Update on mcp750 strangeness (sorry, screwed up my last posting)

[gregory.menke at gsfc.nasa.gov]

OK.  I spent the afternoon narrowing down my problem, though I don't
know whats going on yet.

I moved the config data to its own C file, independently compiled and
linked.  The contents of the file are a simple assignment as shown
here.



=================================
char pmapconf[]="
;
; Midex configuration file
;

;
; specify the absolute home directory for the Midex tree, if omitted
; use the BASEDIR env var, and if that doesn't exist, getcwd()
;
base-dir /


; additional (or other absolute) path from the base-dir to where the
; Midex binaries are located
;
exec-subdir bin


; additional (or other absolute) path from the base-dir to the
; software bus data files.  This item is processed automatically by
; the software bus lib whenever a process starts it up
;
softbus-subdir etc


; log output, if omitted, stdout.  Assumes the file is in the 'log'
; subdir under base-dir unless a pathname is supplied.

log-output logstream


;
; default process mgmt settings, override as needed on individual processes
;
; loglevel=0 suppresses all per-process logging
; loglevel=1 allows error logging, suppresses routine logging
; loglevel>1 allows all logging
;
default restartdelay=3 loglevel=2

start-process hs loglevel=2
#start-process to
#start-process tc
start-process tc1hz
#start-process ci
#start-process it
#start-process ds
start-process xn net-telemetry-port=2001 net-command-port=2051

; eof";

=================================

As I adjust the length of the text, I can make RTEMS either throw an
exception or run fine.  The problem does not appear to be related to
the text itself.  But now that I'm retyping this, and its after 5, it
seems to work just fine.  Is there any remote possibility of ambient
network traffic causing an interrupt at the wrong time- or maybe the
violation of some assumption about the image's section layout which
appears when an application links in global data?

Once a runnable build is found, recompiling it does not cause it to
fail.  Re-running a working build also doesn't fail.



Besides this app, I have another that does dhrystone and whetstone
benchmarks in which I can't initialize the network and ftpd at all or
I get an exception.  I can only make it run when the network
initialization stuff is commented out- and at one point the network
stuff ran just fine.  I just retested this and it still throws the
exception, an example of which is below:


DEC Adapter found !!
DC2114x : driver has been attached
DC2114x 8:0:3e:29:87:9b IRQ 10 IO 10100 M c1041100 .........
configured dexception handler called for exception 7
         Next PC or Address of fault = 0
         Saved MSR = 8b032
         R0 = 0
         R1 = 1fe12d0
         R2 = 0
         R3 = 0
         R4 = 1
         R5 = 0
         R6 = 20000022
         R7 = b032
         R8 = 30
         R9 = 50000
         R10 = b032
         R11 = 0
         R12 = 20000000
         R13 = 0
         R14 = 40000
         R15 = 40000
         R16 = 49
         R17 = 0
         R18 = 0
         R19 = 1fe1370
         R20 = 46912
         R21 = e
         R22 = 40000
         R23 = ffffffff
         R24 = 1
         R25 = 1fe1360
         R26 = 0
         R27 = 1fe1378
         R28 = 46911
         R29 = 1ee9f78
         R30 = 49
         R31 = 1eeae24
         CR = 20000022
         CTR = 0
         XER = 0
         LR = 0
         MSR = 0
unrecoverable exception!!! Push reset button




I fairly sure I'm not fantasizing about whats going on, but I could
sure use some guidance about how to approach discovering the source of
these problems.

Thanks,

Gregm