_VALID_ERR_MISC + i))
+			dev_info(dev,
+				 "ERR_MISC_%d=0x%x\n", i, error->err_misc[i]);
+	}
+
+	/* Recovery for the PCIe controller errors */
+	if (error->err_severity == HISI_ERR_SEV_RECOVERABLE) {
+		/* try reset PCI port for the error recovery */
+		rc = hisi_pcie_port_do_recovery(pdev, error->socket_id,
+			HISI_PCIE_PORT_ID(error->core_id, error->port_id));
+		if (rc) {
+			dev_info(dev, "fail to do hisi pcie port reset\n");
+			return;
+		}
+	}
+}
+
+static int hisi_pcie_notify_error(struct notifier_block *nb,
+				  unsigned long event, void *data)
+{
+	struct acpi_hest_generic_data *gdata = data;
+	const struct hisi_pcie_error_data *error_data =
+				acpi_hest_get_payload(gdata);
+	struct hisi_pcie_error_private *priv =
+			container_of(nb, struct hisi_pcie_error_private, nb);
+	struct platform_device *pdev = priv->pdev;
+	struct device *dev = &pdev->dev;
+	u8 socket;
+
+	if (device_property_read_u8(dev, "socket", &socket))
+		return NOTIFY_DONE;
+
+	if (!guid_equal((guid_t *)gdata->section_type, &hisi_pcie_sec_type) ||
+	    error_data->socket_id != socket)
+		return NOTIFY_DONE;
+
+	hisi_pcie_handle_error(error_data, pdev);
+
+	return NOTIFY_OK;
+}
+
+static int hisi_pcie_error_handler_probe(struct platform_device *pdev)
+{
+	struct hisi_pcie_error_private *priv;
+	int ret;
+
+	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->nb.notifier_call = hisi_pcie_notify_error;
+	priv->pdev = pdev;
+	ret = ghes_register_event_notifier(&priv->nb);
+	if (ret) {
+		dev_err(&pdev->dev, "%s : ghes_register_event_notifier fail\n",
+			__func__);
+		return ret;
+	}
+
+	platform_set_drvdata(pdev, priv);
+
+	return 0;
+}
+
+static int hisi_pcie_error_handler_remove(struct platform_device *pdev)
+{
+	struct hisi_pcie_error_private *priv = platform_get_drvdata(pdev);
+
+	if (priv)
+		ghes_unregister_event_notifier(&priv->nb);
+
+	kfree(priv);
+
+	return 0;
+}
+
+static const struct acpi_device_id hisi_pcie_acpi_match[] = {
+	{ "HISI0361", 0 },
+	{ }
+};
+
+static struct platform_driver hisi_pcie_error_handler_driver = {
+	.driver = {
+		.name	= "hisi-pcie-error-handler",
+		.acpi_match_table = hisi_pcie_acpi_match,
+	},
+	.probe		= hisi_pcie_error_handler_probe,
+	.remove		= hisi_pcie_error_handler_remove,
+};
+module_platform_driver(hisi_pcie_error_handler_driver);
+
+MODULE_DESCRIPTION("HiSilicon HIP PCIe controller error handling driver");
+MODULE_LICENSE("GPL v2");
-- 
2.17.1

Re: [RESEND PATCH v7 6/6] PCI: hip: Add handling of HiSilicon HIP PCIe controller errors

[Dan Carpenter]

On Tue, Apr 21, 2020 at 02:21:36PM +0100, Shiju Jose wrote:
> +static int hisi_pcie_error_handler_probe(struct platform_device *pdev)
> +{
> +	struct hisi_pcie_error_private *priv;
> +	int ret;
> +
> +	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
> +	if (!priv)
> +		return -ENOMEM;
> +
> +	priv->nb.notifier_call = hisi_pcie_notify_error;
> +	priv->pdev = pdev;
> +	ret = ghes_register_event_notifier(&priv->nb);
> +	if (ret) {
> +		dev_err(&pdev->dev, "%s : ghes_register_event_notifier fail\n",
> +			__func__);
> +		return ret;
> +	}
> +
> +	platform_set_drvdata(pdev, priv);
> +
> +	return 0;
> +}
> +
> +static int hisi_pcie_error_handler_remove(struct platform_device *pdev)
> +{
> +	struct hisi_pcie_error_private *priv = platform_get_drvdata(pdev);
> +
> +	if (priv)
> +		ghes_unregister_event_notifier(&priv->nb);

Delete this NULL check.  The remove function isn't called on stuff which
hasn't been allocated.

The "never free things which haven't been allocated" rule is a good rule
to keep in mind because it makes error handling so much simpler.

> +
> +	kfree(priv);
> +
> +	return 0;
> +}

regards,
dan carpenter