>> I'm currently trying to get coda to work on a diskless client
>> the root is /dev/ram0
>>...
>> 06:33:55 brain-wiping recoverable store
>> 06:33:55 fatal error -- Recov_CreateSeg: rvm_create_segment failed (222)
>> 06:33:55 Fatal Signal (11); pid 124 becoming a zombie...
>>...

>I looked at the code of rvm_create_segment, and I can't tell exactly
>where it fails (the errorcode is ETID, bad rvm transaction identifier).
>I'm puzzled, never seen this happen.

I have seen something similar.  I saw this happen on a block device on
NetBSD.  After some thought, but no verification, I came to the conclusion
there may be a bug in the mmap code for block devices. 

For this case, my first thoughts were that possibly /dev/ram0 may not
support mmap.  If it does, then there may be a bug in the mmap support.


-- 
Phil Nelson                    NetBSD: http://www.netbsd.org
e-mail: phil@cs.wwu.edu        Coda: http://www.coda.cs.cmu.edu
http://www.cs.wwu.edu/~phil