Re: error building meta-toolchain for powerpc 405

Re: error building meta-toolchain for powerpc 405

[Ryan Phillips]

[-- Attachment #1: Type: text/plain, Size: 625 bytes --]

On Thu, Apr 1, 2010 at 12:46 AM, Ryan Phillips <ryan@trolocsis.com> wrote:
> Hi All,
>
> I'm trying to compile a meta-toolchain for PowerPC 405 with GCC 4.3.3
> (and I have also tried 4.4.1). Everything goes smoothly until bitbake
> gets to gcc-cross-sdk-4.3.3 and it errors out... I'm not exactly sure
> what the error is.
>
> I have uploaded the log file to:
>   http://hera.kimaker.com/~rphillips/log-results.bz2
>

Patch attached... It looks like gcc-package-sdk needs to have the
strips be optional. Is this the correct fix or should libstdc++
already exist?

Related Commit: 8c4198fc

Regards,
Ryan

[-- Attachment #2: 0001-Fix-meta-toolchain-generation.patch --]
[-- Type: application/octet-stream, Size: 1062 bytes --]

From 11c22ede790d54bd857ed12cf4a0455f1df361f1 Mon Sep 17 00:00:00 2001
From: Ryan Phillips <ryan_d_phillips@dell.com>
Date: Thu, 1 Apr 2010 11:00:32 -0500
Subject: [PATCH] Fix meta-toolchain generation

---
 recipes/gcc/gcc-package-sdk.inc |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/recipes/gcc/gcc-package-sdk.inc b/recipes/gcc/gcc-package-sdk.inc
index dcf2c89..ce7af40 100644
--- a/recipes/gcc/gcc-package-sdk.inc
+++ b/recipes/gcc/gcc-package-sdk.inc
@@ -55,8 +55,8 @@ do_install () {
 	# Manually run the target stripper since we won't get it run by
 	# the packaging.
 	if [ "x${OLD_INHIBIT_PACKAGE_STRIP}" != "x1" ]; then
-		${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/libstdc++.so.*
+		${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/libstdc++.so.* || true
 		${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/libg2c.so.* || true
-		${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/libgcc_s.so.*
+		${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/libgcc_s.so.* || true
 	fi
 }
-- 
1.7.0.1

Re: error building meta-toolchain for powerpc 405

[Bernhard Reutner-Fischer]

On Thu, Apr 01, 2010 at 11:09:56AM -0500, Ryan Phillips wrote:

>Patch attached... It looks like gcc-package-sdk needs to have the
>strips be optional.

Apart from the question below, why not just invoke strip once for all of
them?
${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/*.so.* || :

just curious..

>                    Is this the correct fix or should libstdc++
>already exist?
>
>Related Commit: 8c4198fc

Re: error building meta-toolchain for powerpc 405

[Ryan Phillips]

On Thu, Apr 1, 2010 at 12:21 PM, Bernhard Reutner-Fischer
<rep.dot.nop@gmail.com> wrote:
> On Thu, Apr 01, 2010 at 11:09:56AM -0500, Ryan Phillips wrote:
>
>>Patch attached... It looks like gcc-package-sdk needs to have the
>>strips be optional.
>
> Apart from the question below, why not just invoke strip once for all of
> them?
> ${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/*.so.* || :
>
> just curious..
>

Great point, your fix does appear to be better. I'm just starting OE
development and still learning the system.

Regards,
Ryan

Re: error building meta-toolchain for powerpc 405

[Khem Raj]

On (01/04/10 20:21), Bernhard Reutner-Fischer wrote:
> On Thu, Apr 01, 2010 at 11:09:56AM -0500, Ryan Phillips wrote:
> 
> >Patch attached... It looks like gcc-package-sdk needs to have the
> >strips be optional.
> 
> Apart from the question below, why not just invoke strip once for all of
> them?
> ${TARGET_PREFIX}strip ${D}${prefix}/${TARGET_SYS}/lib/*.so.* || :
> 
> just curious..

While this is certainly possible, I am not in favor of using wildcards in
this case because, its more explicit on what we want to strip and it also 
can tell us problems if a given .so was not built for some reason where we
expected it to.


> 
> >                    Is this the correct fix or should libstdc++
> >already exist?
> >
> >Related Commit: 8c4198fc
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel