Skip to contentSkip to conversation

xfs: new global stats in sysfs

PATCH Superseded billodo 3 participants 13 messages Sep 3, 2015 XFS filesystem

v1 of 10 Superseded Latest is v10 →

lore
Share & tools
lore

billodo

author
#0 lore raw

cover xfs: new global stats in sysfs

Hi all-
Here is a first pass at a series to add new global stats to sysfs. As a part of that, the /proc/fs/xfs/stat file becomes a symlink to the sysfs stats entry. The patch provides the beginnings of the infrastructure for per-fs stats (in addition to global accumulative stats).

Patch 1 handles the bring-up and tear down of xfs/stats directory structure in sysfs when an fs is mounted. The directory contains the stats file and the stats_clear file. The stats file contents mimic those of /proc/fs/xfs/stat. The stats_clear file is empty, and much like the current stat_clear command, handles the zeroing of the stats file when a "1" is echoed to the stats_clear file.

Patch 2 creates the symlink for stats from procfs to sysfs.

Patch 3 removes the now unused portions of procfs for stat.

Comments and questions are welcome.

Thanks-
Bill

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#1 lore raw

PATCH 1/3 xfs: create global stats and stats_clear in sysfs

Currently, xfs global stats are in procfs. This patch introduces (replicates) the global stats in sysfs. Additionally a stats_clear file is introduced in sysfs.

---
 fs/xfs/xfs_stats.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_stats.h |  2 ++
 fs/xfs/xfs_super.c | 17 ++++++++---
 fs/xfs/xfs_sysfs.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_sysfs.h |  1 +
 5 files changed, 177 insertions(+), 4 deletions(-)
fs/xfs/xfs_stats.c +83 -0
index f224038..856cf57 100644
@@ -29,6 +29,89 @@ static int counter_val(int idx)
 	return val;
 }
 
+int xfs_stats_format(char *buf)
+{
+	int		i, j;
+	int len = 0;
+	__uint64_t	xs_xstrat_bytes = 0;
+	__uint64_t	xs_write_bytes = 0;
+	__uint64_t	xs_read_bytes = 0;
+
+	static const struct xstats_entry {
+		char	*desc;
+		int	endpoint;
+	} xstats[] = {
+		{ "extent_alloc",	XFSSTAT_END_EXTENT_ALLOC	},
+		{ "abt",		XFSSTAT_END_ALLOC_BTREE		},
+		{ "blk_map",		XFSSTAT_END_BLOCK_MAPPING	},
+		{ "bmbt",		XFSSTAT_END_BLOCK_MAP_BTREE	},
+		{ "dir",		XFSSTAT_END_DIRECTORY_OPS	},
+		{ "trans",		XFSSTAT_END_TRANSACTIONS	},
+		{ "ig",			XFSSTAT_END_INODE_OPS		},
+		{ "log",		XFSSTAT_END_LOG_OPS		},
+		{ "push_ail",		XFSSTAT_END_TAIL_PUSHING	},
+		{ "xstrat",		XFSSTAT_END_WRITE_CONVERT	},
+		{ "rw",			XFSSTAT_END_READ_WRITE_OPS	},
+		{ "attr",		XFSSTAT_END_ATTRIBUTE_OPS	},
+		{ "icluster",		XFSSTAT_END_INODE_CLUSTER	},
+		{ "vnodes",		XFSSTAT_END_VNODE_OPS		},
+		{ "buf",		XFSSTAT_END_BUF			},
+		{ "abtb2",		XFSSTAT_END_ABTB_V2		},
+		{ "abtc2",		XFSSTAT_END_ABTC_V2		},
+		{ "bmbt2",		XFSSTAT_END_BMBT_V2		},
+		{ "ibt2",		XFSSTAT_END_IBT_V2		},
+		{ "fibt2",		XFSSTAT_END_FIBT_V2		},
+		/* we print both series of quota information together */
+		{ "qm",			XFSSTAT_END_QM			},
+	};
+
+	/* Loop over all stats groups */
+
+	for (i = j = 0; i < ARRAY_SIZE(xstats); i++) {
+		len += snprintf(buf + len, PATH_MAX - len, "%s", xstats[i].desc);
+		/* inner loop does each group */
+		for (; j < xstats[i].endpoint; j++)
+			len += snprintf(buf + len, PATH_MAX - len, " %u", counter_val(j));
+		len += snprintf(buf + len, PATH_MAX - len, "\n");
+	}
+	/* extra precision counters */
+	for_each_possible_cpu(i) {
+		xs_xstrat_bytes += per_cpu(xfsstats, i).xs_xstrat_bytes;
+		xs_write_bytes += per_cpu(xfsstats, i).xs_write_bytes;
+		xs_read_bytes += per_cpu(xfsstats, i).xs_read_bytes;
+	}
+
+	len += snprintf(buf+len, PATH_MAX-len, "xpc %Lu %Lu %Lu\n",
+			xs_xstrat_bytes, xs_write_bytes, xs_read_bytes);
+	len += snprintf(buf+len, PATH_MAX-len, "debug %u\n",
+#if defined(DEBUG)
+		1);
+#else
+		0);
+#endif
+
+return len;
+}
+
+int xfs_stats_clearall(const char *buf)
+{
+	int		c, n;
+	__uint32_t	vn_active;
+
+	n = sizeof(struct xfsstats);
+	xfs_notice(NULL, "Clearing xfsstats");
+	for_each_possible_cpu(c) {
+		preempt_disable();
+		/* save vn_active, it's a universal truth! */
+		vn_active = per_cpu(xfsstats, c).vn_active;
+		memset(&per_cpu(xfsstats, c), 0,
+		       sizeof(struct xfsstats));
+		per_cpu(xfsstats, c).vn_active = vn_active;
+		preempt_enable();
+	}
+	return n;
+}
+
 static int xfs_stat_proc_show(struct seq_file *m, void *v)
 {
 	int		i, j;
fs/xfs/xfs_stats.h +2 -0
index c8f238b..c117afc 100644
@@ -18,6 +18,8 @@
 #ifndef __XFS_STATS_H__
 #define __XFS_STATS_H__
 
+int xfs_stats_format(char *buf);
+int xfs_stats_clearall(const char *buf);
 
 #if defined(CONFIG_PROC_FS) && !defined(XFS_STATS_OFF)
 
fs/xfs/xfs_super.c +13 -4
index 3bf503a..9a794cf 100644
@@ -61,6 +61,7 @@ static kmem_zone_t *xfs_ioend_zone;
 mempool_t *xfs_ioend_pool;
 
 static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
+static struct xfs_kobj xfs_stats_kobj;	/* global stats sysfs attrs */
 #ifdef DEBUG
 static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
 #endif
@@ -1838,15 +1839,20 @@ init_xfs_fs(void)
 	xfs_kset = kset_create_and_add("xfs", NULL, fs_kobj);
 	if (!xfs_kset) {
 		error = -ENOMEM;
-		goto out_sysctl_unregister;;
+		goto out_sysctl_unregister;
 	}
 
+	xfs_stats_kobj.kobject.kset = xfs_kset;
+	error = xfs_sysfs_init(&xfs_stats_kobj, &xfs_stats_ktype, NULL, "stats");
+	if (error)
+		goto out_kset_unregister;
+
 #ifdef DEBUG
 	xfs_dbg_kobj.kobject.kset = xfs_kset;
 	error = xfs_sysfs_init(&xfs_dbg_kobj, &xfs_dbg_ktype, NULL, "debug");
-	if (error)
-		goto out_kset_unregister;
 #endif
+	if (error)
+		goto out_remove_stats_kobj;
 
 	error = xfs_qm_init();
 	if (error)
@@ -1862,8 +1868,10 @@ init_xfs_fs(void)
  out_remove_kobj:
 #ifdef DEBUG
 	xfs_sysfs_del(&xfs_dbg_kobj);
- out_kset_unregister:
 #endif
+ out_remove_stats_kobj:
+	xfs_sysfs_del(&xfs_stats_kobj);
+ out_kset_unregister:
 	kset_unregister(xfs_kset);
  out_sysctl_unregister:
 	xfs_sysctl_unregister();
@@ -1889,6 +1897,7 @@ exit_xfs_fs(void)
 #ifdef DEBUG
 	xfs_sysfs_del(&xfs_dbg_kobj);
 #endif
+	xfs_sysfs_del(&xfs_stats_kobj);
 	kset_unregister(xfs_kset);
 	xfs_sysctl_unregister();
 	xfs_cleanup_procfs();
fs/xfs/xfs_sysfs.c +78 -0

90 lines not on this page · raw message

fs/xfs/xfs_sysfs.h +1 -0
index 240eee3..be692e5 100644
@@ -22,6 +22,7 @@
 extern struct kobj_type xfs_mp_ktype;	/* xfs_mount */
 extern struct kobj_type xfs_dbg_ktype;	/* debug */
 extern struct kobj_type xfs_log_ktype;	/* xlog */
+extern struct kobj_type xfs_stats_ktype;	/* stats */
 
 static inline struct xfs_kobj *
 to_kobj(struct kobject *kobject)
Signature · 6 lines
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#2 lore raw

Re: [PATCH 1/3] xfs: create global stats and stats_clear in sysfs

Quoted 3 lines · billodo

Currently, xfs global stats are in procfs. This patch introduces (replicates) the global stats in sysfs. Additionally a stats_clear file is introduced in sysfs.

Looks pretty good, a few things below.
(FWIW, lots of these came from running scripts/checkpatch.pl against the patch - not everything it says is mandatory, but worth doing to catch some things)

Quoted 47 lines
---
 fs/xfs/xfs_stats.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_stats.h |  2 ++
 fs/xfs/xfs_super.c | 17 ++++++++---
 fs/xfs/xfs_sysfs.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_sysfs.h |  1 +
 5 files changed, 177 insertions(+), 4 deletions(-)
fs/xfs/xfs_stats.c +30 -0
index f224038..856cf57 100644
@@ -29,6 +29,89 @@ static int counter_val(int idx)
 	return val;
 }
 
+int xfs_stats_format(char *buf)
+{
+	int		i, j;
+	int len = 0;
+	__uint64_t	xs_xstrat_bytes = 0;
+	__uint64_t	xs_write_bytes = 0;
+	__uint64_t	xs_read_bytes = 0;
+
+	static const struct xstats_entry {
+		char	*desc;
+		int	endpoint;
+	} xstats[] = {
+		{ "extent_alloc",	XFSSTAT_END_EXTENT_ALLOC	},
+		{ "abt",		XFSSTAT_END_ALLOC_BTREE		},
+		{ "blk_map",		XFSSTAT_END_BLOCK_MAPPING	},
+		{ "bmbt",		XFSSTAT_END_BLOCK_MAP_BTREE	},
+		{ "dir",		XFSSTAT_END_DIRECTORY_OPS	},
+		{ "trans",		XFSSTAT_END_TRANSACTIONS	},
+		{ "ig",			XFSSTAT_END_INODE_OPS		},
+		{ "log",		XFSSTAT_END_LOG_OPS		},
+		{ "push_ail",		XFSSTAT_END_TAIL_PUSHING	},
+		{ "xstrat",		XFSSTAT_END_WRITE_CONVERT	},
+		{ "rw",			XFSSTAT_END_READ_WRITE_OPS	},
+		{ "attr",		XFSSTAT_END_ATTRIBUTE_OPS	},
+		{ "icluster",		XFSSTAT_END_INODE_CLUSTER	},
+		{ "vnodes",		XFSSTAT_END_VNODE_OPS		},
+		{ "buf",		XFSSTAT_END_BUF			},
+		{ "abtb2",		XFSSTAT_END_ABTB_V2		},
+		{ "abtc2",		XFSSTAT_END_ABTC_V2		},
+		{ "bmbt2",		XFSSTAT_END_BMBT_V2		},

fs/xfs/xfs_stats.c static int counter_val(int idx) · line 71

Quoted 10 lines
fs/xfs/xfs_stats.c +10 -0
+		{ "ibt2",		XFSSTAT_END_IBT_V2		},
+		{ "fibt2",		XFSSTAT_END_FIBT_V2		},
+		/* we print both series of quota information together */
+		{ "qm",			XFSSTAT_END_QM			},
+	};
+
+	/* Loop over all stats groups */
+
+	for (i = j = 0; i < ARRAY_SIZE(xstats); i++) {
+		len += snprintf(buf + len, PATH_MAX - len, "%s", xstats[i].desc);

This line is > 80 chars, might wrap the last arg onto the next line

fs/xfs/xfs_stats.c static int counter_val(int idx) · line 74

Quoted 3 lines
fs/xfs/xfs_stats.c +3 -0
+		/* inner loop does each group */
+		for (; j < xstats[i].endpoint; j++)
+			len += snprintf(buf + len, PATH_MAX - len, " %u", counter_val(j));

ditto

Quoted 12 lines
fs/xfs/xfs_stats.c +12 -0
+		len += snprintf(buf + len, PATH_MAX - len, "\n");
+	}
+	/* extra precision counters */
+	for_each_possible_cpu(i) {
+		xs_xstrat_bytes += per_cpu(xfsstats, i).xs_xstrat_bytes;
+		xs_write_bytes += per_cpu(xfsstats, i).xs_write_bytes;
+		xs_read_bytes += per_cpu(xfsstats, i).xs_read_bytes;
+	}
+
+	len += snprintf(buf+len, PATH_MAX-len, "xpc %Lu %Lu %Lu\n",
+			xs_xstrat_bytes, xs_write_bytes, xs_read_bytes);
+	len += snprintf(buf+len, PATH_MAX-len, "debug %u\n",

fs/xfs/xfs_stats.c static int counter_val(int idx) · line 96

Quoted 10 lines
fs/xfs/xfs_stats.c +10 -0
+#if defined(DEBUG)
+		1);
+#else
+		0);
+#endif
+
+return len;
+}
+
+int xfs_stats_clearall(const char *buf)

given that *buf isn't used, I'd make this a (void)

And returning the size of the struct isn't correct, and nothing can really fail, so may as well make return void too,

void xfs_stats_clearall(void)

And finally, given that the sysctl handler for clearing stats is still around, and this dupicates that code, may as well make the sysctl handler just call this function, and eliminate the duplicate code there.

Quoted 47 lines
fs/xfs/xfs_stats.c +18 -0
+{
+	int		c, n;
+	__uint32_t	vn_active;
+
+	n = sizeof(struct xfsstats);
+	xfs_notice(NULL, "Clearing xfsstats");
+	for_each_possible_cpu(c) {
+		preempt_disable();
+		/* save vn_active, it's a universal truth! */
+		vn_active = per_cpu(xfsstats, c).vn_active;
+		memset(&per_cpu(xfsstats, c), 0,
+		       sizeof(struct xfsstats));
+		per_cpu(xfsstats, c).vn_active = vn_active;
+		preempt_enable();
+	}
+	return n;
+}
+
 static int xfs_stat_proc_show(struct seq_file *m, void *v)
 {
 	int		i, j;
fs/xfs/xfs_stats.h +2 -0
index c8f238b..c117afc 100644
@@ -18,6 +18,8 @@
 #ifndef __XFS_STATS_H__
 #define __XFS_STATS_H__
 
+int xfs_stats_format(char *buf);
+int xfs_stats_clearall(const char *buf);
 
 #if defined(CONFIG_PROC_FS) && !defined(XFS_STATS_OFF)
 
fs/xfs/xfs_super.c +1 -0
index 3bf503a..9a794cf 100644
@@ -61,6 +61,7 @@ static kmem_zone_t *xfs_ioend_zone;
 mempool_t *xfs_ioend_pool;
 
 static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
+static struct xfs_kobj xfs_stats_kobj;	/* global stats sysfs attrs */
 #ifdef DEBUG
 static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
 #endif

fs/xfs/xfs_super.c init_xfs_fs(void) · line 1846

Quoted 9 lines
fs/xfs/xfs_super.c +3 -1
@@ -1838,15 +1839,20 @@ init_xfs_fs(void)
 	xfs_kset = kset_create_and_add("xfs", NULL, fs_kobj);
 	if (!xfs_kset) {
 		error = -ENOMEM;
-		goto out_sysctl_unregister;;
+		goto out_sysctl_unregister;
 	}
 
+	xfs_stats_kobj.kobject.kset = xfs_kset;
+	error = xfs_sysfs_init(&xfs_stats_kobj, &xfs_stats_ktype, NULL, "stats");

over 80 chars

fs/xfs/xfs_super.c init_xfs_fs(void) · line 1855

Quoted 11 lines
fs/xfs/xfs_super.c +5 -2
+	if (error)
+		goto out_kset_unregister;
+
 #ifdef DEBUG
 	xfs_dbg_kobj.kobject.kset = xfs_kset;
 	error = xfs_sysfs_init(&xfs_dbg_kobj, &xfs_dbg_ktype, NULL, "debug");
-	if (error)
-		goto out_kset_unregister;
 #endif
+	if (error)
+		goto out_remove_stats_kobj;

this needs to stay inside the #ifdef, otherwise you have two

if (error)
goto BLAH;'s

in a row if DEBUG is off...

fs/xfs/xfs_super.c init_xfs_fs(void) · line 1868

Quoted 5 lines
fs/xfs/xfs_super.c +0 -0
 	error = xfs_qm_init();
 	if (error)
@@ -1862,8 +1868,10 @@ init_xfs_fs(void)
  out_remove_kobj:

I'd make this "out_remove_dbg_kobj" now that there are 2.

fs/xfs/xfs_super.c init_xfs_fs(void) · line 1872

Quoted 5 lines
fs/xfs/xfs_super.c +1 -1
 #ifdef DEBUG
 	xfs_sysfs_del(&xfs_dbg_kobj);
- out_kset_unregister:
 #endif
+ out_remove_stats_kobj:

pretty sure this still needs to be inside the #ifdef too; think it through with and without DEBUG defined, make sure it does the right thing ...

Quoted 38 lines
fs/xfs/xfs_super.c +3 -0
+	xfs_sysfs_del(&xfs_stats_kobj);
+ out_kset_unregister:
 	kset_unregister(xfs_kset);
  out_sysctl_unregister:
 	xfs_sysctl_unregister();
@@ -1889,6 +1897,7 @@ exit_xfs_fs(void)
 #ifdef DEBUG
 	xfs_sysfs_del(&xfs_dbg_kobj);
 #endif
+	xfs_sysfs_del(&xfs_stats_kobj);
 	kset_unregister(xfs_kset);
 	xfs_sysctl_unregister();
 	xfs_cleanup_procfs();
fs/xfs/xfs_sysfs.c +10 -0
index aa03670..ba8d097 100644
@@ -21,6 +21,7 @@
 #include "xfs_log_format.h"
 #include "xfs_log.h"
 #include "xfs_log_priv.h"
+#include "xfs_stats.h"
 
 struct xfs_sysfs_attr {
 	struct attribute attr;
@@ -125,6 +126,83 @@ struct kobj_type xfs_dbg_ktype = {
 
 #endif /* DEBUG */
 
+
+/* stats */
+
+STATIC ssize_t
+stats_show(
+	char	*buf,
+	void	*data)
+{
+	return xfs_stats_format(buf);

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 147

Quoted 10 lines
fs/xfs/xfs_sysfs.c +10 -0
+}
+XFS_SYSFS_ATTR_RO(stats);
+
+STATIC ssize_t
+stats_clear_show(
+	char	*buf,
+	void	*data)
+{
+	return 0;
+}

Ok, this means that if we try to read it we get nothing:

# cat /sys/fs/xfs/stats/stats_clear 
#

I think that's ok. Oh, but see below; we can declare it write-only, then it disappears!

Quoted 6 lines
fs/xfs/xfs_sysfs.c +6 -0
+STATIC ssize_t
+stats_clear_store(
+	const char	*buf,
+	size_t		count,
+	void		*data)
+{

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 164

Quoted 10 lines
fs/xfs/xfs_sysfs.c +10 -0
+	int		ret;
+	int		val;
+
+	ret = kstrtoint(buf, 0, &val);
+	if (ret)
+		return ret;
+
+	if (val != 1)
+		return -EINVAL;
+	return xfs_stats_clearall(buf);

Hm, I think this is supposed to return the number of bytes stored in the file, which is not what the clearall function does. Given that you sanitize the input and return an error if not sane, then by the time we get to something sane, just return "count" - the amt of data we got in. That will make writes to this file return the proper bytes written. i.e.

	xfs_stats_clearall();

	return count;

(and yeah, it doesn't need "buf")

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 165

Quoted 2 lines
fs/xfs/xfs_sysfs.c +2 -0
+}
+XFS_SYSFS_ATTR_RW(stats_clear);

Ooh, I just realized that there is an __ATTR_WO in core kernel code; using that in a new XFS_SYSFS_ATTR_WO would make sense here. Then you can drop the stats_clear_show I think.

Quoted 4 lines
fs/xfs/xfs_sysfs.c +4 -0
+
+static struct attribute *xfs_stats_attrs[] = {
+	ATTR_LIST(stats),
+	ATTR_LIST(stats_clear),

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 180

Quoted 10 lines
fs/xfs/xfs_sysfs.c +10 -0
+	NULL,
+};
+
+STATIC ssize_t
+xfs_stats_show(
+	struct kobject		*kobject,
+	struct attribute	*attr,
+	char			*buf)
+{
+	struct xfs_sysfs_attr *xfs_attr = to_attr(attr);

Add a blank line after the var declaration

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 191

Quoted 11 lines
fs/xfs/xfs_sysfs.c +11 -0
+	return xfs_attr->show ? xfs_attr->show(buf, NULL) : 0;
+}
+
+STATIC ssize_t
+xfs_stats_store(
+       struct kobject          *kobject,
+       struct attribute        *attr,
+       const char              *buf,
+       size_t                  count)
+{
+       struct xfs_sysfs_attr *xfs_attr = to_attr(attr);

Add a blank line after the var declaration

fs/xfs/xfs_sysfs.c struct kobj_type xfs_dbg_ktype = { · line 193

Quoted 2 lines
fs/xfs/xfs_sysfs.c +2 -0
+       return xfs_attr->store ? xfs_attr->store(buf, count, NULL) : 0;
+}

covert all those 8-spaces above to tabs, please

Quoted 21 lines
fs/xfs/xfs_sysfs.c +12 -0
+
+static struct sysfs_ops xfs_stats_ops = {
+	.show = xfs_stats_show,
+	.store = xfs_stats_store,
+};
+
+struct kobj_type xfs_stats_ktype = {
+	.release = xfs_sysfs_release,
+	.sysfs_ops = &xfs_stats_ops,
+	.default_attrs = xfs_stats_attrs,
+};
+
 /* xlog */
 
 STATIC ssize_t
fs/xfs/xfs_sysfs.h +0 -0
index 240eee3..be692e5 100644

fs/xfs/xfs_sysfs.h line 28

Quoted 7 lines
fs/xfs/xfs_sysfs.h +1 -0
@@ -22,6 +22,7 @@
 extern struct kobj_type xfs_mp_ktype;	/* xfs_mount */
 extern struct kobj_type xfs_dbg_ktype;	/* debug */
 extern struct kobj_type xfs_log_ktype;	/* xlog */
+extern struct kobj_type xfs_stats_ktype;	/* stats */
 
 static inline struct xfs_kobj *
 to_kobj(struct kobject *kobject)

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#3 lore raw

Re: [PATCH 1/3] xfs: create global stats and stats_clear in sysfs

In reply to billodo
Quoted 18 lines · billodo

Currently, xfs global stats are in procfs. This patch introduces (replicates) the global stats in sysfs. Additionally a stats_clear file is introduced in sysfs.

---
 fs/xfs/xfs_stats.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_stats.h |  2 ++
 fs/xfs/xfs_super.c | 17 ++++++++---
 fs/xfs/xfs_sysfs.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_sysfs.h |  1 +
 5 files changed, 177 insertions(+), 4 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index f224038..856cf57 100644

fs/xfs/xfs_stats.c static int counter_val(int idx) · line 35

Quoted 7 lines
fs/xfs/xfs_stats.c +4 -0
@@ -29,6 +29,89 @@ static int counter_val(int idx)
 	return val;
 }
 
+int xfs_stats_format(char *buf)
+{
+	int		i, j;
+	int len = 0;

Oh, one other little nitpick, tab over "len" to match the rest.

fs/xfs/xfs_stats.c static int counter_val(int idx) · line 38

Quoted 3 lines
fs/xfs/xfs_stats.c +3 -0
+	__uint64_t	xs_xstrat_bytes = 0;
+	__uint64_t	xs_write_bytes = 0;
+	__uint64_t	xs_read_bytes = 0;

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#4 lore raw

PATCH 2/3 xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

In reply to billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +6 -2
index 856cf57..ad435f1 100644
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 6 lines
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#5 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

Quoted 13 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 248

Quoted 7 lines
fs/xfs/xfs_stats.c +2 -2
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))

Uh.... is it actually guaranteed that sysfs is mounted on /sys now?

I sort of recall gregkh grumbling years ago that sysfs can be mounted anywhere, and that /proc shouldn't hardcode links to it. But that's just handwaving on my part.

--D

Quoted 15 lines
fs/xfs/xfs_stats.c +4 -0
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 6 lines
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#6 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

Quoted 24 lines · Darrick J. Wong
Quoted 13 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 248

Quoted 7 lines
fs/xfs/xfs_stats.c +2 -2
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))

Uh.... is it actually guaranteed that sysfs is mounted on /sys now?

Quoted 3 lines

I sort of recall gregkh grumbling years ago that sysfs can be mounted anywhere, and that /proc shouldn't hardcode links to it. But that's just handwaving on my part.

You can blame me for that idea. At least one other driver does do it, though; of_core_init():

proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");

worst-case scenario, your "legacy" stats file will be a broken symlink...

-Eric

Quoted 12 lines

--D

Quoted 10 lines
fs/xfs/xfs_stats.c +4 -0
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 1 line
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#7 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

Quoted 29 lines · Eric Sandeen
Quoted 24 lines · Darrick J. Wong
Quoted 13 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 248

Quoted 7 lines
fs/xfs/xfs_stats.c +2 -2
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))

Uh.... is it actually guaranteed that sysfs is mounted on /sys now?

Quoted 3 lines

I sort of recall gregkh grumbling years ago that sysfs can be mounted anywhere, and that /proc shouldn't hardcode links to it. But that's just handwaving on my part.

Quoted 9 lines

You can blame me for that idea. At least one other driver does do it, though; of_core_init():

proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");

worst-case scenario, your "legacy" stats file will be a broken symlink...

-Eric

I'm still looking for something in documentation that dictates such a requirement regarding symlinks to sysfs elements.
-Bill

Quoted 13 lines
Quoted 12 lines

--D

Quoted 10 lines
fs/xfs/xfs_stats.c +4 -0
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 1 line
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#8 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

Quoted 35 lines · Bill O'Donnell
Quoted 34 lines · Eric Sandeen
Quoted 24 lines · Darrick J. Wong
Quoted 13 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 248

Quoted 7 lines
fs/xfs/xfs_stats.c +2 -2
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))

Uh.... is it actually guaranteed that sysfs is mounted on /sys now?

Quoted 3 lines

I sort of recall gregkh grumbling years ago that sysfs can be mounted anywhere, and that /proc shouldn't hardcode links to it. But that's just handwaving on my part.

You can blame me for that idea. At least one other driver does do it, though; of_core_init():

proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");

Quoted 8 lines
Quoted 4 lines

worst-case scenario, your "legacy" stats file will be a broken symlink...

-Eric

I'm still looking for something in documentation that dictates such a requirement regarding symlinks to sysfs elements.
-Bill

I did find this in: https://www.kernel.org/doc/Documentation/sysfs-rules.txt
--------------snip--------------------------
- sysfs is always at /sys
Parsing /proc/mounts is a waste of time. Other mount points are a system configuration bug you should not try to solve. For test cases, possibly support a SYSFS_PATH environment variable to overwrite the application's behavior, but never try to search for sysfs. Never try to mount it, if you are not an early boot script.
-------------snip---------------------------

Quoted 15 lines
Quoted 14 lines
Quoted 13 lines

--D

Quoted 10 lines
fs/xfs/xfs_stats.c +4 -0
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 1 line
2.4.3
Quoted 4 lines

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#9 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

Quoted 45 lines · Bill O'Donnell
Quoted 35 lines · Bill O'Donnell
Quoted 34 lines · Eric Sandeen
Quoted 24 lines · Darrick J. Wong
Quoted 13 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 248

Quoted 7 lines
fs/xfs/xfs_stats.c +2 -2
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))

Uh.... is it actually guaranteed that sysfs is mounted on /sys now?

Quoted 3 lines

I sort of recall gregkh grumbling years ago that sysfs can be mounted anywhere, and that /proc shouldn't hardcode links to it. But that's just handwaving on my part.

You can blame me for that idea. At least one other driver does do it, though; of_core_init():

proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");

Quoted 8 lines
Quoted 4 lines

worst-case scenario, your "legacy" stats file will be a broken symlink...

-Eric

I'm still looking for something in documentation that dictates such a requirement regarding symlinks to sysfs elements.
-Bill

Quoted 9 lines

I did find this in: https://www.kernel.org/doc/Documentation/sysfs-rules.txt
--------------snip--------------------------
- sysfs is always at /sys
Parsing /proc/mounts is a waste of time. Other mount points are a system configuration bug you should not try to solve. For test cases, possibly support a SYSFS_PATH environment variable to overwrite the application's behavior, but never try to search for sysfs. Never try to mount it, if you are not an early boot script.
-------------snip---------------------------

Ah, ok, sorry for the noise then.

--D

Quoted 20 lines
Quoted 15 lines
Quoted 14 lines
Quoted 13 lines

--D

Quoted 10 lines
fs/xfs/xfs_stats.c +4 -0
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))
Signature · 1 line
2.4.3
Quoted 4 lines

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

Quoted 4 lines

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#10 lore raw

Re: [PATCH 2/3] xfs: create symlink proc/fs/xfs/stat to sys/fs/xfs/stats

In reply to billodo
Quoted 15 lines · billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch creates the symlink from proc/fs/xfs/stat to sys/fs/xfs/stats.

---
 fs/xfs/xfs_stats.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/xfs/xfs_stats.c +0 -0
index 856cf57..ad435f1 100644
@@ -244,9 +244,13 @@ xfs_init_procfs(void)
 	if (!proc_mkdir("fs/xfs", NULL))
 		goto out;

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 253

Quoted 10 lines
fs/xfs/xfs_stats.c +6 -2
 
-	if (!proc_create("fs/xfs/stat", 0, NULL,
-			 &xfs_stat_proc_fops))
+	if (!proc_symlink("fs/xfs/stat", NULL,
+			  "/sys/fs/xfs/stats/stats"))
+	{
+		printk(KERN_INFO "failed to created fs/xfs/stat symlink\n");
 		goto out_remove_xfs_dir;
+	}
+

I'm not sure we need the printk, we didn't say anything if proc_create(fs/xfs/stat) failed...

I guess the cleanup is fine as it is, right, it can remove a symlink too AFAIK.

-Eric

fs/xfs/xfs_stats.c xfs_init_procfs(void) · line 256

Quoted 4 lines
fs/xfs/xfs_stats.c +0 -0
 #ifdef CONFIG_XFS_QUOTA
 	if (!proc_create("fs/xfs/xqmstat", 0, NULL,
 			 &xqmstat_proc_fops))

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#11 lore raw

PATCH 3/3 xfs: remove unused procfs code

In reply to billodo

As a part of the work to move xfs global stats from procfs to sysfs, this patch removes the now unused procfs code that was xfs stat specific.

---
 fs/xfs/xfs_stats.c | 74 ------------------------------------------------------
 1 file changed, 74 deletions(-)
fs/xfs/xfs_stats.c +0 -74
index ad435f1..9f20d3e 100644
@@ -112,80 +112,6 @@ int xfs_stats_clearall(const char *buf)
 	return n;
 }
 
-static int xfs_stat_proc_show(struct seq_file *m, void *v)
-{
-	int		i, j;
-	__uint64_t	xs_xstrat_bytes = 0;
-	__uint64_t	xs_write_bytes = 0;
-	__uint64_t	xs_read_bytes = 0;
-
-	static const struct xstats_entry {
-		char	*desc;
-		int	endpoint;
-	} xstats[] = {
-		{ "extent_alloc",	XFSSTAT_END_EXTENT_ALLOC	},
-		{ "abt",		XFSSTAT_END_ALLOC_BTREE		},
-		{ "blk_map",		XFSSTAT_END_BLOCK_MAPPING	},
-		{ "bmbt",		XFSSTAT_END_BLOCK_MAP_BTREE	},
-		{ "dir",		XFSSTAT_END_DIRECTORY_OPS	},
-		{ "trans",		XFSSTAT_END_TRANSACTIONS	},
-		{ "ig",			XFSSTAT_END_INODE_OPS		},
-		{ "log",		XFSSTAT_END_LOG_OPS		},
-		{ "push_ail",		XFSSTAT_END_TAIL_PUSHING	},
-		{ "xstrat",		XFSSTAT_END_WRITE_CONVERT	},
-		{ "rw",			XFSSTAT_END_READ_WRITE_OPS	},
-		{ "attr",		XFSSTAT_END_ATTRIBUTE_OPS	},
-		{ "icluster",		XFSSTAT_END_INODE_CLUSTER	},
-		{ "vnodes",		XFSSTAT_END_VNODE_OPS		},
-		{ "buf",		XFSSTAT_END_BUF			},
-		{ "abtb2",		XFSSTAT_END_ABTB_V2		},
-		{ "abtc2",		XFSSTAT_END_ABTC_V2		},
-		{ "bmbt2",		XFSSTAT_END_BMBT_V2		},
-		{ "ibt2",		XFSSTAT_END_IBT_V2		},
-		{ "fibt2",		XFSSTAT_END_FIBT_V2		},
-		/* we print both series of quota information together */
-		{ "qm",			XFSSTAT_END_QM			},
-	};
-
-	/* Loop over all stats groups */
-	for (i = j = 0; i < ARRAY_SIZE(xstats); i++) {
-		seq_printf(m, "%s", xstats[i].desc);
-		/* inner loop does each group */
-		for (; j < xstats[i].endpoint; j++)
-			seq_printf(m, " %u", counter_val(j));
-		seq_putc(m, '\n');
-	}
-	/* extra precision counters */
-	for_each_possible_cpu(i) {
-		xs_xstrat_bytes += per_cpu(xfsstats, i).xs_xstrat_bytes;
-		xs_write_bytes += per_cpu(xfsstats, i).xs_write_bytes;
-		xs_read_bytes += per_cpu(xfsstats, i).xs_read_bytes;
-	}
-
-	seq_printf(m, "xpc %Lu %Lu %Lu\n",
-			xs_xstrat_bytes, xs_write_bytes, xs_read_bytes);
-	seq_printf(m, "debug %u\n",
-#if defined(DEBUG)
-		1);
-#else
-		0);
-#endif
-	return 0;
-}
-
-static int xfs_stat_proc_open(struct inode *inode, struct file *file)
-{
-	return single_open(file, xfs_stat_proc_show, NULL);
-}
-
-static const struct file_operations xfs_stat_proc_fops = {
-	.owner		= THIS_MODULE,
-	.open		= xfs_stat_proc_open,
-	.read		= seq_read,
-	.llseek		= seq_lseek,
-	.release	= single_release,
-};
-
 /* legacy quota interfaces */
 #ifdef CONFIG_XFS_QUOTA
 static int xqm_proc_show(struct seq_file *m, void *v)
Signature · 6 lines
2.4.3

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#12 lore raw
In reply to billodo
Quoted 7 lines · billodo

Hi all-
Here is a first pass at a series to add new global stats to sysfs. As a part of that, the /proc/fs/xfs/stat file becomes a symlink to the sysfs stats entry. The patch provides the beginnings of the infrastructure for per-fs stats (in addition to global accumulative stats).

To flesh this out a little more:

We already have per-fs information in /sys, so it makes sense to have per-fs stats there too. As a first step, moving existing global stats infrastructure to /sys will allow us to re-use that sysfs code for per-fs stats as well.

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs