meta-linaro woes continue

meta-linaro woes continue

[Gary Thomas]

[-- Attachment #1: Type: text/plain, Size: 2200 bytes --]

n.b. this is a continuation of "Kernel Panics on armv4t with gcc.4.5.1"
but the actual discussion has changed enough and there was too much
noise in the previous thread.

Trying to use the meta-linaro layer (to solve my problems
with GCC/4.5.1 on armv5te ), I was able to include and use
the layer with the attached changes.  I added this layer
intact, moving meta-linaro into my poky tree at the same
level as meta, meta-extras, etc.

I also had to copy meta/recipes-devtools/gcc/gcc-4.5.1 into
meta-linaro/recipes-devtools/gcc (I didn't know how to expand
the FILESPATH variable to suck from the main meta tree)

The next problem is this:

| configure: WARNING: unrecognized options: --enable-languages, --enable-threads, --enable-target-optspace, --enable-lto, --enable-libssp, --disable-bootstrap, --disable-libgomp, 
--disable-libmudflap, --with-float, --with-local-prefix, --with-sysroot, --with-build-sysroot, --disable-libunwind-exceptions, --enable-__cxa_atexit
| Checking autotools environment for common misconfiguration
| NOTE: Checking autotools environment for common misconfiguration
| This autoconf log indicates errors, it looked at host includes.
| Rerun configure task after fixing this. The path was 
'/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
| ERROR: This autoconf log indicates errors, it looked at host includes.
| Rerun configure task after fixing this. The path was 
'/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
| Function 'do_qa_configure' failed
| ERROR: Function 'do_qa_configure' failed
NOTE: package gcc-runtime-4.5.1.linaro-r0: task do_qa_configure: Failed

Ideas?  I'm really anxious to see if this fixes my kernel problems.

Thanks

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

[-- Attachment #2: linaro.patch --]
[-- Type: text/plain, Size: 1408 bytes --]

diff --git a/meta-linaro/conf/layer.conf b/meta-linaro/conf/layer.conf
index cce0f43..5c0dc64 100644
--- a/meta-linaro/conf/layer.conf
+++ b/meta-linaro/conf/layer.conf
@@ -5,7 +5,8 @@ RSP_ROOT ?= "sda1"
 BBPATH := "${BBPATH}:${LAYERDIR}"
 
 # We have an images and various recipe-* directories, add to BBFILES
-BBFILES := "${BBFILES} ${LAYERDIR}/images/*.bb ${LAYERDIR}/images/*.bbappend ${LAYERDIR}/recipes-*/*.bb ${LAYERDIR}/recipes-*/*.bbappend"
+BBFILES := "${BBFILES} ${LAYERDIR}/images/*.bb ${LAYERDIR}/images/*.bbappend \
+            ${LAYERDIR}/recipes-*/*/*.bb ${LAYERDIR}/recipes-*/*/*.bbappend"
 	
 BBFILE_COLLECTIONS += "linaro"
 BBFILE_PATTERN_linaro := "^${LAYERDIR}/"
diff --git a/meta-linaro/recipes-devtools/gcc/gcc-4.5.1.linaro.inc b/meta-linaro/recipes-devtools/gcc/gcc-4.5.1.linaro.inc
index 43f5cd2..f145f85 100644
--- a/meta-linaro/recipes-devtools/gcc/gcc-4.5.1.linaro.inc
+++ b/meta-linaro/recipes-devtools/gcc/gcc-4.5.1.linaro.inc
@@ -49,6 +49,14 @@ SRC_URI = "http://launchpad.net/gcc-linaro/4.5/4.5-2010.11-0/+download/gcc-linar
 	   file://gcc-poison-parameters.patch \
 	  "
 
+ 
+do_move_src() {
+	echo moving ${WORKDIR}/gcc-linaro-4.5-2010.11-1 ${WORKDIR}/gcc-${PV}
+	mv ${WORKDIR}/gcc-linaro-4.5-2010.11-1 ${WORKDIR}/gcc-${PV}
+}
+
 do_unpack_append() {
-	mv gcc-linaro-4.5-2010.11-1 gcc-${PV}
+	bb.build.exec_func('do_move_src', d)
 }
+
+

Re: meta-linaro woes continue

[Darren Hart]

On 02/03/2011 03:19 PM, Gary Thomas wrote:
> n.b. this is a continuation of "Kernel Panics on armv4t with gcc.4.5.1"
> but the actual discussion has changed enough and there was too much
> noise in the previous thread.

Hi Gary,

We're both pounding away on the meta-linaro layer at the same time 
unfortunately. I've reorganized things a bit and cleaned up how the 
layer is configured. I believe this layer ran into some trouble when gcc 
and binutils changed, but also suffered from a bad BBFILES variable as 
well (as you noted below). I'll push my changes to 
poky-contrib/dvhart/meta-linaro just as soon as they build.

--
Darren

>
> Trying to use the meta-linaro layer (to solve my problems
> with GCC/4.5.1 on armv5te ), I was able to include and use
> the layer with the attached changes. I added this layer
> intact, moving meta-linaro into my poky tree at the same
> level as meta, meta-extras, etc.
>
> I also had to copy meta/recipes-devtools/gcc/gcc-4.5.1 into
> meta-linaro/recipes-devtools/gcc (I didn't know how to expand
> the FILESPATH variable to suck from the main meta tree)
>
> The next problem is this:
>
> | configure: WARNING: unrecognized options: --enable-languages,
> --enable-threads, --enable-target-optspace, --enable-lto,
> --enable-libssp, --disable-bootstrap, --disable-libgomp,
> --disable-libmudflap, --with-float, --with-local-prefix, --with-sysroot,
> --with-build-sysroot, --disable-libunwind-exceptions, --enable-__cxa_atexit
> | Checking autotools environment for common misconfiguration
> | NOTE: Checking autotools environment for common misconfiguration
> | This autoconf log indicates errors, it looked at host includes.
> | Rerun configure task after fixing this. The path was
> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>
> | ERROR: This autoconf log indicates errors, it looked at host includes.
> | Rerun configure task after fixing this. The path was
> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>
> | Function 'do_qa_configure' failed
> | ERROR: Function 'do_qa_configure' failed
> NOTE: package gcc-runtime-4.5.1.linaro-r0: task do_qa_configure: Failed
>
> Ideas? I'm really anxious to see if this fixes my kernel problems.
>
> Thanks
>
>
>
> _______________________________________________
> yocto mailing list
> yocto@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/yocto


-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel

Re: meta-linaro woes continue

[Darren Hart]

On 02/03/2011 11:08 PM, Darren Hart wrote:
> On 02/03/2011 03:19 PM, Gary Thomas wrote:
>> n.b. this is a continuation of "Kernel Panics on armv4t with gcc.4.5.1"
>> but the actual discussion has changed enough and there was too much
>> noise in the previous thread.
>
> Hi Gary,
>
> We're both pounding away on the meta-linaro layer at the same time
> unfortunately. I've reorganized things a bit and cleaned up how the
> layer is configured. I believe this layer ran into some trouble when gcc
> and binutils changed, but also suffered from a bad BBFILES variable as
> well (as you noted below). I'll push my changes to
> poky-contrib/dvhart/meta-linaro just as soon as they build.
>
> --
> Darren
>
>>
>> Trying to use the meta-linaro layer (to solve my problems
>> with GCC/4.5.1 on armv5te ), I was able to include and use
>> the layer with the attached changes. I added this layer
>> intact, moving meta-linaro into my poky tree at the same
>> level as meta, meta-extras, etc.
>>
>> I also had to copy meta/recipes-devtools/gcc/gcc-4.5.1 into
>> meta-linaro/recipes-devtools/gcc (I didn't know how to expand
>> the FILESPATH variable to suck from the main meta tree)
>>
>> The next problem is this:
>>
>> | configure: WARNING: unrecognized options: --enable-languages,
>> --enable-threads, --enable-target-optspace, --enable-lto,
>> --enable-libssp, --disable-bootstrap, --disable-libgomp,
>> --disable-libmudflap, --with-float, --with-local-prefix, --with-sysroot,
>> --with-build-sysroot, --disable-libunwind-exceptions,
>> --enable-__cxa_atexit
>> | Checking autotools environment for common misconfiguration
>> | NOTE: Checking autotools environment for common misconfiguration
>> | This autoconf log indicates errors, it looked at host includes.
>> | Rerun configure task after fixing this. The path was
>> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>>
>>
>> | ERROR: This autoconf log indicates errors, it looked at host includes.
>> | Rerun configure task after fixing this. The path was
>> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>>
>>
>> | Function 'do_qa_configure' failed
>> | ERROR: Function 'do_qa_configure' failed
>> NOTE: package gcc-runtime-4.5.1.linaro-r0: task do_qa_configure: Failed
>>
>> Ideas? I'm really anxious to see if this fixes my kernel problems.

I'm hitting something similar with libtool. I've pushed my dev branch 
out so you can see how things have changed. I've also asked Nitin to 
take a look at the libtool issues (and if they are related to the linaro 
gcc).

My branch is available here:

poky-extras/dvhart/meta-linaro-dev

I picked up your do_src_move change as it resolves the issue - but I 
can't say that I really understand the problem being solved - can you 
elaborate on why you took this approach? I'll credit you properly in the 
commit in the final set of branches.

Also note, this branch WILL be rebased and otherwise abused as I get it 
in shape for a pull request.

Thanks,

Darren

>>
>> Thanks
>>
>>
>>
>> _______________________________________________
>> yocto mailing list
>> yocto@yoctoproject.org
>> https://lists.yoctoproject.org/listinfo/yocto
>
>


-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel

Re: meta-linaro woes continue

[Gary Thomas]

On 02/04/2011 03:21 PM, Darren Hart wrote:
> On 02/03/2011 11:08 PM, Darren Hart wrote:
>> On 02/03/2011 03:19 PM, Gary Thomas wrote:
>>> n.b. this is a continuation of "Kernel Panics on armv4t with gcc.4.5.1"
>>> but the actual discussion has changed enough and there was too much
>>> noise in the previous thread.
>>
>> Hi Gary,
>>
>> We're both pounding away on the meta-linaro layer at the same time
>> unfortunately. I've reorganized things a bit and cleaned up how the
>> layer is configured. I believe this layer ran into some trouble when gcc
>> and binutils changed, but also suffered from a bad BBFILES variable as
>> well (as you noted below). I'll push my changes to
>> poky-contrib/dvhart/meta-linaro just as soon as they build.
>>
>> --
>> Darren
>>
>>>
>>> Trying to use the meta-linaro layer (to solve my problems
>>> with GCC/4.5.1 on armv5te ), I was able to include and use
>>> the layer with the attached changes. I added this layer
>>> intact, moving meta-linaro into my poky tree at the same
>>> level as meta, meta-extras, etc.
>>>
>>> I also had to copy meta/recipes-devtools/gcc/gcc-4.5.1 into
>>> meta-linaro/recipes-devtools/gcc (I didn't know how to expand
>>> the FILESPATH variable to suck from the main meta tree)
>>>
>>> The next problem is this:
>>>
>>> | configure: WARNING: unrecognized options: --enable-languages,
>>> --enable-threads, --enable-target-optspace, --enable-lto,
>>> --enable-libssp, --disable-bootstrap, --disable-libgomp,
>>> --disable-libmudflap, --with-float, --with-local-prefix, --with-sysroot,
>>> --with-build-sysroot, --disable-libunwind-exceptions,
>>> --enable-__cxa_atexit
>>> | Checking autotools environment for common misconfiguration
>>> | NOTE: Checking autotools environment for common misconfiguration
>>> | This autoconf log indicates errors, it looked at host includes.
>>> | Rerun configure task after fixing this. The path was
>>> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>>>
>>>
>>> | ERROR: This autoconf log indicates errors, it looked at host includes.
>>> | Rerun configure task after fixing this. The path was
>>> '/home/local/efacec_omap_linaro/tmp/work/armv5te-poky-linux-gnueabi/gcc-runtime-4.5.1.linaro-r0/gcc-4.5.1.linaro/build.arm-poky-linux-gnueabi.arm-poky-linux-gnueabi/libstdc++-v3'
>>>
>>>
>>> | Function 'do_qa_configure' failed
>>> | ERROR: Function 'do_qa_configure' failed
>>> NOTE: package gcc-runtime-4.5.1.linaro-r0: task do_qa_configure: Failed
>>>
>>> Ideas? I'm really anxious to see if this fixes my kernel problems.
>
> I'm hitting something similar with libtool. I've pushed my dev branch out so you can see how things have changed. I've also asked Nitin to take a look at the libtool issues (and if
> they are related to the linaro gcc).
>
> My branch is available here:
>
> poky-extras/dvhart/meta-linaro-dev
>
> I picked up your do_src_move change as it resolves the issue - but I can't say that I really understand the problem being solved - can you elaborate on why you took this approach?
> I'll credit you properly in the commit in the final set of branches.

I made this change because do_unpack_append() was being treated
as a Python function, so the syntax was wrong.  By using that
function to run a shell function, the problem is solved.  I patterned
this after similar functions, e.g. in the eglibc recipe.  It took me
a while to find and understand this, but then it became obvious.

>
> Also note, this branch WILL be rebased and otherwise abused as I get it in shape for a pull request.

Thanks for your time

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: meta-linaro woes continue

[Darren Hart]

On 02/04/2011 02:29 PM, Gary Thomas wrote:
> On 02/04/2011 03:21 PM, Darren Hart wrote:

>> I picked up your do_src_move change as it resolves the issue - but I
>> can't say that I really understand the problem being solved - can you
>> elaborate on why you took this approach?
>> I'll credit you properly in the commit in the final set of branches.
>
> I made this change because do_unpack_append() was being treated
> as a Python function, so the syntax was wrong. By using that
> function to run a shell function, the problem is solved. I patterned
> this after similar functions, e.g. in the eglibc recipe. It took me
> a while to find and understand this, but then it became obvious.

Ewww, that's pretty obfuscated. Did you determine why one function is 
treated as python and the other as bash?

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel

Re: meta-linaro woes continue

[Gary Thomas]

On 02/04/2011 04:54 PM, Darren Hart wrote:
> On 02/04/2011 02:29 PM, Gary Thomas wrote:
>> On 02/04/2011 03:21 PM, Darren Hart wrote:
>
>>> I picked up your do_src_move change as it resolves the issue - but I
>>> can't say that I really understand the problem being solved - can you
>>> elaborate on why you took this approach?
>>> I'll credit you properly in the commit in the final set of branches.
>>
>> I made this change because do_unpack_append() was being treated
>> as a Python function, so the syntax was wrong. By using that
>> function to run a shell function, the problem is solved. I patterned
>> this after similar functions, e.g. in the eglibc recipe. It took me
>> a while to find and understand this, but then it became obvious.
>
> Ewww, that's pretty obfuscated. Did you determine why one function is treated as python and the other as bash?
>

No clue, I was just following precedent.  Richard probably
knows the details.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------