This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PR16350 patch ping - gcc doesn't understand big-endian ARM
On Thu, 2005-03-24 at 11:17, Lennert Buytenhek wrote:
> On Thu, Mar 24, 2005 at 10:56:57AM +0000, Richard Earnshaw wrote:
>
> > My primary concern is that a single letter here is just not
> > discriminating enough. Anything less than two letters is just going to
> > cause problems in the future. Overloading this onto the CPU name means
> > we need to take care to avoid potential CPU name extensions that might
> > add 'b' to the processor name. To avoid problems I really think this
> > should be either 'eb' (preferred) or 'be'.
>
> Big-endian target triples that are currently 'out there' are:
>
> armeb-unknown-linux-gnu
> armv4b-unknown-linux-gnu
> armv5teb-unknown-linux-gnu
>
> The first one is the 'generic' big-endian target triple, the second
> and third are what config.guess comes up with on a big-endian StrongARM
> resp. big-endian xscale. Linux 2.4.* reports 'armv5b' instead of
> 'armv5teb' for the xscale.
>
Then config.guess should be fixed. The middle one of these is simply
wrong.
> Given these, arm*b- seemed the obvious choice. 'armeb- OR armv*b-'
> would perhaps also be an option?
>
> binutils 2.15 has:
> ./gas/configure: arm*b|xscale*b|strongarm*b) cpu_type=arm endian=big ;;
And so should binutils.
Lets not propagate this problem any further than it needs. Fixing
config.guess shouldn't break anything that currently matches a single
'b', but it will mean that in time we can solve this problem more
cleanly.