Currently, when the top and bottom 32-bit words are equivalent and the
host is a 32-bit arch, cifs_uniqueid_to_ino_t returns 0 as the ino_t
value. All we're doing to hash the value down to 32 bits is xor'ing the
top and bottom 32-bit words and that obviously results in 0 if they are
equivalent.
The kernel doesn't really care if it returns this value, but some
userland apps (like "ls") will ignore dirents that have a zero d_ino
value.
Change this function to use hash_64 to convert this value to a 31 bit
value and then add 1 to ensure that it doesn't ever return 0. Also,
there's no need to check the sizeof(ino_t) at runtime so create two
different cifs_uniqueid_to_ino_t functions based on whether
BITS_PER_LONG is 64 for not.
This should fix:
https://bugzilla.kernel.org/show_bug.cgi?id=19282
Reported-by: Eric <copet_eric@emc.com>
Reported-by: <per-ola@sadata.se>
Signed-off-by: Jeff Layton <jlayton@poochiereds.net>
Signed-off-by: Steve French <smfrench@gmail.com>
#ifndef _CIFSFS_H
#define _CIFSFS_H
+#include <linux/hash.h>
+
#define ROOT_I 2
/*
* ino_t is 32-bits on 32-bit arch. We have to squash the 64-bit value down
- * so that it will fit.
+ * so that it will fit. We use hash_64 to convert the value to 31 bits, and
+ * then add 1, to ensure that we don't end up with a 0 as the value.
*/
+#if BITS_PER_LONG == 64
static inline ino_t
cifs_uniqueid_to_ino_t(u64 fileid)
{
- ino_t ino = (ino_t) fileid;
- if (sizeof(ino_t) < sizeof(u64))
- ino ^= fileid >> (sizeof(u64)-sizeof(ino_t)) * 8;
- return ino;
+ return (ino_t)fileid;
}
+#else
+static inline ino_t
+cifs_uniqueid_to_ino_t(u64 fileid)
+{
+ return (ino_t)hash_64(fileid, (sizeof(ino_t) * 8) - 1) + 1;
+}
+#endif
extern struct file_system_type cifs_fs_type;
extern const struct address_space_operations cifs_addr_ops;