Original from  Klaus Mahlmann  to All on 11-12-1996
Original Subject: Internal processing error

                         ---------------------------------------

KM> After develloping a OS/2 application which runs in a quite 
KM> complex environment (software from three different parties 
KM> on one machine, OS/2 in a Novell environment) we've got the 
KM> problem that about once a day the OS/2 machine will crash 
KM> and display:
KM> 
KM> The system detected an internal processing error
KM> at location ##0160:fff55890 - 000d:3890
KM> 60001, 5049
KM> 048600b4
KM> Internal revision 8.200, 94/11/09
KM> The system is stopped. Please record...

 I don't trust that location from info in any of the map files 
I can generate from the sym file at this level or the dumps I have
access to. Can you try updating one of the machines to FP 10 level
(BLDLEVEL OS2KRNL should report 8.210) or higher and retest? If there
is any other information on the screen, I would like that also..

 It looks like it is involved in updating or doing something with the 
system file tables if the addresses can be trusted. But I cannot find 
any close calls to the IPE routine in that area - and I usually can 
for these kinds of things. I'll check further later.. 
 If the addresses are correct, it is far inside a routine called 
_IOSFTTRANSPATH and I would guess that it has something to do with the
Novell requester. Does this problem have anything to do with accessing
network files or resources? Is it possible that the "program" is
causing the Novell code to make some errors in it's updating of the
system file table entries?

KM> As far as I understand it, this means that the kernel ran 
KM> into a situation it couldn't handle, and called the "panic" 
KM> routine. 

 Yep. 
 
KM> There have been suggestions that there is a OS2KRNL.MAP or 
KM> OS2KRNL.SYM file which could at least tell me where in the 
KM> kernel 0160:fff55890 is. From that, I might be able to 
KM> deduce what kind of operation was going on. Problem is, I 
KM> haven't found such a map file anywhere.

 There are symbol files around that can be used (in either a dump or 
KDB session - they will be different for a KDB session) that are a 
"result" of some processing of MAP files. I can "reverse" this
processing to get back to a "pseudo-map" file also. I still prefer to
use a dump directly, as I can usually see exactly what the code is 
doing at that address.  


   Denis       

 All opinions are my very own, IBM has no claim upon them
. 
.
--- Maximus/2 3.01