This is the mail archive of the
mailing list for the GCC project.
Re: [Fwd: Current state of MIPS16 support?]
- From: Daniel Jacobowitz <dan at debian dot org>
- To: Dominic Sweetman <dom at algor dot co dot uk>
- Cc: Eric Christopher <echristo at redhat dot com>,Johannes Stezenbach <js at convergence dot de>, gcc at gcc dot gnu dot org,linux-mips at oss dot sgi dot com, sde at algor dot co dot uk
- Date: Wed, 5 Jun 2002 13:53:18 -0400
- Subject: Re: [Fwd: Current state of MIPS16 support?]
- References: <3CBFEAA9.email@example.com> <firstname.lastname@example.org> <email@example.com> <firstname.lastname@example.org>
On Wed, Jun 05, 2002 at 04:39:45PM +0100, Dominic Sweetman wrote:
> > I'm not certain what you are actually fixing here as I've not seen any
> > descriptions of problems here...
> Hmm. Linux/MIPS suffers from widespread and diffuse toolchain
> problems: I thought that much was pretty clear to all involved. I
> agree it seems a pity that the scheme of work laid out above should be
> > Almost 90% of the bug reports I see are against IRIX.
> That does suggest you're missing some pretty large chunks of the
No, that's not what it suggests at all. It suggests to me that the
dubious state of Linux/MIPS toolchains is due to one of two things:
- A pre-existing acceptance of this rather than any real problems
- A reluctance to file bug reports
We've been using GCC for MIPS for several years now and seen relatively
few significant problems, so I suspect the former. There's definitely
some of the latter as well, since we've hit more roadblocks on MIPS
than on, say, x86 or PowerPC; it's not our worst problem architecture,
though. And we've been able to fix them all relatively easily.
I'm sure GCC would benefit from access to your regression tests, though :)
Daniel Jacobowitz Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer