Re: PATCH: Turn mips relaxation if appropriate (Re: MIPS gas relaxation still doesn't work)

["H. J. Lu" <hjl at lucon dot org>]

On Mon, Oct 14, 2002 at 02:17:24PM -0200, Alexandre Oliva wrote:
> On Oct 14, 2002, "H. J. Lu" <hjl@lucon.org> wrote:
> 
> > But it is not required to get branch relaxation.
> 
> I remember having got an error from the IRIX 6.5 assembler when I gave
> it the relax test to assemble.  It even suggested me to use a certain
> option that I suspect was a compiler option, since passing it to the
> assembler had no effect, and failed to assemble the testcase just the
> same.  Unfortunately our IRIX 6.5 box is unavailable ATM so I can't
> double-check, but I think the current implementation matches that of
> IRIX.
> 

Please double check. If I remembered right, g++.dg/opt/longbranch1.C
passes on IRIX 6.5 with n32 ABI.

> > Also, I have a problem with --relax-branch/--no-relax-branch. There is
> > no mention of them with "as --help"
> 
> Oops.  This should be fixed, indeed.
> 
> > Also gas already has a generic option:
> 
> >   --relax                     Relax branches on certain targets
> 
> > Why does mips invent a new option?
> 
> Because mips-gas does far more relaxation than just branches, and the
> rest seems to be what is expected of a mips assembler, so it's enabled
> by default, whereas branch relaxation seems to be not the rule, at
> least for mips.

It makes even less senses. The new mips optio is --relax-branch. Now
are you saying it is not true?


H.J.