Skip to contentSkip to conversation

xfs: load uncached unlinked inodes into memory on demand

PATCH Superseded Darrick J. Wong 5 participants 14 messages Aug 30, 2023 – Sep 1, 2023 XFS filesystem

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 xfs: load uncached unlinked inodes into memory on demand

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +25 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),
Reviewed-by on v2 1/1
#1 lore raw
Quoted 38 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

Quoted 7 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)

This version looks good.

Signature · 2 lines
Dave Chinner
[email protected]
#2 lore raw

PATCH 1/2 xfs_db: dump unlinked buckets

In reply to Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Create a new command to dump the resource usage of files in the unlinked buckets.

---
 db/Makefile              |    2 
 db/command.c             |    1 
 db/command.h             |    1 
 db/iunlink.c             |  204 ++++++++++++++++++++++++++++++++++++++++++++++
 libxfs/libxfs_api_defs.h |    1 
 man/man8/xfs_db.8        |   19 ++++
 6 files changed, 227 insertions(+), 1 deletion(-)
 create mode 100644 db/iunlink.c
db/Makefile +1 -1
index 2f95f670..d00801ab 100644
@@ -14,7 +14,7 @@ HFILES = addr.h agf.h agfl.h agi.h attr.h attrshort.h bit.h block.h bmap.h \
 	io.h logformat.h malloc.h metadump.h output.h print.h quit.h sb.h \
 	sig.h strvec.h text.h type.h write.h attrset.h symlink.h fsmap.h \
 	fuzz.h obfuscate.h
-CFILES = $(HFILES:.h=.c) btdump.c btheight.c convert.c info.c namei.c \
+CFILES = $(HFILES:.h=.c) btdump.c btheight.c convert.c info.c iunlink.c namei.c \
 	timelimit.c
 LSRCFILES = xfs_admin.sh xfs_ncheck.sh xfs_metadump.sh
 
db/command.c +1 -0
index 02f778b9..b4021c86 100644
@@ -127,6 +127,7 @@ init_commands(void)
 	info_init();
 	inode_init();
 	input_init();
+	iunlink_init();
 	logres_init();
 	logformat_init();
 	io_init();
db/command.h +1 -0
index 498983ff..a89e7150 100644
@@ -34,3 +34,4 @@ extern void		info_init(void);
 extern void		btheight_init(void);
 extern void		timelimit_init(void);
 extern void		namei_init(void);
+extern void		iunlink_init(void);
db/iunlink.c new file +204 -0

204 lines not on this page · raw message

libxfs/libxfs_api_defs.h +1 -0
index 026aa510..ddba5c7c 100644
@@ -125,6 +125,7 @@
 #define xfs_idestroy_fork		libxfs_idestroy_fork
 #define xfs_iext_lookup_extent		libxfs_iext_lookup_extent
 #define xfs_ifork_zap_attr		libxfs_ifork_zap_attr
+#define xfs_imap_to_bp			libxfs_imap_to_bp
 #define xfs_initialize_perag		libxfs_initialize_perag
 #define xfs_initialize_perag_data	libxfs_initialize_perag_data
 #define xfs_init_local_fork		libxfs_init_local_fork
man/man8/xfs_db.8 +19 -0
index 60dcdc52..2d6d0da4 100644
@@ -579,6 +579,25 @@ print the current debug option bits. These are for the use of the implementor.
 .BI "dquot [" \-g | \-p | \-u ] " id"
 Set current address to a group, project or user quota block for the given ID. Defaults to user quota.
 .TP
+.BI "dump_iunlinked [-a " agno " ] [-b " bucket " ] [-q] [-v]"
+Dump the contents of unlinked buckets.
+
+Options include:
+.RS 1.0i
+.TP 0.4i
+.B \-a
+Print only this AG's unlinked buckets.
+.TP 0.4i
+.B \-b
+Print only this bucket within each AGI.
+.TP 0.4i
+.B \-q
+Only print the essentials.
+.TP 0.4i
+.B \-v
+Print resource usage of each file on the unlinked lists.
+.RE
+.TP
 .BI "echo [" arg "] ..."
 Echo the arguments to the output.
 .TP
Reviewed-by on v1 1/2
#3 lore raw

Re: [PATCH 1/2] xfs_db: dump unlinked buckets

Quoted 6 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Create a new command to dump the resource usage of files in the unlinked buckets.

Quoted 297 lines
---
 db/Makefile              |    2 
 db/command.c             |    1 
 db/command.h             |    1 
 db/iunlink.c             |  204 ++++++++++++++++++++++++++++++++++++++++++++++
 libxfs/libxfs_api_defs.h |    1 
 man/man8/xfs_db.8        |   19 ++++
 6 files changed, 227 insertions(+), 1 deletion(-)
 create mode 100644 db/iunlink.c
db/Makefile +1 -1
index 2f95f670..d00801ab 100644
@@ -14,7 +14,7 @@ HFILES = addr.h agf.h agfl.h agi.h attr.h attrshort.h bit.h block.h bmap.h \
 	io.h logformat.h malloc.h metadump.h output.h print.h quit.h sb.h \
 	sig.h strvec.h text.h type.h write.h attrset.h symlink.h fsmap.h \
 	fuzz.h obfuscate.h
-CFILES = $(HFILES:.h=.c) btdump.c btheight.c convert.c info.c namei.c \
+CFILES = $(HFILES:.h=.c) btdump.c btheight.c convert.c info.c iunlink.c namei.c \
 	timelimit.c
 LSRCFILES = xfs_admin.sh xfs_ncheck.sh xfs_metadump.sh
 
db/command.c +1 -0
index 02f778b9..b4021c86 100644
@@ -127,6 +127,7 @@ init_commands(void)
 	info_init();
 	inode_init();
 	input_init();
+	iunlink_init();
 	logres_init();
 	logformat_init();
 	io_init();
db/command.h +1 -0
index 498983ff..a89e7150 100644
@@ -34,3 +34,4 @@ extern void		info_init(void);
 extern void		btheight_init(void);
 extern void		timelimit_init(void);
 extern void		namei_init(void);
+extern void		iunlink_init(void);
db/iunlink.c new file +204 -0
new file mode 100644
index 00000000..303b5daf
@@ -0,0 +1,204 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2022-2023 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <[email protected]>
+ */
+#include "libxfs.h"
+#include "command.h"
+#include "output.h"
+#include "init.h"
+
+static xfs_filblks_t
+count_rtblocks(
+	struct xfs_inode	*ip)
+{
+	struct xfs_iext_cursor	icur;
+	struct xfs_bmbt_irec	got;
+	xfs_filblks_t		count = 0;
+	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, XFS_DATA_FORK);
+	int			error;
+
+	error = -libxfs_iread_extents(NULL, ip, XFS_DATA_FORK);
+	if (error) {
+		dbprintf(
+_("could not read AG %u agino %u extents, err=%d\n"),
+				XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino),
+				XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino),
+				error);
+		return 0;
+	}
+
+	for_each_xfs_iext(ifp, &icur, &got)
+		if (!isnullstartblock(got.br_startblock))
+			count += got.br_blockcount;
+	return count;
+}
+
+static xfs_agino_t
+get_next_unlinked(
+	xfs_agnumber_t		agno,
+	xfs_agino_t		agino,
+	bool			verbose)
+{
+	struct xfs_buf		*ino_bp;
+	struct xfs_dinode	*dip;
+	struct xfs_inode	*ip;
+	xfs_ino_t		ino;
+	xfs_agino_t		ret;
+	int			error;
+
+	ino = XFS_AGINO_TO_INO(mp, agno, agino);
+	error = -libxfs_iget(mp, NULL, ino, 0, &ip);
+	if (error)
+		goto bad;
+
+	if (verbose) {
+		xfs_filblks_t	blocks, rtblks = 0;
+
+		if (XFS_IS_REALTIME_INODE(ip))
+			rtblks = count_rtblocks(ip);
+		blocks = ip->i_nblocks - rtblks;
+
+		dbprintf(_(" blocks %llu rtblocks %llu\n"),
+				blocks, rtblks);
+	} else {
+		dbprintf("\n");
+	}
+
+	error = -libxfs_imap_to_bp(mp, NULL, &ip->i_imap, &ino_bp);
+	if (error)
+		goto bad;
+
+	dip = xfs_buf_offset(ino_bp, ip->i_imap.im_boffset);
+	ret = be32_to_cpu(dip->di_next_unlinked);
+	libxfs_buf_relse(ino_bp);
+
+	return ret;
+bad:
+	dbprintf(_("AG %u agino %u: %s\n"), agno, agino, strerror(error));
+	return NULLAGINO;
+}
+
+static void
+dump_unlinked_bucket(
+	xfs_agnumber_t	agno,
+	struct xfs_buf	*agi_bp,
+	unsigned int	bucket,
+	bool		quiet,
+	bool		verbose)
+{
+	struct xfs_agi	*agi = agi_bp->b_addr;
+	xfs_agino_t	agino;
+	unsigned int	i = 0;
+
+	agino = be32_to_cpu(agi->agi_unlinked[bucket]);
+	if (agino != NULLAGINO)
+		dbprintf(_("AG %u bucket %u agino %u"), agno, bucket, agino);
+	else if (!quiet && agino == NULLAGINO)
+		dbprintf(_("AG %u bucket %u agino NULL\n"), agno, bucket);
+
+	while (agino != NULLAGINO) {
+		agino = get_next_unlinked(agno, agino, verbose);
+		if (agino != NULLAGINO)
+			dbprintf(_("    [%u] agino %u"), i++, agino);
+		else if (!quiet && agino == NULLAGINO)
+			dbprintf(_("    [%u] agino NULL\n"), i++);
+	}
+}
+
+static void
+dump_unlinked(
+	struct xfs_perag	*pag,
+	unsigned int		bucket,
+	bool			quiet,
+	bool			verbose)
+{
+	struct xfs_buf		*agi_bp;
+	xfs_agnumber_t		agno = pag->pag_agno;
+	int			error;
+
+	error = -libxfs_ialloc_read_agi(pag, NULL, &agi_bp);
+	if (error) {
+		dbprintf(_("AGI %u: %s\n"), agno, strerror(errno));
+		return;
+	}
+
+	if (bucket != -1U) {
+		dump_unlinked_bucket(agno, agi_bp, bucket, quiet, verbose);
+		goto relse;
+	}
+
+	for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++) {
+		dump_unlinked_bucket(agno, agi_bp, bucket, quiet, verbose);
+	}
+
+relse:
+	libxfs_buf_relse(agi_bp);
+}
+
+static int
+dump_iunlinked_f(
+	int			argc,
+	char			**argv)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno = NULLAGNUMBER;
+	unsigned int		bucket = -1U;
+	bool			quiet = false;
+	bool			verbose = false;
+	int			c;
+
+	while ((c = getopt(argc, argv, "a:b:qv")) != EOF) {
+		switch (c) {
+		case 'a':
+			agno = atoi(optarg);
+			if (agno >= mp->m_sb.sb_agcount) {
+				dbprintf(_("Unknown AG %u, agcount is %u.\n"),
+						agno, mp->m_sb.sb_agcount);
+				return 0;
+			}
+			break;
+		case 'b':
+			bucket = atoi(optarg);
+			if (bucket >= XFS_AGI_UNLINKED_BUCKETS) {
+				dbprintf(_("Unknown bucket %u, max is 63.\n"),
+						bucket);
+				return 0;
+			}
+			break;
+		case 'q':
+			quiet = true;
+			break;
+		case 'v':
+			verbose = true;
+			break;
+		default:
+			dbprintf(_("Bad option for dump_iunlinked command.\n"));
+			return 0;
+		}
+	}
+
+	if (agno != NULLAGNUMBER) {
+		struct xfs_perag	*pag = libxfs_perag_get(mp, agno);
+
+		dump_unlinked(pag, bucket, quiet, verbose);
+		libxfs_perag_put(pag);
+		return 0;
+	}
+
+	for_each_perag(mp, agno, pag)
+		dump_unlinked(pag, bucket, quiet, verbose);
+
+	return 0;
+}
+
+static const cmdinfo_t	dump_iunlinked_cmd =
+	{ "dump_iunlinked", NULL, dump_iunlinked_f, 0, -1, 0,
+	  N_("[-a agno] [-b bucket] [-q] [-v]"),
+	  N_("dump chain of unlinked inode buckets"), NULL };
+
+void
+iunlink_init(void)
+{
+	add_command(&dump_iunlinked_cmd);
+}
libxfs/libxfs_api_defs.h +1 -0
index 026aa510..ddba5c7c 100644
@@ -125,6 +125,7 @@
 #define xfs_idestroy_fork		libxfs_idestroy_fork
 #define xfs_iext_lookup_extent		libxfs_iext_lookup_extent
 #define xfs_ifork_zap_attr		libxfs_ifork_zap_attr
+#define xfs_imap_to_bp			libxfs_imap_to_bp
 #define xfs_initialize_perag		libxfs_initialize_perag
 #define xfs_initialize_perag_data	libxfs_initialize_perag_data
 #define xfs_init_local_fork		libxfs_init_local_fork
man/man8/xfs_db.8 +19 -0
index 60dcdc52..2d6d0da4 100644
@@ -579,6 +579,25 @@ print the current debug option bits. These are for the use of the implementor.
 .BI "dquot [" \-g | \-p | \-u ] " id"
 Set current address to a group, project or user quota block for the given ID. Defaults to user quota.
 .TP
+.BI "dump_iunlinked [-a " agno " ] [-b " bucket " ] [-q] [-v]"
+Dump the contents of unlinked buckets.
+
+Options include:
+.RS 1.0i
+.TP 0.4i
+.B \-a
+Print only this AG's unlinked buckets.
+.TP 0.4i
+.B \-b
+Print only this bucket within each AGI.
+.TP 0.4i
+.B \-q
+Only print the essentials.
+.TP 0.4i
+.B \-v
+Print resource usage of each file on the unlinked lists.
+.RE
+.TP
 .BI "echo [" arg "] ..."
 Echo the arguments to the output.
 .TP
#4 lore raw

PATCH 2/2 xfs_db: create unlinked inodes

In reply to Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Create an expert-mode debugger command to create unlinked inodes. This will hopefully aid in simulation of leaked unlinked inode handling in the kernel and elsewhere.

---
 db/iunlink.c             |  196 ++++++++++++++++++++++++++++++++++++++++++++++
 libxfs/libxfs_api_defs.h |    1 
 man/man8/xfs_db.8        |   11 +++
 3 files changed, 208 insertions(+)
db/iunlink.c +196 -0
index 303b5daf..d87562e3 100644
@@ -197,8 +197,204 @@ static const cmdinfo_t	dump_iunlinked_cmd =
 	  N_("[-a agno] [-b bucket] [-q] [-v]"),
 	  N_("dump chain of unlinked inode buckets"), NULL };
 
+/*
+ * Look up the inode cluster buffer and log the on-disk unlinked inode change
+ * we need to make.
+ */
+static int
+iunlink_log_dinode(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_dinode	*dip;
+	struct xfs_buf		*ibp;
+	int			offset;
+	int			error;
+
+	error = -libxfs_imap_to_bp(mp, tp, &ip->i_imap, &ibp);
+	if (error)
+		return error;
+
+	dip = xfs_buf_offset(ibp, ip->i_imap.im_boffset);
+
+	dip->di_next_unlinked = cpu_to_be32(next_agino);
+	offset = ip->i_imap.im_boffset +
+			offsetof(struct xfs_dinode, di_next_unlinked);
+
+	libxfs_dinode_calc_crc(mp, dip);
+	libxfs_trans_log_buf(tp, ibp, offset, offset + sizeof(xfs_agino_t) - 1);
+	return 0;
+}
+
+static int
+iunlink_insert_inode(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	struct xfs_buf		*agibp,
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_agi		*agi = agibp->b_addr;
+	xfs_agino_t		next_agino;
+	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
+	short			bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
+	int			offset;
+	int			error;
+
+	/*
+	 * Get the index into the agi hash table for the list this inode will
+	 * go on.  Make sure the pointer isn't garbage and that this inode
+	 * isn't already on the list.
+	 */
+	next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
+	if (next_agino == agino || !xfs_verify_agino_or_null(pag, next_agino))
+		return EFSCORRUPTED;
+
+	if (next_agino != NULLAGINO) {
+		/*
+		 * There is already another inode in the bucket, so point this
+		 * inode to the current head of the list.
+		 */
+		error = iunlink_log_dinode(tp, ip, pag, next_agino);
+		if (error)
+			return error;
+	}
+
+	/* Update the bucket. */
+	agi->agi_unlinked[bucket_index] = cpu_to_be32(agino);
+	offset = offsetof(struct xfs_agi, agi_unlinked) +
+			(sizeof(xfs_agino_t) * bucket_index);
+	libxfs_trans_log_buf(tp, agibp, offset,
+			offset + sizeof(xfs_agino_t) - 1);
+	return 0;
+}
+
+/*
+ * This is called when the inode's link count has gone to 0 or we are creating
+ * a tmpfile via O_TMPFILE.  The inode @ip must have nlink == 0.
+ *
+ * We place the on-disk inode on a list in the AGI.  It will be pulled from this
+ * list when the inode is freed.
+ */
+static int
+iunlink(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
+	struct xfs_buf		*agibp;
+	int			error;
+
+	ASSERT(VFS_I(ip)->i_nlink == 0);
+	ASSERT(VFS_I(ip)->i_mode != 0);
+
+	pag = libxfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+
+	/* Get the agi buffer first.  It ensures lock ordering on the list. */
+	error = -libxfs_read_agi(pag, tp, &agibp);
+	if (error)
+		goto out;
+
+	error = iunlink_insert_inode(tp, pag, agibp, ip);
+out:
+	libxfs_perag_put(pag);
+	return error;
+}
+
+static int
+create_unlinked(
+	struct xfs_mount	*mp)
+{
+	struct cred		cr = { };
+	struct fsxattr		fsx = { };
+	struct xfs_inode	*ip;
+	struct xfs_trans	*tp;
+	unsigned int		resblks;
+	int			error;
+
+	resblks = XFS_IALLOC_SPACE_RES(mp);
+	error = -libxfs_trans_alloc(mp, &M_RES(mp)->tr_create_tmpfile, resblks,
+			0, 0, &tp);
+	if (error) {
+		dbprintf(_("alloc trans: %s\n"), strerror(error));
+		return error;
+	}
+
+	error = -libxfs_dir_ialloc(&tp, NULL, S_IFREG | 0600, 0, 0, &cr, &fsx,
+			&ip);
+	if (error) {
+		dbprintf(_("create inode: %s\n"), strerror(error));
+		goto out_cancel;
+	}
+
+	error = iunlink(tp, ip);
+	if (error) {
+		dbprintf(_("unlink inode: %s\n"), strerror(error));
+		goto out_rele;
+	}
+
+	error = -libxfs_trans_commit(tp);
+	if (error)
+		dbprintf(_("commit inode: %s\n"), strerror(error));
+
+	dbprintf(_("Created unlinked inode %llu in agno %u\n"),
+			(unsigned long long)ip->i_ino,
+			XFS_INO_TO_AGNO(mp, ip->i_ino));
+	libxfs_irele(ip);
+	return error;
+out_rele:
+	libxfs_irele(ip);
+out_cancel:
+	libxfs_trans_cancel(tp);
+	return error;
+}
+
+static int
+iunlink_f(
+	int		argc,
+	char		**argv)
+{
+	int		nr = 1;
+	int		c;
+	int		error;
+
+	while ((c = getopt(argc, argv, "n:")) != EOF) {
+		switch (c) {
+		case 'n':
+			nr = atoi(optarg);
+			if (nr <= 0) {
+				dbprintf(_("%s: need positive number\n"));
+				return 0;
+			}
+			break;
+		default:
+			dbprintf(_("Bad option for iunlink command.\n"));
+			return 0;
+		}
+	}
+
+	for (c = 0; c < nr; c++) {
+		error = create_unlinked(mp);
+		if (error)
+			return 1;
+	}
+
+	return 0;
+}
+
+static const cmdinfo_t	iunlink_cmd =
+	{ "iunlink", NULL, iunlink_f, 0, -1, 0,
+	  N_("[-n nr]"),
+	  N_("allocate inodes and put them on the unlinked list"), NULL };
+
 void
 iunlink_init(void)
 {
 	add_command(&dump_iunlinked_cmd);
+	if (expert_mode)
+		add_command(&iunlink_cmd);
 }
libxfs/libxfs_api_defs.h +1 -0
index ddba5c7c..04277c00 100644
@@ -149,6 +149,7 @@
 #define xfs_prealloc_blocks		libxfs_prealloc_blocks
 
 #define xfs_read_agf			libxfs_read_agf
+#define xfs_read_agi			libxfs_read_agi
 #define xfs_refc_block			libxfs_refc_block
 #define xfs_refcountbt_calc_reserves	libxfs_refcountbt_calc_reserves
 #define xfs_refcountbt_calc_size	libxfs_refcountbt_calc_size
man/man8/xfs_db.8 +11 -0
index 2d6d0da4..f53ddd67 100644
@@ -840,6 +840,17 @@ Set the current inode number. If no
 .I inode#
 is given, print the current inode number.
 .TP
+.BI "iunlink [-n " nr " ]"
+Allocate inodes and put them on the unlinked list.
+
+Options include:
+.RS 1.0i
+.TP 0.4i
+.B \-n
+Create this number of unlinked inodes.
+If not specified, 1 inode will be created.
+.RE
+.TP
 .BI "label [" label ]
 Set the filesystem label. The filesystem label can be used by
 .BR mount (8)
Reviewed-by on v1 2/2
#5 lore raw

Re: [PATCH 2/2] xfs_db: create unlinked inodes

Quoted 7 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Create an expert-mode debugger command to create unlinked inodes. This will hopefully aid in simulation of leaked unlinked inode handling in the kernel and elsewhere.

Quoted 250 lines
---
 db/iunlink.c             |  196 ++++++++++++++++++++++++++++++++++++++++++++++
 libxfs/libxfs_api_defs.h |    1 
 man/man8/xfs_db.8        |   11 +++
 3 files changed, 208 insertions(+)
db/iunlink.c +196 -0
index 303b5daf..d87562e3 100644
@@ -197,8 +197,204 @@ static const cmdinfo_t	dump_iunlinked_cmd =
 	  N_("[-a agno] [-b bucket] [-q] [-v]"),
 	  N_("dump chain of unlinked inode buckets"), NULL };
 
+/*
+ * Look up the inode cluster buffer and log the on-disk unlinked inode change
+ * we need to make.
+ */
+static int
+iunlink_log_dinode(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_dinode	*dip;
+	struct xfs_buf		*ibp;
+	int			offset;
+	int			error;
+
+	error = -libxfs_imap_to_bp(mp, tp, &ip->i_imap, &ibp);
+	if (error)
+		return error;
+
+	dip = xfs_buf_offset(ibp, ip->i_imap.im_boffset);
+
+	dip->di_next_unlinked = cpu_to_be32(next_agino);
+	offset = ip->i_imap.im_boffset +
+			offsetof(struct xfs_dinode, di_next_unlinked);
+
+	libxfs_dinode_calc_crc(mp, dip);
+	libxfs_trans_log_buf(tp, ibp, offset, offset + sizeof(xfs_agino_t) - 1);
+	return 0;
+}
+
+static int
+iunlink_insert_inode(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	struct xfs_buf		*agibp,
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_agi		*agi = agibp->b_addr;
+	xfs_agino_t		next_agino;
+	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
+	short			bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
+	int			offset;
+	int			error;
+
+	/*
+	 * Get the index into the agi hash table for the list this inode will
+	 * go on.  Make sure the pointer isn't garbage and that this inode
+	 * isn't already on the list.
+	 */
+	next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
+	if (next_agino == agino || !xfs_verify_agino_or_null(pag, next_agino))
+		return EFSCORRUPTED;
+
+	if (next_agino != NULLAGINO) {
+		/*
+		 * There is already another inode in the bucket, so point this
+		 * inode to the current head of the list.
+		 */
+		error = iunlink_log_dinode(tp, ip, pag, next_agino);
+		if (error)
+			return error;
+	}
+
+	/* Update the bucket. */
+	agi->agi_unlinked[bucket_index] = cpu_to_be32(agino);
+	offset = offsetof(struct xfs_agi, agi_unlinked) +
+			(sizeof(xfs_agino_t) * bucket_index);
+	libxfs_trans_log_buf(tp, agibp, offset,
+			offset + sizeof(xfs_agino_t) - 1);
+	return 0;
+}
+
+/*
+ * This is called when the inode's link count has gone to 0 or we are creating
+ * a tmpfile via O_TMPFILE.  The inode @ip must have nlink == 0.
+ *
+ * We place the on-disk inode on a list in the AGI.  It will be pulled from this
+ * list when the inode is freed.
+ */
+static int
+iunlink(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
+	struct xfs_buf		*agibp;
+	int			error;
+
+	ASSERT(VFS_I(ip)->i_nlink == 0);
+	ASSERT(VFS_I(ip)->i_mode != 0);
+
+	pag = libxfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+
+	/* Get the agi buffer first.  It ensures lock ordering on the list. */
+	error = -libxfs_read_agi(pag, tp, &agibp);
+	if (error)
+		goto out;
+
+	error = iunlink_insert_inode(tp, pag, agibp, ip);
+out:
+	libxfs_perag_put(pag);
+	return error;
+}
+
+static int
+create_unlinked(
+	struct xfs_mount	*mp)
+{
+	struct cred		cr = { };
+	struct fsxattr		fsx = { };
+	struct xfs_inode	*ip;
+	struct xfs_trans	*tp;
+	unsigned int		resblks;
+	int			error;
+
+	resblks = XFS_IALLOC_SPACE_RES(mp);
+	error = -libxfs_trans_alloc(mp, &M_RES(mp)->tr_create_tmpfile, resblks,
+			0, 0, &tp);
+	if (error) {
+		dbprintf(_("alloc trans: %s\n"), strerror(error));
+		return error;
+	}
+
+	error = -libxfs_dir_ialloc(&tp, NULL, S_IFREG | 0600, 0, 0, &cr, &fsx,
+			&ip);
+	if (error) {
+		dbprintf(_("create inode: %s\n"), strerror(error));
+		goto out_cancel;
+	}
+
+	error = iunlink(tp, ip);
+	if (error) {
+		dbprintf(_("unlink inode: %s\n"), strerror(error));
+		goto out_rele;
+	}
+
+	error = -libxfs_trans_commit(tp);
+	if (error)
+		dbprintf(_("commit inode: %s\n"), strerror(error));
+
+	dbprintf(_("Created unlinked inode %llu in agno %u\n"),
+			(unsigned long long)ip->i_ino,
+			XFS_INO_TO_AGNO(mp, ip->i_ino));
+	libxfs_irele(ip);
+	return error;
+out_rele:
+	libxfs_irele(ip);
+out_cancel:
+	libxfs_trans_cancel(tp);
+	return error;
+}
+
+static int
+iunlink_f(
+	int		argc,
+	char		**argv)
+{
+	int		nr = 1;
+	int		c;
+	int		error;
+
+	while ((c = getopt(argc, argv, "n:")) != EOF) {
+		switch (c) {
+		case 'n':
+			nr = atoi(optarg);
+			if (nr <= 0) {
+				dbprintf(_("%s: need positive number\n"));
+				return 0;
+			}
+			break;
+		default:
+			dbprintf(_("Bad option for iunlink command.\n"));
+			return 0;
+		}
+	}
+
+	for (c = 0; c < nr; c++) {
+		error = create_unlinked(mp);
+		if (error)
+			return 1;
+	}
+
+	return 0;
+}
+
+static const cmdinfo_t	iunlink_cmd =
+	{ "iunlink", NULL, iunlink_f, 0, -1, 0,
+	  N_("[-n nr]"),
+	  N_("allocate inodes and put them on the unlinked list"), NULL };
+
 void
 iunlink_init(void)
 {
 	add_command(&dump_iunlinked_cmd);
+	if (expert_mode)
+		add_command(&iunlink_cmd);
 }
libxfs/libxfs_api_defs.h +1 -0
index ddba5c7c..04277c00 100644
@@ -149,6 +149,7 @@
 #define xfs_prealloc_blocks		libxfs_prealloc_blocks
 
 #define xfs_read_agf			libxfs_read_agf
+#define xfs_read_agi			libxfs_read_agi
 #define xfs_refc_block			libxfs_refc_block
 #define xfs_refcountbt_calc_reserves	libxfs_refcountbt_calc_reserves
 #define xfs_refcountbt_calc_size	libxfs_refcountbt_calc_size
man/man8/xfs_db.8 +11 -0
index 2d6d0da4..f53ddd67 100644
@@ -840,6 +840,17 @@ Set the current inode number. If no
 .I inode#
 is given, print the current inode number.
 .TP
+.BI "iunlink [-n " nr " ]"
+Allocate inodes and put them on the unlinked list.
+
+Options include:
+.RS 1.0i
+.TP 0.4i
+.B \-n
+Create this number of unlinked inodes.
+If not specified, 1 inode will be created.
+.RE
+.TP
 .BI "label [" label ]
 Set the filesystem label. The filesystem label can be used by
 .BR mount (8)
#6 lore raw

RFC PATCH 1/1 fstests: test unlinked inode list repair on demand

In reply to Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Create a test to exercise recovery of unlinked inodes on a clean filesystem. This was definitely possible on old kernels that on an ro mount would clean the log without processing the iunlink list.

---
 common/rc          |    4 +
 tests/xfs/1872     |   89 +++++++++++++++++++++
 tests/xfs/1872.out |    3 +
 tests/xfs/1873     |  216 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/xfs/1873.out |    6 +
 5 files changed, 317 insertions(+), 1 deletion(-)
 create mode 100755 tests/xfs/1872
 create mode 100644 tests/xfs/1872.out
 create mode 100755 tests/xfs/1873
 create mode 100644 tests/xfs/1873.out
common/rc +3 -1
index 45d488b2aa..982dd8c312 100644
@@ -2674,9 +2674,11 @@ _require_xfs_io_command()
 		param_checked="$pwrite_opts $param"
 		;;
 	"scrub"|"repair")
-		testio=`$XFS_IO_PROG -x -c "$command probe" $TEST_DIR 2>&1`
+		test -z "$param" && param="probe"
+		testio=`$XFS_IO_PROG -x -c "$command $param" $TEST_DIR 2>&1`
 		echo $testio | grep -q "Inappropriate ioctl" && \
 			_notrun "xfs_io $command support is missing"
+		param_checked="$param"
 		;;
 	"startupdate"|"commitupdate"|"cancelupdate")
 		$XFS_IO_PROG -f -c 'pwrite -S 0x58 0 128k -b 128k' $testfile > /dev/null
tests/xfs/1872 new file +89 -0

89 lines not on this page · raw message

tests/xfs/1872.out new file +3 -0
new file mode 100644
index 0000000000..0aa4b5ae5c
@@ -0,0 +1,3 @@
+QA output created by 1872
++ Part 0: See if runtime can recover the unlinked list
+Silence is golden
tests/xfs/1873 new file +216 -0

216 lines not on this page · raw message

tests/xfs/1873.out new file +6 -0
new file mode 100644
index 0000000000..0e36bd2304
@@ -0,0 +1,6 @@
+QA output created by 1873
++ Part 1: See if scrub can recover the unlinked list
++ Part 2: Corrupt the first inode in the bucket
++ Part 3: Corrupt the middle inode in the bucket
++ Part 4: Corrupt the last inode in the bucket
+Silence is golden
#7 lore raw
In reply to Darrick J. Wong
Quoted 38 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

Quoted 7 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)

Hi Darrick,

Thanks for taking a look at this. I tested this patch on the setup where Shrikanth earlier saw the crash. I still can see a problem. I saw it is taking the branch from

+ /* If this is not an unlinked inode, something is very wrong. */
+ if (VFS_I(next_ip)->i_nlink != 0) {
+ error = -EFSCORRUPTED;
+ goto rele;
+ }

Here are the logs of reference -

[   21.399573] XFS (dm-0): Found unrecovered unlinked inode 0x2ec44d in AG 0x0.  Initiating recovery.
[   21.400150] XFS (dm-0): Internal error xfs_trans_cancel at line 1104 of file fs/xfs/xfs_trans.c.  Caller xfs_remove+0x1a0/0x310 [xfs]
[   21.400222] CPU: 0 PID: 1629 Comm: systemd-tmpfile Not tainted 6.5.0+ #2
[   21.400226] Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1010.22 (NH1010_122) hv:phyp pSeries
[   21.400230] Call Trace:
[   21.400231] [c000000014cdbb70] [c000000000f377b8] dump_stack_lvl+0x6c/0x9c (unreliable)
[   21.400239] [c000000014cdbba0] [c008000000f7c204] xfs_error_report+0x5c/0x80 [xfs]
[   21.400303] [c000000014cdbc00] [c008000000fab320] xfs_trans_cancel+0x178/0x1b0 [xfs]
[   21.400371] [c000000014cdbc50] [c008000000f999d8] xfs_remove+0x1a0/0x310 [xfs]
[   21.400432] [c000000014cdbcc0] [c008000000f93eb0] xfs_vn_unlink+0x68/0xf0 [xfs]
[   21.400493] [c000000014cdbd20] [c0000000005b8038] vfs_rmdir+0x178/0x300
[   21.400498] [c000000014cdbd60] [c0000000005be444] do_rmdir+0x124/0x240
[   21.400502] [c000000014cdbdf0] [c0000000005be594] sys_rmdir+0x34/0x50
[   21.400506] [c000000014cdbe10] [c000000000033c38] system_call_exception+0x148/0x3a0
[   21.400511] [c000000014cdbe50] [c00000000000c6d4] system_call_common+0xf4/0x258
[   21.400515] --- interrupt: c00 at 0x7fff9ad230d4
[   21.400518] NIP:  00007fff9ad230d4 LR: 00007fff9b12ff74 CTR: 0000000000000000
[   21.400521] REGS: c000000014cdbe80 TRAP: 0c00   Not tainted  (6.5.0+)
[   21.400524] MSR:  800000000280f033 <SF,VEC,VSX,EE,PR,FP,ME,IR,DR,RI,LE>  CR: 28004204  XER: 00000000
[   21.400534] IRQMASK: 0
[   21.400534] GPR00: 0000000000000028 00007fffc57f9aa0 00007fff9ae07300 000000014e651df0
[   21.400534] GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[   21.400534] GPR08: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[   21.400534] GPR12: 0000000000000000 00007fff9b4f5e60 000000013ebbee38 000000013ebc05d8
[   21.400534] GPR16: 000000013ebbee60 000000013ebc0818 000000013ebbee70 000000013ebbfe68
[   21.400534] GPR20: 000000013ebe003b 000000013ebe0051 000000013ebe003c 0000000000000000
[   21.400534] GPR24: 000000013ebc07d8 000000014e6496c0 000000013ebe0078 00007fffc57f9c80
[   21.400534] GPR28: 0000000000000000 000000014e651df0 0000000000000000 000000000000000e
[   21.400569] NIP [00007fff9ad230d4] 0x7fff9ad230d4
[   21.400571] LR [00007fff9b12ff74] 0x7fff9b12ff74
[   21.400573] --- interrupt: c00
[   21.402551] XFS (dm-0): Corruption of in-memory data (0x8) detected at xfs_trans_cancel+0x190/0x1b0 [xfs] (fs/xfs/xfs_trans.c:1105).  Shutting down filesystem.
[   21.402615] XFS (dm-0): Please unmount the filesystem and rectify the problem(s)

Do you know any possible reason/explaination on what could be going wrong?

Quoted 144 lines
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +18 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",

fs/xfs/xfs_trace.h TRACE_EVENT(xfs_iunlink_update_dinode, · line 3862

Quoted 10 lines
fs/xfs/xfs_trace.h +7 -0
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),

-ritesh

#8 lore raw
Quoted 60 lines · Ritesh Harjani (IBM)
Quoted 38 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

Quoted 7 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)

Hi Darrick,

Thanks for taking a look at this. I tested this patch on the setup where Shrikanth earlier saw the crash. I still can see a problem. I saw it is taking the branch from

fs/xfs/xfs_inode.c +5 -0
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
Quoted 4 lines

Here are the logs of reference -

[   21.399573] XFS (dm-0): Found unrecovered unlinked inode 0x2ec44d in AG 0x0.  Initiating recovery.
[   21.400150] XFS (dm-0): Internal error xfs_trans_cancel at line 1104 of file fs/xfs/xfs_trans.c.  Caller xfs_remove+0x1a0/0x310 [xfs]

Do you have a metadump for that filesystem, to examine that inode?

-Eric

Quoted 13 lines
[   21.400222] CPU: 0 PID: 1629 Comm: systemd-tmpfile Not tainted 6.5.0+ #2
[   21.400226] Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1010.22 (NH1010_122) hv:phyp pSeries
[   21.400230] Call Trace:
[   21.400231] [c000000014cdbb70] [c000000000f377b8] dump_stack_lvl+0x6c/0x9c (unreliable)
[   21.400239] [c000000014cdbba0] [c008000000f7c204] xfs_error_report+0x5c/0x80 [xfs]
[   21.400303] [c000000014cdbc00] [c008000000fab320] xfs_trans_cancel+0x178/0x1b0 [xfs]
[   21.400371] [c000000014cdbc50] [c008000000f999d8] xfs_remove+0x1a0/0x310 [xfs]
[   21.400432] [c000000014cdbcc0] [c008000000f93eb0] xfs_vn_unlink+0x68/0xf0 [xfs]
[   21.400493] [c000000014cdbd20] [c0000000005b8038] vfs_rmdir+0x178/0x300
[   21.400498] [c000000014cdbd60] [c0000000005be444] do_rmdir+0x124/0x240
[   21.400502] [c000000014cdbdf0] [c0000000005be594] sys_rmdir+0x34/0x50
[   21.400506] [c000000014cdbe10] [c000000000033c38] system_call_exception+0x148/0x3a0
[   21.400511] [c000000014cdbe50] [c00000000000c6d4] system_call_common+0xf4/0x258
#9 lore raw
Quoted 61 lines · Eric Sandeen
Quoted 60 lines · Ritesh Harjani (IBM)
Quoted 38 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

Quoted 7 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)

Hi Darrick,

Thanks for taking a look at this. I tested this patch on the setup where Shrikanth earlier saw the crash. I still can see a problem. I saw it is taking the branch from

fs/xfs/xfs_inode.c +5 -0
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
Quoted 6 lines
Quoted 4 lines

Here are the logs of reference -

[   21.399573] XFS (dm-0): Found unrecovered unlinked inode 0x2ec44d in AG 0x0.  Initiating recovery.
[   21.400150] XFS (dm-0): Internal error xfs_trans_cancel at line 1104 of file fs/xfs/xfs_trans.c.  Caller xfs_remove+0x1a0/0x310 [xfs]

Do you have a metadump for that filesystem, to examine that inode?

IIRC, Ritesh's problem was that there were inodes on the unlinked list *and* they had nonzero i_nlink. This patch doesn't address that; you'll have to wait for the online repair version.

--D

Quoted 18 lines

-Eric

Quoted 13 lines
[   21.400222] CPU: 0 PID: 1629 Comm: systemd-tmpfile Not tainted 6.5.0+ #2
[   21.400226] Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1010.22 (NH1010_122) hv:phyp pSeries
[   21.400230] Call Trace:
[   21.400231] [c000000014cdbb70] [c000000000f377b8] dump_stack_lvl+0x6c/0x9c (unreliable)
[   21.400239] [c000000014cdbba0] [c008000000f7c204] xfs_error_report+0x5c/0x80 [xfs]
[   21.400303] [c000000014cdbc00] [c008000000fab320] xfs_trans_cancel+0x178/0x1b0 [xfs]
[   21.400371] [c000000014cdbc50] [c008000000f999d8] xfs_remove+0x1a0/0x310 [xfs]
[   21.400432] [c000000014cdbcc0] [c008000000f93eb0] xfs_vn_unlink+0x68/0xf0 [xfs]
[   21.400493] [c000000014cdbd20] [c0000000005b8038] vfs_rmdir+0x178/0x300
[   21.400498] [c000000014cdbd60] [c0000000005be444] do_rmdir+0x124/0x240
[   21.400502] [c000000014cdbdf0] [c0000000005be594] sys_rmdir+0x34/0x50
[   21.400506] [c000000014cdbe10] [c000000000033c38] system_call_exception+0x148/0x3a0
[   21.400511] [c000000014cdbe50] [c00000000000c6d4] system_call_common+0xf4/0x258
Reviewed-by Brian Foster Reviewed-by Christoph Hellwig Reviewed-by Darrick J. Wong on v2 1/1
#10 lore raw
In reply to Darrick J. Wong
Quoted 9 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

Quoted 7 lines
	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Thank you for working on this Darrick!

I'll point out that at least one way to end up in this situation is to have at one point run a very old kernel which did not contain this commit, merged in kernel v4.14:

commit 6f4a1eefdd0ad4561543270a7fceadabcca075dd

Author: Eric Sandeen <[email protected]>
Date:   Tue Aug 8 18:21:49 2017 -0700

xfs: toggle readonly state around xfs_log_mount_finish

When we do log recovery on a readonly mount, unlinked inode processing does not happen due to the readonly checks in xfs_inactive(), which are trying to prevent any I/O on a readonly mount.

This is misguided - we do I/O on readonly mounts all the time, for consistency; for example, log recovery. So do the same RDONLY flag twiddling around xfs_log_mount_finish() as we do around xfs_log_mount(), for the same reason.

This all cries out for a big rework but for now this is a simple fix to an obvious problem.

so if you:

1) Crash with unlinked inodes
2) mount -o ro <recovers log but skips unlinked inode recovery>
3) mount -o remount,rw
4) umount <writes clean log record>

you now have a filesystem with on-disk unlinked inodes and a clean log, and those inodes won't get cleaned up until log recovery runs again or xfs_repair is run.

And in testing an old OS (RHEL7) it does seem that the root filesystem goes through a mount -o ro, mount -o remount,rw transition at boot time. So this situation may be somewhat common.

-Eric

Quoted 183 lines

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +25 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),
Reviewed-by on v2 1/1
#11 lore raw
In reply to Darrick J. Wong
Quoted 31 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Quoted 7 lines

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

This LGTM. Thanks!

Quoted 162 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +25 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),
#12 lore raw

I jumped the gun and only reviewed the userspace patches. Sorry, I'll actually look at the kernel patch now. ;)
-Bill

Quoted 206 lines · Bill O'Donnell
Quoted 31 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Quoted 7 lines

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

This LGTM. Thanks!

Quoted 162 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +25 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),
Reviewed-by on v2 1/1
#13 lore raw
In reply to Darrick J. Wong
Quoted 38 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

shrikanth hegde reports that filesystems fail shortly after mount with the following failure:

	WARNING: CPU: 56 PID: 12450 at fs/xfs/xfs_inode.c:1839 xfs_iunlink_lookup+0x58/0x80 [xfs]

This of course is the WARN_ON_ONCE in xfs_iunlink_lookup:

	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
	if (WARN_ON_ONCE(!ip || !ip->i_ino)) { ... }

From diagnostic data collected by the bug reporters, it would appear that we cleanly mounted a filesystem that contained unlinked inodes. Unlinked inodes are only processed as a final step of log recovery, which means that clean mounts do not process the unlinked list at all.

Prior to the introduction of the incore unlinked lists, this wasn't a problem because the unlink code would (very expensively) traverse the entire ondisk metadata iunlink chain to keep things up to date. However, the incore unlinked list code complains when it realizes that it is out of sync with the ondisk metadata and shuts down the fs, which is bad.

Ritesh proposed to solve this problem by unconditionally parsing the unlinked lists at mount time, but this imposes a mount time cost for every filesystem to catch something that should be very infrequent. Instead, let's target the places where we can encounter a next_unlinked pointer that refers to an inode that is not in cache, and load it into cache.

Note: This patch does not address the problem of iget loading an inode from the middle of the iunlink list and needing to set i_prev_unlinked correctly.

Quoted 8 lines

---
v2: log that we're doing runtime recovery, dont mess with DONTCACHE,
and actually return ENOLINK

---
 fs/xfs/xfs_inode.c |   75 +++++++++++++++++++++++++++++++++++++++++++++++++---
 fs/xfs/xfs_trace.h |   25 +++++++++++++++++
 2 files changed, 96 insertions(+), 4 deletions(-)

Makes sense. Feel free to add my RB.

Quoted 154 lines
fs/xfs/xfs_inode.c +71 -4
index 6ee266be45d4..2942002560b5 100644
@@ -1829,12 +1829,17 @@ xfs_iunlink_lookup(
 
 	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	if (!ip) {
+		/* Caller can handle inode not being in memory. */
+		rcu_read_unlock();
+		return NULL;
+	}
 
 	/*
-	 * Inode not in memory or in RCU freeing limbo should not happen.
-	 * Warn about this and let the caller handle the failure.
+	 * Inode in RCU freeing limbo should not happen.  Warn about this and
+	 * let the caller handle the failure.
 	 */
-	if (WARN_ON_ONCE(!ip || !ip->i_ino)) {
+	if (WARN_ON_ONCE(!ip->i_ino)) {
 		rcu_read_unlock();
 		return NULL;
 	}
@@ -1858,7 +1863,8 @@ xfs_iunlink_update_backref(
 
 	ip = xfs_iunlink_lookup(pag, next_agino);
 	if (!ip)
-		return -EFSCORRUPTED;
+		return -ENOLINK;
+
 	ip->i_prev_unlinked = prev_agino;
 	return 0;
 }
@@ -1902,6 +1908,62 @@ xfs_iunlink_update_bucket(
 	return 0;
 }
 
+/*
+ * Load the inode @next_agino into the cache and set its prev_unlinked pointer
+ * to @prev_agino.  Caller must hold the AGI to synchronize with other changes
+ * to the unlinked list.
+ */
+STATIC int
+xfs_iunlink_reload_next(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agibp,
+	xfs_agino_t		prev_agino,
+	xfs_agino_t		next_agino)
+{
+	struct xfs_perag	*pag = agibp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
+	struct xfs_inode	*next_ip = NULL;
+	xfs_ino_t		ino;
+	int			error;
+
+	ASSERT(next_agino != NULLAGINO);
+
+#ifdef DEBUG
+	rcu_read_lock();
+	next_ip = radix_tree_lookup(&pag->pag_ici_root, next_agino);
+	ASSERT(next_ip == NULL);
+	rcu_read_unlock();
+#endif
+
+	xfs_info_ratelimited(mp,
+ "Found unrecovered unlinked inode 0x%x in AG 0x%x.  Initiating recovery.",
+			next_agino, pag->pag_agno);
+
+	/*
+	 * Use an untrusted lookup to be cautious in case the AGI has been
+	 * corrupted and now points at a free inode.  That shouldn't happen,
+	 * but we'd rather shut down now since we're already running in a weird
+	 * situation.
+	 */
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, next_agino);
+	error = xfs_iget(mp, tp, ino, XFS_IGET_UNTRUSTED, 0, &next_ip);
+	if (error)
+		return error;
+
+	/* If this is not an unlinked inode, something is very wrong. */
+	if (VFS_I(next_ip)->i_nlink != 0) {
+		error = -EFSCORRUPTED;
+		goto rele;
+	}
+
+	next_ip->i_prev_unlinked = prev_agino;
+	trace_xfs_iunlink_reload_next(next_ip);
+rele:
+	ASSERT(!(VFS_I(next_ip)->i_state & I_DONTCACHE));
+	xfs_irele(next_ip);
+	return error;
+}
+
 static int
 xfs_iunlink_insert_inode(
 	struct xfs_trans	*tp,
@@ -1933,6 +1995,8 @@ xfs_iunlink_insert_inode(
 	 * inode.
 	 */
 	error = xfs_iunlink_update_backref(pag, agino, next_agino);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, agino, next_agino);
 	if (error)
 		return error;
 
@@ -2027,6 +2091,9 @@ xfs_iunlink_remove_inode(
 	 */
 	error = xfs_iunlink_update_backref(pag, ip->i_prev_unlinked,
 			ip->i_next_unlinked);
+	if (error == -ENOLINK)
+		error = xfs_iunlink_reload_next(tp, agibp, ip->i_prev_unlinked,
+				ip->i_next_unlinked);
 	if (error)
 		return error;
 
fs/xfs/xfs_trace.h +25 -0
index 36bd42ed9ec8..f4e46bac9b91 100644
@@ -3832,6 +3832,31 @@ TRACE_EVENT(xfs_iunlink_update_dinode,
 		  __entry->new_ptr)
 );
 
+TRACE_EVENT(xfs_iunlink_reload_next,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, agino)
+		__field(xfs_agino_t, prev_agino)
+		__field(xfs_agino_t, next_agino)
+	),
+	TP_fast_assign(
+		__entry->dev = ip->i_mount->m_super->s_dev;
+		__entry->agno = XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino);
+		__entry->agino = XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino);
+		__entry->prev_agino = ip->i_prev_unlinked;
+		__entry->next_agino = ip->i_next_unlinked;
+	),
+	TP_printk("dev %d:%d agno 0x%x agino 0x%x prev_unlinked 0x%x next_unlinked 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agino,
+		  __entry->prev_agino,
+		  __entry->next_agino)
+);
+
 DECLARE_EVENT_CLASS(xfs_ag_inode_class,
 	TP_PROTO(struct xfs_inode *ip),
 	TP_ARGS(ip),