[okl4-developer] memory address conversion from process to wombat

[Damien Schulz]

Hi there,

currently I'm trying to execute OKL4 1.5.2 including Wombat on an Samsung s3c2440 SoC.
After some time I always get the following error (console output):

<6>alignment fault:  @ 1032c, faddr=35ff75a2 in: dummyapp (50)
Segmentation fault

Code around the faulty instruction (in l4kernel):

f00102a8 <__udivsi3>:
f00102a8:   e2512001    subs    r2, r1, #1  ; 0x1
f00102ac:   01a0f00e    moveq   pc, lr
f00102b0:   3a000036    bcc f0010390 <__udivsi3+0xe8>
f00102b4:   e1500001    cmp r0, r1
f00102b8:   9a000022    bls f0010348 <__udivsi3+0xa0>
f00102bc:   e1110002    tst r1, r2
f00102c0:   0a000023    beq f0010354 <__udivsi3+0xac>
f00102c4:   e311020e    tst r1, #-536870912 ; 0xe0000000
f00102c8:   01a01181    moveq   r1, r1, lsl #3
f00102cc:   03a03008    moveq   r3, #8  ; 0x8
f00102d0:   13a03001    movne   r3, #1  ; 0x1
f00102d4:   e3510201    cmp r1, #268435456  ; 0x10000000
f00102d8:   31510000    cmpcc   r1, r0
f00102dc:   31a01201    movcc   r1, r1, lsl #4
f00102e0:   31a03203    movcc   r3, r3, lsl #4
f00102e4:   3afffffa    bcc f00102d4 <__udivsi3+0x2c>
f00102e8:   e3510102    cmp r1, #-2147483648    ; 0x80000000
f00102ec:   31510000    cmpcc   r1, r0
f00102f0:   31a01081    movcc   r1, r1, lsl #1
f00102f4:   31a03083    movcc   r3, r3, lsl #1
f00102f8:   3afffffa    bcc f00102e8 <__udivsi3+0x40>
f00102fc:   e3a02000    mov r2, #0  ; 0x0
f0010300:   e1500001    cmp r0, r1
f0010304:   20400001    subcs   r0, r0, r1
f0010308:   21822003    orrcs   r2, r2, r3
f001030c:   e15000a1    cmp r0, r1, lsr #1
f0010310:   204000a1    subcs   r0, r0, r1, lsr #1
f0010314:   218220a3    orrcs   r2, r2, r3, lsr #1
f0010318:   e1500121    cmp r0, r1, lsr #2
f001031c:   20400121    subcs   r0, r0, r1, lsr #2
f0010320:   21822123    orrcs   r2, r2, r3, lsr #2
f0010324:   e15001a1    cmp r0, r1, lsr #3
f0010328:   204001a1    subcs   r0, r0, r1, lsr #3
f001032c:   218221a3    orrcs   r2, r2, r3, lsr #3                 <-- crash
f0010330:   e3500000    cmp r0, #0  ; 0x0
f0010334:   11b03223    movnes  r3, r3, lsr #4
f0010338:   11a01221    movne   r1, r1, lsr #4
f001033c:   1affffef    bne f0010300 <__udivsi3+0x58>
f0010340:   e1a00002    mov r0, r2
f0010344:   e1a0f00e    mov pc, lr

And this is what I google says to this problem:
  https://lists.ertos.nicta.com.au/archives/kenge-users/2006-September/000352.html

It seems that a similar problem has already been fixed some time ago.

Has anyone an idea what could cause this problem?

Thank you,
~Damien