FAILED: patch "[PATCH] nilfs2: ignore non-fatal signals during synchronous log write wait" failed to apply to 6.1-stable tree

FAILED: patch "[PATCH] nilfs2: ignore non-fatal signals during synchronous log write wait" failed to apply to 6.1-stable tree

[Sasha Levin]

The patch below does not apply to the 6.1-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.1.y
git checkout FETCH_HEAD
git cherry-pick -x 7131aaa74bcc7c9da1f1d1c8724da75c775557a0
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable@vger.kernel.org>' --in-reply-to '20261107054332-failed-6.1-7131aaa74bcc@kernel.org' --subject-prefix 'PATCH 6.1.y' 'HEAD^..'

Thanks,
Sasha

------------------ original commit in Linus's tree ------------------

From 7131aaa74bcc7c9da1f1d1c8724da75c775557a0 Mon Sep 17 00:00:00 2001
From: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Date: Wed, 30 Sep 2026 03:26:54 +0900
Subject: [PATCH] nilfs2: ignore non-fatal signals during synchronous log write
 wait

Based on a syzbot report, Wang Jianjian pointed out that following a
successful directory operation in nilfs_rename(), a synchronous write
requested within nilfs_transaction_commit(), triggered by a sync flag
set by nilfs_commit_chunk(), could be interrupted by a user signal.
This interruption leaves in-memory metadata updates intact while
returning an error, causing inconsistency with the VFS dentry cache and
ultimately triggering a kernel warning.

According to "signal(7)", local disks are not classified as "slow
devices" (such as terminals, FIFOs, or pipes) and their I/O operations
on disk devices are not interrupted by signals.  In kernel space,
waiting for such disk I/O should therefore not be aborted by ordinary
(non-fatal) signals.

Suppress this issue by modifying nilfs_segctor_sync(), which waits for
the log writer thread to complete synchronous writes, to catch only
fatal signals via fatal_signal_pending() and return -EINTR rather than
-ERESTARTSYS.  To complement this change and prevent the wait loop from
busy-looping when schedule() returns immediately due to a pending
non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
TASK_KILLABLE.

This serves as an effective interim stabilization measure without
structural changes until a comprehensive fix for metadata consistency
is implemented.

Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Cc: Wang Jianjian <wangjianjian3@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
---
 fs/nilfs2/ioctl.c    |  2 +-
 fs/nilfs2/recovery.c |  2 +-
 fs/nilfs2/segment.c  | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index 01a04080ef704..b2b67d6a1f097 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -965,10 +965,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
  * * %-EFAULT		- Failure during execution of requested operation.
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index 9a70c28ec94c1..abe4101f7550c 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -790,11 +790,11 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EINVAL		- Inconsistent filesystem state.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  */
 int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
 			      struct super_block *sb,
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index 2f0e59847d026..af0c3560444f5 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -2259,7 +2259,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 	wake_up(&sci->sc_wait_daemon);
 
 	for (;;) {
-		set_current_state(TASK_INTERRUPTIBLE);
+		set_current_state(TASK_KILLABLE);
 
 		/*
 		 * Synchronize only while the log writer thread is alive.
@@ -2273,11 +2273,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 			err = wait_req.err;
 			break;
 		}
-		if (!signal_pending(current)) {
+		if (!fatal_signal_pending(current)) {
 			schedule();
 			continue;
 		}
-		err = -ERESTARTSYS;
+		err = -EINTR;
 		break;
 	}
 	finish_wait(&sci->sc_wait_request, &wait_req.wq);
@@ -2311,10 +2311,10 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error (including metadata corruption).
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_segment(struct super_block *sb)
@@ -2341,10 +2341,10 @@ int nilfs_construct_segment(struct super_block *sb)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error (including metadata corruption).
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_dsync_segment(struct super_block *sb, struct inode *inode,
-- 
2.53.0

[PATCH 6.1.y 1/2] nilfs2: correct return value kernel-doc descriptions for the rest

[Sasha Levin]

From: Ryusuke Konishi <konishi.ryusuke@gmail.com>

[ Upstream commit 343d4a33f26d048fde4b02cfc3681f766b3f1b12 ]

Similar to the previous changes to fix return value descriptions, this
fixes the format of the return value descriptions of functions for the
rest.

Link: https://lkml.kernel.org/r/20250110010530.21872-6-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: "Brian G ." <gissf1@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Stable-dep-of: 7131aaa74bcc ("nilfs2: ignore non-fatal signals during synchronous log write wait")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/nilfs2/gcinode.c   | 24 +++++++++-------------
 fs/nilfs2/inode.c     | 12 +++--------
 fs/nilfs2/recovery.c  | 32 +++++++++++------------------
 fs/nilfs2/segbuf.c    | 12 ++---------
 fs/nilfs2/segment.c   | 48 ++++++++++++++++---------------------------
 fs/nilfs2/the_nilfs.c | 12 +++++------
 6 files changed, 50 insertions(+), 90 deletions(-)

diff --git a/fs/nilfs2/gcinode.c b/fs/nilfs2/gcinode.c
index 8cdb1dc76f612..1e7de19e58306 100644
--- a/fs/nilfs2/gcinode.c
+++ b/fs/nilfs2/gcinode.c
@@ -46,14 +46,11 @@
  * specified by @pbn to the GC pagecache with the key @blkoff.
  * This function sets @vbn (@pbn if @vbn is zero) in b_blocknr of the buffer.
  *
- * Return Value: On success, 0 is returned. On Error, one of the following
- * negative error code is returned.
- *
- * %-EIO - I/O error.
- *
- * %-ENOMEM - Insufficient amount of memory available.
- *
- * %-ENOENT - The block specified with @pbn does not exist.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- The block specified with @pbn does not exist.
+ * * %-ENOMEM	- Insufficient memory available.
  */
 int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
 				   sector_t pbn, __u64 vbn,
@@ -114,12 +111,11 @@ int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
  * specified by @vbn to the GC pagecache.  @pbn can be supplied by the
  * caller to avoid translation of the disk block address.
  *
- * Return Value: On success, 0 is returned. On Error, one of the following
- * negative error code is returned.
- *
- * %-EIO - I/O error.
- *
- * %-ENOMEM - Insufficient amount of memory available.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Invalid virtual block address.
+ * * %-ENOMEM	- Insufficient memory available.
  */
 int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
 				   __u64 vbn, struct buffer_head **out_bh)
diff --git a/fs/nilfs2/inode.c b/fs/nilfs2/inode.c
index a20d52951f1ef..eaf4e61113afd 100644
--- a/fs/nilfs2/inode.c
+++ b/fs/nilfs2/inode.c
@@ -676,10 +676,7 @@ struct inode *nilfs_iget_for_gc(struct super_block *sb, unsigned long ino,
  * or does nothing if the inode already has it.  This function allocates
  * an additional inode to maintain page cache of B-tree nodes one-on-one.
  *
- * Return Value: On success, 0 is returned. On errors, one of the following
- * negative error code is returned.
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: 0 on success, or %-ENOMEM if memory is insufficient.
  */
 int nilfs_attach_btree_node_cache(struct inode *inode)
 {
@@ -740,11 +737,8 @@ void nilfs_detach_btree_node_cache(struct inode *inode)
  * in one inode and the one for b-tree node pages is set up in the
  * other inode, which is attached to the former inode.
  *
- * Return Value: On success, a pointer to the inode for data pages is
- * returned. On errors, one of the following negative error code is returned
- * in a pointer type.
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: a pointer to the inode for data pages on success, or %-ENOMEM
+ * if memory is insufficient.
  */
 struct inode *nilfs_iget_for_shadow(struct inode *inode)
 {
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index ce30b51ac593c..b5207e4c50007 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -741,18 +741,13 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
  * @sb: super block instance
  * @ri: pointer to a nilfs_recovery_info struct to store search results.
  *
- * Return Value: On success, 0 is returned.  On error, one of the following
- * negative error code is returned.
- *
- * %-EINVAL - Inconsistent filesystem state.
- *
- * %-EIO - I/O error
- *
- * %-ENOSPC - No space left on device (only in a panic state).
- *
- * %-ERESTARTSYS - Interrupted.
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL		- Inconsistent filesystem state.
+ * * %-EIO		- I/O error.
+ * * %-ENOMEM		- Insufficient memory available.
+ * * %-ENOSPC		- No space left on device (only in a panic state).
+ * * %-ERESTARTSYS	- Interrupted.
  */
 int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
 			      struct super_block *sb,
@@ -817,14 +812,11 @@ int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
  * segment pointed by the superblock.  It sets up struct the_nilfs through
  * this search. It fills nilfs_recovery_info (ri) required for recovery.
  *
- * Return Value: On success, 0 is returned.  On error, one of the following
- * negative error code is returned.
- *
- * %-EINVAL - No valid segment found
- *
- * %-EIO - I/O error
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- No valid segment found.
+ * * %-EIO	- I/O error.
+ * * %-ENOMEM	- Insufficient memory available.
  */
 int nilfs_search_super_root(struct the_nilfs *nilfs,
 			    struct nilfs_recovery_info *ri)
diff --git a/fs/nilfs2/segbuf.c b/fs/nilfs2/segbuf.c
index 6e59dc19a7324..7416fb6e3fd4b 100644
--- a/fs/nilfs2/segbuf.c
+++ b/fs/nilfs2/segbuf.c
@@ -403,12 +403,7 @@ static int nilfs_segbuf_submit_bh(struct nilfs_segment_buffer *segbuf,
  * @segbuf: buffer storing a log to be written
  * @nilfs: nilfs object
  *
- * Return Value: On Success, 0 is returned. On Error, one of the following
- * negative error code is returned.
- *
- * %-EIO - I/O error
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: Always 0.
  */
 static int nilfs_segbuf_write(struct nilfs_segment_buffer *segbuf,
 			      struct the_nilfs *nilfs)
@@ -449,10 +444,7 @@ static int nilfs_segbuf_write(struct nilfs_segment_buffer *segbuf,
  * nilfs_segbuf_wait - wait for completion of requested BIOs
  * @segbuf: segment buffer
  *
- * Return Value: On Success, 0 is returned. On Error, one of the following
- * negative error code is returned.
- *
- * %-EIO - I/O error
+ * Return: 0 on success, or %-EIO if I/O error is detected.
  */
 static int nilfs_segbuf_wait(struct nilfs_segment_buffer *segbuf)
 {
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index a412abdfbbf3e..92631394c8d29 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -191,12 +191,10 @@ static int nilfs_prepare_segment_lock(struct super_block *sb,
  * When @vacancy_check flag is set, this function will check the amount of
  * free space, and will wait for the GC to reclaim disk space if low capacity.
  *
- * Return Value: On success, 0 is returned. On error, one of the following
- * negative error code is returned.
- *
- * %-ENOMEM - Insufficient memory available.
- *
- * %-ENOSPC - No space left on device
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-ENOSPC	- No space left on device (if checking free space).
  */
 int nilfs_transaction_begin(struct super_block *sb,
 			    struct nilfs_transaction_info *ti,
@@ -2286,18 +2284,13 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
  * nilfs_construct_segment - construct a logical segment
  * @sb: super block
  *
- * Return Value: On success, 0 is returned. On errors, one of the following
- * negative error code is returned.
- *
- * %-EROFS - Read only filesystem.
- *
- * %-EIO - I/O error
- *
- * %-ENOSPC - No space left on device (only in a panic state).
- *
- * %-ERESTARTSYS - Interrupted.
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO		- I/O error (including metadata corruption).
+ * * %-ENOMEM		- Insufficient memory available.
+ * * %-ENOSPC		- No space left on device (only in a panic state).
+ * * %-ERESTARTSYS	- Interrupted.
+ * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_segment(struct super_block *sb)
 {
@@ -2321,18 +2314,13 @@ int nilfs_construct_segment(struct super_block *sb)
  * @start: start byte offset
  * @end: end byte offset (inclusive)
  *
- * Return Value: On success, 0 is returned. On errors, one of the following
- * negative error code is returned.
- *
- * %-EROFS - Read only filesystem.
- *
- * %-EIO - I/O error
- *
- * %-ENOSPC - No space left on device (only in a panic state).
- *
- * %-ERESTARTSYS - Interrupted.
- *
- * %-ENOMEM - Insufficient memory available.
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO		- I/O error (including metadata corruption).
+ * * %-ENOMEM		- Insufficient memory available.
+ * * %-ENOSPC		- No space left on device (only in a panic state).
+ * * %-ERESTARTSYS	- Interrupted.
+ * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_dsync_segment(struct super_block *sb, struct inode *inode,
 				  loff_t start, loff_t end)
diff --git a/fs/nilfs2/the_nilfs.c b/fs/nilfs2/the_nilfs.c
index f9b10b4e22c01..013aea9dc355d 100644
--- a/fs/nilfs2/the_nilfs.c
+++ b/fs/nilfs2/the_nilfs.c
@@ -50,8 +50,8 @@ void nilfs_set_last_segment(struct the_nilfs *nilfs,
  * alloc_nilfs - allocate a nilfs object
  * @sb: super block instance
  *
- * Return Value: On success, pointer to the_nilfs is returned.
- * On error, NULL is returned.
+ * Return: a pointer to the allocated nilfs object on success, or NULL on
+ * failure.
  */
 struct the_nilfs *alloc_nilfs(struct super_block *sb)
 {
@@ -202,8 +202,7 @@ static int nilfs_store_log_cursor(struct the_nilfs *nilfs,
  * exponent information written in @sbp and stores it in @blocksize,
  * or aborts with an error message if it's too large.
  *
- * Return Value: On success, 0 is returned. If the block size is too
- * large, -EINVAL is returned.
+ * Return: 0 on success, or %-EINVAL if the block size is too large.
  */
 static int nilfs_get_blocksize(struct super_block *sb,
 			       struct nilfs_super_block *sbp, int *blocksize)
@@ -546,7 +545,7 @@ static int nilfs_valid_sb(struct nilfs_super_block *sbp)
  * area, or if the parameters themselves are not normal, it is
  * determined to be invalid.
  *
- * Return Value: true if invalid, false if valid.
+ * Return: true if invalid, false if valid.
  */
 static bool nilfs_sb2_bad_offset(struct nilfs_super_block *sbp, u64 offset)
 {
@@ -677,8 +676,7 @@ static int nilfs_load_super_block(struct the_nilfs *nilfs,
  * reading the super block, getting disk layout information, initializing
  * shared fields in the_nilfs).
  *
- * Return Value: On success, 0 is returned. On error, a negative error
- * code is returned.
+ * Return: 0 on success, or a negative error code on failure.
  */
 int init_nilfs(struct the_nilfs *nilfs, struct super_block *sb, char *data)
 {
-- 
2.53.0

[PATCH 6.1.y 2/2] nilfs2: ignore non-fatal signals during synchronous log write wait

[Sasha Levin]

From: Ryusuke Konishi <konishi.ryusuke@gmail.com>

[ Upstream commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0 ]

Based on a syzbot report, Wang Jianjian pointed out that following a
successful directory operation in nilfs_rename(), a synchronous write
requested within nilfs_transaction_commit(), triggered by a sync flag
set by nilfs_commit_chunk(), could be interrupted by a user signal.
This interruption leaves in-memory metadata updates intact while
returning an error, causing inconsistency with the VFS dentry cache and
ultimately triggering a kernel warning.

According to "signal(7)", local disks are not classified as "slow
devices" (such as terminals, FIFOs, or pipes) and their I/O operations
on disk devices are not interrupted by signals.  In kernel space,
waiting for such disk I/O should therefore not be aborted by ordinary
(non-fatal) signals.

Suppress this issue by modifying nilfs_segctor_sync(), which waits for
the log writer thread to complete synchronous writes, to catch only
fatal signals via fatal_signal_pending() and return -EINTR rather than
-ERESTARTSYS.  To complement this change and prevent the wait loop from
busy-looping when schedule() returns immediately due to a pending
non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
TASK_KILLABLE.

This serves as an effective interim stabilization measure without
structural changes until a comprehensive fix for metadata consistency
is implemented.

Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Cc: Wang Jianjian <wangjianjian3@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/nilfs2/ioctl.c    |  2 +-
 fs/nilfs2/recovery.c |  2 +-
 fs/nilfs2/segment.c  | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index 4e964549bc4ce..cbbe49bb4926b 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
  * * %-EFAULT		- Failure during execution of requested operation.
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index b5207e4c50007..af1b2a6a93db7 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -743,11 +743,11 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EINVAL		- Inconsistent filesystem state.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  */
 int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
 			      struct super_block *sb,
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index 92631394c8d29..e1f4859b2db54 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -2234,7 +2234,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 	wake_up(&sci->sc_wait_daemon);
 
 	for (;;) {
-		set_current_state(TASK_INTERRUPTIBLE);
+		set_current_state(TASK_KILLABLE);
 
 		/*
 		 * Synchronize only while the log writer thread is alive.
@@ -2248,11 +2248,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 			err = wait_req.err;
 			break;
 		}
-		if (!signal_pending(current)) {
+		if (!fatal_signal_pending(current)) {
 			schedule();
 			continue;
 		}
-		err = -ERESTARTSYS;
+		err = -EINTR;
 		break;
 	}
 	finish_wait(&sci->sc_wait_request, &wait_req.wq);
@@ -2286,10 +2286,10 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error (including metadata corruption).
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_segment(struct super_block *sb)
@@ -2316,10 +2316,10 @@ int nilfs_construct_segment(struct super_block *sb)
  *
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error (including metadata corruption).
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 int nilfs_construct_dsync_segment(struct super_block *sb, struct inode *inode,
-- 
2.53.0

[PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[Wang Jianjian]

From: Ryusuke Konishi <konishi.ryusuke@gmail.com>

Based on a syzbot report, Wang Jianjian pointed out that following a
successful directory operation in nilfs_rename(), a synchronous write
requested within nilfs_transaction_commit(), triggered by a sync flag
set by nilfs_commit_chunk(), could be interrupted by a user signal.
This interruption leaves in-memory metadata updates intact while
returning an error, causing inconsistency with the VFS dentry cache and
ultimately triggering a kernel warning.

According to "signal(7)", local disks are not classified as "slow
devices" (such as terminals, FIFOs, or pipes) and their I/O operations
on disk devices are not interrupted by signals.  In kernel space,
waiting for such disk I/O should therefore not be aborted by ordinary
(non-fatal) signals.

Suppress this issue by modifying nilfs_segctor_sync(), which waits for
the log writer thread to complete synchronous writes, to catch only
fatal signals via fatal_signal_pending() and return -EINTR rather than
-ERESTARTSYS.  To complement this change and prevent the wait loop from
busy-looping when schedule() returns immediately due to a pending
non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
TASK_KILLABLE.

This serves as an effective interim stabilization measure without
structural changes until a comprehensive fix for metadata consistency
is implemented.

Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Cc: Wang Jianjian <wangjianjian3@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
(cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
---
 fs/nilfs2/ioctl.c    |  2 +-
 fs/nilfs2/recovery.c |  2 ++
 fs/nilfs2/segment.c  | 10 +++++++---
 3 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index 4e964549bc4c..cbbe49bb4926 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
  * * %-EFAULT		- Failure during execution of requested operation.
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index ce30b51ac593..34462039ae7d 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -744,6 +744,8 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
  * Return Value: On success, 0 is returned.  On error, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EINVAL - Inconsistent filesystem state.
  *
  * %-EIO - I/O error
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index a412abdfbbf3..3a256f786db2 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -2236,7 +2236,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 	wake_up(&sci->sc_wait_daemon);
 
 	for (;;) {
-		set_current_state(TASK_INTERRUPTIBLE);
+		set_current_state(TASK_KILLABLE);
 
 		/*
 		 * Synchronize only while the log writer thread is alive.
@@ -2250,11 +2250,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 			err = wait_req.err;
 			break;
 		}
-		if (!signal_pending(current)) {
+		if (!fatal_signal_pending(current)) {
 			schedule();
 			continue;
 		}
-		err = -ERESTARTSYS;
+		err = -EINTR;
 		break;
 	}
 	finish_wait(&sci->sc_wait_request, &wait_req.wq);
@@ -2289,6 +2289,8 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
  * Return Value: On success, 0 is returned. On errors, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EROFS - Read only filesystem.
  *
  * %-EIO - I/O error
@@ -2324,6 +2326,8 @@ int nilfs_construct_segment(struct super_block *sb)
  * Return Value: On success, 0 is returned. On errors, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EROFS - Read only filesystem.
  *
  * %-EIO - I/O error
-- 
2.34.1

[PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[Wang Jianjian]

From: Ryusuke Konishi <konishi.ryusuke@gmail.com>

Based on a syzbot report, Wang Jianjian pointed out that following a
successful directory operation in nilfs_rename(), a synchronous write
requested within nilfs_transaction_commit(), triggered by a sync flag
set by nilfs_commit_chunk(), could be interrupted by a user signal.
This interruption leaves in-memory metadata updates intact while
returning an error, causing inconsistency with the VFS dentry cache and
ultimately triggering a kernel warning.

According to "signal(7)", local disks are not classified as "slow
devices" (such as terminals, FIFOs, or pipes) and their I/O operations
on disk devices are not interrupted by signals.  In kernel space,
waiting for such disk I/O should therefore not be aborted by ordinary
(non-fatal) signals.

Suppress this issue by modifying nilfs_segctor_sync(), which waits for
the log writer thread to complete synchronous writes, to catch only
fatal signals via fatal_signal_pending() and return -EINTR rather than
-ERESTARTSYS.  To complement this change and prevent the wait loop from
busy-looping when schedule() returns immediately due to a pending
non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
TASK_KILLABLE.

This serves as an effective interim stabilization measure without
structural changes until a comprehensive fix for metadata consistency
is implemented.

Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
Cc: Wang Jianjian <wangjianjian3@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
(cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
---
 fs/nilfs2/ioctl.c    |  2 +-
 fs/nilfs2/recovery.c |  4 ++--
 fs/nilfs2/segment.c  | 14 +++++++-------
 3 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index 4e964549bc4c..cbbe49bb4926 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
  * Return: 0 on success, or one of the following negative error codes on
  * failure:
  * * %-EFAULT		- Failure during execution of requested operation.
+ * * %-EINTR		- Interrupted.
  * * %-EIO		- I/O error.
  * * %-ENOMEM		- Insufficient memory available.
  * * %-ENOSPC		- No space left on device (only in a panic state).
- * * %-ERESTARTSYS	- Interrupted.
  * * %-EROFS		- Read only filesystem.
  */
 static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index ce30b51ac593..249be54f7732 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -744,14 +744,14 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
  * Return Value: On success, 0 is returned.  On error, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EINVAL - Inconsistent filesystem state.
  *
  * %-EIO - I/O error
  *
  * %-ENOSPC - No space left on device (only in a panic state).
  *
- * %-ERESTARTSYS - Interrupted.
- *
  * %-ENOMEM - Insufficient memory available.
  */
 int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index a412abdfbbf3..2355ed02045c 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -2236,7 +2236,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 	wake_up(&sci->sc_wait_daemon);
 
 	for (;;) {
-		set_current_state(TASK_INTERRUPTIBLE);
+		set_current_state(TASK_KILLABLE);
 
 		/*
 		 * Synchronize only while the log writer thread is alive.
@@ -2250,11 +2250,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
 			err = wait_req.err;
 			break;
 		}
-		if (!signal_pending(current)) {
+		if (!fatal_signal_pending(current)) {
 			schedule();
 			continue;
 		}
-		err = -ERESTARTSYS;
+		err = -EINTR;
 		break;
 	}
 	finish_wait(&sci->sc_wait_request, &wait_req.wq);
@@ -2289,14 +2289,14 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
  * Return Value: On success, 0 is returned. On errors, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EROFS - Read only filesystem.
  *
  * %-EIO - I/O error
  *
  * %-ENOSPC - No space left on device (only in a panic state).
  *
- * %-ERESTARTSYS - Interrupted.
- *
  * %-ENOMEM - Insufficient memory available.
  */
 int nilfs_construct_segment(struct super_block *sb)
@@ -2324,14 +2324,14 @@ int nilfs_construct_segment(struct super_block *sb)
  * Return Value: On success, 0 is returned. On errors, one of the following
  * negative error code is returned.
  *
+ * %-EINTR - Interrupted.
+ *
  * %-EROFS - Read only filesystem.
  *
  * %-EIO - I/O error
  *
  * %-ENOSPC - No space left on device (only in a panic state).
  *
- * %-ERESTARTSYS - Interrupted.
- *
  * %-ENOMEM - Insufficient memory available.
  */
 int nilfs_construct_dsync_segment(struct super_block *sb, struct inode *inode,
-- 
2.34.1

Re: [PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[Ryusuke Konishi]

On Wed, Nov 11, 2026 at 10:33 AM Wang Jianjian wrote:
>
> From: Ryusuke Konishi <konishi.ryusuke@gmail.com>
>
> Based on a syzbot report, Wang Jianjian pointed out that following a
> successful directory operation in nilfs_rename(), a synchronous write
> requested within nilfs_transaction_commit(), triggered by a sync flag
> set by nilfs_commit_chunk(), could be interrupted by a user signal.
> This interruption leaves in-memory metadata updates intact while
> returning an error, causing inconsistency with the VFS dentry cache and
> ultimately triggering a kernel warning.
>
> According to "signal(7)", local disks are not classified as "slow
> devices" (such as terminals, FIFOs, or pipes) and their I/O operations
> on disk devices are not interrupted by signals.  In kernel space,
> waiting for such disk I/O should therefore not be aborted by ordinary
> (non-fatal) signals.
>
> Suppress this issue by modifying nilfs_segctor_sync(), which waits for
> the log writer thread to complete synchronous writes, to catch only
> fatal signals via fatal_signal_pending() and return -EINTR rather than
> -ERESTARTSYS.  To complement this change and prevent the wait loop from
> busy-looping when schedule() returns immediately due to a pending
> non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
> TASK_KILLABLE.
>
> This serves as an effective interim stabilization measure without
> structural changes until a comprehensive fix for metadata consistency
> is implemented.
>
> Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
> Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
> Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> Cc: Wang Jianjian <wangjianjian3@huawei.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
> (cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
> ---

Hi Wang Jianjian,

The patches that the fix patch depends on have also been submitted to
the stable tree, so the conflicts should be resolved.

Considering future backports, I think Sasha Levin's approach of
simultaneously sending acceptable dependency patches to resolve the
issue is preferable.

Could you please check the link below?

[6.12.y] https://lore.kernel.org/all/20261110131241.1664282-1-sashal@kernel.org/
[6.6.y] https://lore.kernel.org/all/20261110131728.1706995-1-sashal@kernel.org/
[6.1.y] https://lore.kernel.org/all/20261110132237.1722151-1-sashal@kernel.org/
[5.15.y] https://lore.kernel.org/all/20261110132713.1736389-1-sashal@kernel.org

Please let me know if your intention was different.

Thanks,
Ryusuke Konishi



>  fs/nilfs2/ioctl.c    |  2 +-
>  fs/nilfs2/recovery.c |  2 ++
>  fs/nilfs2/segment.c  | 10 +++++++---
>  3 files changed, 10 insertions(+), 4 deletions(-)
>
> diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
> index 4e964549bc4c..cbbe49bb4926 100644
> --- a/fs/nilfs2/ioctl.c
> +++ b/fs/nilfs2/ioctl.c
> @@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
>   * Return: 0 on success, or one of the following negative error codes on
>   * failure:
>   * * %-EFAULT          - Failure during execution of requested operation.
> + * * %-EINTR           - Interrupted.
>   * * %-EIO             - I/O error.
>   * * %-ENOMEM          - Insufficient memory available.
>   * * %-ENOSPC          - No space left on device (only in a panic state).
> - * * %-ERESTARTSYS     - Interrupted.
>   * * %-EROFS           - Read only filesystem.
>   */
>  static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
> diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
> index ce30b51ac593..34462039ae7d 100644
> --- a/fs/nilfs2/recovery.c
> +++ b/fs/nilfs2/recovery.c
> @@ -744,6 +744,8 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
>   * Return Value: On success, 0 is returned.  On error, one of the following
>   * negative error code is returned.
>   *
> + * %-EINTR - Interrupted.
> + *
>   * %-EINVAL - Inconsistent filesystem state.
>   *
>   * %-EIO - I/O error
> diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
> index a412abdfbbf3..3a256f786db2 100644
> --- a/fs/nilfs2/segment.c
> +++ b/fs/nilfs2/segment.c
> @@ -2236,7 +2236,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>         wake_up(&sci->sc_wait_daemon);
>
>         for (;;) {
> -               set_current_state(TASK_INTERRUPTIBLE);
> +               set_current_state(TASK_KILLABLE);
>
>                 /*
>                  * Synchronize only while the log writer thread is alive.
> @@ -2250,11 +2250,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>                         err = wait_req.err;
>                         break;
>                 }
> -               if (!signal_pending(current)) {
> +               if (!fatal_signal_pending(current)) {
>                         schedule();
>                         continue;
>                 }
> -               err = -ERESTARTSYS;
> +               err = -EINTR;
>                 break;
>         }
>         finish_wait(&sci->sc_wait_request, &wait_req.wq);
> @@ -2289,6 +2289,8 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
>   * Return Value: On success, 0 is returned. On errors, one of the following
>   * negative error code is returned.
>   *
> + * %-EINTR - Interrupted.
> + *
>   * %-EROFS - Read only filesystem.
>   *
>   * %-EIO - I/O error
> @@ -2324,6 +2326,8 @@ int nilfs_construct_segment(struct super_block *sb)
>   * Return Value: On success, 0 is returned. On errors, one of the following
>   * negative error code is returned.
>   *
> + * %-EINTR - Interrupted.
> + *
>   * %-EROFS - Read only filesystem.
>   *
>   * %-EIO - I/O error
> --
> 2.34.1
>

Re: [PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[wangjianjian (C)]

Hi,
Please ignore this patch(I resend a new version). Sorry for this.
在 2026/11/11 9:31, Wang Jianjian 写道:
> From: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> 
> Based on a syzbot report, Wang Jianjian pointed out that following a
> successful directory operation in nilfs_rename(), a synchronous write
> requested within nilfs_transaction_commit(), triggered by a sync flag
> set by nilfs_commit_chunk(), could be interrupted by a user signal.
> This interruption leaves in-memory metadata updates intact while
> returning an error, causing inconsistency with the VFS dentry cache and
> ultimately triggering a kernel warning.
> 
> According to "signal(7)", local disks are not classified as "slow
> devices" (such as terminals, FIFOs, or pipes) and their I/O operations
> on disk devices are not interrupted by signals.  In kernel space,
> waiting for such disk I/O should therefore not be aborted by ordinary
> (non-fatal) signals.
> 
> Suppress this issue by modifying nilfs_segctor_sync(), which waits for
> the log writer thread to complete synchronous writes, to catch only
> fatal signals via fatal_signal_pending() and return -EINTR rather than
> -ERESTARTSYS.  To complement this change and prevent the wait loop from
> busy-looping when schedule() returns immediately due to a pending
> non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
> TASK_KILLABLE.
> 
> This serves as an effective interim stabilization measure without
> structural changes until a comprehensive fix for metadata consistency
> is implemented.
> 
> Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
> Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
> Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> Cc: Wang Jianjian <wangjianjian3@huawei.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
> (cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
> ---
>   fs/nilfs2/ioctl.c    |  2 +-
>   fs/nilfs2/recovery.c |  2 ++
>   fs/nilfs2/segment.c  | 10 +++++++---
>   3 files changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
> index 4e964549bc4c..cbbe49bb4926 100644
> --- a/fs/nilfs2/ioctl.c
> +++ b/fs/nilfs2/ioctl.c
> @@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
>    * Return: 0 on success, or one of the following negative error codes on
>    * failure:
>    * * %-EFAULT		- Failure during execution of requested operation.
> + * * %-EINTR		- Interrupted.
>    * * %-EIO		- I/O error.
>    * * %-ENOMEM		- Insufficient memory available.
>    * * %-ENOSPC		- No space left on device (only in a panic state).
> - * * %-ERESTARTSYS	- Interrupted.
>    * * %-EROFS		- Read only filesystem.
>    */
>   static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
> diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
> index ce30b51ac593..34462039ae7d 100644
> --- a/fs/nilfs2/recovery.c
> +++ b/fs/nilfs2/recovery.c
> @@ -744,6 +744,8 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
>    * Return Value: On success, 0 is returned.  On error, one of the following
>    * negative error code is returned.
>    *
> + * %-EINTR - Interrupted.
> + *
>    * %-EINVAL - Inconsistent filesystem state.
>    *
>    * %-EIO - I/O error
> diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
> index a412abdfbbf3..3a256f786db2 100644
> --- a/fs/nilfs2/segment.c
> +++ b/fs/nilfs2/segment.c
> @@ -2236,7 +2236,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>   	wake_up(&sci->sc_wait_daemon);
>   
>   	for (;;) {
> -		set_current_state(TASK_INTERRUPTIBLE);
> +		set_current_state(TASK_KILLABLE);
>   
>   		/*
>   		 * Synchronize only while the log writer thread is alive.
> @@ -2250,11 +2250,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>   			err = wait_req.err;
>   			break;
>   		}
> -		if (!signal_pending(current)) {
> +		if (!fatal_signal_pending(current)) {
>   			schedule();
>   			continue;
>   		}
> -		err = -ERESTARTSYS;
> +		err = -EINTR;
>   		break;
>   	}
>   	finish_wait(&sci->sc_wait_request, &wait_req.wq);
> @@ -2289,6 +2289,8 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
>    * Return Value: On success, 0 is returned. On errors, one of the following
>    * negative error code is returned.
>    *
> + * %-EINTR - Interrupted.
> + *
>    * %-EROFS - Read only filesystem.
>    *
>    * %-EIO - I/O error
> @@ -2324,6 +2326,8 @@ int nilfs_construct_segment(struct super_block *sb)
>    * Return Value: On success, 0 is returned. On errors, one of the following
>    * negative error code is returned.
>    *
> + * %-EINTR - Interrupted.
> + *
>    * %-EROFS - Read only filesystem.
>    *
>    * %-EIO - I/O error

-- 
Regards

Re: [PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[wangjianjian (C)]

在 2026/11/11 9:59, Ryusuke Konishi 写道:
> On Wed, Nov 11, 2026 at 10:33 AM Wang Jianjian wrote:
>>
>> From: Ryusuke Konishi <konishi.ryusuke@gmail.com>
>>
>> Based on a syzbot report, Wang Jianjian pointed out that following a
>> successful directory operation in nilfs_rename(), a synchronous write
>> requested within nilfs_transaction_commit(), triggered by a sync flag
>> set by nilfs_commit_chunk(), could be interrupted by a user signal.
>> This interruption leaves in-memory metadata updates intact while
>> returning an error, causing inconsistency with the VFS dentry cache and
>> ultimately triggering a kernel warning.
>>
>> According to "signal(7)", local disks are not classified as "slow
>> devices" (such as terminals, FIFOs, or pipes) and their I/O operations
>> on disk devices are not interrupted by signals.  In kernel space,
>> waiting for such disk I/O should therefore not be aborted by ordinary
>> (non-fatal) signals.
>>
>> Suppress this issue by modifying nilfs_segctor_sync(), which waits for
>> the log writer thread to complete synchronous writes, to catch only
>> fatal signals via fatal_signal_pending() and return -EINTR rather than
>> -ERESTARTSYS.  To complement this change and prevent the wait loop from
>> busy-looping when schedule() returns immediately due to a pending
>> non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
>> TASK_KILLABLE.
>>
>> This serves as an effective interim stabilization measure without
>> structural changes until a comprehensive fix for metadata consistency
>> is implemented.
>>
>> Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
>> Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
>> Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
>> Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
>> Cc: Wang Jianjian <wangjianjian3@huawei.com>
>> Cc: stable@vger.kernel.org
>> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
>> Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
>> (cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
>> ---
> 
> Hi Wang Jianjian,
> 
> The patches that the fix patch depends on have also been submitted to
> the stable tree, so the conflicts should be resolved.
> 
> Considering future backports, I think Sasha Levin's approach of
> simultaneously sending acceptable dependency patches to resolve the
> issue is preferable.
> 
> Could you please check the link below?
> 
> [6.12.y] https://lore.kernel.org/all/20261110131241.1664282-1-sashal@kernel.org/
> [6.6.y] https://lore.kernel.org/all/20261110131728.1706995-1-sashal@kernel.org/
> [6.1.y] https://lore.kernel.org/all/20261110132237.1722151-1-sashal@kernel.org/
> [5.15.y] https://lore.kernel.org/all/20261110132713.1736389-1-sashal@kernel.org
> 
> Please let me know if your intention was different.

Hi,
ok, I misunderstood this. sorry about this confuse.>
> Thanks,
> Ryusuke Konishi
> 
> 
> 
>>   fs/nilfs2/ioctl.c    |  2 +-
>>   fs/nilfs2/recovery.c |  2 ++
>>   fs/nilfs2/segment.c  | 10 +++++++---
>>   3 files changed, 10 insertions(+), 4 deletions(-)
>>
>> diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
>> index 4e964549bc4c..cbbe49bb4926 100644
>> --- a/fs/nilfs2/ioctl.c
>> +++ b/fs/nilfs2/ioctl.c
>> @@ -960,10 +960,10 @@ static int nilfs_ioctl_clean_segments(struct inode *inode, struct file *filp,
>>    * Return: 0 on success, or one of the following negative error codes on
>>    * failure:
>>    * * %-EFAULT          - Failure during execution of requested operation.
>> + * * %-EINTR           - Interrupted.
>>    * * %-EIO             - I/O error.
>>    * * %-ENOMEM          - Insufficient memory available.
>>    * * %-ENOSPC          - No space left on device (only in a panic state).
>> - * * %-ERESTARTSYS     - Interrupted.
>>    * * %-EROFS           - Read only filesystem.
>>    */
>>   static int nilfs_ioctl_sync(struct inode *inode, struct file *filp,
>> diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
>> index ce30b51ac593..34462039ae7d 100644
>> --- a/fs/nilfs2/recovery.c
>> +++ b/fs/nilfs2/recovery.c
>> @@ -744,6 +744,8 @@ static void nilfs_abort_roll_forward(struct the_nilfs *nilfs)
>>    * Return Value: On success, 0 is returned.  On error, one of the following
>>    * negative error code is returned.
>>    *
>> + * %-EINTR - Interrupted.
>> + *
>>    * %-EINVAL - Inconsistent filesystem state.
>>    *
>>    * %-EIO - I/O error
>> diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
>> index a412abdfbbf3..3a256f786db2 100644
>> --- a/fs/nilfs2/segment.c
>> +++ b/fs/nilfs2/segment.c
>> @@ -2236,7 +2236,7 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>>          wake_up(&sci->sc_wait_daemon);
>>
>>          for (;;) {
>> -               set_current_state(TASK_INTERRUPTIBLE);
>> +               set_current_state(TASK_KILLABLE);
>>
>>                  /*
>>                   * Synchronize only while the log writer thread is alive.
>> @@ -2250,11 +2250,11 @@ static int nilfs_segctor_sync(struct nilfs_sc_info *sci)
>>                          err = wait_req.err;
>>                          break;
>>                  }
>> -               if (!signal_pending(current)) {
>> +               if (!fatal_signal_pending(current)) {
>>                          schedule();
>>                          continue;
>>                  }
>> -               err = -ERESTARTSYS;
>> +               err = -EINTR;
>>                  break;
>>          }
>>          finish_wait(&sci->sc_wait_request, &wait_req.wq);
>> @@ -2289,6 +2289,8 @@ static void nilfs_segctor_wakeup(struct nilfs_sc_info *sci, int err, bool force)
>>    * Return Value: On success, 0 is returned. On errors, one of the following
>>    * negative error code is returned.
>>    *
>> + * %-EINTR - Interrupted.
>> + *
>>    * %-EROFS - Read only filesystem.
>>    *
>>    * %-EIO - I/O error
>> @@ -2324,6 +2326,8 @@ int nilfs_construct_segment(struct super_block *sb)
>>    * Return Value: On success, 0 is returned. On errors, one of the following
>>    * negative error code is returned.
>>    *
>> + * %-EINTR - Interrupted.
>> + *
>>    * %-EROFS - Read only filesystem.
>>    *
>>    * %-EIO - I/O error
>> --
>> 2.34.1
>>

-- 
Regards

Re: [PATCH 6.1.y] nilfs2: ignore non-fatal signals during synchronous log write wait

[Ryusuke Konishi]

On Wed, Nov 11, 2026 at 11:10 AM wangjianjian (C) wrote:
>
>
> 在 2026/11/11 9:59, Ryusuke Konishi 写道:
> > On Wed, Nov 11, 2026 at 10:33 AM Wang Jianjian wrote:
> >>
> >> From: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> >>
> >> Based on a syzbot report, Wang Jianjian pointed out that following a
> >> successful directory operation in nilfs_rename(), a synchronous write
> >> requested within nilfs_transaction_commit(), triggered by a sync flag
> >> set by nilfs_commit_chunk(), could be interrupted by a user signal.
> >> This interruption leaves in-memory metadata updates intact while
> >> returning an error, causing inconsistency with the VFS dentry cache and
> >> ultimately triggering a kernel warning.
> >>
> >> According to "signal(7)", local disks are not classified as "slow
> >> devices" (such as terminals, FIFOs, or pipes) and their I/O operations
> >> on disk devices are not interrupted by signals.  In kernel space,
> >> waiting for such disk I/O should therefore not be aborted by ordinary
> >> (non-fatal) signals.
> >>
> >> Suppress this issue by modifying nilfs_segctor_sync(), which waits for
> >> the log writer thread to complete synchronous writes, to catch only
> >> fatal signals via fatal_signal_pending() and return -EINTR rather than
> >> -ERESTARTSYS.  To complement this change and prevent the wait loop from
> >> busy-looping when schedule() returns immediately due to a pending
> >> non-fatal signal, switch the wait state from TASK_INTERRUPTIBLE to
> >> TASK_KILLABLE.
> >>
> >> This serves as an effective interim stabilization measure without
> >> structural changes until a comprehensive fix for metadata consistency
> >> is implemented.
> >>
> >> Reported-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> >> Closes: https://syzkaller.appspot.com/bug?extid=f6c7e1f1809f235eeb90
> >> Fixes: 9ff05123e3bf ("nilfs2: segment constructor")
> >> Tested-by: syzbot+f6c7e1f1809f235eeb90@syzkaller.appspotmail.com
> >> Cc: Wang Jianjian <wangjianjian3@huawei.com>
> >> Cc: stable@vger.kernel.org
> >> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> >> Signed-off-by: Viacheslav Dubeyko <slava@dubeyko.com>
> >> (cherry picked from commit 7131aaa74bcc7c9da1f1d1c8724da75c775557a0)
> >> ---
> >
> > Hi Wang Jianjian,
> >
> > The patches that the fix patch depends on have also been submitted to
> > the stable tree, so the conflicts should be resolved.
> >
> > Considering future backports, I think Sasha Levin's approach of
> > simultaneously sending acceptable dependency patches to resolve the
> > issue is preferable.
> >
> > Could you please check the link below?
> >
> > [6.12.y] https://lore.kernel.org/all/20261110131241.1664282-1-sashal@kernel.org/
> > [6.6.y] https://lore.kernel.org/all/20261110131728.1706995-1-sashal@kernel.org/
> > [6.1.y] https://lore.kernel.org/all/20261110132237.1722151-1-sashal@kernel.org/
> > [5.15.y] https://lore.kernel.org/all/20261110132713.1736389-1-sashal@kernel.org
> >
> > Please let me know if your intention was different.
>
> Hi,
> ok, I misunderstood this. sorry about this confuse.>

No worries.
Thanks for taking the time to put together patches for multiple branches anyway.

Thanks,
Ryusuke Konishi