Skip to contentSkip to conversation

fs: Return EAGAIN when O_NONBLOCK write should block on frozen fs

PATCH In review Jan Kara 3 participants 6 messages Nov 8, 2012 – Nov 12, 2012

Subsystems (8)
lore
Share & tools
lore

Jan Kara

author maintainer
#0 lore raw

PATCH v2 1/2 fs: Return EAGAIN when O_NONBLOCK write should block on frozen fs

When user asks for O_NONBLOCK behavior for a file descriptor, return EAGAIN instead of blocking on a frozen filesystem.

---
 fs/btrfs/file.c    |    3 ++-
 fs/cifs/file.c     |    3 ++-
 fs/fuse/file.c     |    3 ++-
 fs/ntfs/file.c     |    3 ++-
 fs/ocfs2/file.c    |    3 ++-
 fs/xfs/xfs_file.c  |    3 ++-
 include/linux/fs.h |   10 ++++++++++
 mm/filemap.c       |    3 ++-
 8 files changed, 24 insertions(+), 7 deletions(-)

This is a second version of patches to fix BSD process accounting deadlocks
on frozen filesystem. This time I used O_NONBLOCK to indicate write shouldn't block on frozen filesystem as Dave Chinner suggested.

fs/btrfs/file.c +2 -1
index 9ab1bed..6eb2e30 100644
@@ -1411,7 +1411,8 @@ static ssize_t btrfs_file_aio_write(struct kiocb *iocb,
 	ssize_t err = 0;
 	size_t count, ocount;
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 
 	mutex_lock(&inode->i_mutex);
 
fs/cifs/file.c +2 -1
index edb25b4..1629e47 100644
@@ -2448,7 +2448,8 @@ cifs_writev(struct kiocb *iocb, const struct iovec *iov,
 
 	BUG_ON(iocb->ki_pos != pos);
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 
 	/*
 	 * We need to hold the sem to be sure nobody modifies lock list
fs/fuse/file.c +2 -1
index 78d2837..641df9e 100644
@@ -947,7 +947,8 @@ static ssize_t fuse_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 		return err;
 
 	count = ocount;
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 	mutex_lock(&inode->i_mutex);
 
 	/* We can write back this queue in page reclaim */
fs/ntfs/file.c +2 -1
index 1ecf464..028b349 100644
@@ -2118,7 +2118,8 @@ static ssize_t ntfs_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 
 	BUG_ON(iocb->ki_pos != pos);
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 	mutex_lock(&inode->i_mutex);
 	ret = ntfs_file_aio_write_nolock(iocb, iov, nr_segs, &iocb->ki_pos);
 	mutex_unlock(&inode->i_mutex);
fs/ocfs2/file.c +2 -1
index 5a4ee77..93ef34d 100644
@@ -2265,7 +2265,8 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 	if (iocb->ki_left == 0)
 		return 0;
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 
 	appending = file->f_flags & O_APPEND ? 1 : 0;
 	direct_io = file->f_flags & O_DIRECT ? 1 : 0;
fs/xfs/xfs_file.c +2 -1
index aa473fa..7d8af61 100644
@@ -771,7 +771,8 @@ xfs_file_aio_write(
 	if (ocount == 0)
 		return 0;
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
 		ret = -EIO;
include/linux/fs.h +10 -0
index b33cfc9..2325e95 100644
@@ -1401,6 +1401,16 @@ static inline int sb_start_write_trylock(struct super_block *sb)
 	return __sb_start_write(sb, SB_FREEZE_WRITE, false);
 }
 
+/*
+ * sb_start_write() for writing into a file. When file has O_NONBLOCK set,
+ * we use trylock semantics, otherwise we block on frozen filesystem.
+ */
+static inline int sb_start_file_write(struct file *file)
+{
+	return __sb_start_write(file->f_mapping->host->i_sb, SB_FREEZE_WRITE,
+				!(file->f_flags & O_NONBLOCK));
+}
+
 /**
  * sb_start_pagefault - get write access to a superblock from a page fault
  * @sb: the super we write to
mm/filemap.c +2 -1
index 83efee7..3b2812b 100644
@@ -2527,7 +2527,8 @@ ssize_t generic_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 
 	BUG_ON(iocb->ki_pos != pos);
 
-	sb_start_write(inode->i_sb);
+	if (!sb_start_file_write(file))
+		return -EAGAIN;
 	mutex_lock(&inode->i_mutex);
 	ret = __generic_file_aio_write(iocb, iov, nr_segs, &iocb->ki_pos);
 	mutex_unlock(&inode->i_mutex);
Signature · 1 line
1.7.1
author maintainer
#1 lore raw

PATCH v2 2/2 fs: Fix hang with BSD accounting on frozen filesystem

When BSD process accounting is enabled and logs information to a filesystem which gets frozen, system easily becomes unusable because each attempt to account process information blocks. Thus e.g. every task gets blocked in exit.

It seems better to drop accounting information (which can already happen when filesystem is running out of space) instead of locking system up. So we open the accounting file with O_NONBLOCK.

---
 kernel/acct.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
kernel/acct.c +2 -1
index 051e071..a061116 100644
@@ -201,7 +201,8 @@ static int acct_on(struct filename *pathname)
 	struct bsd_acct_struct *acct = NULL;
 
 	/* Difference from BSD - they don't do O_APPEND */
-	file = file_open_name(pathname, O_WRONLY|O_APPEND|O_LARGEFILE, 0);
+	file = file_open_name(pathname,
+			      O_WRONLY|O_APPEND|O_LARGEFILE|O_NONBLOCK, 0);
 	if (IS_ERR(file))
 		return PTR_ERR(file);
 
Signature · 1 line
1.7.1
Reviewed-by on v2 2/2
#2 lore raw

Re: [PATCH 2/2 v2] fs: Fix hang with BSD accounting on frozen filesystem

Quoted 19 lines · Jan Kara

When BSD process accounting is enabled and logs information to a filesystem which gets frozen, system easily becomes unusable because each attempt to account process information blocks. Thus e.g. every task gets blocked in exit.

It seems better to drop accounting information (which can already happen when filesystem is running out of space) instead of locking system up. So we open the accounting file with O_NONBLOCK.

---
 kernel/acct.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
kernel/acct.c +0 -0
index 051e071..a061116 100644

kernel/acct.c static int acct_on(struct filename *pathname) · line 207

Quoted 8 lines
kernel/acct.c +2 -1
@@ -201,7 +201,8 @@ static int acct_on(struct filename *pathname)
 	struct bsd_acct_struct *acct = NULL;
 
 	/* Difference from BSD - they don't do O_APPEND */
-	file = file_open_name(pathname, O_WRONLY|O_APPEND|O_LARGEFILE, 0);
+	file = file_open_name(pathname,
+			      O_WRONLY|O_APPEND|O_LARGEFILE|O_NONBLOCK, 0);
 	if (IS_ERR(file))
 		return PTR_ERR(file);

Looks good.

Signature · 2 lines
Dave Chinner
[email protected]
#3 lore raw
In reply to Jan Kara

Il 08/11/2012 13:41, Jan Kara ha scritto:

Quoted 5 lines

When user asks for O_NONBLOCK behavior for a file descriptor, return EAGAIN instead of blocking on a frozen filesystem.

---

It seems good. Since we have changed a bit the behavior, what do you think to replace the uninterruptible sleeping with killable sleeping in this context?

Marco

#4 lore raw
In reply to Jan Kara
Quoted 4 lines · Jan Kara

When user asks for O_NONBLOCK behavior for a file descriptor, return EAGAIN instead of blocking on a frozen filesystem.

Quoted 10 lines
---
 fs/btrfs/file.c    |    3 ++-
 fs/cifs/file.c     |    3 ++-
 fs/fuse/file.c     |    3 ++-
 fs/ntfs/file.c     |    3 ++-
 fs/ocfs2/file.c    |    3 ++-
 fs/xfs/xfs_file.c  |    3 ++-
 include/linux/fs.h |   10 ++++++++++
 mm/filemap.c       |    3 ++-
 8 files changed, 24 insertions(+), 7 deletions(-)

Looks good, though you missed generic_file_splice_write(). Shouldn't this also return EAGAIN?

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
author maintainer
#5 lore raw
Quoted 16 lines · Dave Chinner
Quoted 4 lines · Jan Kara

When user asks for O_NONBLOCK behavior for a file descriptor, return EAGAIN instead of blocking on a frozen filesystem.

Quoted 10 lines
---
 fs/btrfs/file.c    |    3 ++-
 fs/cifs/file.c     |    3 ++-
 fs/fuse/file.c     |    3 ++-
 fs/ntfs/file.c     |    3 ++-
 fs/ocfs2/file.c    |    3 ++-
 fs/xfs/xfs_file.c  |    3 ++-
 include/linux/fs.h |   10 ++++++++++
 mm/filemap.c       |    3 ++-
 8 files changed, 24 insertions(+), 7 deletions(-)
Quoted 2 lines

Looks good, though you missed generic_file_splice_write(). Shouldn't this also return EAGAIN?

Yes, it should. Thanks for spotting this. I've also noticed ocfs2 misses
freeze protection in splice_write at all. I'll fix that in a separate patch and send v3.

Honza

Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR