Re: soft float on a hard flat capable machine ?

On Wed, Sep 23, 2015 at 4:19 PM, Jérémy Lal <[email protected]> wrote:
> 2015-09-22 10:43 GMT+02:00 James Cowgill <[email protected]>:
>>
>> On Tue, 2015-09-22 at 10:30 +0200, Jérémy Lal wrote:
>> > 2015-09-22 10:22 GMT+02:00 James Cowgill <[email protected]>:
>> > > On Tue, 2015-09-22 at 10:21 +0200, Jérémy Lal wrote:
>> > > > 2015-09-22 10:14 GMT+02:00 James Cowgill <[email protected]>:
>> > > > > Hi,
>> > > > >
>> > > > > On Tue, 2015-09-22 at 08:59 +0200, Jérémy Lal wrote:
>> > > > > > Hi,
>> > > > > >
>> > > > > > while building nodejs 4.1.0~dfsg-2 (not -3) it seems i can't
>> > > > > > enable
>> > > > > >   -D__mips_soft_float
>> > > > > > It errors out with [0]
>> > > > > >   gnu/stubs-o32_soft.h: No such file or directory
>> > > > > > Indeed, there is only one header in that directory.
>> > > > > > Is it not possible to enable soft float where hard float is
>> > > > > > available ?
>> > > > > > Is this a bug ?
>> > > > >
>> > > > > On MIPS, soft-float and hard-float use totally incompatible ABIs
>> > > > > so you
>> > > > > would need to recompile the toolchain to use it.
>> > > >
>> > > > So does it imply that mips/mipsel architectures are hard-float ?
>> > >
>> > > Yes
>> > >
>> >
>> > Ok then, shouldn't this be written somewhere ?
>> >
>> > https://www.debian.org/ports/mips/
>> > https://wiki.debian.org/MIPSPort
>>
>> I've added it to the wiki page. Those pages don't get a lot of updates
>> though.
>
>
> Good, thank you.
> Now about another problem,
> https://github.com/paul99/v8m-rb/issues/192#issuecomment-142479915
> the question is
> Do mips/mipsel use only fp32 ?
>

Previously we use FP-32, and we are now in the progress of transiting
to FP-XX. After the transition is done we will discuss about FP-64.

Regards,
Aron