Miklos, This series fixes the append-only violations found by the new xfstest overlay/030. I did it to see how bad it would be compared to propagating the S_APPEND/S_IMMUTABLE flags to overlay inode. IMO, the [f]truncate patches look ok, but the last vfs_open() patch less so (?). Do you think we can get away without propagating the flags? If getflags()/setflags() were standard vfs ops, we could have handled this nicer... Those "fs specific flags" have long been a standard, so perhaps its time to make the setflags/getflags a vfs api? Amir. Amir Goldstein (4): vfs: ftruncate freeze protect backing inode vfs: ftruncate check IS_APPEND() on backing inode vfs: truncate check IS_APPEND() on backing inode vfs: open check IS_APPEND() on backing inode fs/open.c | 34 +++++++++++++++++++++------------- 1 file changed, 21 insertions(+), 13 deletions(-) -- 2.7.4
