ra@is.uu.no (Robert Andersson) (11/19/89)
Some things I noticed while porting gcc 1.36 to a new 680x0 based
machine:
The m68k.md file has a lot of conditionals to create the correct
assembler output:
ISI_OV
HPUX_ASM
MOTOROLA
SONY_ASM
SGS
SGS_3B1
NO_ADDSUB_Q
ASM_OUTPUT_CASE_LABEL
These all seem to be created sort of ad hoc? I'm I wrong?
Also, a couple of the tm-*.h files define ASM_OUTPUT_OPCODE to
transform the opcode syntax after they have been generated by m68k.md.
This is done in tm-3b1.h and tm-hp9k320.h
Now, the assembler on the machine I'm porting to has quite a few of
the same quirks as the 3b1 and hpux assemblers. I could of course
#define ASM_OUTPUT_OPCODE in my tm-*.h file as well, but I find the
#ASM_OUTPUT_OPCODE hack gross.
Shouldn't the m68k.md file be changed instead? There's already 8
preprocessor conditionals it act's upon, maybe we should add one or
two more and get rid of ASM_OUTPUT_OPCODE definitions in the tm-*.h
files?
Is there any interest in me doing this and providing context diffs?
---
Robert Andersson, International Systems A/S, Oslo, Norway.
Internet: ra@is.uu.no
UUCP: ...!{uunet,mcvax,ifi}!is.uu.no!ra