Tom wrote:

> Here is the dump:
>
>  **** INTERNAL TRAP
>  Data PAGE Fault
>  [VSM] Undecoded status.info = -300
>  ABORT: B08I040.Y2KPRG.DEVELOP
>         PC=b6.00cd89b0 sqlx_p:sqlx+$a8
>  NM* 0) SP=418607b8 RP=b6.00cd8874 sqlx_p:?sqlx+$8
>           export stub: b6.00ce62d0 sqlxpp_p:sqlxpp+$154
>  NM  1) SP=41853fb8 RP=b6.00ce6168 sqlxpp_p:?sqlxpp+$8
>           export stub: 13e.0012b518 spwfdbe_cst_srch+$1e4
>  NM  2) SP=41853f78 RP=13e.0012b16c ?spwfdbe_cst_srch+$8
>           export stub: 3a3.00016988 a08i0404+$1330
>  NM  3) SP=418532f8 RP=3a3.00009028 _start+$1080
>  NM  4) SP=41852338 RP=3a3.00000000
>       (end of NM stack)
>
>  R0 =00000000 418547b8 00cd8877 41853fb8 R4 =41853b94 00d12800 418547b8
> 41853b94
>
>  R8 =0012a90c 30303030 20202020 00000000 R12=00000000 00000000 00000000
> 00000000
>
>  R16=00000000 00000000 00000000 41202020 R20=000000b6 53514c43 53514c43
> 00d12a18
>
>  R24=41853508 00002332 41853b94 4164a000 R28=00000004 41853730 418607b8
> 61367236
>
>
>  IPSW=0004fe0f=thlnxbCvmrQPDI PRIV=2
>   SAR=18  PCQF=b6.cd89b2  b6.cd89b6
>
>  SR0=0000013e 00000000 00000000 00000000 SR4=000000b6 0000039f 0000000b
> 0000000a
>
>  TR0=0000fa40 05744db4 01a0240e 81d8d3d0 TR4=ffffff88 80000000 000264d3
> 0012f160
>
>  PID1=00000222=00000111(W)               PID2=00000000=00000000(W)
>  PID3=00000000=00000000(W)               PID4=00000000=00000000(W)
>  RCTR=00000000 ISR=0000039f IOR=61367712 IIR=4bf409b8 IVA=00142000
> ITMR=5c7eb1b8
>
>  EIEM=ffffffff EIRR=00000000 CCR=c0
>

The instruction that causes the trap is 4bf409b8 (the IIR above):

$5 ($1a1) nmdebug > dis 4bf409b8
LDW      1244(0,31),20

The address it doesn't like is 39f.61367712 (ISR.IOR) and is arrived  by adding
1244 to R31 which was 61367236.

This value seems a bit unusual as an SR5 (process local) pointer so one thing I do
is see what else it might be, for example:

$3 ($1a1) nmdebug > =61367236,a
a6r6

So that pointer _MIGHT_ also be an ASCII value "a6r6" and if that were true it
could be the result of stack corruption caused, perhaps by overflowing a buffer.

This might be consistent with what you provided later that the subroutine works
sucessfully 6 times and fails on the seventh and _may_ even explain why adding
the extra display makes the problem stop. However, offhand I can't think of why
adding a display makes the problem stop but I've seen similar odd things.

Probably not a lot of help...

Bill
HP/CSY

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *