AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Robert Wörle]

Dear All

I am still fighting to get a clean build for the avr32
The build i try is on a x86_64 host (if that makes a differecen)

First i have to "fake" external-toolchin do_populate_staging with 
touching the stamp.

Then it goes on and it stops at gcc-cross-4.2.1 with this configure error.

Please advice !!!

NOTE: Resolving missing task queue dependencies
NOTE: multiple providers are available for 
virtual/avr32-angstrom-linux-uclibc-libc-for-gcc (external-toolchain, 
uclibc);
NOTE: consider defining 
PREFERRED_PROVIDER_virtual/avr32-angstrom-linux-uclibc-libc-for-gcc
NOTE: Preparing runqueue



  checking for library containing strerror... configure: error: Link 
tests are not allowed after GCC_NO_EXECUTABLES.
| checking for enabled wchar_t specializations... yes
| checking for ISO C99 support in <math.h>... yes
| make[1]: *** [configure-target-libiberty] Error 1
| make[1]: *** Waiting for unfinished jobs....
| checking complex.h usability... yes
| checking complex.h presence... yes
| checking for complex.h... yes
| checking for ISO C99 support in <complex.h>... no
| checking for ISO C99 support in <stdio.h>... yes
| checking for ISO C99 support in <stdlib.h>... yes
| checking for ISO C99 support in <wchar.h>... yes
| checking for fully enabled ISO C99 support... no
| configure: Debug build flags set to -g3 -O0
| checking for additional debug build... no
| checking for extra compiler flags for building...
| checking for thread model used by GCC... posix
| checking for atomic builtins... yes
| checking for g++ that supports -ffunction-sections -fdata-sections... yes
| checking nan.h usability... no
| checking nan.h presence... no
| checking for nan.h... no
| checking ieeefp.h usability... no
| checking ieeefp.h presence... no
| checking for ieeefp.h... no
| checking endian.h usability... yes
| checking endian.h presence... yes
| checking for endian.h... yes
| checking sys/isa_defs.h usability... no
| checking sys/isa_defs.h presence... no
| checking for sys/isa_defs.h... no
| checking machine/endian.h usability... no
| checking machine/endian.h presence... no
| checking for machine/endian.h... no
| checking machine/param.h usability... no
| checking machine/param.h presence... no
| checking for machine/param.h... no
| checking sys/machine.h usability... no
| checking sys/machine.h presence... no
| checking for sys/machine.h... no
| checking for sys/types.h... (cached) yes
| checking fp.h usability... no
| checking fp.h presence... no
| checking for fp.h... no
| checking float.h usability... yes
| checking float.h presence... yes
| checking for float.h... yes
| checking for endian.h... (cached) yes
| checking for inttypes.h... (cached) yes
| checking locale.h usability... yes
| checking locale.h presence... yes
| checking for locale.h... yes
| checking for float.h... (cached) yes
| checking for stdint.h... (cached) yes
| checking for g++ that supports -ffunction-sections -fdata-sections... yes
| checking for ld version... 21700
| checking for ld that supports -Wl,--gc-sections... configure: error: 
Link tests are not allowed after GCC_NO_EXECUTABLES.
| make[1]: *** [configure-target-libstdc++-v3] Error 1
| make[1]: Leaving directory 
`/srv/home/bob/oe/avr32/tmp/work/avr32-angstrom-linux-uclibc/gcc-cross-4.2.1-r12/gcc-4.2.1/build.x86_64-linux.avr32-angstrom-linux-uclibc'
| make: *** [all] Error 2
| FATAL: oe_runmake failed
NOTE: Task failed: 
/srv/home/bob/oe/avr32/tmp/work/avr32-angstrom-linux-uclibc/gcc-cross-4.2.1-r12/temp/log.do_compile.14611
NOTE: package gcc-cross-4.2.1-r12: task do_compile: failed
ERROR: TaskFailed event exception, aborting
NOTE: package gcc-cross-4.2.1: failed
ERROR: Build of 
/srv/home/bob/oe/org.openembedded.dev/packages/gcc/gcc-cross_4.2.1.bb 
do_compile failed

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Geoffrey Wossum]

On Monday 24 March 2008 04:33:46 am Robert Wörle wrote:
> Dear All
>
> I am still fighting to get a clean build for the avr32
> The build i try is on a x86_64 host (if that makes a differecen)

Could be.  I have a Core 2 Duo, but I use it as a 32-bit system.  Too many 
little gotchas with x86_64 without any advantage for what I do, IMO.  I don't 
think that's the problem here, though.

> First i have to "fake" external-toolchin do_populate_staging with
> touching the stamp.

Not sure why you need to use the external-toolchain anymore.  I submitted some 
patches that got applied on March 14 that allow OE to build the same 
toolchain that Atmel's latest AVR32 buildroot system uses.

Here's my local.conf file, stripped of all comments and blank lines:

---begin---
DL_DIR = "${PRJROOT}/sources"
BBFILES = "${PRJROOT}/org.openembedded.dev/packages/*/*.bb"
BBMASK = ""
PREFERRED_PROVIDERS = "virtual/qte:qte virtual/libqpe:libqpe-opie"
PREFERRED_PROVIDERS += " virtual/libsdl:libsdl-x11"
PREFERRED_PROVIDERS += " 
virtual/${TARGET_PREFIX}gcc-initial:gcc-cross-initial"
PREFERRED_PROVIDERS += " virtual/${TARGET_PREFIX}gcc:gcc-cross"
PREFERRED_PROVIDERS += " virtual/${TARGET_PREFIX}g++:gcc-cross"
PREFERRED_VERSION_linux = "2.6.23"
MACHINE = "at32stk1000"
TARGET_FPU_avr32 = "soft"
TARGET_OS = "linux-uclibc"
DISTRO = "angstrom-2008.1"
ANGSTROM_MODE = "uclibc"
IMAGE_FSTYPES = "ext3 jffs2 tar"
BBDEBUG = "yes"
PARALLEL_MAKE = "-j 3"
BBINCLUDELOGS = "yes"
---end---

I have some more patches that I'm working on, but they're for stuff much after 
getting minimal-image working.  I can almost get through the entire 
base-image build now.  I'm looking at an ICE while building OpenSSH right 
now.

---
Geoffrey

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Robert Wörle]

> 
> Not sure why you need to use the external-toolchain anymore.

why is it then pulled into the image .. ?? hmm maybe i am still at a 
home mode host problem .


   I submitted some
> patches that got applied on March 14 that allow OE to build the same 
> toolchain that Atmel's latest AVR32 buildroot system uses.
> 
I pulled those the 2nd they where in.

I copied your local.conf ( which was very similar to mine ) and restart 
but same error.

Maybe you could ( if you dare ) rebuild your gcc-cross-4.2.1 and send me 
the configure output log. then i could compare and see what it did 
detect and those certain points ?
Would you be so kind ?

Robert

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Geoffrey Wossum]

On Monday 24 March 2008 12:18:59 pm Robert Wörle wrote:
> > Not sure why you need to use the external-toolchain anymore.
>
> why is it then pulled into the image .. ?? hmm maybe i am still at a
> home mode host problem .

I don't know enough about OE to answer that.  Maybe if you move your work 
directory (tmp) out of the way and start over.  I've had a couple of issues 
that were fixed by that.  The magic of a good "make clean" :)

> Maybe you could ( if you dare ) rebuild your gcc-cross-4.2.1 and send me
> the configure output log. then i could compare and see what it did
> detect and those certain points ?
> Would you be so kind ?

I'm actually trying to rebuild it now with a patch for the ICE I was getting 
while building OpenSSH.  I'm getting some kind of error while building 
gcc-cross-initial.  Doesn't look like the same error you were getting, 
though.  Looking into it...

---
Geoffrey

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Geoffrey Wossum]

On Monday 24 March 2008 01:21:48 pm Geoffrey Wossum wrote:
> On Monday 24 March 2008 12:18:59 pm Robert Wörle wrote:
> > > Not sure why you need to use the external-toolchain anymore.

> > Maybe you could ( if you dare ) rebuild your gcc-cross-4.2.1 and send me
> > the configure output log. then i could compare and see what it did
> > detect and those certain points ?
> > Would you be so kind ?
>
> I'm actually trying to rebuild it now with a patch for the ICE I was
> getting while building OpenSSH.  I'm getting some kind of error while
> building gcc-cross-initial.  Doesn't look like the same error you were
> getting, though.  Looking into it...

Okay, figured that one out.  Don't use TARGET_FPU_avr32 = "soft".  The AVR32 
target in gcc doesn't like that.  I added that because I thought I needed it 
to get libmad to compile, but that wasn't the issue with libmad anyway.  I 
needed some sort of optimizations turned on.  angstrom-uclibc.inc was turning 
off all optimizations for the AVR32, which was unnecessary.

Waiting for a full rebuild right now.

---
Geoffrey

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Robert Wörle]

Okay, figured that one out. Don't use TARGET_FPU_avr32 = "soft". The AVR32
> target in gcc doesn't like that.  I added that because I thought I needed it 
> to get libmad to compile, but that wasn't the issue with libmad anyway.  I 
> needed some sort of optimizations turned on.  angstrom-uclibc.inc was turning 
> off all optimizations for the AVR32, which was unnecessary.
>
> Waiting for a full rebuild right now.
>
>   
I had a succes now on my intel core 2 duo laptop ..  that build the 
current oe chain nice ( beside some bzero() problems which can be easly 
replaced by memset() )

But the opteron server x86_64 as host still fails to build ..  on 
gcc-cross configure ..
so seems we need some site.conf entry `s for it !!

Robert

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Geoffrey Wossum]

On Tuesday 25 March 2008 05:53:15 am Robert Wörle wrote:
> Okay, figured that one out. Don't use TARGET_FPU_avr32 = "soft". The AVR32
>
> > target in gcc doesn't like that.  I added that because I thought I needed
> > it to get libmad to compile, but that wasn't the issue with libmad
> > anyway.  I needed some sort of optimizations turned on. 
> > angstrom-uclibc.inc was turning off all optimizations for the AVR32,
> > which was unnecessary.
> >
> > Waiting for a full rebuild right now.
>
> I had a succes now on my intel core 2 duo laptop ..  that build the
> current oe chain nice ( beside some bzero() problems which can be easly
> replaced by memset() )

I think I submitted a patch to OE for the bzero() problem.  I know I submitted 
it to the uClibc mailing list, and it was picked up by the Atmel guy working 
on the port.  I'll put at the bottom of this email so you can check if you 
have it.


I can now build base-image for the AVR32.  Haven't tried out all the apps yet, 
but it seems to all be working.  I'm going to work on getting the patch set 
together and submitted this morning.

---
Geoffrey


file: uClibc-0.9.29-avr32-bzero.patch
--- uClibc-0.9.29-orig/libc/string/avr32/bzero.S        2008-03-10 
15:02:14.000000000 -0500
+++ uClibc-0.9.29/libc/string/avr32/bzero.S     2008-03-10 
15:05:39.000000000 -0500
@@ -6,6 +6,8 @@
  * archive for more details.
  */

+#include <features.h>
+
 #ifdef __UCLIBC_SUSV3_LEGACY__

        .text

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Geoffrey Wossum]

On Tuesday 25 March 2008 08:55:45 am Geoffrey Wossum wrote:
> On Tuesday 25 March 2008 05:53:15 am Robert Wörle wrote:

> I think I submitted a patch to OE for the bzero() problem.  I know I
> submitted it to the uClibc mailing list, and it was picked up by the Atmel
> guy working on the port.  I'll put at the bottom of this email so you can
> check if you have it.

I submitted the actual patch to uClibc, but I must not have submitted the 
patch that makes the patch get used :)  But I've submitted that now.

---
Geoffrey

Re: AVR32 - stops at gcc-cross-4.2.1 configure GCC_NO_EXECUTABLES

[Robert Wörle]

> On Tuesday 25 March 2008 08:55:45 am Geoffrey Wossum wrote:
>> On Tuesday 25 March 2008 05:53:15 am Robert Wörle wrote:
> 
>> I think I submitted a patch to OE for the bzero() problem.  I know I
>> submitted it to the uClibc mailing list, and it was picked up by the Atmel
>> guy working on the port.  I'll put at the bottom of this email so you can
>> check if you have it.
> 
it where not too many apps that still use bzero() .. google told me that 
using memset() is the better way. One should ping those apps ups
tream  and let them change that ( if its truly the "more standard (c)" way )

> I submitted the actual patch to uClibc, but I must not have submitted the 
> patch that makes the patch get used :)  But I've submitted that now.
that sounds complicated ;-)



So far the rest of the chain cooks nicely. Many thanks for that work.
It still brakes on x86_64 hosts to me.
I hope i can identify this at one point.

If i have some work worth sharing i gladly help.
Until then my thanks for the community and the positive feedback from a 
"heavy user" is all i can give.

i need to prepare some custom board code which not too many might need.

Maybe i paste some howto`s to quickly connect buttons as keyboard to a 
ngw100 board.
J5 has quite some open pins which hardly no one might use for the 
beginning and gpio-key is impressively simple for that.

Rob