Re: Patch: Updates on mips testcases

[Daniel Jacobowitz <dmj+ at andrew dot cmu dot edu>]

On Sun, Jun 10, 2001 at 11:12:03PM -0700, Geoff Keating wrote:
> > Date: Sat, 9 Jun 2001 23:01:11 -0700
> > From: Daniel Jacobowitz <dmj+@andrew.cmu.edu>
> 
> ...
> >         Reword conditional branch patterns to use
> >         mips_output_conditional_branch.
> > 
> > That's really relatively recent.  I'd been using GCC 2.95.3 for MIPS,
> > with a few appropriate patches - as far as I can tell, 2.95.3 has no
> > equivalent code at all.  Right?
> 
> I would be surprised if that was the case.  This patch seems to just
> have moved the code out of the .md file into a function.

Nope.  It's full of chunks like:

-  mips_branch_likely = (final_sequence && INSN_ANNULLED_BRANCH_P (insn));
-  return (operands[1] != pc_rtx) ? \"%*bc1t%?\\t%Z0%1\" : \"%*bc1f%?\\t%Z0%2\";
+  return mips_output_conditional_branch (insn,
+                                        operands,
+                                        /*two_operands_p=*/0,
+                                        /*float_p=*/1,
+                                        /*inverted_p=*/0,
+                                        get_attr_length (insn));

So before we would always get a branch conditional, and afterwards we
can get anything up to a 256MB jump.

Thanks for pointing me at the right place, although I don't think I'm
going to try backporting this massive patch at this point - not with
GCC 3.0 due out in under a week.

-- 
Daniel Jacobowitz                           Debian GNU/Linux Developer
Monta Vista Software                              Debian Security Team