Well, I appear to have gotten coda to work on linuxppc (and most likely
mklinux). Both venus and vice start up and connect to each other. I also
started a venus on a 486 box I have, and it happily connects and can
create directories and appears to work fine (althoug I haven't done much
testing).

(If you want to build coda on a linuxppc system, let me know. Getting it
working involves applying some (rather rough) patches I have, and using
an older version of the c++ standard library, since the one that comes
with egcs (and linuxppc) is *not* thread safe.)

However, venus (and the coda kernel module) don't appear to be working
correctly on PPC. On the machine running both the server and client, if I
try to do 'ls /coda/' I get the following error:

[root_at_altus coda-4.6.1]# ls /coda/
ls: /coda/: No such device or address

Any ideas what is going on? Could this have something to do with some
incorrect byteswapping that is happening? PPC Linux uses a 'little-endian'
ext2fs format to be compatible with Intel linux.. Could this be causeing a
problem? Anyway, any help as to where I should look for problems or what
debugging message to turn on would be quite appreciated.

/var/log/messages also shows the following messages:

Aug 18 01:10:29 altus kernel: Coda Kernel/Venus communications (module),
v4.6.0, braam_at_cs.cmu.edu
Aug 18 01:10:29 altus kernel: coda_psdev_write: downcall, no SB!
Aug 18 01:10:35 altus kernel: coda_read_super: rootfid is
(0x1000001,0x1,0x1)
Aug 18 01:10:36 altus kernel: coda_read_super: rootinode is 1049601 dev 3
Aug 18 01:10:37 altus kernel: coda_inode_grab: coda_find_super returns
NULL.
Aug 18 01:10:39 altus kernel: coda_open: coda_inode_grab error
-6.coda_inode_grab: coda_find_super returns NULL.