ewerlid@kuling.UUCP (Ove Ewerlid) (09/13/88)
This may seem like a rather odd question but as I'm currently working on a ns32k-assembler (porting GAS actually) the following question arose: Why has national decided on a new syntax form for optionlists for the ns32532? (This may not be the case, as the datasheet I have is marked preliminary.) In APPENDIX C, page 89 of this datasheet the CINV instruction is described. Assembler syntax: CINV options,src where options are combinations of A,I or D separated by ',' ie there are 3 bits in the machine instruction that are to be set if the corresponding letter EXIST in the operand(s). compare this with the old instruction SETCFG Assembler syntax: SETCFG [option list] where option-list is combinations of C,M,F or I separated by ',' this list is enfolded within two squarebrackets. Ie there are 4 bits in the machine instruction that are to be set if the corresponding letter EXIST in the operand. (just one operand) Now, at the machine level those instructions have the same needs. Ie the setting of a bitmap of options according to the letters in the operands. I think this should be reflected with two identical syntax-forms and not introduce a second form that leads to another specialcase in the assembler. Furthermore I think that the syntax of SETCFG gives a cleaner modell for operand disassembly in the assembler. This means that the CINV should have exactly two operands and not 1 to 4. Again, the above may not be the case if my preliminary datasheet not is up to date and as I think like a man that is writeing an assembler I may have missed some subtle point that justifies the new syntax. /Ove
schultz@mmm.UUCP (John C Schultz) (09/16/88)
In article <828@kuling.UUCP> ewerlid@kuling.UUCP (Ove Ewerlid) writes: > > This may seem like a rather odd question but as I'm currently working on >a ns32k-assembler (porting GAS actually) the following question arose: Has anyone ported GCC for the ICM3216? I have been trying but get the following errors: no ranlib for Sys V - run it anyway and pray that I can figure out the link errors. insn-codes.h line 153 compiler error:parameter stack overflow - patch - I removed a couple hopefully esoteric mnemonics lots of files - too many #defines - apparently the Nat'l compiler does not like long (10-20 lines) definitions. I have not solved this one yet. -- john c. schultz schultz@mmm.UUCP (612) 733-4047 3M Center, Bldg 518-1-1, St. Paul, MN 55144-1000 The opinions expressed herein are, as always, my own and not 3M's.
rfg@nsc.nsc.com (Ron Guilmette) (09/20/88)
In article <1115@mmm.UUCP> schultz@mmm.UUCP (John C Schultz) writes: > >Has anyone ported GCC for the ICM3216? No, but I have ported it to the vme532 which ought to be quite similar. For instance it is also a 32000 processor and also runs System V. Actually, I also have a number of patches for the vme532 (which might also be applicable to your ICM3216) which I have not yet sent to Stallman. If you want, I will be glag to E-mail them to you. >I have been trying but get the following errors: > >no ranlib for Sys V - run it anyway and pray that I can figure out the > link errors. You obviously have a somewhat out-of-date copy of GCC. Stallman fixed the Makefile so that it doesn't try to do ranlib's as of about 1.26. Get a newer version. The latest is 1.28. >insn-codes.h line 153 compiler error:parameter stack overflow > - patch - I removed a couple hopefully esoteric mnemonics > >lots of files - too many #defines - apparently the Nat'l compiler does > not like long (10-20 lines) definitions. I have not > solved this one yet. One possible solution to the second problem would be for me to send you a set of ".o" files for just the GNU C-preprocessor. You could then try linking these in your environment with your libraries to get an initial executable copy of the GNU preprocessor which could be used in place of the normal C-preprocessor for further compilations. Let me know if you would like me to do this. Also let me know if you have executable versions of uudecode and uncompress on your system. If not then the transfer via E-mail will be harder. -- Ron Guilmette National SemiConductor, 1135 Kern Ave. M/S 7C-266; Sunnyvale, CA 94086 Internet: rfg@nsc.nsc.com or amdahl!nsc!rfg@ames.arc.nasa.gov Uucp: ...{pyramid,sun,amdahl,apple}!nsc!rfg