This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] missing #ifndef inhibit_libc in gcc/config/rs6000/linux.h


On Tuesday 28 October 2003 22:21, Dan Kegel wrote:
> Franz Sirl wrote:
> > On Tuesday 28 October 2003 17:05, Gernot Hillier wrote:
> >>Hi!
> >>
> >>gcc/config/rs6000/linux.h misses support for the libchack thus making it
> >>impossible (or at least not nice) to bootstrap a cross compiler.
> >>
> >>This also solves http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12813
> >>
> >>Ok for mainline? TIA!
> >>
> >>Changelog:
> >>2003-10-28  Gernot Hillier  <gernot.hillier@siemens.com>
> >>	* config/rs6000/linux.h (MD_FALLBACK_FRAME_STATE_FOR): Don't
> >>	define this at all when -Dinhibit_libc is used.
> >
> > Hmm, in my eyes this patch just makes it easier to build broken cross
> > compilers and we shouldn't do that. There are proper linux cross build
> > instructions on the web, eg. http://kegel.com/crosstool/, that don't use
> > and need inhibit_libc IIRC.
>
> inhibit_libc isn't a hack per se; it's an internal part of the
> implementation of --without-headers and --with-newlib.  Perhaps the
> proposed patch
> solves a problem even for users who are simply setting --without-headers
> and/or --with-newlib?
>
> See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8180 if you're interested
> in why some users feel they have to set -Dinhibit_libc
> (gcc's configure used to be broken, and didn't always set it when it
> should). - Dan

You are right, it is not a hack per se, but I consider it a hack if you use it 
on _linux_. As you know building a proper linux toolchain from scratch 
requires a proper procedure if you don't want to end up with subtle ABI 
differences. The linux headers (kernel and glibc) are freely available and I 
see no use in avoiding them, just to end up with a compiler that isn't a 
fully compatible linux platform compiler.
But I'm in no position to reject the patch myself, just my 2ct.

Franz.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]