Corinna Vinschen <corinna-cygwin <at> cygwin.com> writes: > > FYI, I've applied a patch today which has two purposes: > > - For directories, which are used as mount points, the d_ino was the one > returned by Windows for this very directory entry. However, for mount > points, the inode number which should be returned (and which *is* > returned by stat) is the inode number of the target directory/ target > drive, the mount point points to. This is (hopefully correctly) > accomplished by the patch.
Worked for me on a self-built dll! Thanks for an awesome patch. > > Now, with the patch, they are listed even though they don't exists in > reality on the drive. The additional advantage is that the shell's > file completion now also works for these paths. Now, what would be really cool is if the link count of directories with mount points could be updated, to match the fact that readdir can now see mount points: $ cd /usr/ $ ls -dl . drwxr-xr-x+ 21 eblake Domain Users 0 Mar 1 09:22 ./ $ mkdir /cygdrive/c/cygwin/usr/lib $ ls -dl . drwxr-xr-x+ 22 eblake Domain Users 0 Mar 1 09:26 ./ Oops - see how the link count changed from 21 to 22 when I had a physical backing directory? find (and any other tool that optimizes a recursive search to look at (link count - 2) subdirectories and then quit) might get confused by the fact that the number of subdirs given by readdir is no longer consistent with the link count. -- Eric Blake -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
