Problems with linuxppc_2_2 on an MVME-2400

[Michael Lundkvist <ml@epact.se>]

I rsync:ed against FSMLabs repository yesterday and got linuxppc_2_2
to play with on my MVME-2400.

I've got a MVME-2400 VME-board with a Ramix PMC-disk (based an a
CMD-646 controller with an IBM disk).

I can get it to boot but as soon as I start any moderate disk activity
I get a panic in find_buffer().

NIP: C0030594 XER: 00000000 LR: C00305D4 REGS: cf99fc50 TRAP: 0300
MSR: 00009032 [EEIRDRME]
TASK = cf99e000[26] 'bzip2' mm->pgd cfcca000 Last syscall: 3
last math cf9d4000
GPR00: 0003FFFF CF99FD00 CF99E000 0180C200 00010CE2 00001000 00000302 C0100000
GPR08: 00433880 0180C200 CFF00000 C0100000 24244858 1001D3C4 00000000 C015DDE0
GPR16: C0109A78 C00243B8 C0100000 00000001 C0030000 00000001 00000000 00000302
GPR24: CF99FD8C 00000000 C04FD198 CCAFEEA0 00001000 CCAFEEB8 CCAFEEA0 00010CE2
Call backtrace:
00000001 C0031718 C0031AD8 C0023B88 C0024050 C00244AC C002E75C
C0003924 FFFFFFFF 0FF13E10 0FF151B8 0FF149EC 0FF15594 0FF0FAE0
0FFDD464 10001064 1000296C 10003AD8 0FEDF75C 00000000
Kernel panic: kernel access of bad area pc c0030594 lr c00305d4 address 180C204 tsk bzip2/26
Rebooting in 180 seconds..
>>NIP; c0030594 <find_buffer+64/94>   <=====
Trace; 00000001 Before first symbol
Trace; c0031718 <brw_page+134/3a0>
Trace; c0031ad8 <generic_readpage+a8/c0>
Trace; c0023b88 <try_to_read_ahead+17c/1a0>
Trace; c0024050 <do_generic_file_read+300/668>
Trace; c00244ac <generic_file_read+68/a0>
Trace; c002e75c <sys_read+fc/120>
Trace; c0003924 <syscall_ret_1+0/a0>
Trace; ffffffff <END_OF_CODE+3fe99f63/????>
Trace; 0ff13e10 Before first symbol
Trace; 0ff151b8 Before first symbol
Trace; 0ff149ec Before first symbol
Trace; 0ff15594 Before first symbol
Trace; 0ff0fae0 Before first symbol
Trace; 0ffdd464 Before first symbol
Trace; 10001064 Before first symbol
Trace; 1000296c Before first symbol
Trace; 10003ad8 Before first symbol
Trace; 0fedf75c Before first symbol
Trace; 00000000 Before first symbol


I've tried several times and the panic is always in the same place.

This is the disassemble of the find_buffer function:
c0030530 <find_buffer>:
c0030530:       3d 20 c0 10     lis     r9,-16368
c0030534:       81 69 74 38     lwz     r11,29752(r9)
c0030538:       7c 66 1b 78     mr      r6,r3
c003053c:       39 2b ff fa     addi    r9,r11,-6
c0030540:       39 0b ff f4     addi    r8,r11,-12
c0030544:       7c 8a 48 30     slw     r10,r4,r9
c0030548:       7c 80 6e 70     srawi   r0,r4,13
c003054c:       39 6b ff f7     addi    r11,r11,-9
c0030550:       7d 4a 02 78     xor     r10,r10,r0
c0030554:       7c cb 58 30     slw     r11,r6,r11
c0030558:       7c 88 40 30     slw     r8,r4,r8
c003055c:       7c c9 48 30     slw     r9,r6,r9
c0030560:       3c e0 c0 10     lis     r7,-16368
c0030564:       7d 29 5a 78     xor     r9,r9,r11
c0030568:       7d 4a 42 78     xor     r10,r10,r8
c003056c:       80 07 74 34     lwz     r0,29748(r7)
c0030570:       7d 29 52 78     xor     r9,r9,r10
c0030574:       3d 60 c0 10     lis     r11,-16368
c0030578:       7d 29 00 38     and     r9,r9,r0
c003057c:       81 4b 74 3c     lwz     r10,29756(r11)
c0030580:       55 29 10 3a     rlwinm  r9,r9,2,0,29
c0030584:       7c 69 50 2e     lwzx    r3,r9,r10
c0030588:       2c 03 00 00     cmpwi   r3,0
c003058c:       7c 69 1b 78     mr      r9,r3
c0030590:       4d 82 00 20     beqlr
c0030594:       80 09 00 04     lwz     r0,4(r9)
c0030598:       80 63 00 00     lwz     r3,0(r3)
c003059c:       7c 00 20 00     cmpw    r0,r4
c00305a0:       40 82 ff e8     bne     c0030588 <find_buffer+0x58>
c00305a4:       80 09 00 08     lwz     r0,8(r9)
c00305a8:       7c 00 28 00     cmpw    r0,r5
c00305ac:       40 82 ff dc     bne     c0030588 <find_buffer+0x58>
c00305b0:       a0 09 00 0c     lhz     r0,12(r9)
c00305b4:       7c 00 30 00     cmpw    r0,r6
c00305b8:       40 82 ff d0     bne     c0030588 <find_buffer+0x58>
c00305bc:       7d 23 4b 78     mr      r3,r9
c00305c0:       4e 80 00 20     blr


I have a pre-compiled Debian kernel that works, but I can't seem to
find the source for exactly that version.

This is what the working version claims to be:
Linux version 2.2.16pre4 (drow@crack.them.org) (gcc version 2.95.2 20000313 (Debian GNU/Linux)) #1 Tue Jun 6 17:34:50 CDT 2000

Does anyone have any hints on how to solve this?

/Micke

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/