From: Harald W. <hwa...@ha...> - 2003-02-24 10:23:36
|
Am Freitag, 21.02.03 um 14:27 Uhr schrieb Alexey Kotovich: > Hi there; > > I have the same experience some time ago. > I wrote a patch for netatalk-1.5.0/3 that can help you to workaround > problem. > If my memory doesn't fail me the issue is as follows (as short as > possible): > > AFP uses unique file number for internal needs which is two bytes of > length. > Netatalk does something like this ((st->st_dev << 16) | (st->st_ino & > 0x0000ffff)) > to get such a number. I'd like to mention that it must be unique per > request only. > As xfs has 64bits inode number and it has inconsequent order, > sometimes we get > the same file numbers in the same request that causes "error -50" on > MacOS. > > kind regards, > Alexey Kotovich ICQ:97715595 Is this still true with cnid (netatalk 1.6.0)? Or does cnid decouple did from file inode information? If no, could it be made into doing so? Regards, Harald -- Harald Wagener * FCB/Wilkens * An der Alster 42 * 20099 Hamburg |