Merge branch 'apei-release' into release

4 files changed, 233 insertions, 51 deletions

diff --git a/drivers/acpi/apei/Kconfig b/drivers/acpi/apei/Kconfig
index fca34ccfd29..9ecf6feae83 100644
--- a/drivers/acpi/apei/Kconfig
+++ b/drivers/acpi/apei/Kconfig
@@ -21,6 +21,13 @@ config ACPI_APEI_GHES
 	  by firmware to produce more valuable hardware error
 	  information for Linux.
 
+config ACPI_APEI_PCIEAER
+	bool "APEI PCIe AER logging/recovering support"
+	depends on ACPI_APEI && PCIEAER
+	help
+	  PCIe AER errors may be reported via APEI firmware first mode.
+	  Turn on this option to enable the corresponding support.
+
 config ACPI_APEI_EINJ
 	tristate "APEI Error INJection (EINJ)"
 	depends on ACPI_APEI && DEBUG_FS
diff --git a/drivers/acpi/apei/cper.c b/drivers/acpi/apei/cper.c
index 31464a006d7..5d4189464d6 100644
--- a/drivers/acpi/apei/cper.c
+++ b/drivers/acpi/apei/cper.c
@@ -29,6 +29,7 @@
 #include <linux/time.h>
 #include <linux/cper.h>
 #include <linux/acpi.h>
+#include <linux/aer.h>
 
 /*
  * CPER record ID need to be unique even after reboot, because record
@@ -70,8 +71,8 @@ static const char *cper_severity_str(unsigned int severity)
  * If the output length is longer than 80, multiple line will be
  * printed, with @pfx is printed at the beginning of each line.
  */
-static void cper_print_bits(const char *pfx, unsigned int bits,
-			    const char *strs[], unsigned int strs_size)
+void cper_print_bits(const char *pfx, unsigned int bits,
+		     const char *strs[], unsigned int strs_size)
 {
 	int i, len = 0;
 	const char *str;
@@ -81,6 +82,8 @@ static void cper_print_bits(const char *pfx, unsigned int bits,
 		if (!(bits & (1U << i)))
 			continue;
 		str = strs[i];
+		if (!str)
+			continue;
 		if (len && len + strlen(str) + 2 > 80) {
 			printk("%s\n", buf);
 			len = 0;
@@ -243,7 +246,8 @@ static const char *cper_pcie_port_type_strs[] = {
 	"root complex event collector",
 };
 
-static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie)
+static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
+			    const struct acpi_hest_generic_data *gdata)
 {
 	if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE)
 		printk("%s""port_type: %d, %s\n", pfx, pcie->port_type,
@@ -276,6 +280,12 @@ static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie)
 		printk(
 	"%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
 	pfx, pcie->bridge.secondary_status, pcie->bridge.control);
+#ifdef CONFIG_ACPI_APEI_PCIEAER
+	if (pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) {
+		struct aer_capability_regs *aer_regs = (void *)pcie->aer_info;
+		cper_print_aer(pfx, gdata->error_severity, aer_regs);
+	}
+#endif
 }
 
 static const char *apei_estatus_section_flag_strs[] = {
@@ -322,7 +332,7 @@ static void apei_estatus_print_section(
 		struct cper_sec_pcie *pcie = (void *)(gdata + 1);
 		printk("%s""section_type: PCIe error\n", pfx);
 		if (gdata->error_data_length >= sizeof(*pcie))
-			cper_print_pcie(pfx, pcie);
+			cper_print_pcie(pfx, pcie, gdata);
 		else
 			goto err_section_too_small;
 	} else
diff --git a/drivers/acpi/apei/erst-dbg.c b/drivers/acpi/apei/erst-dbg.c
index de73caf3ceb..a4cfb64c86a 100644
--- a/drivers/acpi/apei/erst-dbg.c
+++ b/drivers/acpi/apei/erst-dbg.c
@@ -43,12 +43,27 @@ static DEFINE_MUTEX(erst_dbg_mutex);
 
 static int erst_dbg_open(struct inode *inode, struct file *file)
 {
+	int rc, *pos;
+
 	if (erst_disable)
 		return -ENODEV;
 
+	pos = (int *)&file->private_data;
+
+	rc = erst_get_record_id_begin(pos);
+	if (rc)
+		return rc;
+
 	return nonseekable_open(inode, file);
 }
 
+static int erst_dbg_release(struct inode *inode, struct file *file)
+{
+	erst_get_record_id_end();
+
+	return 0;
+}
+
 static long erst_dbg_ioctl(struct file *f, unsigned int cmd, unsigned long arg)
 {
 	int rc;
@@ -79,18 +94,20 @@ static long erst_dbg_ioctl(struct file *f, unsigned int cmd, unsigned long arg)
 static ssize_t erst_dbg_read(struct file *filp, char __user *ubuf,
 			     size_t usize, loff_t *off)
 {
-	int rc;
+	int rc, *pos;
 	ssize_t len = 0;
 	u64 id;
 
-	if (*off != 0)
+	if (*off)
 		return -EINVAL;
 
 	if (mutex_lock_interruptible(&erst_dbg_mutex) != 0)
 		return -EINTR;
 
+	pos = (int *)&filp->private_data;
+
 retry_next:
-	rc = erst_get_next_record_id(&id);
+	rc = erst_get_record_id_next(pos, &id);
 	if (rc)
 		goto out;
 	/* no more record */
@@ -181,6 +198,7 @@ out:
 static const struct file_operations erst_dbg_ops = {
 	.owner		= THIS_MODULE,
 	.open		= erst_dbg_open,
+	.release	= erst_dbg_release,
 	.read		= erst_dbg_read,
 	.write		= erst_dbg_write,
 	.unlocked_ioctl	= erst_dbg_ioctl,
diff --git a/drivers/acpi/apei/erst.c b/drivers/acpi/apei/erst.c
index cf6db6b7662..8ff8c32fef5 100644
--- a/drivers/acpi/apei/erst.c
+++ b/drivers/acpi/apei/erst.c
@@ -429,6 +429,22 @@ ssize_t erst_get_record_count(void)
 }
 EXPORT_SYMBOL_GPL(erst_get_record_count);
 
+#define ERST_RECORD_ID_CACHE_SIZE_MIN	16
+#define ERST_RECORD_ID_CACHE_SIZE_MAX	1024
+
+struct erst_record_id_cache {
+	struct mutex lock;
+	u64 *entries;
+	int len;
+	int size;
+	int refcount;
+};
+
+static struct erst_record_id_cache erst_record_id_cache = {
+	.lock = __MUTEX_INITIALIZER(erst_record_id_cache.lock),
+	.refcount = 0,
+};
+
 static int __erst_get_next_record_id(u64 *record_id)
 {
 	struct apei_exec_context ctx;
@@ -443,26 +459,179 @@ static int __erst_get_next_record_id(u64 *record_id)
 	return 0;
 }
 
+int erst_get_record_id_begin(int *pos)
+{
+	int rc;
+
+	if (erst_disable)
+		return -ENODEV;
+
+	rc = mutex_lock_interruptible(&erst_record_id_cache.lock);
+	if (rc)
+		return rc;
+	erst_record_id_cache.refcount++;
+	mutex_unlock(&erst_record_id_cache.lock);
+
+	*pos = 0;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(erst_get_record_id_begin);
+
+/* erst_record_id_cache.lock must be held by caller */
+static int __erst_record_id_cache_add_one(void)
+{
+	u64 id, prev_id, first_id;
+	int i, rc;
+	u64 *entries;
+	unsigned long flags;
+
+	id = prev_id = first_id = APEI_ERST_INVALID_RECORD_ID;
+retry:
+	raw_spin_lock_irqsave(&erst_lock, flags);
+	rc = __erst_get_next_record_id(&id);
+	raw_spin_unlock_irqrestore(&erst_lock, flags);
+	if (rc == -ENOENT)
+		return 0;
+	if (rc)
+		return rc;
+	if (id == APEI_ERST_INVALID_RECORD_ID)
+		return 0;
+	/* can not skip current ID, or loop back to first ID */
+	if (id == prev_id || id == first_id)
+		return 0;
+	if (first_id == APEI_ERST_INVALID_RECORD_ID)
+		first_id = id;
+	prev_id = id;
+
+	entries = erst_record_id_cache.entries;
+	for (i = 0; i < erst_record_id_cache.len; i++) {
+		if (entries[i] == id)
+			break;
+	}
+	/* record id already in cache, try next */
+	if (i < erst_record_id_cache.len)
+		goto retry;
+	if (erst_record_id_cache.len >= erst_record_id_cache.size) {
+		int new_size, alloc_size;
+		u64 *new_entries;
+
+		new_size = erst_record_id_cache.size * 2;
+		new_size = clamp_val(new_size, ERST_RECORD_ID_CACHE_SIZE_MIN,
+				     ERST_RECORD_ID_CACHE_SIZE_MAX);
+		if (new_size <= erst_record_id_cache.size) {
+			if (printk_ratelimit())
+				pr_warning(FW_WARN ERST_PFX
+					   "too many record ID!\n");
+			return 0;
+		}
+		alloc_size = new_size * sizeof(entries[0]);
+		if (alloc_size < PAGE_SIZE)
+			new_entries = kmalloc(alloc_size, GFP_KERNEL);
+		else
+			new_entries = vmalloc(alloc_size);
+		if (!new_entries)
+			return -ENOMEM;
+		memcpy(new_entries, entries,
+		       erst_record_id_cache.len * sizeof(entries[0]));
+		if (erst_record_id_cache.size < PAGE_SIZE)
+			kfree(entries);
+		else
+			vfree(entries);
+		erst_record_id_cache.entries = entries = new_entries;
+		erst_record_id_cache.size = new_size;
+	}
+	entries[i] = id;
+	erst_record_id_cache.len++;
+
+	return 1;
+}
+
 /*
  * Get the record ID of an existing error record on the persistent
  * storage. If there is no error record on the persistent storage, the
  * returned record_id is APEI_ERST_INVALID_RECORD_ID.
  */
-int erst_get_next_record_id(u64 *record_id)
+int erst_get_record_id_next(int *pos, u64 *record_id)
 {
-	int rc;
-	unsigned long flags;
+	int rc = 0;
+	u64 *entries;
 
 	if (erst_disable)
 		return -ENODEV;
 
-	raw_spin_lock_irqsave(&erst_lock, flags);
-	rc = __erst_get_next_record_id(record_id);
-	raw_spin_unlock_irqrestore(&erst_lock, flags);
+	/* must be enclosed by erst_get_record_id_begin/end */
+	BUG_ON(!erst_record_id_cache.refcount);
+	BUG_ON(*pos < 0 || *pos > erst_record_id_cache.len);
+
+	mutex_lock(&erst_record_id_cache.lock);
+	entries = erst_record_id_cache.entries;
+	for (; *pos < erst_record_id_cache.len; (*pos)++)
+		if (entries[*pos] != APEI_ERST_INVALID_RECORD_ID)
+			break;
+	/* found next record id in cache */
+	if (*pos < erst_record_id_cache.len) {
+		*record_id = entries[*pos];
+		(*pos)++;
+		goto out_unlock;
+	}
+
+	/* Try to add one more record ID to cache */
+	rc = __erst_record_id_cache_add_one();
+	if (rc < 0)
+		goto out_unlock;
+	/* successfully add one new ID */
+	if (rc == 1) {
+		*record_id = erst_record_id_cache.entries[*pos];
+		(*pos)++;
+		rc = 0;
+	} else {
+		*pos = -1;
+		*record_id = APEI_ERST_INVALID_RECORD_ID;
+	}
+out_unlock:
+	mutex_unlock(&erst_record_id_cache.lock);
 
 	return rc;
 }
-EXPORT_SYMBOL_GPL(erst_get_next_record_id);
+EXPORT_SYMBOL_GPL(erst_get_record_id_next);
+
+/* erst_record_id_cache.lock must be held by caller */
+static void __erst_record_id_cache_compact(void)
+{
+	int i, wpos = 0;
+	u64 *entries;
+
+	if (erst_record_id_cache.refcount)
+		return;
+
+	entries = erst_record_id_cache.entries;
+	for (i = 0; i < erst_record_id_cache.len; i++) {
+		if (entries[i] == APEI_ERST_INVALID_RECORD_ID)
+			continue;
+		if (wpos != i)
+			memcpy(&entries[wpos], &entries[i], sizeof(entries[i]));
+		wpos++;
+	}
+	erst_record_id_cache.len = wpos;
+}
+
+void erst_get_record_id_end(void)
+{
+	/*
+	 * erst_disable != 0 should be detected by invoker via the
+	 * return value of erst_get_record_id_begin/next, so this
+	 * function should not be called for erst_disable != 0.
+	 */
+	BUG_ON(erst_disable);
+
+	mutex_lock(&erst_record_id_cache.lock);
+	erst_record_id_cache.refcount--;
+	BUG_ON(erst_record_id_cache.refcount < 0);
+	__erst_record_id_cache_compact();
+	mutex_unlock(&erst_record_id_cache.lock);
+}
+EXPORT_SYMBOL_GPL(erst_get_record_id_end);
 
 static int __erst_write_to_storage(u64 offset)
 {
@@ -703,56 +872,34 @@ ssize_t erst_read(u64 record_id, struct cper_record_header *record,
 }
 EXPORT_SYMBOL_GPL(erst_read);
 
-/*
- * If return value > buflen, the buffer size is not big enough,
- * else if return value = 0, there is no more record to read,
- * else if return value < 0, something goes wrong,
- * else everything is OK, and return value is record length
- */
-ssize_t erst_read_next(struct cper_record_header *record, size_t buflen)
-{
-	int rc;
-	ssize_t len;
-	unsigned long flags;
-	u64 record_id;
-
-	if (erst_disable)
-		return -ENODEV;
-
-	raw_spin_lock_irqsave(&erst_lock, flags);
-	rc = __erst_get_next_record_id(&record_id);
-	if (rc) {
-		raw_spin_unlock_irqrestore(&erst_lock, flags);
-		return rc;
-	}
-	/* no more record */
-	if (record_id == APEI_ERST_INVALID_RECORD_ID) {
-		raw_spin_unlock_irqrestore(&erst_lock, flags);
-		return 0;
-	}
-
-	len = __erst_read(record_id, record, buflen);
-	raw_spin_unlock_irqrestore(&erst_lock, flags);
-
-	return len;
-}
-EXPORT_SYMBOL_GPL(erst_read_next);
-
 int erst_clear(u64 record_id)
 {
-	int rc;
+	int rc, i;
 	unsigned long flags;
+	u64 *entries;
 
 	if (erst_disable)
 		return -ENODEV;
 
+	rc = mutex_lock_interruptible(&erst_record_id_cache.lock);
+	if (rc)
+		return rc;
 	raw_spin_lock_irqsave(&erst_lock, flags);
 	if (erst_erange.attr & ERST_RANGE_NVRAM)
 		rc = __erst_clear_from_nvram(record_id);
 	else
 		rc = __erst_clear_from_storage(record_id);
 	raw_spin_unlock_irqrestore(&erst_lock, flags);
-
+	if (rc)
+		goto out;
+	entries = erst_record_id_cache.entries;
+	for (i = 0; i < erst_record_id_cache.len; i++) {
+		if (entries[i] == record_id)
+			entries[i] = APEI_ERST_INVALID_RECORD_ID;
+	}
+	__erst_record_id_cache_compact();
+out:
+	mutex_unlock(&erst_record_id_cache.lock);
 	return rc;
 }
 EXPORT_SYMBOL_GPL(erst_clear);