IEEE 1003.1-2004 says that open() must not have created or modified a
file if -1 is returned.
But this happens with the released Linux kernel 3.0.0, x86_64.
This issue happens when you run test 079 from xfstests on an ext2, ext3,
ext4 or btrfs filesystem. This test 079 makes use of the
append-only flag. It is set on the directory and inherited to the
created file. The test 079 creates the situation that during a call
to open(..., O_CREAT | O_RDWR, ...) the creation of the file is
allowed but the opening for writing is not allowed.
The expected behaviour would be that open() returns -1 and that the
file _is_ _not_ created.
The recognized behaviour is that -1 is returned but the file _is_ created.
mkdir("/mnt3/foo/append-only.d", 0777) = 0
open("/mnt3/foo/append-only.d", O_RDONLY) = 3
ioctl(3, FS_IOC32_SETFLAGS or FS_IOC_SETFLAGS, 0x7fffaf60b07c) = 0
(this ioctl enables FS_APPEND_FL for the directory)
close(3)
open("/mnt3/foo/append-only.d/newfile-0", O_RDWR|O_CREAT, 0666) = -1
EPERM (Operation not permitted)
But you can see that an empty file was created although -1 was
returned.
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
