On Fri, 21 Apr 1995 22:47:44 -0400 
 der Mouse <mouse@Collatz.McRCIM.McGill.EDU> wrote:

 > I'm having trouble booting NetBSD/sun3 diskless.  Something inside the
 > kernel seems to be returning ENOBUFS.

I had this exact problem with a 3/50.  It's assuming that the rpc reply 
can fit into one mbuf.  It's even XXX'd in the code (the last time I 
looked...)  cgd suggested shortening the pathname...that didn't help, so 
I recompiled the kernel with double-size mbufs.  That worked, but it had 
other `interesting' side effects...

 > 
 > In the below, 132.206.82.2 is another Sun running SunOS 3.5; I'm using
 > it as rarp and tftp server for the boot (it has a SunOS 4.x bootblock
 > in its tftp area).  132.206.82.1 is collatz, a NeXT (which does not
 > support rarp as far as I can tell, or I'd be using it for everything),
 > which I am using as NFS server.  132.206.82.4 is the machine I'm trying
 > to boot.  I had to write my own bootparamd, since neither NeXT nor
 > SunOS 3.5 provides one; it's rudimentary (eg, the tables are hardcoded
 > at present rather than reading a config file), but it does seem to
 > return correct packets.  At least, all the stuff that should come from
 > the bootparam server and gets printed shows up as what I want it to be.
 > Logging from the bootparamd indicates that three queries arrive during
 > the boot process: a WHOAMI, and a GETFILE for "root", when the diskless
 > bootblock has control, and a WHOAMI once the kernel takes over.  The
 > kernel apparently never gets far enough to do any GETFILE queries.
 > 
 > Incidentally, the model number is wrong: it's a -3/150, not /160.  But
 > as far as I know the only differences are mechanical (for example, the
 > VME cage has only six slots instead of 12), so it's not surprising it
 > gets it wrong.
 > 
 > ftp://ftp.netbsd.org/pub/NetBSD/arch/sun3/netbsd-gen.gz is where I got
 > the kernel, and I just now checked and the date reported by sun-lamp is
 > over a week older than the date on the kernel file I have, and the size
 > of the .gz file is identical, so I expect it's the same one.  (My
 > netlink is sufficiently slow I'd rather not fetch it just for a
 > compare.  The md5 checksum of the gunzipped kernel I'm using is
 > 6b9a54935a7c05155e7dd669a6226098.)
 > 
 > Here's what happens when I try to boot.  (This is a ten-finger copy, so
 > please forgive any obvious typos.)  Any advice on stuff to try would be
 > welcomed.  As far as I know the open and short reported by TDR are
 > lies; the number of clocks in the open report varies from boot to boot,
 > but the short is always 0 clocks away.
 > 
 > > b ie()netbsd -s
 > Boot: ie(0,0,0)netbsd -s
 > Using IP Address 132.206.82.4 = 84CE5204
 > Booting from tftp server at 132.206.82.2 = 84CE5202
 > Downloaded 101800 bytes from tftp server.
 > 
 > Using IP Address 132.206.82.4 = 84CE5204
 > hostname: Daily-Planet.Rodents.Montreal.QC.CA
 > domainname: a4d0e0f1bb28e8e6db9dc68b26197e20
 > server name 'collatz'
 > root pathname '/export/root/daily-planet'
 > root on collatz:/export/root/daily-planet fstype nfs
 > Size: 583904+125864+81248 bytes
 > console on zs0 (ttya)
 > DDB: found symbols [46944 + 50408 bytes]
 > Copyright (c) 1982, 1986, 1989, 1991, 1993
 > The Regents of the University of California.  All rights reserved.
 > 
 > NetBSD 1.0A (GENERIC) #208: Sat Apr  8 00:13:12 EDT 1995
 >     gwr@saturn2:/home/gwr/netbsd/src/sys/arch/sun3/compile/GENERIC
 > Model: Sun 3/160
 > real mem  = 8372224
 > avail mem = 6979584
 > using 51 buffers contaning 417792 bytes of memory
 > mainbus0 (root)
 > obctl0 at mainbus0
 > idprom0 at obctl0 hostid 0x11007895
 > fpu0 at mainbus0 (mc68881)
 > obmem0 at mainbus0
 > bwtwo0 at obmem0 addr 0xff000000 (1152x900)
 > obio0 at mainbus0
 > zs0 at obio0 addr 0x20000 level 6 softpr 3
 > zs1 at obio0 addr 0x0 level 6 softpri 3
 > eeprom0 at obio0 addr 0x40000
 > clock0 at obio0 addr 0x60000 level 5
 > ie0 at obio0 addr 0xc0000 level 3 hwaddr 08:00:20:06:c8:a9
 > vmes0 at mainbus0
 > si0 at vmes0 addr 0xff200000 level 2 vector 0x40
 > scsibus0 at si0
 > vmel0 at mainbus0
 > root on ie0
 > swap on ie0
 > dump on ie0
 > nfs_boot: using network interface 'ie0'
 > ie0: TDR detected an open 53 clocks away
 > nfs_boot: client_addr=0x84ce5204
 > ie0: TDR detected a short 0 clocks away
 > panic: nfs_boot: bootparam whoami, error=55
 > Stopped at      _Debugger+0x6:  unlk    a6
 > db>
 > 
 > 					der Mouse
 > 
 > 			    mouse@collatz.mcrcim.mcgill.edu

--------------------------------------------------------------------------
Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                               Home: 408.866.1912
NAS: M/S 258-6                                          Work: 415.604.0935
Moffet Field, CA 94035                                 Pager: 415.428.6939