boost fails to build on MIPS

boost fails to build on MIPS

[Mike Looijmans]

On current OE master, boost fails to build, because something attempts 
to add "-m32" to the gcc commandline for no apparent reason.

Can anyone confirm this, or am I the only MIPS user in the world?

For example:

| gcc.compile.c++ 
bin.v2/libs/atomic/build/gcc-4.3.1/release/threading-multi/lockpool.o
| mipsel-oe-linux-g++: error: unrecognized command line option '-m32'
|
|     "mipsel-oe-linux-g++" "-mel" "-mabi=32" "-mhard-float" 
"-march=mips32" "-Wl,-O1" "-Wl,--as-needed" 
"--sysroot=/home/milo/work/openpli-next/build/tmp/sysroots/vuduo2" 
-ftemplate-depth-128  -Os -pipe -g -feliminate-unused-debug-types 
-fvisibility-inlines-hidden -O3 -finline-functions -Wno-inline -Wall 
-pthread -fPIC -m32  -DBOOST_ALL_NO_LIB=1 -DBOOST_ATOMIC_DYN_LINK=1 
-DBOOST_ATOMIC_SOURCE -DNDEBUG  -I"." -c -o 
"bin.v2/libs/atomic/build/gcc-4.3.1/release/threading-multi/lockpool.o" 
"libs/atomic/src/lockpool.cpp"

-- 
Mike Looijmans

Re: boost fails to build on MIPS

[Phil Blundell]

On Fri, 2015-05-29 at 20:24 +0200, Mike Looijmans wrote:
> On current OE master, boost fails to build, because something attempts 
> to add "-m32" to the gcc commandline for no apparent reason.
> 
> Can anyone confirm this, or am I the only MIPS user in the world?

This is an upstream boost bug.  See the entertaining history at:

https://github.com/boostorg/build/commit/9576ca216cae888d7efb5da97c009c245c5ba567

p.

Re: boost fails to build on MIPS

[Mike Looijmans]

On 30-05-15 00:09, Phil Blundell wrote:
> On Fri, 2015-05-29 at 20:24 +0200, Mike Looijmans wrote:
>> On current OE master, boost fails to build, because something attempts
>> to add "-m32" to the gcc commandline for no apparent reason.
>>
>> Can anyone confirm this, or am I the only MIPS user in the world?
>
> This is an upstream boost bug.  See the entertaining history at:
>
> https://github.com/boostorg/build/commit/9576ca216cae888d7efb5da97c009c245c5ba567

Strange, I added this fix as a patch to bjam (which seems to be the one 
causing all the troubles), but still get the same build error, even 
after cleansstate both bjam-native and boost.

Maybe some cross-compiler stupidity like that it looks at the build 
machine's arch instead of the target host's?



-- 
Mike Looijmans

Re: boost fails to build on MIPS

[Robert Yang]

On 05/31/2015 12:45 AM, Mike Looijmans wrote:
> On 30-05-15 00:09, Phil Blundell wrote:
>> On Fri, 2015-05-29 at 20:24 +0200, Mike Looijmans wrote:
>>> On current OE master, boost fails to build, because something attempts
>>> to add "-m32" to the gcc commandline for no apparent reason.
>>>
>>> Can anyone confirm this, or am I the only MIPS user in the world?
>>
>> This is an upstream boost bug.  See the entertaining history at:
>>
>> https://github.com/boostorg/build/commit/9576ca216cae888d7efb5da97c009c245c5ba567
>
> Strange, I added this fix as a patch to bjam (which seems to be the one causing
> all the troubles), but still get the same build error, even after cleansstate
> both bjam-native and boost.
>
> Maybe some cross-compiler stupidity like that it looks at the build machine's
> arch instead of the target host's?

Hello,

I verified the following 2 patches work:

https://github.com/boostorg/build/commit/c0634341d9ee2c02d3a55c91dafb988afc066c49
https://github.com/boostorg/build/commit/9576ca216cae888d7efb5da97c009c245c5ba567

I will send a fix for boost.

// Robert


>
>
>