boost breaks on rebuild

boost breaks on rebuild

[Steve Sakoman]

Boost is another package when a rebuild is triggered:

ERROR: Logfile of failure stored in:
/media/Work/yocto/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/boost-1.49.0-r2.0/temp/log.do_compile.22475
Log data follows:
| ERROR: Function failed: do_compile (see
/media/Work/yocto/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/boost-1.49.0-r2.0/temp/log.do_compile.22475
for further information)
| + bjam -sTOOLS=gcc '-sGCC=ccache arm-poky-linux-gnueabi-gcc
-march=armv7-a -fno-tree-vectorize      -mthumb-interwork
-mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8  -Wl,-O1
-Wl,--hash-style=gnu -Wl,--as-needed
--sysroot=/media/Work/yocto/tmp/sysroots/omap4-multi
'\''-DBOOST_PLATFORM_CONFIG="boost/config/platform/linux-gnueabi.hpp"'\'''
'-sGXX=ccache arm-poky-linux-gnueabi-g++  -march=armv7-a
-fno-tree-vectorize      -mthumb-interwork -mfloat-abi=softfp
-mfpu=neon -mtune=cortex-a8  -Wl,-O1 -Wl,--hash-style=gnu
-Wl,--as-needed --sysroot=/media/Work/yocto/tmp/sysroots/omap4-multi
'\''-DBOOST_PLATFORM_CONFIG="boost/config/platform/linux-gnueabi.hpp"'\'''
-sGCC_INCLUDE_DIRECTORY=/media/Work/yocto/tmp/sysroots/omap4-multi/usr/include
-sGCC_STDLIB_DIRECTORY=/media/Work/yocto/tmp/sysroots/omap4-multi/usr/lib
-sNO_BZIP2=1 -sNO_ZLIB=1 '-sBUILD=release <optimization>space
<threading>multi <inlining>on <debug-symbols>off'
'-sPYTHON_VERSION=${PYTHON_VERSION}' --layout=system
--builddir=/media/Work/yocto/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/boost-1.49.0-r2.0/boost_1_49_0/arm-poky-linux-gnueabi
--disable-icu --with-date_time --with-filesystem --with-graph
--with-iostreams --with-program_options --with-regex --with-signals
--with-system --with-test --with-thread --prefix=/usr
--exec-prefix=/usr --libdir=/usr/lib --includedir=/usr/include
| error: at /media/Work/yocto/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/boost-1.49.0-r2.0/boost_1_49_0/tools/build/v2/user-config.jam:94
| error: duplicate initialization of gcc with the following parameters:
| error: version = 4.3.1
| error: previous initialization at
/media/Work/yocto/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/boost-1.49.0-r2.0/boost_1_49_0/tools/build/v2/user-config.jam:93
|
NOTE: package boost-1.49.0-r2.0: task do_compile: Failed

I don't have a fix for this but wanted to document the failure.

As with all the other failures of this type, a -c cleansstate and
rebuild is the workaround.

It feels to me like there are quite a few recipes that don't handle
rebuilds -- I seem to run into a few after every pull.

Steve

Re: boost breaks on rebuild

[Richard Purdie]

On Wed, 2012-03-14 at 15:28 -0700, Steve Sakoman wrote:
> NOTE: package boost-1.49.0-r2.0: task do_compile: Failed
> 
> I don't have a fix for this but wanted to document the failure.
> 
> As with all the other failures of this type, a -c cleansstate and
> rebuild is the workaround.
> 
> It feels to me like there are quite a few recipes that don't handle
> rebuilds -- I seem to run into a few after every pull.

Agreed, there is a problem here. I've filed the bug:

https://bugzilla.yoctoproject.org/show_bug.cgi?id=2123

since I think we systematically need to work through these and eradicate
them.

An alternative to fix some of the compile cases would be to run a "make
clean" before the make. I suspect this would annoy people and its a
shame to do that because of a few broken recipes though...

Cheers,

Richard

Re: boost breaks on rebuild

[Mark Hatle]

On 3/15/12 8:39 AM, Richard Purdie wrote:
> On Wed, 2012-03-14 at 15:28 -0700, Steve Sakoman wrote:
>> NOTE: package boost-1.49.0-r2.0: task do_compile: Failed
>>
>> I don't have a fix for this but wanted to document the failure.
>>
>> As with all the other failures of this type, a -c cleansstate and
>> rebuild is the workaround.
>>
>> It feels to me like there are quite a few recipes that don't handle
>> rebuilds -- I seem to run into a few after every pull.

I've observed the same thing.  There seem to be a lot of packages that once 
they've been configured, compiled and installed, won't rebuild without being 
cleaned in someway.

One suggestion, maybe a new flag for a recipe that says that recipe rebuilds 
need to start from a clean patched tree?  Then if the do_configure -> do_install 
steps have been run, it knows it needs to clean and fetch/unpack/patch before it 
tries to rebuild the component...  (Maybe sstate-cache can help speed this up in 
some way...)

Then over time we can find the items with the flags and attempt to fix them in 
other ways...

--Mark

> Agreed, there is a problem here. I've filed the bug:
>
> https://bugzilla.yoctoproject.org/show_bug.cgi?id=2123
>
> since I think we systematically need to work through these and eradicate
> them.
>
> An alternative to fix some of the compile cases would be to run a "make
> clean" before the make. I suspect this would annoy people and its a
> shame to do that because of a few broken recipes though...
>
> Cheers,
>
> Richard
>
>
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core

Re: boost breaks on rebuild

[Steve Sakoman]

On Thu, Mar 15, 2012 at 9:21 AM, Mark Hatle <mark.hatle@windriver.com> wrote:
> On 3/15/12 8:39 AM, Richard Purdie wrote:
>>
>> On Wed, 2012-03-14 at 15:28 -0700, Steve Sakoman wrote:
>>>
>>> NOTE: package boost-1.49.0-r2.0: task do_compile: Failed
>>>
>>> I don't have a fix for this but wanted to document the failure.
>>>
>>> As with all the other failures of this type, a -c cleansstate and
>>> rebuild is the workaround.
>>>
>>> It feels to me like there are quite a few recipes that don't handle
>>> rebuilds -- I seem to run into a few after every pull.
>
>
> I've observed the same thing.  There seem to be a lot of packages that once
> they've been configured, compiled and installed, won't rebuild without being
> cleaned in someway.

This is the one that really pisses me off, happens after every pull on
my Ubuntu 64 bit build machine:

Summary: 1 task failed:
  virtual:native:/home/steve/source/yocto/poky/meta/recipes-devtools/pseudo/pseudo_1.2.bb,
do_compile
Summary: There were 8 WARNING messages shown.
Summary: There was 1 ERROR message shown, returning a non-zero exit code.

No simple fix since any bitbake command complains about needing to
rebuild psuedo:

Pseudo is not present but is required, building this first before the main build

And of course the psuedo rebuild instantly fails!

Steve

> One suggestion, maybe a new flag for a recipe that says that recipe rebuilds
> need to start from a clean patched tree?  Then if the do_configure ->
> do_install steps have been run, it knows it needs to clean and
> fetch/unpack/patch before it tries to rebuild the component...  (Maybe
> sstate-cache can help speed this up in some way...)
>
> Then over time we can find the items with the flags and attempt to fix them
> in other ways...
>
> --Mark
>
>
>> Agreed, there is a problem here. I've filed the bug:
>>
>> https://bugzilla.yoctoproject.org/show_bug.cgi?id=2123
>>
>> since I think we systematically need to work through these and eradicate
>> them.
>>
>> An alternative to fix some of the compile cases would be to run a "make
>> clean" before the make. I suspect this would annoy people and its a
>> shame to do that because of a few broken recipes though...
>>
>> Cheers,
>>
>> Richard
>>
>>
>>
>>
>> _______________________________________________
>> Openembedded-core mailing list
>> Openembedded-core@lists.openembedded.org
>> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core

Re: boost breaks on rebuild

[Richard Purdie]

On Thu, 2012-03-15 at 12:14 -0700, Steve Sakoman wrote:
> On Thu, Mar 15, 2012 at 9:21 AM, Mark Hatle <mark.hatle@windriver.com> wrote:
> > On 3/15/12 8:39 AM, Richard Purdie wrote:
> >>
> >> On Wed, 2012-03-14 at 15:28 -0700, Steve Sakoman wrote:
> >>>
> >>> NOTE: package boost-1.49.0-r2.0: task do_compile: Failed
> >>>
> >>> I don't have a fix for this but wanted to document the failure.
> >>>
> >>> As with all the other failures of this type, a -c cleansstate and
> >>> rebuild is the workaround.
> >>>
> >>> It feels to me like there are quite a few recipes that don't handle
> >>> rebuilds -- I seem to run into a few after every pull.
> >
> >
> > I've observed the same thing.  There seem to be a lot of packages that once
> > they've been configured, compiled and installed, won't rebuild without being
> > cleaned in someway.
> 
> This is the one that really pisses me off, happens after every pull on
> my Ubuntu 64 bit build machine:
> 
> Summary: 1 task failed:
>   virtual:native:/home/steve/source/yocto/poky/meta/recipes-devtools/pseudo/pseudo_1.2.bb,
> do_compile
> Summary: There were 8 WARNING messages shown.
> Summary: There was 1 ERROR message shown, returning a non-zero exit code.
> 
> No simple fix since any bitbake command complains about needing to
> rebuild psuedo:
> 
> Pseudo is not present but is required, building this first before the main build
> 
> And of course the psuedo rebuild instantly fails!

Sounds like we need to teach the wrapper script to allow a pseudo-native
to be passed -c options directly (and fix pseudo's makefile if needed).

Cheers,

Richard