[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: My progress on armhf: xf86-video-msm for armhf attempt. Please test.



> On Fri, Sep 02, 2011 at 03:10:50AM +0100, Paul Brook wrote:
> > Your understanding is incorrect.  Thumb interworking is always enabled
> > when generating armv5 or later code, plus it is enabled by default for
> > all EABI based targets.  Both of which are true for armhf.
> 
> Well in that case, there was no reason for the armel package to specify
> it in the makefile.

Right.  That's almost certainly a bug.  My guess would be cargo-cult copying 
or an old hack to workaround borkenness elsewhere.
 
> > I don't see how gcc specs are relevant.  They just control commandline
> > option mangling.  Absence of an explicit -mthumb-interwork option
> > doesn't mean what you seem to think it does.
> 
> So what DOES this mean:
> 
> gcc -dumpspecs
> 
> ...
> *version:
> 4.6.1
> 
> *multilib:
> .::arm-linux-gnueabihf ;
> 
> *multilib_defaults:
> marm mlittle-endian mhard-float mno-thumb-interwork

Very little.

Your toolchain config is slightly confused. The multilib selection bits don't 
quite match what your compiler is actually generating.  However you only have 
a single set of libraries (i.e. aren't using multilibs), so that's completely 
irrelevant and harmless.

In the unlikely event that you did decide to build multilibs with/without 
interworking (which for any vaguely recent core would be completely pointless 
and have no effect as interworking is enabled whether you want it or not) it 
might cause some subtle breakage.  Until then the only effect is to confuse 
people who don't understand what they're reading (i.e. you).  GCC spec files, 
and multilib selection in partiular, is a horribly crufty mess which tends to 
acquire historic warts like this.  It's also in the process of being 
rewritten, so noone cares.

Paul


Reply to: