[Simh] pdp11: 11/70 mbrk register and losize,hisize register

[Walter F.J. Mueller]

Dear all,

the EKBE xxdp test, see previous posting why I'm into that, gives
also

   MICR0 BREAK REG LOADED INCORRECTLY. ONLY LOWER BYTE SHOULD BE WRITABLE
   REGADDR DATAREC DATAEXP TESTNO  ERRORPC
   177770  000400  000000  000004  041276

EKBE tests that bit<7:0> of mbrk are writeable, and that bit<15:8> is
always returned as zero.

simh in pdp11_mod.c implements mbrk as a full 16 bit register, an
MBRK_IMP mask is missing.

the EKBE test also gives the messages

   UNEXPECTED CPU TRAP OR ABORT THRU 'ERRVEC' (004)
   RECEIVD TESTNO  PC AT ABORT
   000060  000002  041036
   UNEXPECTED CPU TRAP OR ABORT THRU 'ERRVEC' (004)
   RECEIVD TESTNO  PC AT ABORT
   000020  000002  041104

Behind those traps is the code

   mov  #177760,r0
   ...
   clr  (r0)           <- first trap
   ...
   mov  #177762,r0
   ...
   mov  #177777,(r0)   <- second trap

where it checks the behaviour the losize and hisize registers and
expects, that both allow a write access, but are immutable.

simh, in CPU70_wr in pdp11_mod.c, returns SCPE_NXM for case 010 and
011, thus causes a trap when 177760 and 177762 are write accessed.

I agree that all three difference don't play a role for normal
application programs, but they prevent that diagnostics run properly.


		With best regards,	Walter




-- 
Dr. Walter F.J. Müller       Mail:  W.F.J.Mueller at gsi.de
GSI,  Abteilung CBM          Phone: +49-6159-71-2766
Planckstr. 1                 FAX:   +49-6159-71-3762
D-64291 Darmstadt            URL: http://www-linux.gsi.de/~mueller/

Gesellschaft für Schwerionenforschung mbH; Amtsgericht Darmstadt, HRB 1528
Geschäftsführer: Prof. Dr. Horst Stoecker
Vorsitzende des Aufsichtsrates: Dr. Beatrix Vierkorn-Rudolph