Booting a SPARC machine off a Linux NFS server

uFu · 02-20-2007, 09:52 PM

Hi,

I have a Voyager box that I salvaged from a company house-cleaning (it's a luggable UltraSPARC IIi box). I'm trying to get it to boot over a LAN from a Fedora Core 5 x86 box that's set up as a boot server, the ultimate goal being to install Solaris 10 on the Voyager. (I don't have access to a SCSI DVD-ROM reader, so the easy way is out

)

All of the ARP/RARP bootparam and TFTP stuff is working, to the point where the Voyager is loading its initial boot image and trying to NFS-mount its root filesystem from the Linux box. The symptom on the Voyager is a not-very-helpful "boot: cannot open path/to/kernel". The output of tcpdump suggests that there may be a permission issue of some sort:

tcpdump output:
22:11:09.795562 IP (tos 0x0, ttl 255, id 1, offset 0, flags [DF], proto: UDP (17), length: 104) tadpole.biff > 255.255.255.255.sunrpc: [udp sum ok] UDP, length 76
22:11:10.098129 IP (tos 0x0, ttl 64, id 47, offset 0, flags [DF], proto: UDP (17), length: 104) 192.168.1.201.sunrpc > tadpole.biff: [bad udp cksum f945!] UDP, length 76
22:11:10.109304 arp who-has 192.168.1.201 (Broadcast) tell tadpole
22:11:10.109333 arp reply 192.168.1.201 is-at 00:50:ba:d6:35:29 (oui Unknown)

This looks like the Voyager (tadpole) is doing a broadcast to find an RPC server, it gets a response from x.x.x.201 (my Linux box), then ARPs it to get its MAC address (I'm making educated guesses here; I'm not all that familiar with RPC innards). The only thing that looks strange to me so far is that all of the UDP packets from the Linux box to tadpole are flagged for a bad UDP checksum.

22:11:10.114850 IP (tos 0x0, ttl 255, id 2, offset 0, flags [DF], proto: UDP (17), length: 88) tadpole.who > 192.168.1.201.ggf-ncp: [udp sum ok] UDP, length 60
22:11:10.115459 IP (tos 0x0, ttl 64, id 19, offset 0, flags [DF], proto: UDP (17), length: 136) 192.168.1.201.ggf-ncp > tadpole.who: [bad udp cksum f8c6!] UDP, length 108
22:11:10.129835 IP (tos 0x0, ttl 255, id 3, offset 0, flags [DF], proto: UDP (17), length: 92) tadpole.syslog > 192.168.1.201.ggf-ncp: [udp sum ok] [|syslog]
22:11:10.130373 IP (tos 0x0, ttl 64, id 20, offset 0, flags [DF], proto: UDP (17), length: 108) 192.168.1.201.ggf-ncp > tadpole.syslog: [bad udp cksum 3beb!] [|syslog]
22:11:10.145027 IP (tos 0x0, ttl 64, id 4, offset 0, flags [none], proto: TCP (6), length: 64) tadpole.822083584 > 192.168.1.201.nfs: 0 proc-4294967241

This looks like some initial NFS chatter, but that "proc-4294967241" looks suspicious. That number in hex is FFFFFFC9, or decimal 201 sign-extended out to 32-bits?

22:11:10.145087 IP (tos 0x0, ttl 64, id 54409, offset 0, flags [DF], proto: TCP (6), length: 52) 192.168.1.201.nfs > tadpole.printer: ., cksum 0x5d15 (correct), 2653093931:2653093931(0) ack 204995 win 2252 <nop,nop,timestamp 1784132 35200>
22:11:10.159975 IP (tos 0x0, ttl 64, id 5, offset 0, flags [none], proto: TCP (6), length: 40) tadpole.printer > 192.168.1.201.nfs: R, cksum 0x3e4a (correct), 204995:204995(0) win 49640
22:11:13.476121 IP (tos 0x0, ttl 64, id 6, offset 0, flags [none], proto: TCP (6), length: 64) tadpole.822083584 > 192.168.1.201.nfs: 0 proc-4294967241
22:11:13.476178 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto: TCP (6), length: 60) 192.168.1.201.nfs > tadpole.825111088: reply ERR 0

Same proc-4294967241, but this one generates an error?

22:11:13.491054 IP (tos 0x0, ttl 64, id 7, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x12c2 (correct), 44084:44084(0) ack 1670348019 win 50680 <nop,nop,timestamp 34480 1784965>
22:11:13.585797 IP (tos 0x0, ttl 64, id 8, offset 0, flags [none], proto: TCP (6), length: 128) tadpole.34492 > 192.168.1.201.nfs: 76 null
22:11:13.585832 IP (tos 0x0, ttl 64, id 16791, offset 0, flags [DF], proto: TCP (6), length: 52) 192.168.1.201.nfs > tadpole.printer: ., cksum 0xd23c (correct), 1670348019:1670348019(0) ack 44160 win 1448 <nop,nop,timestamp 1784993 34590>
22:11:13.586245 IP (tos 0x0, ttl 64, id 16792, offset 0, flags [DF], proto: TCP (6), length: 80) 192.168.1.201.nfs > tadpole.34492: reply ok 28 null
22:11:13.603051 IP (tos 0x0, ttl 64, id 9, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x11d8 (correct), 44160:44160(0) ack 1670348047 win 50652 <nop,nop,timestamp 34610 1784993>
22:11:13.606565 IP (tos 0x0, ttl 64, id 10, offset 0, flags [none], proto: TCP (6), length: 196) tadpole.34493 > 192.168.1.201.nfs: 144 getattr fh 0,15/543977327
22:11:13.623878 IP (tos 0x0, ttl 64, id 16793, offset 0, flags [DF], proto: TCP (6), length: 116) 192.168.1.201.nfs > tadpole.34493: reply ok 64 getattr ERROR: Operation not permitted

Some more chatter, then a getattr request which triggers a permission error. The rest of this basically looks like more of the same...

22:11:13.635440 IP (tos 0x0, ttl 64, id 11, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x1121 (correct), 44304:44304(0) ack 1670348111 win 50588 <nop,nop,timestamp 34640 1785002>
22:11:13.639235 IP (tos 0x0, ttl 64, id 12, offset 0, flags [none], proto: TCP (6), length: 188) tadpole.34494 > 192.168.1.201.nfs: 136 getattr fh 0,15/543977327
22:11:13.639420 IP (tos 0x0, ttl 64, id 16794, offset 0, flags [DF], proto: TCP (6), length: 116) 192.168.1.201.nfs > tadpole.34494: reply ok 64 getattr ERROR: Operation not permitted
22:11:13.654227 IP (tos 0x0, ttl 64, id 13, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x1081 (correct), 44440:44440(0) ack 1670348175 win 50524 <nop,nop,timestamp 34660 1785006>
22:11:13.657734 IP (tos 0x0, ttl 64, id 14, offset 0, flags [none], proto: TCP (6), length: 176) tadpole.34495 > 192.168.1.201.nfs: 124 getattr fh 0,16/543647092
22:11:13.658000 IP (tos 0x0, ttl 64, id 16795, offset 0, flags [DF], proto: TCP (6), length: 116) 192.168.1.201.nfs > tadpole.34495: reply ok 64 getattr ERROR: Operation not permitted
22:11:13.672718 IP (tos 0x0, ttl 64, id 15, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x0fec (correct), 44564:44564(0) ack 1670348239 win 50460 <nop,nop,timestamp 34680 1785011>
22:11:13.676344 IP (tos 0x0, ttl 64, id 16, offset 0, flags [none], proto: TCP (6), length: 188) tadpole.34496 > 192.168.1.201.nfs: 136 getattr fh 0,15/543977327
22:11:13.677212 IP (tos 0x0, ttl 64, id 16796, offset 0, flags [DF], proto: TCP (6), length: 116) 192.168.1.201.nfs > tadpole.34496: reply ok 64 getattr ERROR: Operation not permitted
22:11:13.691330 IP (tos 0x0, ttl 64, id 17, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x0f4c (correct), 44700:44700(0) ack 1670348303 win 50396 <nop,nop,timestamp 34700 1785015>
22:11:13.694821 IP (tos 0x0, ttl 64, id 18, offset 0, flags [none], proto: TCP (6), length: 176) tadpole.34497 > 192.168.1.201.nfs: 124 getattr fh 0,16/543647092
22:11:13.696490 IP (tos 0x0, ttl 64, id 16797, offset 0, flags [DF], proto: TCP (6), length: 116) 192.168.1.201.nfs > tadpole.34497: reply ok 64 getattr ERROR: Operation not permitted
22:11:13.709811 IP (tos 0x0, ttl 64, id 19, offset 0, flags [none], proto: TCP (6), length: 52) tadpole.printer > 192.168.1.201.nfs: ., cksum 0x0eb7 (correct), 44824:44824(0) ack 1670348367 win 50332 <nop,nop,timestamp 34720 1785020>

- hosts.allow and hosts.deny are empty
- I've tried various settings in /etc/exports. Currently the entry looks like:

/solaris-install/Solaris_10/Tools/Boot *(rw,insecure,sync,no_subtree_check,insecure_locks,nohide,no_root_squash)

- The output of "exportfs" is:
/solaris-install/Solaris_10/Tools/Boot
<world>

- It seems that all of the necessary RPC services are available:

program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32768 status
100024 1 tcp 44533 status
100011 1 udp 651 rquotad
100011 2 udp 651 rquotad
100011 1 tcp 654 rquotad
100011 2 tcp 654 rquotad
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100003 4 udp 2049 nfs
100003 2 tcp 2049 nfs
100003 3 tcp 2049 nfs
100003 4 tcp 2049 nfs
100021 1 udp 32770 nlockmgr
100021 3 udp 32770 nlockmgr
100021 4 udp 32770 nlockmgr
100021 1 tcp 35225 nlockmgr
100021 3 tcp 35225 nlockmgr
100021 4 tcp 35225 nlockmgr
100005 1 udp 689 mountd
100005 1 tcp 692 mountd
100005 2 udp 689 mountd
100005 2 tcp 692 mountd
100005 3 udp 689 mountd
100005 3 tcp 692 mountd
100026 1 udp 678 bootparam

Running strace on the portmapper shows some activity when the Voyager tries to boot, but I don't see anything when stracing mountd, which suggests that it's never getting that far.

I'm at the point now where I'm basically spinning my wheels. I've googled around quite a bit and have been unable to find any references to this problem, or any clues on how to proceed beyond what I've documented here.

Does anybody have any thoughts?

TIA!