This is the mail archive of the
mailing list for the binutils project.
Re: Macro name and comment fixes for gas/tc-mips.c
Richard Sandiford <email@example.com> writes:
> Ian Lance Taylor <firstname.lastname@example.org> writes:
> > I spent some time being confused by the macro ISA_HAS_COPROC_DELAYS
> > because I thought it meant something like ``this ISA requires delays
> > for co-processor instructions.'' In fact, as the comment explains, it
> > means something rather different, namely that the ISA requires delays
> > in a variety of different situations.
> > I propose that we rename the macro to ISA_HAS_NONBRANCH_DELAYS to
> > avoid such confusion. This patch does that rename, and also fixes the
> > comment for this macro and for the gpr_interlocks macro.
> ISA_HAS_NONBRANCH_DELAYS seems a bit too general. I'd have
> expected !ISA_HAS_NONBRANCH_DELAYS to imply the ISA needs no
> non-branch delays, but that isn't necessarily true. You still
> might need hi/lo-related nops.
True. I expressed that in the comment, but perhaps you're right that
the name is still too ambiguous.
Maybe CGD is right and we should just drop the macro entirely and
incorporate the test into other macros--it's likely that all the
required macros already exist.