vfs patches, beginning
Assorted stuff for that cycle, part 1. There will be more tomorrow.
Most of that coincides with the last for-next, with several more fixes added. Two patches (do_lookup() ones) are dropped from queue, due to unfixable problems they cause for autofs4; broken autofs4 patch that went in during the last cycle gets reverted. write_inode branch is *not* included into today's queue due to conflicts in XFS.
do_filp_open() patch series will go in tomorrow pile.
Please, pull from the usual place:git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6.git/ for-linus
Shortlog:
Al Viro (37):
Lose the new_name argument of fsnotify_move()
Lose the first argument of audit_inode_child()
switch inotify_user to anon_inode
Revert "autofs4: always use lookup for lookup"
Switch may_open() and break_lease() to passing O_...
Switch gfs2 to nd_set_link()
Kill CL_PROPAGATION, sanitize fs/pnode.c:get_source()
add several pieces to shared subtree documentation
switch infiniband uverbs to anon_inodes
Sanitize autofs_dev_ioctl_ismountpoint()
Get rid of mnt_mountpoint abuses in ext4
Use kill_litter_super() in autofs4 ->kill_sb()
Don't bother with d_genocide in rpc_pipe
Simplify failure exits in s390/hypfs fill_super()
New helper: path_is_under(path1, path2)
Switch alloc_nfs_open_context() to struct path
a couple of mntget+dget -> path_get in nfs4proc
fix mnt_mountpoint abuse in smack
Clean follow_dotdot() up a bit
fix NFS4 handling of mountpoint stat
new helper: iterate_mounts()
sanitize signedness/const for pointers to char in hpfs a bit sanitize const/signedness of ufs a bit
nilfs really shouldn't slap struct dentry on stack...
nilfs: sanitize const/signedness in dealing with ->d_name.name sanitize const/signedness for udf
Don't mess with generic_permission() under ->d_lock in hpfs
take check for new events in namespace (guts of mounts_poll()) to namespace.c get rid of insanity with namespace roots in tomoyo
Take vfsmount_lock to fs/internal.h
get rid of useless vfsmount_lock use in put_mnt_ns()
Mirror MS_KERNMOUNT in ->mnt_flags
hppfs can use existing proc_mnt, no need for do_kern_mount() in there get rid of ->mnt_parent in tomoyo/realpath
set S_DEAD on unlink() and non-directory rename() victims
fix race in d_splice_alias()
Merge branch 'for-fsnotify' into for-linus
André Goddard Rosa (6):
mqueue: fix mq_open() file descriptor leak on user-space processes
mqueue: remove unneeded info->messages initialization
mqueue: apply mathematics distributivity on mq_bytes calculation
mqueue: simplify do_open() error handling
mqueue: only set error codes if they are really necessary
mqueue: fix typo "failues" -> "failures"
Boaz Harrosh (2):
libfs: Open code simple_commit_write into only user
libfs: Unexport and kill simple_prepare_write
Christoph Hellwig (1):
kill unused invalidate_inode_pages helper
Eric W. Biederman (1):
init: Open /dev/console from rootfs
H Hartley Sweeten (1):
fs/dcache.c: CodingStyle cleanup
Helight.Xu (1):
some clean up in fs/proc
Miklos Szeredi (1):
vfs: add NOFOLLOW flag to umount(2)
Nick Piggin (1):
fs: improve remount,ro vs buffercache coherency
Richard Kennedy (1):
fs: re-order super_block to remove 16 bytes of padding on 64bit builds
Valerie Aurora (1):
VFS: Clean up shared mount flag propagation
Diffstat:
Documentation/filesystems/sharedsubtree.txt | 16 +- arch/s390/hypfs/inode.c | 42 +-- arch/um/drivers/mconsole_kern.c | 2 +- drivers/infiniband/Kconfig | 1 + drivers/infiniband/core/uverbs.h | 2 +- drivers/infiniband/core/uverbs_cmd.c | 25 ++- drivers/infiniband/core/uverbs_main.c | 82 +----- drivers/usb/gadget/f_mass_storage.c | 2 +- drivers/usb/gadget/file_storage.c | 2 +- fs/autofs4/autofs_i.h | 7 - fs/autofs4/dev-ioctl.c | 11 +- fs/autofs4/expire.c | 6 +- fs/autofs4/inode.c | 63 +---- fs/autofs4/root.c | 474 +++++++++------------------ fs/cifs/file.c | 4 +- fs/dcache.c | 70 +++-- fs/debugfs/inode.c | 2 +- fs/ext4/file.c | 6 +- fs/gfs2/ops_inode.c | 113 ++----- fs/hpfs/anode.c | 2 +- fs/hpfs/dentry.c | 14 +- fs/hpfs/dir.c | 14 +- fs/hpfs/dnode.c | 21 +- fs/hpfs/ea.c | 7 +- fs/hpfs/hpfs_fn.h | 30 +- fs/hpfs/inode.c | 4 +- fs/hpfs/map.c | 6 +- fs/hpfs/name.c | 21 +- fs/hpfs/namei.c | 75 +++-- fs/hppfs/hppfs.c | 2 +- fs/internal.h | 2 + fs/libfs.c | 77 +++--- fs/locks.c | 5 +- fs/namei.c | 44 +-- fs/namespace.c | 53 +++- fs/nfs/inode.c | 8 +- fs/nfs/nfs4proc.c | 8 +- fs/nfsctl.c | 5 +- fs/nfsd/nfs4xdr.c | 12 +- fs/nfsd/vfs.c | 4 +- fs/nilfs2/dir.c | 14 +- fs/nilfs2/namei.c | 13 +- fs/nilfs2/nilfs.h | 4 +- fs/notify/inotify/inotify_user.c | 59 +--- fs/open.c | 2 +- fs/pnode.c | 28 +- fs/pnode.h | 5 +- fs/proc/base.c | 10 +- fs/proc/generic.c | 5 + fs/proc/root.c | 6 - fs/super.c | 21 +- fs/udf/balloc.c | 2 +- fs/udf/dir.c | 4 +- fs/udf/inode.c | 2 +- fs/udf/namei.c | 20 +- fs/udf/symlink.c | 10 +- fs/ufs/dir.c | 10 +- fs/ufs/ufs.h | 4 +- include/linux/audit.h | 11 +- include/linux/fs.h | 26 +- include/linux/fsnotify.h | 11 +- include/linux/magic.h | 1 - include/linux/mnt_namespace.h | 1 + include/linux/mount.h | 14 +- init/do_mounts_initrd.c | 4 - init/main.c | 11 +- ipc/mqueue.c | 120 ++++--- kernel/audit_tree.c | 100 ++---- kernel/auditsc.c | 7 +- kernel/sysctl_binary.c | 7 +- mm/filemap.c | 2 +- net/sunrpc/rpc_pipe.c | 9 +- security/smack/smack_lsm.c | 2 +- security/tomoyo/realpath.c | 23 +- 74 files changed, 777 insertions(+), 1135 deletions(-)
Al Viro:
Quoted 1 line
Assorted stuff for that cycle, part 1. There will be more tomorrow.
:::
Quoted 1 line
set
S_DEADonunlink()and non-directoryrename()victims
If the inode is hardlinked, it will survive.
Is setting S_DEAD correct in such case?
J. R. Okajima