[PATCH 26/27] fs: Refuse to freeze filesystem with open but unlinked files

From: Jan Kara
Date: Mon Apr 16 2012 - 12:17:42 EST


Filesystem with frozen but unlinked files cannot be forced into
a fully consistent state. Also handling of closing such files on frozen
filesystem is problematic since fput() can be called with mmap_sem held
and protection against frozen filesystem ranks above it.

BugLink: https://bugs.launchpad.net/bugs/897421
Tested-by: Kamal Mostafa <kamal@xxxxxxxxxxxxx>
Tested-by: Peter M. Petrakis <peter.petrakis@xxxxxxxxxxxxx>
Tested-by: Dann Frazier <dann.frazier@xxxxxxxxxxxxx>
Tested-by: Massimo Morana <massimo.morana@xxxxxxxxxxxxx>
Signed-off-by: Jan Kara <jack@xxxxxxx>
---
fs/super.c | 26 +++++++++++++++++++-------
1 files changed, 19 insertions(+), 7 deletions(-)

diff --git a/fs/super.c b/fs/super.c
index 7b2c81c..2bd3632 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -1485,16 +1485,22 @@ int freeze_super(struct super_block *sb)
smp_wmb();
sb_wait_write(sb, SB_FREEZE_FS);

+ /*
+ * Check whether there are some open unlinked files (since all writers
+ * are blocked, new ones cannot be created now so the check is
+ * reliable). In that case filesystem cannot be forced into a
+ * consistent state so we just bail out.
+ */
+ if (atomic_long_read(&sb->s_remove_count)) {
+ ret = -EBUSY;
+ goto bail;
+ }
+
if (sb->s_op->freeze_fs) {
ret = sb->s_op->freeze_fs(sb);
if (ret) {
- printk(KERN_ERR
- "VFS:Filesystem freeze failed\n");
- sb->s_writers.frozen = SB_UNFROZEN;
- smp_wmb();
- wake_up(&sb->s_writers.wait_unfrozen);
- deactivate_locked_super(sb);
- return ret;
+ printk(KERN_ERR "VFS: Filesystem freeze failed\n");
+ goto bail;
}
}
/*
@@ -1504,6 +1510,12 @@ int freeze_super(struct super_block *sb)
sb->s_writers.frozen = SB_FREEZE_COMPLETE;
up_write(&sb->s_umount);
return 0;
+bail:
+ sb->s_writers.frozen = SB_UNFROZEN;
+ smp_wmb();
+ wake_up(&sb->s_writers.wait_unfrozen);
+ deactivate_locked_super(sb);
+ return ret;
}
EXPORT_SYMBOL(freeze_super);

--
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/