Hi Michael, On 03/19/2013 10:18 AM, Michael Tokarev wrote: > 19.03.2013 01:49, Helge Deller wrote: >> The second part (the "if arch" check) can be dropped. > > Now I'm completely confused. So do we need any changes or not? > Does it work as-is now or doesn't?
It does work as-is, because parisc is currently only 32bit userspace with either 32- or 64-bit kernel. _Nevertheless_: The patch for the auto_fs.h header file was necessary to Linux upstream kernel to build a correct 64bit kernel which behaves exactly similiar to a 32bit Linux kernel regarding to communication to userspace: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=4f4ffc3a5398ef9bdbb32db04756d7d34e356fcf Since the autofs debian package includes a copy of this header file the autofs package would need the same patch as soon as we would build a 64bit autofs userspace package (which isn't the case up to now). Nevertheless, applying this patch to the autofs packages now would not hurt. The second part of my patch: diff -up ./daemon/automount.c.org ./daemon/automount.c --- ./daemon/automount.c.org 2013-03-06 18:09:10.000000000 +0100 +++ ./daemon/automount.c 2013-03-06 18:09:31.000000000 +0100 @@ -610,6 +610,7 @@ static size_t get_kpkt_len(void) if (strcmp(un.machine, "alpha") == 0 || strcmp(un.machine, "ia64") == 0 || strcmp(un.machine, "x86_64") == 0 || + strcmp(un.machine, "parisc64") == 0 || strcmp(un.machine, "ppc64") == 0) pkt_len += 4; isn't actually needed, because this code would only be executed if the kernel version is <= 3.3 *and* if "pkt_len" isn't a multiple of 8. In the parisc case (with the upstream patch applied) the size is always 304, and as such the check for "parisc64" would never be executed. I think think I ran into this misunderstandings during my original testing because of other issues I had during my tests. E.g. my automounter setup tried to connect only via nfsv4, while the servers only allowed nfsv3 (or even nfsv2) and as such automounting failed which I interpreted as wrong coding. In summary: - applying both patches would not hurt at all - applying the header-file patch is currently not needed, but would be needed if we have at some point a 64bit userspace. At least I think it makes sense to keep the copy of the header file in sync with upstream kernel, so I'd propose to apply the header-patch. - nevertheless, the currently available pre-built autofs package works. Sorry for the confusion, I hope I could explain it a little better this time. Thanks, Helge -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
