The patch titled
r/o bind mounts: add vfsmount writer counts
has been added to the -mm tree. Its filename is
r-o-bind-mounts-add-vfsmount-writer-counts.patch
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: r/o bind mounts: add vfsmount writer counts
From: Dave Hansen <hansendc@xxxxxxxxxx>
This patch actually adds the mount and superblock writer counts, and the
mnt_want/drop_write() functions that use them.
Before these can become useful, we must first cover each place in the VFS
where writes are performed with a want/drop pair. When that is complete, we
can actually introduce code that will safely check the counts before allowing
r/w<->r/o transitions to occur.
Signed-off-by: Dave Hansen <haveblue@xxxxxxxxxx>
Cc: Miklos Szeredi <miklos@xxxxxxxxxx>
Cc: Ram Pai <linuxram@xxxxxxxxxx>
Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
fs/namespace.c | 53 ++++++++++++++++++++++++++++++++++++++++
fs/super.c | 18 ++++++++++---
include/linux/fs.h | 2 +
include/linux/mount.h | 28 ++++++++++++++++++---
4 files changed, 94 insertions(+), 7 deletions(-)
diff -puN fs/namespace.c~r-o-bind-mounts-add-vfsmount-writer-counts fs/namespace.c
--- a/fs/namespace.c~r-o-bind-mounts-add-vfsmount-writer-counts
+++ a/fs/namespace.c
@@ -56,6 +56,7 @@ struct vfsmount *alloc_vfsmnt(const char
struct vfsmount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL);
if (mnt) {
atomic_set(&mnt->mnt_count, 1);
+ mnt->mnt_writers = 0;
INIT_LIST_HEAD(&mnt->mnt_hash);
INIT_LIST_HEAD(&mnt->mnt_child);
INIT_LIST_HEAD(&mnt->mnt_mounts);
@@ -76,6 +77,56 @@ struct vfsmount *alloc_vfsmnt(const char
return mnt;
}
+int mnt_make_readonly(struct vfsmount *mnt)
+{
+ int ret = 0;
+
+ WARN_ON(__mnt_is_readonly(mnt));
+
+ /*
+ * This flag set is actually redundant with what
+ * happens in do_remount(), but since we do this
+ * under the lock, anyone attempting to get a write
+ * on it after this will fail.
+ */
+ spin_lock(&mnt->mnt_sb->s_mnt_writers_lock);
+ if (!mnt->mnt_writers)
+ mnt->mnt_flags |= MNT_READONLY;
+ else
+ ret = -EBUSY;
+ spin_unlock(&mnt->mnt_sb->s_mnt_writers_lock);
+ return ret;
+}
+
+int mnt_want_write(struct vfsmount *mnt)
+{
+ int ret = 0;
+
+ spin_lock(&mnt->mnt_sb->s_mnt_writers_lock);
+ if (mnt->mnt_writers)
+ goto out;
+
+ if (__mnt_is_readonly(mnt)) {
+ ret = -EROFS;
+ goto out;
+ }
+ mnt->mnt_sb->s_writers++;
+ mnt->mnt_writers++;
+out:
+ spin_unlock(&mnt->mnt_sb->s_mnt_writers_lock);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(mnt_want_write);
+
+void mnt_drop_write(struct vfsmount *mnt)
+{
+ spin_lock(&mnt->mnt_sb->s_mnt_writers_lock);
+ mnt->mnt_sb->s_writers--;
+ mnt->mnt_writers--;
+ spin_unlock(&mnt->mnt_sb->s_mnt_writers_lock);
+}
+EXPORT_SYMBOL_GPL(mnt_drop_write);
+
int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
{
mnt->mnt_sb = sb;
@@ -1395,6 +1446,8 @@ long do_mount(char *dev_name, char *dir_
((char *)data_page)[PAGE_SIZE - 1] = 0;
/* Separate the per-mountpoint flags */
+ if (flags & MS_RDONLY)
+ mnt_flags |= MNT_READONLY;
if (flags & MS_NOSUID)
mnt_flags |= MNT_NOSUID;
if (flags & MS_NODEV)
diff -puN fs/super.c~r-o-bind-mounts-add-vfsmount-writer-counts fs/super.c
--- a/fs/super.c~r-o-bind-mounts-add-vfsmount-writer-counts
+++ a/fs/super.c
@@ -93,6 +93,8 @@ static struct super_block *alloc_super(s
s->s_qcop = sb_quotactl_ops;
s->s_op = &default_op;
s->s_time_gran = 1000000000;
+ s->s_writers = 0;
+ spin_lock_init(&s->s_mnt_writers_lock);
}
out:
return s;
@@ -576,6 +578,11 @@ static void mark_files_ro(struct super_b
file_list_unlock();
}
+static int sb_remount_ro(struct super_block *sb)
+{
+ return fs_may_remount_ro(sb);
+}
+
/**
* do_remount_sb - asks filesystem to change mount options.
* @sb: superblock in question
@@ -587,7 +594,8 @@ static void mark_files_ro(struct super_b
*/
int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
{
- int retval;
+ int retval = 0;
+ int sb_started_ro = (sb->s_flags & MS_RDONLY);
#ifdef CONFIG_BLOCK
if (!(flags & MS_RDONLY) && bdev_read_only(sb->s_bdev))
@@ -600,11 +608,13 @@ int do_remount_sb(struct super_block *sb
/* If we are remounting RDONLY and current sb is read/write,
make sure there are no rw files opened */
- if ((flags & MS_RDONLY) && !(sb->s_flags & MS_RDONLY)) {
+ if ((flags & MS_RDONLY) && !sb_started_ro) {
if (force)
mark_files_ro(sb);
- else if (!fs_may_remount_ro(sb))
- return -EBUSY;
+ else
+ retval = sb_remount_ro(sb);
+ if (retval)
+ return retval;
}
if (sb->s_op->remount_fs) {
diff -puN include/linux/fs.h~r-o-bind-mounts-add-vfsmount-writer-counts include/linux/fs.h
--- a/include/linux/fs.h~r-o-bind-mounts-add-vfsmount-writer-counts
+++ a/include/linux/fs.h
@@ -970,6 +970,8 @@ struct super_block {
struct list_head s_io; /* parked for writeback */
struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
struct list_head s_files;
+ int s_writers; /* number of files open for write */
+ spinlock_t s_mnt_writers_lock; /* taken when mounts change rw state */
struct block_device *s_bdev;
struct list_head s_instances;
diff -puN include/linux/mount.h~r-o-bind-mounts-add-vfsmount-writer-counts include/linux/mount.h
--- a/include/linux/mount.h~r-o-bind-mounts-add-vfsmount-writer-counts
+++ a/include/linux/mount.h
@@ -28,6 +28,7 @@ struct mnt_namespace;
#define MNT_NOATIME 0x08
#define MNT_NODIRATIME 0x10
#define MNT_RELATIME 0x20
+#define MNT_READONLY 0x40 /* does the user want this to be r/o? */
#define MNT_SHRINKABLE 0x100
@@ -54,12 +55,14 @@ struct vfsmount {
struct vfsmount *mnt_master; /* slave is on master->mnt_slave_list */
struct mnt_namespace *mnt_ns; /* containing namespace */
/*
- * We put mnt_count & mnt_expiry_mark at the end of struct vfsmount
- * to let these frequently modified fields in a separate cache line
- * (so that reads of mnt_flags wont ping-pong on SMP machines)
+ * We put mnt_count, mnt_expiry_mark, and mnt_writers at the end of
+ * struct vfsmount to let these frequently modified fields in a
+ * separate cache line (so that reads of mnt_flags wont ping-pong
+ * on SMP machines)
*/
atomic_t mnt_count;
int mnt_expiry_mark; /* true if marked for expiry */
+ int mnt_writers; /* nr files open for write */
int mnt_pinned;
};
@@ -69,7 +72,26 @@ static inline struct vfsmount *mntget(st
atomic_inc(&mnt->mnt_count);
return mnt;
}
+/*
+ * This is temporary for now. We also don't want to check
+ * the SB in because it is already checked in other
+ * code paths. We'll have a better way to do this in
+ * the end of this series
+ */
+static inline int __mnt_is_readonly(struct vfsmount *mnt)
+{
+ return mnt->mnt_flags & MNT_READONLY;
+}
+
+static inline void __mnt_unmake_readonly(struct vfsmount *mnt)
+{
+ WARN_ON(!__mnt_is_readonly(mnt));
+ mnt->mnt_flags &= ~MNT_READONLY;
+}
+extern int mnt_make_readonly(struct vfsmount *mnt);
+extern int mnt_want_write(struct vfsmount *mnt);
+extern void mnt_drop_write(struct vfsmount *mnt);
extern void mntput_no_expire(struct vfsmount *mnt);
extern void mnt_pin(struct vfsmount *mnt);
extern void mnt_unpin(struct vfsmount *mnt);
_
Patches currently in -mm which might be from hansendc@xxxxxxxxxx are
r-o-bind-mounts-filesystem-helpers-for-custom-struct-files.patch
r-o-bind-mounts-add-vfsmount-writer-counts.patch
r-o-bind-mounts-record-when-sb_writer_count-elevated-for-inode.patch
r-o-bind-mounts-elevate-writer-count-for-chown-and-friends.patch
r-o-bind-mounts-elevate-mnt-writers-for-callers-of-vfs_mkdir.patch
r-o-bind-mounts-elevate-write-count-during-entire-ncp_ioctl.patch
r-o-bind-mounts-elevate-write-count-for-link-and-symlink-calls.patch
r-o-bind-mounts-elevate-mount-count-for-extended-attributes.patch
r-o-bind-mounts-mount_is_safe-add-comment.patch
r-o-bind-mounts-unix_find_other-elevate-write-count-for-touch_atime.patch
r-o-bind-mounts-elevate-write-count-over-calls-to-vfs_rename.patch
r-o-bind-mounts-elevate-write-count-files-are-opened.patch
r-o-bind-mounts-elevate-writer-count-for-do_sys_truncate.patch
r-o-bind-mounts-elevate-write-count-for-do_utimes.patch
r-o-bind-mounts-elevate-write-count-for-do_sys_utime-and-touch_atime.patch
r-o-bind-mounts-sys_mknodat-elevate-write-count-for-vfs_mknod-create.patch
r-o-bind-mounts-elevate-mnt-writers-for-vfs_unlink-callers.patch
r-o-bind-mounts-do_rmdir-elevate-write-count.patch
r-o-bind-mounts-elevate-writer-count-for-custom-struct_file.patch
r-o-bind-mounts-gfs-check-nlink-count.patch
r-o-bind-mounts-honor-r-w-changes-at-do_remount-time.patch
r-o-bind-mounts-kill-open-files-traverse-on-remount-ro.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
