[PATCH v5 07/12] pstore/blk: Add ftrace frontend support

[WeiXiong Liao <liaoweixiong@xxxxxxxxxxxxxxxxx>]

Support backend for ftrace. To enable ftrace backend, just make
ftrace_size be greater than 0 and a multiple of 4096.

Signed-off-by: WeiXiong Liao <liaoweixiong@xxxxxxxxxxxxxxxxx>
Link: https://lore.kernel.org/r/1585126506-18635-7-git-send-email-liaoweixiong@xxxxxxxxxxxxxxxxx
Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
---
 fs/pstore/Kconfig           |  12 ++++
 fs/pstore/blk.c             |   9 +++
 fs/pstore/zone.c            | 170 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/pstore_zone.h |   2 +
 4 files changed, 193 insertions(+)

diff --git a/fs/pstore/Kconfig b/fs/pstore/Kconfig
index f1484f751c5e..16a0440d8d5a 100644
--- a/fs/pstore/Kconfig
+++ b/fs/pstore/Kconfig
@@ -248,3 +248,15 @@ config PSTORE_BLK_CONSOLE_SIZE
 
 	  NOTE that, both Kconfig and module parameters can configure
 	  pstore/blk, but module parameters have priority over Kconfig.
+
+config PSTORE_BLK_FTRACE_SIZE
+	int "Size in Kbytes of ftarce to store"
+	depends on PSTORE_BLK
+	depends on PSTORE_FTRACE
+	default 64
+	help
+	  This just sets size of ftrace (ftrace_size) for pstore/blk. The
+	  size is in KB and must be a multiple of 4.
+
+	  NOTE that, both Kconfig and module parameters can configure
+	  pstore/blk, but module parameters have priority over Kconfig.
diff --git a/fs/pstore/blk.c b/fs/pstore/blk.c
index acdeb5d7b5f3..84ca98923d8a 100644
--- a/fs/pstore/blk.c
+++ b/fs/pstore/blk.c
@@ -40,6 +40,14 @@
 module_param(console_size, long, 0400);
 MODULE_PARM_DESC(console_size, "console size in kbytes");
 
+#if IS_ENABLED(CONFIG_PSTORE_FTRACE)
+static long ftrace_size = CONFIG_PSTORE_BLK_FTRACE_SIZE;
+#else
+static long ftrace_size = -1;
+#endif
+module_param(ftrace_size, long, 0400);
+MODULE_PARM_DESC(ftrace_size, "ftrace size in kbytes");
+
 /*
  * blkdev - The block device to use.
  *
@@ -143,6 +151,7 @@ static int psblk_register_do(struct psblk_device *dev)
 	verify_size(kmsg_size, 4096, dev->flags & PSTORE_FLAGS_DMESG);
 	verify_size(pmsg_size, 4096, dev->flags & PSTORE_FLAGS_PMSG);
 	verify_size(console_size, 4096, dev->flags & PSTORE_FLAGS_CONSOLE);
+	verify_size(ftrace_size, 4096, dev->flags & PSTORE_FLAGS_FTRACE);
 #undef verify_size
 
 	pstore_zone_info->total_size = dev->total_size;
diff --git a/fs/pstore/zone.c b/fs/pstore/zone.c
index 427b458aa675..a51594f95347 100644
--- a/fs/pstore/zone.c
+++ b/fs/pstore/zone.c
@@ -88,11 +88,14 @@ struct pstore_zone {
  * @opszs: oops/panic storage zones
  * @ppsz: pmsg storage zone
  * @cpsz: console storage zone
+ * @fpszs: ftrace storage zones
  * @oops_max_cnt: max count of @opszs
  * @oops_read_cnt: counter to read oops zone
  * @oops_write_cnt: counter to write
  * @pmsg_read_cnt: counter to read pmsg zone
  * @console_read_cnt: counter to read console zone
+ * @ftrace_max_cnt: max count of @fpszs
+ * @ftrace_read_cnt: counter to read ftrace zone
  * @oops_counter: counter to oops
  * @panic_counter: counter to panic
  * @recovered: whether finish recovering data from storage
@@ -105,11 +108,14 @@ struct psz_context {
 	struct pstore_zone **opszs;
 	struct pstore_zone *ppsz;
 	struct pstore_zone *cpsz;
+	struct pstore_zone **fpszs;
 	unsigned int oops_max_cnt;
 	unsigned int oops_read_cnt;
 	unsigned int oops_write_cnt;
 	unsigned int pmsg_read_cnt;
 	unsigned int console_read_cnt;
+	unsigned int ftrace_max_cnt;
+	unsigned int ftrace_read_cnt;
 	/*
 	 * the counter should be recovered when recover.
 	 * It records the oops/panic times after burning rather than booting.
@@ -312,6 +318,8 @@ static void psz_flush_all_dirty_zones(struct work_struct *work)
 		ret |= psz_flush_dirty_zone(cxt->cpsz);
 	if (cxt->opszs)
 		ret |= psz_flush_dirty_zones(cxt->opszs, cxt->oops_max_cnt);
+	if (cxt->fpszs)
+		ret |= psz_flush_dirty_zones(cxt->fpszs, cxt->ftrace_max_cnt);
 	if (ret && cxt->pstore_zone_info)
 		schedule_delayed_work(&psz_cleaner, msecs_to_jiffies(1000));
 }
@@ -548,6 +556,31 @@ static int psz_recover_zone(struct psz_context *cxt, struct pstore_zone *zone)
 	return ret;
 }
 
+static int psz_recover_zones(struct psz_context *cxt,
+		struct pstore_zone **zones, unsigned int cnt)
+{
+	int ret;
+	unsigned int i;
+	struct pstore_zone *zone;
+
+	if (!zones)
+		return 0;
+
+	for (i = 0; i < cnt; i++) {
+		zone = zones[i];
+		if (unlikely(!zone))
+			continue;
+		ret = psz_recover_zone(cxt, zone);
+		if (ret)
+			goto recover_fail;
+	}
+
+	return 0;
+recover_fail:
+	pr_debug("recover %s[%u] failed\n", zone->name, i);
+	return ret;
+}
+
 /**
  * psz_recovery() - recover data from storage
  * @cxt: the context of pstore/zone
@@ -575,6 +608,10 @@ static inline int psz_recovery(struct psz_context *cxt)
 	if (ret)
 		goto recover_fail;
 
+	ret = psz_recover_zones(cxt, cxt->fpszs, cxt->ftrace_max_cnt);
+	if (ret)
+		goto recover_fail;
+
 	pr_debug("recover end!\n");
 	atomic_set(&cxt->recovered, 1);
 	return 0;
@@ -591,6 +628,7 @@ static int psz_pstore_open(struct pstore_info *psi)
 	cxt->oops_read_cnt = 0;
 	cxt->pmsg_read_cnt = 0;
 	cxt->console_read_cnt = 0;
+	cxt->ftrace_read_cnt = 0;
 	return 0;
 }
 
@@ -657,6 +695,10 @@ static int psz_pstore_erase(struct pstore_record *record)
 		return psz_record_erase(cxt, cxt->ppsz);
 	case PSTORE_TYPE_CONSOLE:
 		return psz_record_erase(cxt, cxt->cpsz);
+	case PSTORE_TYPE_FTRACE:
+		if (record->id >= cxt->ftrace_max_cnt)
+			return -EINVAL;
+		return psz_record_erase(cxt, cxt->fpszs[record->id]);
 	default: return -EINVAL;
 	}
 }
@@ -799,6 +841,13 @@ static int notrace psz_pstore_write(struct pstore_record *record)
 		return psz_record_write(cxt->cpsz, record);
 	case PSTORE_TYPE_PMSG:
 		return psz_record_write(cxt->ppsz, record);
+	case PSTORE_TYPE_FTRACE: {
+		int zonenum = smp_processor_id();
+
+		if (!cxt->fpszs)
+			return -ENOSPC;
+		return psz_record_write(cxt->fpszs[zonenum], record);
+	}
 	default:
 		return -EINVAL;
 	}
@@ -814,6 +863,14 @@ static struct pstore_zone *psz_read_next_zone(struct psz_context *cxt)
 			return zone;
 	}
 
+	if (cxt->ftrace_read_cnt < cxt->ftrace_max_cnt)
+		/*
+		 * No need psz_old_ok(). Let psz_ftrace_read() do so for
+		 * combination. psz_ftrace_read() should traverse over
+		 * all zones in case of some zone without data.
+		 */
+		return cxt->fpszs[cxt->ftrace_read_cnt++];
+
 	if (cxt->pmsg_read_cnt == 0) {
 		cxt->pmsg_read_cnt++;
 		zone = cxt->ppsz;
@@ -888,6 +945,98 @@ static ssize_t psz_oops_read(struct pstore_zone *zone,
 	return size + hlen;
 }
 
+static int psz_ftrace_combine(char *src1_buf, size_t src1_size,
+		char *src2_buf, size_t src2_size,
+		char **dest_buf, size_t *dest_size)
+{
+	size_t src1_off, src2_off, total;
+	size_t src1_idx = 0, src2_idx = 0, merged_idx = 0;
+	void *merged_buf;
+	struct pstore_ftrace_record *mrec, *s1rec, *s2rec;
+	size_t record_size = sizeof(struct pstore_ftrace_record);
+
+	src1_off = src1_size % record_size;
+	src1_size -= src1_off;
+
+	src2_off = src2_size % record_size;
+	src2_size -= src2_off;
+
+	total = src1_size + src2_size;
+	merged_buf = kmalloc(total, GFP_KERNEL);
+	if (!merged_buf)
+		return -ENOMEM;
+
+	s1rec = (struct pstore_ftrace_record *)(src1_buf + src1_off);
+	s2rec = (struct pstore_ftrace_record *)(src2_buf + src2_off);
+	mrec = (struct pstore_ftrace_record *)(merged_buf);
+
+	while (src1_size > 0 && src2_size > 0) {
+		u64 s1_ts, s2_ts;
+
+		s1_ts = pstore_ftrace_read_timestamp(&s1rec[src1_idx]);
+		s2_ts = pstore_ftrace_read_timestamp(&s2rec[src2_idx]);
+		if (s1_ts < s2_ts) {
+			mrec[merged_idx++] = s1rec[src1_idx++];
+			src1_size -= record_size;
+		} else {
+			mrec[merged_idx++] = s2rec[src2_idx++];
+			src2_size -= record_size;
+		}
+	}
+
+	while (src1_size > 0) {
+		mrec[merged_idx++] = s1rec[src1_idx++];
+		src1_size -= record_size;
+	}
+
+	while (src2_size > 0) {
+		mrec[merged_idx++] = s2rec[src2_idx++];
+		src2_size -= record_size;
+	}
+
+	*dest_buf = merged_buf;
+	*dest_size = total;
+	return 0;
+}
+
+/* try to combine all ftrace zones */
+static ssize_t psz_ftrace_read(struct pstore_zone *zone,
+		struct pstore_record *record)
+{
+	struct psz_context *cxt = record->psi->data;
+	struct psz_buffer *buf;
+	char *dest;
+	size_t dest_size;
+	int ret;
+
+	if (!zone || !record)
+		return -ENOSPC;
+
+	if (!psz_old_ok(zone))
+		goto out;
+
+	buf = (struct psz_buffer *)zone->oldbuf;
+	if (!buf)
+		return -ENOMSG;
+
+	ret = psz_ftrace_combine(record->buf, record->size,
+			(char *)buf->data, atomic_read(&buf->datalen),
+			&dest, &dest_size);
+	if (unlikely(ret))
+		return ret;
+
+	kfree(record->buf);
+	record->buf = dest;
+	record->size = dest_size;
+
+out:
+	if (cxt->ftrace_read_cnt < cxt->ftrace_max_cnt)
+		/* then, read next ftrace zone */
+		return -ENOMSG;
+	record->id = 0;
+	return record->size ? record->size : -ENOMSG;
+}
+
 static ssize_t psz_record_read(struct pstore_zone *zone,
 		struct pstore_record *record)
 {
@@ -938,6 +1087,9 @@ static ssize_t psz_pstore_read(struct pstore_record *record)
 		readop = psz_oops_read;
 		record->id = cxt->oops_read_cnt - 1;
 		break;
+	case PSTORE_TYPE_FTRACE:
+		readop = psz_ftrace_read;
+		break;
 	case PSTORE_TYPE_CONSOLE:
 		fallthrough;
 	case PSTORE_TYPE_PMSG:
@@ -1090,6 +1242,8 @@ static void psz_free_all_zones(struct psz_context *cxt)
 		psz_free_zone(&cxt->ppsz);
 	if (cxt->cpsz)
 		psz_free_zone(&cxt->cpsz);
+	if (cxt->fpszs)
+		psz_free_zones(&cxt->fpszs, &cxt->ftrace_max_cnt);
 }
 
 static int psz_alloc_zones(struct psz_context *cxt)
@@ -1114,6 +1268,16 @@ static int psz_alloc_zones(struct psz_context *cxt)
 		goto free_out;
 	}
 
+	off_size += info->ftrace_size;
+	cxt->fpszs = psz_init_zones(PSTORE_TYPE_FTRACE, &off,
+			info->ftrace_size,
+			info->ftrace_size / nr_cpu_ids,
+			&cxt->ftrace_max_cnt);
+	if (IS_ERR(cxt->fpszs)) {
+		err = PTR_ERR(cxt->fpszs);
+		goto free_out;
+	}
+
 	cxt->opszs = psz_init_zones(PSTORE_TYPE_DMESG, &off,
 			info->total_size - off_size,
 			info->kmsg_size, &cxt->oops_max_cnt);
@@ -1176,6 +1340,7 @@ int register_pstore_zone(struct pstore_zone_info *info)
 	check_size(kmsg_size, SECTOR_SIZE);
 	check_size(pmsg_size, SECTOR_SIZE);
 	check_size(console_size, SECTOR_SIZE);
+	check_size(ftrace_size, SECTOR_SIZE);
 
 #undef check_size
 
@@ -1204,6 +1369,7 @@ int register_pstore_zone(struct pstore_zone_info *info)
 	pr_debug("\toops size : %ld Bytes\n", info->kmsg_size);
 	pr_debug("\tpmsg size : %ld Bytes\n", info->pmsg_size);
 	pr_debug("\tconsole size : %ld Bytes\n", info->console_size);
+	pr_debug("\tftrace size : %ld Bytes\n", info->ftrace_size);
 
 	err = psz_alloc_zones(cxt);
 	if (err) {
@@ -1240,6 +1406,10 @@ int register_pstore_zone(struct pstore_zone_info *info)
 		cxt->pstore.flags |= PSTORE_FLAGS_CONSOLE;
 		pr_cont(" console");
 	}
+	if (info->ftrace_size) {
+		cxt->pstore.flags |= PSTORE_FLAGS_FTRACE;
+		pr_cont(" ftrace");
+	}
 	pr_cont("\n");
 
 	err = pstore_register(&cxt->pstore);
diff --git a/include/linux/pstore_zone.h b/include/linux/pstore_zone.h
index da294e6d7661..94f441b8b616 100644
--- a/include/linux/pstore_zone.h
+++ b/include/linux/pstore_zone.h
@@ -19,6 +19,7 @@
  * @max_reason: Maximum kmsg dump reason to store.
  * @pmsg_size:	The size of pmsg zone which is the same as @kmsg_size.
  * @console_size:The size of console zone which is the same as @kmsg_size.
+ * @ftrace_size:The size of ftrace zone which is the same as @kmsg_size.
  * @read:	The general read operation. Both of the function parameters
  *		@size and @offset are relative value to storage.
  *		On success, the number of bytes should be returned, others
@@ -37,6 +38,7 @@ struct pstore_zone_info {
 	int max_reason;
 	unsigned long pmsg_size;
 	unsigned long console_size;
+	unsigned long ftrace_size;
 	psz_read_op read;
 	psz_write_op write;
 	psz_write_op panic_write;
-- 
1.9.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/