This version is mainly a collection of patches for minixfs pl10. Details are contained below. Many thanks to all concerned! The only addition I have made is a -z option to fsck to clear troublesome inodes. See fsck.doc for more details. ************************************** initial revision ************************************** makefile fixes ************************************** increase cache sizes ************************************** From: inf03@Uni-Trier.de (Sascha Blank) Message-Id: <9409051059.AA07720@Uni-Trier.De> Subject: Better personae management for MiNT 1.11beta use effective instead of real uid for access checking ************************************** From: Andreas Schwab Date: Mon, 17 Oct 94 10:46:00 +0100 Message-Id: <9410170946.AA02849@issan.informatik.uni-dortmund.de> To: mint@atari.archive.umich.edu Subject: MinixFS: directory creation und symlinks There are two other places where MinixFS uses the ruid instead of the euid, that is when it creates a directory or a symlink (although for the latter it does not matter). Also, the size of a symlink node is set to the length + 1, but Linux (and probably Minix too) expects it to be the length without the terminating null. And neither Linux nor Minix knows anything about drive specifiers, but MiNTlib always adds U: to an absolute path, making it difficult to, for example, extract the Linux root filesystem with its symlinks under MiNT. I have fixed this by always stripping the U: prefix on symlink creation and adding it back in readlink. This isn't fully backward compatible since before an absolute link target without drive has been interpreted relative to the current drive, but i don't thinks this is much of a problem. Another difference is that the link size reported by getxattr may be off by two, but MiNT always uses PATH_MAX when reading a link. If this creates a problem then getxattr will have to check if the symlink name starts with a slash and add two to the size in this case. ************************************** From: Ulrich Kuehn Message-Id: <9411021016.AA21609@math.uni-muenster.de> Subject: Re: Sync() and shutdown() system call config.h, filesys.h, minixfs.c: use Sync() and Shutdown() system calls ************************************** From: Juergen Lock Subject: Re: MinixFS 0.60 PL10 Patch Collection Message-Id: <9411122007.AA00358@jelal.north.de> . fix bitmaps Kmalloc calculation for larger filesystems . a hack for fsck to check for sparse directories . clear bitmap dirty bits on `unmounts' (Dlock), you'll get warnings about bitmaps not written out after fsck runs sometimes, don't ask me why... as long as fsck found no errors you can ignore them . make sure symlinks always go in the same cache too ************************************** From: Ulrich Kuehn Message-Id: <9411180940.AA27680@math.uni-muenster.de> Subject: Re: sync patch for minixfs pl 10 don't run the update daemon if SYSUPDATE def'd ************************************** From: scott@telle.dmi.stevens-tech.edu (Scott Kolodzieski) Message-Id: <9411250353.AA09142@telle.dmi.stevens-tech.edu> Subject: minor diff for minixfs pl10 really don't run the update daemon if MiNT >= 1.12 ************************************** From: woodford@esca.com (Steven Woodford) Message-Id: <9411290411.AA05910@flash.esca.com> Subject: One more patch for MinixFs PL10 don't delete "." or ".." even if dir increment > 1 ************************************** From: Andreas Schwab Message-Id: <9412160954.AA28843@issan.informatik.uni-dortmund.de> Subject: MinixFS: completed support for V2 filesystem add support for v2 triple indirection blocks ------------------------------------------------------------------------------