This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: egcs mips patch
- To: law at cygnus dot com, knobi at knobisoft dot de
- Subject: Re: egcs mips patch
- From: Martin Knoblauch <knobi at rocketmail dot com>
- Date: Wed, 9 Sep 1998 01:10:35 -0700 (PDT)
- Cc: Gavin Romig-Koch <gavin at cygnus dot com>, Jim Wilson <wilson at cygnus dot com>, egcs-patches at cygnus dot com
- Reply-To: knobi at knobisoft dot de
---Jeffrey A Law <law@cygnus.com> wrote:
>
>
> In message
<19980908221320.10313.rocketmail@attach1.rocketmail.com>you write:
> > Gavin,
> >
> > be careful on mips-sgi-irix6.x systems. 64-bit implies
> > mips3/mips4 in this case. This is probably true for all
> > configs, as (to my knowledge) 64-bit pointers are only
> > supported in mips3 and above. It may be better to make
> > mips1/mips2 illegal for 64-bit (address) code.
> How we define LONG_MAX should not effect how we deal with pointers
> in any way shape or form.
>
> In the past the MIPS port assumed that pointers and longs were the
> same size, and thus, maybe via some obscure dependency a change to
> LONG_MAX could have an effect on pointers. We removed that assumption
> for the MIPS port a while ago.
>
Again, be careful for the mips-sgi-irix targets. The IRIX runtime
assumes/forces sizeof(long) == sizeof(void *). Other mips targets
may have weaker/no dependencies.
> Note that -mips1 and -mips2 can still work on 64bit systems, you just
> can't use the entire 64bit address space.
>
Of course, they should run if the system knows how to deal
with 32-bit pointers :-) Maybe I just confused "64 bit systems"
and "64 bit binaries". Under IRIX, there are definitely no
64-bit mips1/mips2 binaries.
===
------------------------------------------------------
Martin Knoblauch
email: knobi@knobisoft.de or knobi@rocketmail.com
www: http://www.knobisoft.de
_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com