> Hm. strace suggests my pwd is walking up the path. But WTF would it
> break? 2.4.7 was fine. What did I break?
Walking the path works so:
open ..
read it, entry by entry
find an entry that would have inode number equal to that of our directory.
We had found the last component of our name. Lather, rinse, repeat.
It relies on numbers from stat() and numbers from readdir() being
in sync. It's not true on so many filesystems that this algorithm
is laughable. Anything with synthetic inode numbers breaks it.
IOW, your pwd(1) behaviour is b0rken. Aside of being unable to work on a
lot of filesystems, it's doing a lot of extra work even on the filesystems
where it happens to work.
Check what your getcwd(3) is doing. It should at least try to call
getcwd(2) before reverting to that horror. Notice that you need
to walk the path only on 2.0 - 2.2 and above have a syscall that
works without any IO and regardless of the inumber assignment policy.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/