Skip to contentSkip to conversation

xfs: old lost patches

PATCH In review Eric Sandeen 6 participants 19 messages Oct 11, 2013 – Oct 17, 2013 XFS filesystem

lore
Share & tools
lore
#1 lore raw

PATCH 1/4 xfs: remove newlines from 3 xfs_alert_tag error strings

xfs_alert_tag passes the format string to __xfs_printk, which adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages.

Most format strings have no newline, but these 3 do, leading to i.e.:

[ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119911] 
[ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119919] 

---

fs/xfs/xfs_bmap.c +1 -1
index f47e65c..e6b4202 100644
@@ -1482,7 +1482,7 @@ xfs_bmap_search_extents(
 		xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
 				"Access to block zero in inode %llu "
 				"start_block: %llx start_off: %llx "
-				"blkcnt: %llx extent-state: %x lastx: %x\n",
+				"blkcnt: %llx extent-state: %x lastx: %x",
 			(unsigned long long)ip->i_ino,
 			(unsigned long long)gotp->br_startblock,
 			(unsigned long long)gotp->br_startoff,
fs/xfs/xfs_error.c +1 -1
index 1123d93..40f2985 100644
@@ -159,7 +159,7 @@ xfs_error_report(
 {
 	if (level <= xfs_error_level) {
 		xfs_alert_tag(mp, XFS_PTAG_ERROR_REPORT,
-		"Internal error %s at line %d of file %s.  Caller 0x%p\n",
+		"Internal error %s at line %d of file %s.  Caller 0x%p",
 			    tag, linenum, filename, ra);
 
 		xfs_stack_trace();
fs/xfs/xfs_iomap.c +1 -1
index 8d4d49b..43e31b0 100644
@@ -110,7 +110,7 @@ xfs_alert_fsblock_zero(
 	xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
 			"Access to block zero in inode %llu "
 			"start_block: %llx start_off: %llx "
-			"blkcnt: %llx extent-state: %x\n",
+			"blkcnt: %llx extent-state: %x",
 		(unsigned long long)ip->i_ino,
 		(unsigned long long)imap->br_startblock,
 		(unsigned long long)imap->br_startoff,

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

#6 lore raw

PATCH 2/4 xfs: reject completely bogus remount options

In reply to Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

if we try to remount with something ridiculous.

In the long run we should probably revamp a lot of the mount option handling...

---

Note, not really sure what KM_* flag would be appropriate here, if it fails, it really is ok, other than missing the verification. But maybe that's too "nice?"

commit 72e6ddd901dc8a8ecb835324eb4e11b0d7ad8cf8

Author: Eric Sandeen <[email protected]>
Date:   Fri Oct 11 14:03:59 2013 -0500

xfs: reject completely bogus remount options

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in seq 1 10; do mount -o remount,noacl /mnt/test2; done
# for I in seq 1 10; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

if we try to remount with something ridiculous.

In the long run we should probably revamp a lot of the mount option handling...

fs/xfs/xfs_super.c +15 -1
index 15188cc..00a06d6 100644
@@ -1202,11 +1202,25 @@ xfs_fs_remount(
 	int			*flags,
 	char			*options)
 {
-	struct xfs_mount	*mp = XFS_M(sb);
+	struct xfs_mount	*mp = XFS_M(sb), *dummy_mp;
 	substring_t		args[MAX_OPT_ARGS];
 	char			*p;
 	int			error;
 
+	/*
+	 * Check all the mount options presented to be sure
+	 * there's nothing too crazy in there.  Non-remountable
+	 * but valid options are a different issue.
+	 */
+	dummy_mp = kmem_zalloc(sizeof(*dummy_mp), KM_MAYFAIL);
+	if (dummy_mp) {
+		dummy_mp->m_super = sb;
+		error = xfs_parseargs(dummy_mp, options);
+		kfree(dummy_mp);
+		if (error)
+			return -error;
+	}
+
 	while ((p = strsep(&options, ",")) != NULL) {
 		int token;
 

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

#14 lore raw

PATCH 3/4 xfs: don't emit corruption noise on fs probes

In reply to Eric Sandeen

If we get EWRONGFS due to probing of non-xfs filesystems, there's no need to issue the scary corruption error and backtrace.

---

fs/xfs/xfs_sb.c +3 -2
index a5b59d9..d327b2c 100644
@@ -624,8 +624,9 @@ xfs_sb_read_verify(
 
 out_error:
 	if (error) {
-		XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
-				     mp, bp->b_addr);
+		if (error != EWRONGFS)
+			XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+					     mp, bp->b_addr);
 		xfs_buf_ioerror(bp, error);
 	}
 }

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

#17 lore raw

PATCH 4/4 xfs: don't break from growfs ag update loop on error

In reply to Eric Sandeen

When xfs_growfs_data_private() is updating backup superblocks, it bails out on the first error encountered, whether reading or writing:

* If we get an error writing out the alternate superblocks,
* just issue a warning and continue. The real work is
* already done and committed.

This can cause a problem later during repair, because repair looks at all superblocks, and picks the most prevalent one as correct. If we bail out early in the backup superblock loop, we can end up with more "bad" matching superblocks than good, and a post-growfs repair may revert the filesystem to the old geometry.

With the combination of superblock verifiers and old bugs, we're more likely to encounter read errors due to verification.

And perhaps even worse, we don't even properly write any of the newly-added superblocks in the new AGs.

Even with this change, growfs will still say:

xfs_growfs: XFS_IOC_FSGROWFSDATA xfsctl failed: Structure needs cleaning data blocks changed from 319815680 to 335216640

which might be confusing to the user, but it at least communicates that something has gone wrong, and dmesg will probably highlight the need for an xfs_repair.

And this is still best-effort; if verifiers fail on more than half the backup supers, they may still "win" - but that's probably best left to repair to more gracefully handle by doing its own strict verification as part of the backup super "voting."

---

fs/xfs/xfs_fsops.c +13 -9
index e64ee52..634000c 100644
@@ -153,7 +153,7 @@ xfs_growfs_data_private(
 	xfs_buf_t		*bp;
 	int			bucket;
 	int			dpct;
-	int			error;
+	int			error, saved_error = 0;
 	xfs_agnumber_t		nagcount;
 	xfs_agnumber_t		nagimax = 0;
 	xfs_rfsblock_t		nb, nb_mod;
@@ -496,29 +496,33 @@ xfs_growfs_data_private(
 				error = ENOMEM;
 		}
 
+		/*
+		 * If we get an error reading or writing alternate superblocks,
+		 * continue.  xfs_repair chooses the "best" superblock based
+		 * on most matches; if we break early, we'll leave more
+		 * superblocks un-updated than updated, and xfs_repair may
+		 * pick them over the properly-updated primary.
+		 */
 		if (error) {
 			xfs_warn(mp,
 		"error %d reading secondary superblock for ag %d",
 				error, agno);
-			break;
+			saved_error = error;
+			continue;
 		}
 		xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb, XFS_SB_ALL_BITS);
 
-		/*
-		 * If we get an error writing out the alternate superblocks,
-		 * just issue a warning and continue.  The real work is
-		 * already done and committed.
-		 */
 		error = xfs_bwrite(bp);
 		xfs_buf_relse(bp);
 		if (error) {
 			xfs_warn(mp,
 		"write error %d updating secondary superblock for ag %d",
 				error, agno);
-			break; /* no point in continuing */
+			saved_error = error;
+			continue;
 		}
 	}
-	return error;
+	return saved_error ? saved_error : error;
 
  error0:
 	xfs_trans_cancel(tp, XFS_TRANS_ABORT);

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

Reviewed-by on v1 3/4
#15 lore raw

Re: [PATCH 3/4] xfs: don't emit corruption noise on fs probes

In reply to Eric Sandeen
Quoted 5 lines · Eric Sandeen

If we get EWRONGFS due to probing of non-xfs filesystems, there's no need to issue the scary corruption error and backtrace.

---

Looks good.

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

#7 lore raw

Re: [PATCH 2/4] xfs: reject completely bogus remount options

In reply to Eric Sandeen
Quoted 17 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value
Quoted 7 lines

if we try to remount with something ridiculous.

In the long run we should probably revamp a lot of the mount option handling...

---

I don't seem to get the duplicate mtab entries on a top of tree kernel. Is this still appropriate?

--Mark.

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

#2 lore raw

Re: [PATCH 1/4] xfs: remove newlines from 3 xfs_alert_tag error strings

In reply to Eric Sandeen
Quoted 7 lines · Eric Sandeen

xfs_alert_tag passes the format string to __xfs_printk, which adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages.

Most format strings have no newline, but these 3 do, leading to i.e.:

Quoted 8 lines
[ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119911]
[ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119919]

---

Is this true of xfs_alert() too? ie the newline in xfs_alert in xfs_dir2_leafn_rebalance(). The newline in xfs_alert() in xlog_unpack_data_crc() looks intentional.

--Mark.

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

#8 lore raw

Re: [PATCH 2/4] xfs: reject completely bogus remount options

In reply to Mark Tinguely
Quoted 23 lines · Mark Tinguely
Quoted 22 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

if we try to remount with something ridiculous.

In the long run we should probably revamp a lot of the mount option handling...

Quoted 6 lines
Quoted 2 lines

---

I don't seem to get the duplicate mtab entries on a top of tree kernel. Is this still appropriate?

Maybe different mount(8) behavior on your system? (probably symlinked to /proc/mounts)

On RHEL6:

# mount /dev/sdb1 /mnt/test
# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test; done 
# mount | grep sdb1
/dev/sdb1 on /mnt/test type xfs (rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl)
# uname -a
Linux hostname 3.12.0-rc4+ #41 SMP Fri Oct 11 19:43:01 CDT 2013 x86_64 x86_64 x86_64 GNU/Linux

-Eric

Quoted 7 lines

--Mark.

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

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

#3 lore raw

Re: [PATCH 1/4] xfs: remove newlines from 3 xfs_alert_tag error strings

In reply to Mark Tinguely
Quoted 13 lines · Mark Tinguely
Quoted 12 lines · Eric Sandeen

xfs_alert_tag passes the format string to __xfs_printk, which adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages.

Most format strings have no newline, but these 3 do, leading to i.e.:

[ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119911]
[ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119919]
Quoted 5 lines
Quoted 3 lines

---

Is this true of xfs_alert() too? ie the newline in xfs_alert in xfs_dir2_leafn_rebalance(). The newline in xfs_alert() in xlog_unpack_data_crc() looks intentional.

All of these:

define_xfs_printk_level(xfs_emerg, KERN_EMERG);
define_xfs_printk_level(xfs_alert, KERN_ALERT);
define_xfs_printk_level(xfs_crit, KERN_CRIT);
define_xfs_printk_level(xfs_err, KERN_ERR);
define_xfs_printk_level(xfs_warn, KERN_WARNING);
define_xfs_printk_level(xfs_notice, KERN_NOTICE);
define_xfs_printk_level(xfs_info, KERN_INFO);
#ifdef DEBUG
define_xfs_printk_level(xfs_debug, KERN_DEBUG);
#endif

go through __xfs_printk(), which adds a newline...

seems like more fixes are in order, yeah.

-Eric

Quoted 7 lines

--Mark.

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

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

#4 lore raw

PATCH v2 1/4 xfs: remove newlines from strings passed to __xfs_printk

In reply to Eric Sandeen

__xfs_printk adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages.

Most format strings have no newline, but a few do, leading to i.e.:

[ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119911] 
[ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119919] 

Fix them all.

---

fs/xfs/xfs_bmap.c +1 -1
index f47e65c..e6b4202 100644
@@ -1482,7 +1482,7 @@ xfs_bmap_search_extents(
 		xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
 				"Access to block zero in inode %llu "
 				"start_block: %llx start_off: %llx "
-				"blkcnt: %llx extent-state: %x lastx: %x\n",
+				"blkcnt: %llx extent-state: %x lastx: %x",
 			(unsigned long long)ip->i_ino,
 			(unsigned long long)gotp->br_startblock,
 			(unsigned long long)gotp->br_startoff,
fs/xfs/xfs_buf.c +3 -3
index 2634700..75e4ea7 100644
@@ -590,7 +590,7 @@ found:
 		error = _xfs_buf_map_pages(bp, flags);
 		if (unlikely(error)) {
 			xfs_warn(target->bt_mount,
-				"%s: failed to map pages\n", __func__);
+				"%s: failed to map pagesn", __func__);
 			xfs_buf_relse(bp);
 			return NULL;
 		}
@@ -809,7 +809,7 @@ xfs_buf_get_uncached(
 	error = _xfs_buf_map_pages(bp, 0);
 	if (unlikely(error)) {
 		xfs_warn(target->bt_mount,
-			"%s: failed to map pages\n", __func__);
+			"%s: failed to map pages", __func__);
 		goto fail_free_mem;
 	}
 
@@ -1618,7 +1618,7 @@ xfs_setsize_buftarg_flags(
 		bdevname(btp->bt_bdev, name);
 
 		xfs_warn(btp->bt_mount,
-			"Cannot set_blocksize to %u on device %s\n",
+			"Cannot set_blocksize to %u on device %s",
 			sectorsize, name);
 		return EINVAL;
 	}
fs/xfs/xfs_dir2_node.c +1 -1
index 4c3dba7..34df052 100644
@@ -1101,7 +1101,7 @@ xfs_dir2_leafn_rebalance(
 		state->inleaf = 1;
 		blk2->index = 0;
 		xfs_alert(args->dp->i_mount,
-	"%s: picked the wrong leaf? reverting original leaf: blk1->index %d\n",
+	"%s: picked the wrong leaf? reverting original leaf: blk1->index %d",
 			__func__, blk1->index);
 	}
 }
fs/xfs/xfs_error.c +1 -1
index 1123d93..40f2985 100644
@@ -159,7 +159,7 @@ xfs_error_report(
 {
 	if (level <= xfs_error_level) {
 		xfs_alert_tag(mp, XFS_PTAG_ERROR_REPORT,
-		"Internal error %s at line %d of file %s.  Caller 0x%p\n",
+		"Internal error %s at line %d of file %s.  Caller 0x%p",
 			    tag, linenum, filename, ra);
 
 		xfs_stack_trace();
fs/xfs/xfs_iomap.c +1 -1
index 8d4d49b..43e31b0 100644
@@ -110,7 +110,7 @@ xfs_alert_fsblock_zero(
 	xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
 			"Access to block zero in inode %llu "
 			"start_block: %llx start_off: %llx "
-			"blkcnt: %llx extent-state: %x\n",
+			"blkcnt: %llx extent-state: %x",
 		(unsigned long long)ip->i_ino,
 		(unsigned long long)imap->br_startblock,
 		(unsigned long long)imap->br_startoff,
fs/xfs/xfs_log.c +1 -1
index a2dea10..ea02b48 100644
@@ -1979,7 +1979,7 @@ xlog_print_tic_res(
 
 	for (i = 0; i < ticket->t_res_num; i++) {
 		uint r_type = ticket->t_res_arr[i].r_type;
-		xfs_warn(mp, "region[%u]: %s - %u bytes\n", i,
+		xfs_warn(mp, "region[%u]: %s - %u bytes", i,
 			    ((r_type <= 0 || r_type > XLOG_REG_TYPE_MAX) ?
 			    "bad-rtype" : res_type_str[r_type-1]),
 			    ticket->t_res_arr[i].r_len);
fs/xfs/xfs_log_recover.c +3 -3
index 3979749..e7ca48e 100644
@@ -305,9 +305,9 @@ xlog_header_check_dump(
 	xfs_mount_t		*mp,
 	xlog_rec_header_t	*head)
 {
-	xfs_debug(mp, "%s:  SB : uuid = %pU, fmt = %d\n",
+	xfs_debug(mp, "%s:  SB : uuid = %pU, fmt = %d",
 		__func__, &mp->m_sb.sb_uuid, XLOG_FMT);
-	xfs_debug(mp, "    log : uuid = %pU, fmt = %d\n",
+	xfs_debug(mp, "    log : uuid = %pU, fmt = %d",
 		&head->h_fs_uuid, be32_to_cpu(head->h_fmt));
 }
 #else
@@ -4077,7 +4077,7 @@ xlog_unpack_data_crc(
 	if (crc != rhead->h_crc) {
 		if (rhead->h_crc || xfs_sb_version_hascrc(&log->l_mp->m_sb)) {
 			xfs_alert(log->l_mp,
-		"log record CRC mismatch: found 0x%x, expected 0x%x.\n",
+		"log record CRC mismatch: found 0x%x, expected 0x%x.",
 					le32_to_cpu(rhead->h_crc),
 					le32_to_cpu(crc));
 			xfs_hex_dump(dp, 32);
fs/xfs/xfs_qm_syscalls.c +6 -6
index 8174aad..fb67091 100644
@@ -287,7 +287,7 @@ xfs_qm_scall_trunc_qfiles(
 	int		error = 0, error2 = 0;
 
 	if (!xfs_sb_version_hasquota(&mp->m_sb) || flags == 0) {
-		xfs_debug(mp, "%s: flags=%x m_qflags=%x\n",
+		xfs_debug(mp, "%s: flags=%x m_qflags=%x",
 			__func__, flags, mp->m_qflags);
 		return XFS_ERROR(EINVAL);
 	}
@@ -325,7 +325,7 @@ xfs_qm_scall_quotaon(
 	sbflags = 0;
 
 	if (flags == 0) {
-		xfs_debug(mp, "%s: zero flags, m_qflags=%x\n",
+		xfs_debug(mp, "%s: zero flags, m_qflags=%x",
 			__func__, mp->m_qflags);
 		return XFS_ERROR(EINVAL);
 	}
@@ -348,7 +348,7 @@ xfs_qm_scall_quotaon(
 	     (mp->m_sb.sb_qflags & XFS_PQUOTA_ACCT) == 0 &&
 	     (flags & XFS_PQUOTA_ENFD))) {
 		xfs_debug(mp,
-			"%s: Can't enforce without acct, flags=%x sbflags=%x\n",
+			"%s: Can't enforce without acct, flags=%x sbflags=%x",
 			__func__, flags, mp->m_sb.sb_qflags);
 		return XFS_ERROR(EINVAL);
 	}
@@ -648,7 +648,7 @@ xfs_qm_scall_setqlim(
 			q->qi_bsoftlimit = soft;
 		}
 	} else {
-		xfs_debug(mp, "blkhard %Ld < blksoft %Ld\n", hard, soft);
+		xfs_debug(mp, "blkhard %Ld < blksoft %Ld", hard, soft);
 	}
 	hard = (newlim->d_fieldmask & FS_DQ_RTBHARD) ?
 		(xfs_qcnt_t) XFS_BB_TO_FSB(mp, newlim->d_rtb_hardlimit) :
@@ -664,7 +664,7 @@ xfs_qm_scall_setqlim(
 			q->qi_rtbsoftlimit = soft;
 		}
 	} else {
-		xfs_debug(mp, "rtbhard %Ld < rtbsoft %Ld\n", hard, soft);
+		xfs_debug(mp, "rtbhard %Ld < rtbsoft %Ld", hard, soft);
 	}
 
 	hard = (newlim->d_fieldmask & FS_DQ_IHARD) ?
@@ -681,7 +681,7 @@ xfs_qm_scall_setqlim(
 			q->qi_isoftlimit = soft;
 		}
 	} else {
-		xfs_debug(mp, "ihard %Ld < isoft %Ld\n", hard, soft);
+		xfs_debug(mp, "ihard %Ld < isoft %Ld", hard, soft);
 	}
 
 	/*
fs/xfs/xfs_sb.c +2 -2
index a5b59d9..a89d0bc 100644
@@ -249,13 +249,13 @@ xfs_mount_validate_sb(
 	if (xfs_sb_version_has_pquotino(sbp)) {
 		if (sbp->sb_qflags & (XFS_OQUOTA_ENFD | XFS_OQUOTA_CHKD)) {
 			xfs_notice(mp,
-			   "Version 5 of Super block has XFS_OQUOTA bits.\n");
+			   "Version 5 of Super block has XFS_OQUOTA bits.");
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 	} else if (sbp->sb_qflags & (XFS_PQUOTA_ENFD | XFS_GQUOTA_ENFD |
 				XFS_PQUOTA_CHKD | XFS_GQUOTA_CHKD)) {
 			xfs_notice(mp,
-"Superblock earlier than Version 5 has XFS_[PQ]UOTA_{ENFD|CHKD} bits.\n");
+"Superblock earlier than Version 5 has XFS_[PQ]UOTA_{ENFD|CHKD} bits.");
 			return XFS_ERROR(EFSCORRUPTED);
 	}
 
fs/xfs/xfs_super.c +1 -1
index 15188cc..77093de 100644
@@ -1246,7 +1246,7 @@ xfs_fs_remount(
 			 */
 #if 0
 			xfs_info(mp,
-		"mount option \"%s\" not supported for remount\n", p);
+		"mount option \"%s\" not supported for remount", p);
 			return -EINVAL;
 #else
 			break;

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

Reviewed-by on v2 1/4
#5 lore raw

Re: [PATCH 1/4 V2] xfs: remove newlines from strings passed to __xfs_printk

Quoted 11 lines · Eric Sandeen

__xfs_printk adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages.

Most format strings have no newline, but a few do, leading to i.e.:

[ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119911]
[ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05
[ 7347.119919]
Quoted 4 lines

Fix them all.

---

Thanks for the updated patch. It looks good.

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

Reviewed-by
#9 lore raw

Re: [PATCH 2/4] xfs: reject completely bogus remount options

In reply to Eric Sandeen
Quoted 43 lines · Eric Sandeen
Quoted 23 lines · Mark Tinguely
Quoted 22 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

if we try to remount with something ridiculous.

In the long run we should probably revamp a lot of the mount option handling...

Quoted 6 lines
Quoted 2 lines

---

I don't seem to get the duplicate mtab entries on a top of tree kernel. Is this still appropriate?

Maybe different mount(8) behavior on your system? (probably symlinked to /proc/mounts)

On RHEL6:

# mount /dev/sdb1 /mnt/test
# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test; done
# mount | grep sdb1
/dev/sdb1 on /mnt/test type xfs (rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl)
# uname -a
Linux hostname 3.12.0-rc4+ #41 SMP Fri Oct 11 19:43:01 CDT 2013 x86_64 x86_64 x86_64 GNU/Linux
Quoted 1 line

-Eric

Yep, confirmed the described behavior on a RHEL 6 box without patch. The patch looks good.

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

#10 lore raw

Re: [PATCH 2/4] xfs: reject completely bogus remount options

In reply to Eric Sandeen
Quoted 10 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0
Quoted 6 lines

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

ext4? Really? :)

Quoted 13 lines

+++ b/fs/xfs/xfs_super.c

fs/xfs/xfs_super.c +3 -1
@@ -1202,11 +1202,25 @@ xfs_fs_remount(
 	int			*flags,
 	char			*options)
 {
-	struct xfs_mount	*mp = XFS_M(sb);
+	struct xfs_mount	*mp = XFS_M(sb), *dummy_mp;
 	substring_t		args[MAX_OPT_ARGS];
 	char			*p;
 	int			error;
 
+	/*
+	 * Check all the mount options presented to be sure

fs/xfs/xfs_super.c xfs_fs_remount( · line 1221

Quoted 10 lines
fs/xfs/xfs_super.c +10 -0
+	 * there's nothing too crazy in there.  Non-remountable
+	 * but valid options are a different issue.
+	 */
+	dummy_mp = kmem_zalloc(sizeof(*dummy_mp), KM_MAYFAIL);
+	if (dummy_mp) {
+		dummy_mp->m_super = sb;
+		error = xfs_parseargs(dummy_mp, options);
+		kfree(dummy_mp);
+		if (error)
+			return -error;

This, at minimum, leaks dummy_mp->m_fsname, and it will leak other strings that are also kstrdup()d by xfs_parseargs().

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

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

Re: [PATCH 2/4] xfs: reject completely bogus remount options

Quoted 11 lines · Dave Chinner
Quoted 10 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0
Quoted 8 lines
Quoted 6 lines

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

ext4? Really? :)

uhhh ;)

Quoted 11 lines
Quoted 11 lines

+++ b/fs/xfs/xfs_super.c

fs/xfs/xfs_super.c +1 -1
@@ -1202,11 +1202,25 @@ xfs_fs_remount(
 	int			*flags,
 	char			*options)
 {
-	struct xfs_mount	*mp = XFS_M(sb);
+	struct xfs_mount	*mp = XFS_M(sb), *dummy_mp;
 	substring_t		args[MAX_OPT_ARGS];
 	char			*p;
 	int			error;
 
Quoted 15 lines

fs/xfs/xfs_super.c xfs_fs_remount( · line 1221

Quoted 12 lines
fs/xfs/xfs_super.c +12 -0
+	/*
+	 * Check all the mount options presented to be sure
+	 * there's nothing too crazy in there.  Non-remountable
+	 * but valid options are a different issue.
+	 */
+	dummy_mp = kmem_zalloc(sizeof(*dummy_mp), KM_MAYFAIL);
+	if (dummy_mp) {
+		dummy_mp->m_super = sb;
+		error = xfs_parseargs(dummy_mp, options);
+		kfree(dummy_mp);
+		if (error)
+			return -error;

This, at minimum, leaks dummy_mp->m_fsname, and it will leak other strings that are also kstrdup()d by xfs_parseargs().

nnngh. Forgot about that side effect, sorry. Dammit.

Think it's still worth doing this if I handle freeing them all up?

-Eric

Quoted 4 lines

Cheers,

Dave.

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

#12 lore raw

Re: [PATCH 2/4] xfs: reject completely bogus remount options

Quoted 51 lines · Eric Sandeen
Quoted 11 lines · Dave Chinner
Quoted 10 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done
# grep sdb4 /etc/mtab
/dev/sdb4 /mnt/test2 xfs rw,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,noacl,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption,badoption 0 0
Quoted 8 lines
Quoted 6 lines

This is a bit of a hack, but we can re-use xfs_parseargs() with a dummy mount struct to just vet all of the remount options which were passed in. With this, we get a saner result:

[44898.102990] EXT4-fs (sdb4): Unrecognized mount option "badoption" or missing value

ext4? Really? :)

uhhh ;)

Quoted 11 lines
Quoted 11 lines

+++ b/fs/xfs/xfs_super.c

fs/xfs/xfs_super.c +1 -1
@@ -1202,11 +1202,25 @@ xfs_fs_remount(
 	int			*flags,
 	char			*options)
 {
-	struct xfs_mount	*mp = XFS_M(sb);
+	struct xfs_mount	*mp = XFS_M(sb), *dummy_mp;
 	substring_t		args[MAX_OPT_ARGS];
 	char			*p;
 	int			error;
 
Quoted 15 lines

fs/xfs/xfs_super.c xfs_fs_remount( · line 1221

Quoted 12 lines
fs/xfs/xfs_super.c +12 -0
+	/*
+	 * Check all the mount options presented to be sure
+	 * there's nothing too crazy in there.  Non-remountable
+	 * but valid options are a different issue.
+	 */
+	dummy_mp = kmem_zalloc(sizeof(*dummy_mp), KM_MAYFAIL);
+	if (dummy_mp) {
+		dummy_mp->m_super = sb;
+		error = xfs_parseargs(dummy_mp, options);
+		kfree(dummy_mp);
+		if (error)
+			return -error;

This, at minimum, leaks dummy_mp->m_fsname, and it will leak other strings that are also kstrdup()d by xfs_parseargs().

Quoted 3 lines

nnngh. Forgot about that side effect, sorry. Dammit.

Think it's still worth doing this if I handle freeing them all up?

If you wrap it all in a helper function (xfs_check-args()?) that does all the temporary structure allocation and freeing, I think it will be fine.

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

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

Re: [PATCH 2/4] xfs: reject completely bogus remount options

Quoted 10 lines · Dave Chinner
Quoted 9 lines · Eric Sandeen
Quoted 8 lines · Dave Chinner
Quoted 7 lines · Eric Sandeen

There's a long comment about handling non-remountable options in xfs_fs_remount, but nothing addresses the case of completely bogus mount options at remount time, which can lead to some severe strangeness:

# for I in `seq 1 10`; do mount -o remount,noacl /mnt/test2; done
# for I in `seq 1 10`; do mount -o remount,badoption /mnt/test2; done

I'm going to just drop this patch for now; it seems too hacky.

really, mount option handling just needs a big rework.

-Eric

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

Reviewed-by on v1 3/4
#16 lore raw

Re: [PATCH 3/4] xfs: don't emit corruption noise on fs probes

In reply to Eric Sandeen
Quoted 4 lines · Eric Sandeen

If we get EWRONGFS due to probing of non-xfs filesystems, there's no need to issue the scary corruption error and backtrace.

Looks good,

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