tags 645692 + upstream fixed-upstream pending
thanks
After some more work to support relocations that gcc generate when compiling
for armv6 or greater and Thumb-2, armhf is now finally supported upstream.
There is still a few bits missing to make R_ARM_THM_JUMP24 work when the
target is PLT but
Le samedi 31 mars 2012 22:58:17, Thomas Preud'homme a écrit :
[SNIP]
>
> I fixed the regressions (a few alignment issues) brought by the last few
> changes in the code. While doing that, I also spotted one place which can
> be improved and two bugs. It is easy code to fix though, maybe 10 minute
Le mercredi 21 mars 2012 02:03:57, Thomas Preud'homme a écrit :
> Le mardi 13 mars 2012 00:07:33, Thomas Preud'homme a écrit :
> > > * homogeneous aggregate
> >
> > IN PROGRESS
>
> DONE
>
> So only correct assignment of float register is missing. There might also
> be a regression somewhere but
Le mardi 13 mars 2012 00:07:33, Thomas Preud'homme a écrit :
>
> > * homogeneous aggregate
>
> IN PROGRESS
DONE
So only correct assignment of float register is missing. There might also be a
regression somewhere but I'm not sure yet. I guess if I continue on this rythm
a patch should be ready
[SNIP]
>
> * bug when manipulating some structure lead to wrong register being used to
> pass parameter -> FIXED
> * space allocated on stack for passing parameter is not freed correctly
> after function call -> FIXED
> * The way alignment is done works for armel but not armhf -> WORK IN
> PROGRES
retitle 645692 Support for hard-float calling convention
thanks
Time for another update.
When trying to add support for homogeneous aggregate (which are now passed in
VFP register with armhf calling convention), I realized a few things in the
current code are wrong for armhf:
* bug when manipu
Le lundi 12 décembre 2011 16:16:18, Thomas Preud'homme a écrit :
Here is another update:
> 1) Return float result in VFP register(s)
Done
> 2) Use VFP registers for homogeneous aggregate (see §4.3.5 and §6.1.2.1 in
> AAPCS)
> 3) Pack 2 consecutive float arguments into one VFP register
Done
> 4) R
Small update about the progress so far:
I built a patch which add basic support for hardfloat ABI. It correctly place
float argument in VFP registers when calling a function and can fall back to
softfp if the function called is variadic. However, still a lot remains to be
done:
1) Return float
8 matches
Mail list logo