Re: gcc-cross-kernel-3.3.4

Re: gcc-cross-kernel-3.3.4

[Tom Rini]

On Wed, Apr 01, 2009 at 09:48:56AM -0000, =?utf-8?b?SsO2cmcgQmFra2VyIDxqb2VyZ0BoYWtrZXIuZGU+?= wrote:

> Yes, you're right. Changed patch is attached.
[snip]
> diff -Nur openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb
> --- openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 10:44:02.000000000 +0100
> +++ build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 17:39:11.000000000 +0100
> @@ -1,2 +1,3 @@
>  require gcc-cross-initial_${PV}.bb
>  require gcc-cross-kernel.inc
> +SRC_URI_append = "file://gcc334-collect2_file_open_args.patch;patch=1"

Grabbed this as mbox from patchwork, so I hope this gets followed-up
right.  The patch is good, but we should be doing this for all 3.3.4's
we have left.  I'll take care of that and push shortly.  Thanks.

-- 
Tom Rini

Re: gcc-cross-kernel-3.3.4

[Khem Raj]

On (03/04/09 10:59), Tom Rini wrote:
> On Wed, Apr 01, 2009 at 09:48:56AM -0000, =?utf-8?b?SsO2cmcgQmFra2VyIDxqb2VyZ0BoYWtrZXIuZGU+?= wrote:
> 
> > Yes, you're right. Changed patch is attached.
> [snip]
> > diff -Nur openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb
> > --- openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 10:44:02.000000000 +0100
> > +++ build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 17:39:11.000000000 +0100
> > @@ -1,2 +1,3 @@
> >  require gcc-cross-initial_${PV}.bb
> >  require gcc-cross-kernel.inc
> > +SRC_URI_append = "file://gcc334-collect2_file_open_args.patch;patch=1"
> 
> Grabbed this as mbox from patchwork, so I hope this gets followed-up
> right.  The patch is good, but we should be doing this for all 3.3.4's
> we have left.  I'll take care of that and push shortly.  Thanks.

I have a slightly different opinion. This only solves one problem which
is due to FORTIFY_SOURCES on ubuntu native gcc. On newer host distributions (with
gcc 4.x). When we chose a OE distro (target) which uses old compilers (
3.4 or below) we end up compiling the cross compilers with relatively
newer compiler which may have problems. Same could be true for other
host utilities which are old. 

I think best solution is to build a native compiler from the same
version first. Now it will add to compilation time but
then we can have ASSUME_PROVIDED.

> 
> -- 
> Tom Rini
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: gcc-cross-kernel-3.3.4

[Tom Rini]

On Fri, Apr 03, 2009 at 12:17:33PM -0700, Khem Raj wrote:
> On (03/04/09 10:59), Tom Rini wrote:
> > On Wed, Apr 01, 2009 at 09:48:56AM -0000, =?utf-8?b?SsO2cmcgQmFra2VyIDxqb2VyZ0BoYWtrZXIuZGU+?= wrote:
> > 
> > > Yes, you're right. Changed patch is attached.
> > [snip]
> > > diff -Nur openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb
> > > --- openembedded/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 10:44:02.000000000 +0100
> > > +++ build/recipes/gcc/gcc-cross-kernel-3.3.4_3.3.4.bb	2009-03-21 17:39:11.000000000 +0100
> > > @@ -1,2 +1,3 @@
> > >  require gcc-cross-initial_${PV}.bb
> > >  require gcc-cross-kernel.inc
> > > +SRC_URI_append = "file://gcc334-collect2_file_open_args.patch;patch=1"
> > 
> > Grabbed this as mbox from patchwork, so I hope this gets followed-up
> > right.  The patch is good, but we should be doing this for all 3.3.4's
> > we have left.  I'll take care of that and push shortly.  Thanks.
> 
> I have a slightly different opinion. This only solves one problem which
> is due to FORTIFY_SOURCES on ubuntu native gcc. On newer host distributions (with
> gcc 4.x). When we chose a OE distro (target) which uses old compilers (
> 3.4 or below) we end up compiling the cross compilers with relatively
> newer compiler which may have problems. Same could be true for other
> host utilities which are old. 
> 
> I think best solution is to build a native compiler from the same
> version first. Now it will add to compilation time but
> then we can have ASSUME_PROVIDED.

Note that there's also a few other problems needing to be patched, at
least on Ubuntu 8.10, to start with, needing
https://dev.openwrt.org/browser/trunk/toolchain/gcc/patches/3.4.6/820-i386_inhibit_libc.patch?rev=14370
and I'm trying to nail down a glibc that builds with all of this as
well (2.2.5 had some problem for DISTRO=angstrom-2008.1 MACHINE=epia).
All in all, I'm about to give up on 3.* (I was building for some SDK
patches I'm going to post shortly, need to get back to testing that
stuff..)

-- 
Tom Rini