[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[ninevoltz at uclibc.org]

Author: ninevoltz
Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
New Revision: 21439

Log:
current avr32 gcc patches break x86 and possibly other systems. moving bad patches to separate directory until a proper fix is made. discovered by jacmet and Dr. Nigel Kukard 

Added:
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/900-avr32-atmel.1.3.2.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/901-avr32-add-mno-pic.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/902-avr32-fix-pic-redefinition-warning.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/903-avr32-fix-expanding-builtin-xchg.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/905-avr32-fix-folding-machine-reorg-optimizations.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/906-avr32-use-rjmp-instead-of-got-when-jumping.patch
   trunk/buildroot/toolchain/gcc/4.2.1/avr32_patches_break_x86/908-avr32-fix-invalid-assembler-code-on-64bit-hosts.patch

Removed:
   trunk/buildroot/toolchain/gcc/4.2.1/900-avr32-atmel.1.3.2.patch
   trunk/buildroot/toolchain/gcc/4.2.1/901-avr32-add-mno-pic.patch
   trunk/buildroot/toolchain/gcc/4.2.1/902-avr32-fix-pic-redefinition-warning.patch
   trunk/buildroot/toolchain/gcc/4.2.1/903-avr32-fix-expanding-builtin-xchg.patch
   trunk/buildroot/toolchain/gcc/4.2.1/905-avr32-fix-folding-machine-reorg-optimizations.patch
   trunk/buildroot/toolchain/gcc/4.2.1/906-avr32-use-rjmp-instead-of-got-when-jumping.patch
   trunk/buildroot/toolchain/gcc/4.2.1/908-avr32-fix-invalid-assembler-code-on-64bit-hosts.patch


Changeset:

Sorry, the patch is too large to include (43673 lines).
Please use ViewCVS to see it!

http://uclibc.org/cgi-bin/viewcvs.cgi?view=rev&root=svn&rev=21439

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Peter Korsgaard]

>>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:

 ninevoltz> Author: ninevoltz
 ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
 ninevoltz> New Revision: 21439

 ninevoltz> Log:
 ninevoltz> current avr32 gcc patches break x86 and possibly other
 ninevoltz> systems. moving bad patches to separate directory until a
 ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
 ninevoltz> Kukard 

Thanks!

So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you could
add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to apply
these patches for AVR32?

-- 
Bye, Peter Korsgaard

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

I'd rather leave it unusable until Atmel (or someone else) can fix it
properly. It has caused way too many people headaches and wasted LOTS of
people's time tracking down the causes of bugs and weirdness.

John


On Fri, Mar 21, 2008 at 7:39 AM, Peter Korsgaard <jacmet@uclibc.org> wrote:

> >>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:
>
>  ninevoltz> Author: ninevoltz
>  ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
>  ninevoltz> New Revision: 21439
>
>  ninevoltz> Log:
>  ninevoltz> current avr32 gcc patches break x86 and possibly other
>  ninevoltz> systems. moving bad patches to separate directory until a
>  ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
>  ninevoltz> Kukard
>
> Thanks!
>
> So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you could
> add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to apply
> these patches for AVR32?
>
> --
> Bye, Peter Korsgaard
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/28a264c6/attachment.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Thiago A. Corrêa]

Actually, it was unstable before.... I lost the a whole week's work
trying to figure out why init crashes, and still haven't figured out.
Even tried different gcc versions.


VFS: Mounted root (jffs2 filesystem).
Freeing init memory: 64K (90000000 - 90010000)
init[1]: segfault at 00000000 pc 00000000 sp 7f9aee9c ecr 20
init has generated signal 11 but has no handler for it
Kernel panic - not syncing: Attempted to kill init!

I'm going back revisions to find one that works (atmel's buildroot
fork works). Perhaps we could use the external toolchain for now, as
it was before, where it downloads a pre-patched gcc for AVR32 instead
of applying patches. (?)

Kind Regards,
   Thiago A. Correa

On Fri, Mar 21, 2008 at 8:55 AM, John Voltz <john.voltz@gmail.com> wrote:
> I'd rather leave it unusable until Atmel (or someone else) can fix it
> properly. It has caused way too many people headaches and wasted LOTS of
> people's time tracking down the causes of bugs and weirdness.
>
> John
>
>
>
>
> On Fri, Mar 21, 2008 at 7:39 AM, Peter Korsgaard <jacmet@uclibc.org> wrote:
> > >>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:
> >
> >  ninevoltz> Author: ninevoltz
> >  ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
> >  ninevoltz> New Revision: 21439
> >
> >  ninevoltz> Log:
> >  ninevoltz> current avr32 gcc patches break x86 and possibly other
> >  ninevoltz> systems. moving bad patches to separate directory until a
> >  ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
> >  ninevoltz> Kukard
> >
> > Thanks!
> >
> > So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you could
> > add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to apply
> > these patches for AVR32?
> >
> > --
> > Bye, Peter Korsgaard
> > _______________________________________________
> > buildroot mailing list
> > buildroot at uclibc.org
> > http://busybox.net/mailman/listinfo/buildroot
> >
>
>
> _______________________________________________
>  buildroot mailing list
>  buildroot at uclibc.org
>  http://busybox.net/mailman/listinfo/buildroot
>

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

I'm doing a full build for x86 right now with gcc 4.2.1 and kernel 2.6.24.3.
I'm using all of the patches except for the AVR32 900 series. I want to see
how many of the problems remain or disappear.  All of the AVR32 patches came
from Atmel's buildroot, so I don't think their version will work any
differently.

Where is the prepatched gcc available from? Is this a binary or source code?


John


On Fri, Mar 21, 2008 at 11:14 AM, Thiago A. Corr?a <thiago.correa@gmail.com>
wrote:

> Actually, it was unstable before.... I lost the a whole week's work
> trying to figure out why init crashes, and still haven't figured out.
> Even tried different gcc versions.
>
>
> VFS: Mounted root (jffs2 filesystem).
> Freeing init memory: 64K (90000000 - 90010000)
> init[1]: segfault at 00000000 pc 00000000 sp 7f9aee9c ecr 20
> init has generated signal 11 but has no handler for it
> Kernel panic - not syncing: Attempted to kill init!
>
> I'm going back revisions to find one that works (atmel's buildroot
> fork works). Perhaps we could use the external toolchain for now, as
> it was before, where it downloads a pre-patched gcc for AVR32 instead
> of applying patches. (?)
>
> Kind Regards,
>   Thiago A. Correa
>
> On Fri, Mar 21, 2008 at 8:55 AM, John Voltz <john.voltz@gmail.com> wrote:
> > I'd rather leave it unusable until Atmel (or someone else) can fix it
> > properly. It has caused way too many people headaches and wasted LOTS of
> > people's time tracking down the causes of bugs and weirdness.
> >
> > John
> >
> >
> >
> >
> > On Fri, Mar 21, 2008 at 7:39 AM, Peter Korsgaard <jacmet@uclibc.org>
> wrote:
> > > >>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:
> > >
> > >  ninevoltz> Author: ninevoltz
> > >  ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
> > >  ninevoltz> New Revision: 21439
> > >
> > >  ninevoltz> Log:
> > >  ninevoltz> current avr32 gcc patches break x86 and possibly other
> > >  ninevoltz> systems. moving bad patches to separate directory until a
> > >  ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
> > >  ninevoltz> Kukard
> > >
> > > Thanks!
> > >
> > > So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you could
> > > add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to apply
> > > these patches for AVR32?
> > >
> > > --
> > > Bye, Peter Korsgaard
> > > _______________________________________________
> > > buildroot mailing list
> > > buildroot at uclibc.org
> > > http://busybox.net/mailman/listinfo/buildroot
> > >
> >
> >
> > _______________________________________________
> >  buildroot mailing list
> >  buildroot at uclibc.org
> >  http://busybox.net/mailman/listinfo/buildroot
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/264b9302/attachment-0001.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Thiago A. Corrêa]

It was a source download, I suspect the behavior might still be in
there, I haven't tried yet. I was focusing on figuring out in which
revision things got broken... unfortunally it takes over an hour to
try each revision. It was a download from the same repository as the
Atmel's ARM stuff, so, I guess it's now at www.at91.com ftp

On Fri, Mar 21, 2008 at 12:24 PM, John Voltz <john.voltz@gmail.com> wrote:
> I'm doing a full build for x86 right now with gcc 4.2.1 and kernel 2.6.24.3.
> I'm using all of the patches except for the AVR32 900 series. I want to see
> how many of the problems remain or disappear.  All of the AVR32 patches came
> from Atmel's buildroot, so I don't think their version will work any
> differently.
>
> Where is the prepatched gcc available from? Is this a binary or source code?
>
> John
>
>
>
>
> On Fri, Mar 21, 2008 at 11:14 AM, Thiago A. Corr?a <thiago.correa@gmail.com>
> wrote:
>
> > Actually, it was unstable before.... I lost the a whole week's work
> > trying to figure out why init crashes, and still haven't figured out.
> > Even tried different gcc versions.
> >
> >
> > VFS: Mounted root (jffs2 filesystem).
> > Freeing init memory: 64K (90000000 - 90010000)
> > init[1]: segfault at 00000000 pc 00000000 sp 7f9aee9c ecr 20
> > init has generated signal 11 but has no handler for it
> > Kernel panic - not syncing: Attempted to kill init!
> >
> > I'm going back revisions to find one that works (atmel's buildroot
> > fork works). Perhaps we could use the external toolchain for now, as
> > it was before, where it downloads a pre-patched gcc for AVR32 instead
> > of applying patches. (?)
> >
> > Kind Regards,
> >   Thiago A. Correa
> >
> > On Fri, Mar 21, 2008 at 8:55 AM, John Voltz <john.voltz@gmail.com> wrote:
> > > I'd rather leave it unusable until Atmel (or someone else) can fix it
> > > properly. It has caused way too many people headaches and wasted LOTS of
> > > people's time tracking down the causes of bugs and weirdness.
> > >
> > > John
> > >
> > >
> > >
> > >
> > > On Fri, Mar 21, 2008 at 7:39 AM, Peter Korsgaard <jacmet@uclibc.org>
> wrote:
> > > > >>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:
> > > >
> > > >  ninevoltz> Author: ninevoltz
> > > >  ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
> > > >  ninevoltz> New Revision: 21439
> > > >
> > > >  ninevoltz> Log:
> > > >  ninevoltz> current avr32 gcc patches break x86 and possibly other
> > > >  ninevoltz> systems. moving bad patches to separate directory until a
> > > >  ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
> > > >  ninevoltz> Kukard
> > > >
> > > > Thanks!
> > > >
> > > > So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you could
> > > > add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to apply
> > > > these patches for AVR32?
> > > >
> > > > --
> > > > Bye, Peter Korsgaard
> > > > _______________________________________________
> > > > buildroot mailing list
> > > > buildroot at uclibc.org
> > > > http://busybox.net/mailman/listinfo/buildroot
> > > >
> > >
> > >
> > > _______________________________________________
> > >  buildroot mailing list
> > >  buildroot at uclibc.org
> > >  http://busybox.net/mailman/listinfo/buildroot
> > >
> >
>
>

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

For the record, I started all of my AVR32 work using the toolset rpms from
Atmel's website, without using buildroot at all. Tedious, yes. But
everything I built was stable for the most part. They have a source download
now, I'm checking it out right now:

http://www.atmel.com/dyn/products/tools_card.asp?tool_id=4118

I'm not blaming buildroot for my problems, it's a fantastic tool that I will
continue to use and improve.

John


On Fri, Mar 21, 2008 at 11:33 AM, Thiago A. Corr?a <thiago.correa@gmail.com>
wrote:

> It was a source download, I suspect the behavior might still be in
> there, I haven't tried yet. I was focusing on figuring out in which
> revision things got broken... unfortunally it takes over an hour to
> try each revision. It was a download from the same repository as the
> Atmel's ARM stuff, so, I guess it's now at www.at91.com ftp
>
> On Fri, Mar 21, 2008 at 12:24 PM, John Voltz <john.voltz@gmail.com> wrote:
> > I'm doing a full build for x86 right now with gcc 4.2.1 and kernel
> 2.6.24.3.
> > I'm using all of the patches except for the AVR32 900 series. I want to
> see
> > how many of the problems remain or disappear.  All of the AVR32 patches
> came
> > from Atmel's buildroot, so I don't think their version will work any
> > differently.
> >
> > Where is the prepatched gcc available from? Is this a binary or source
> code?
> >
> > John
> >
> >
> >
> >
> > On Fri, Mar 21, 2008 at 11:14 AM, Thiago A. Corr?a <
> thiago.correa at gmail.com>
> > wrote:
> >
> > > Actually, it was unstable before.... I lost the a whole week's work
> > > trying to figure out why init crashes, and still haven't figured out.
> > > Even tried different gcc versions.
> > >
> > >
> > > VFS: Mounted root (jffs2 filesystem).
> > > Freeing init memory: 64K (90000000 - 90010000)
> > > init[1]: segfault at 00000000 pc 00000000 sp 7f9aee9c ecr 20
> > > init has generated signal 11 but has no handler for it
> > > Kernel panic - not syncing: Attempted to kill init!
> > >
> > > I'm going back revisions to find one that works (atmel's buildroot
> > > fork works). Perhaps we could use the external toolchain for now, as
> > > it was before, where it downloads a pre-patched gcc for AVR32 instead
> > > of applying patches. (?)
> > >
> > > Kind Regards,
> > >   Thiago A. Correa
> > >
> > > On Fri, Mar 21, 2008 at 8:55 AM, John Voltz <john.voltz@gmail.com>
> wrote:
> > > > I'd rather leave it unusable until Atmel (or someone else) can fix
> it
> > > > properly. It has caused way too many people headaches and wasted
> LOTS of
> > > > people's time tracking down the causes of bugs and weirdness.
> > > >
> > > > John
> > > >
> > > >
> > > >
> > > >
> > > > On Fri, Mar 21, 2008 at 7:39 AM, Peter Korsgaard <jacmet@uclibc.org>
> > wrote:
> > > > > >>>>> "ninevoltz" == ninevoltz  <ninevoltz@uclibc.org> writes:
> > > > >
> > > > >  ninevoltz> Author: ninevoltz
> > > > >  ninevoltz> Date: 2008-03-21 04:21:02 -0700 (Fri, 21 Mar 2008)
> > > > >  ninevoltz> New Revision: 21439
> > > > >
> > > > >  ninevoltz> Log:
> > > > >  ninevoltz> current avr32 gcc patches break x86 and possibly other
> > > > >  ninevoltz> systems. moving bad patches to separate directory
> until a
> > > > >  ninevoltz> proper fix is made. discovered by jacmet and Dr. Nigel
> > > > >  ninevoltz> Kukard
> > > > >
> > > > > Thanks!
> > > > >
> > > > > So I take it that gcc 4.2.1 is broken on AVR32 now? Perhaps you
> could
> > > > > add a conditional in gcc-uclibc-4.x.mk like the ones for ARM to
> apply
> > > > > these patches for AVR32?
> > > > >
> > > > > --
> > > > > Bye, Peter Korsgaard
> > > > > _______________________________________________
> > > > > buildroot mailing list
> > > > > buildroot at uclibc.org
> > > > > http://busybox.net/mailman/listinfo/buildroot
> > > > >
> > > >
> > > >
> > > > _______________________________________________
> > > >  buildroot mailing list
> > > >  buildroot at uclibc.org
> > > >  http://busybox.net/mailman/listinfo/buildroot
> > > >
> > >
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/640e6727/attachment.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

The Atmel sources are available at:
ftp://81.80.104.162/AT91_Third_Party_Design_Flow/Linux_Host/Source
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/5ec25b6c/attachment.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Thiago A. Corrêa]

That's what's puzzling, everything was running ok for quite sometime
here as well.  At first I thought it could be the optimization issues,
as sometimes I get a different error saying that it can't find the
entry point of several standard c libraries, but it seams that earlier
revisions also segfaults. :(

The earlier I tried was 180 (google code) and still has the issues
(3rd march). Now I have updated to HEAD and I'm trying to build with
the external toolchain prepatched sources I've mentioned.

> I'm not blaming buildroot for my problems, it's a fantastic tool that I will
> continue to use and improve.

I'm not blaming anyone, or saying that I will stop sending patches or
anything. Since you mention that the AVR32 patches were causing issues
to x86 users, I'm just sharing that there are problems with the AVR32
as well, at least on my end, and that I've been trying to figure it
out.

Kind Regards,
   Thiago A. Correa

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1:avr32_patches_break_x86

[Ulf Samuelsson]

> It was a source download, I suspect the behavior might still be in
> there, I haven't tried yet. I was focusing on figuring out in which
> revision things got broken... unfortunally it takes over an hour to
> try each revision. It was a download from the same repository as the
> Atmel's ARM stuff, so, I guess it's now at www.at91.com ftp

You do not need to worry about where the external sources are located.
They are relative to $(BR2_ATMEL_MIRROR)
and you can configure to use the prepatched toolchain inside buildroot 
using make menuconfig

The Config.in for the prepatched toolchain is in "target/device".
Any patches for the prepatched toolchain will be in a subdirectory
of "target/device/Atmel".

Best Regards
Ulf Samuelsson 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Nigel Kukard]

> Actually, it was unstable before.... I lost the a whole week's work
> trying to figure out why init crashes, and still haven't figured out.
> Even tried different gcc versions.

I lost a week of work aswell .... kernel debugging because what I got
first was an OOPS in a very obscure position.

The my busybox gave me corruption .... must of built buildroot over a
hundred times at least  ;)

Even straces showed up half blank data.

-N
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://busybox.net/lists/buildroot/attachments/20080321/01a29566/attachment.pgp 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

ftp://at91dist:distrib at 81.80.104.162/AT91_Third_Party_Design_Flow/Linux_Host/Source
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/67e7f759/attachment.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[John Voltz]

I noticed, looking through that patch, there are lots of little changes that
affect the code path for all arches. That can't be good, or right. Also,
there are huge chunks of unchanged code that probably got white-spaced
somehow and don't even belong in there. I truly am sorry for unleashing that
mess on you guys. I'm cranking away at finding a stable fix, along with
bugging the Atmel guys about it. I think Atmel Norway must be on holiday or
something, because they've been awfully quiet lately.

John


On Fri, Mar 21, 2008 at 1:23 PM, John Voltz <john.voltz@gmail.com> wrote:

>
> ftp://at91dist:distrib at 81.80.104.162/AT91_Third_Party_Design_Flow/Linux_Host/Source
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080321/66ded0a0/attachment-0001.htm 

[Buildroot] svn commit: trunk/buildroot/toolchain/gcc/4.2.1: avr32_patches_break_x86

[Nigel Kukard]

> I noticed, looking through that patch, there are lots of little
> changes that affect the code path for all arches. That can't be good,
> or right. Also, there are huge chunks of unchanged code that probably
> got white-spaced somehow and don't even belong in there. I truly am
> sorry for unleashing that mess on you guys. I'm cranking away at
> finding a stable fix, along with bugging the Atmel guys about it. I
> think Atmel Norway must be on holiday or something, because they've
> been awfully quiet lately.

After about 10 million pgdwn's in the patch I kinda gave up ... some
pretty nasty changes they've made, I agree!

-N
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://busybox.net/lists/buildroot/attachments/20080321/5a8688e0/attachment.pgp