[PATCH] scripts: add softfp support

Thu Oct 20 00:30:00 GMT 2011

On Thu, Oct 20, 2011 at 10:02 AM, Yann E. MORIN
<yann.morin.1998@anciens.enib.fr> wrote:
> Michael, All,
>
> On Wednesday 19 October 2011 04:29:20 Michael Hope wrote:
>> # HG changeset patch
>> # User Michael Hope <michael.hope@linaro.org>
>> # Date 1318991252 -46800
>> # Node ID a31d097e28cd73d07a5484129929a500b4d58efa
>> # Parent  a32156bd31c0d395e8d346431b123a7d2caa14cd
>> scripts: add softfp support
>>
>> ARM compilers can be built for soft float (software only, floats in
>> core registers), hard float (uses floating point instructions, floats
>> in FPU registers), or the half-way house softfp (uses floating point
>> instructions, floats in core registers).
>
> Feature definitely a nice addition, but too close to the release to
> add it now (which reminds me I should document the release plan on
> the website...). FYI, it's a release every three months, with about
> a 15-day slack before, used to stabilise the stuff. Next release is
> due by October the 31st, so we just entered the 15-day delay...
>
> I was thinking about cutting the release branch ahead of time, but
> handling both the relase and the devel at the same time is a bit
> complicated in my head, and does colide a bit on the schedule...
> I'd like to try, but this release is special: it also colides with
> the Prague events. Probably I'll do it for the next release (Feb`12)...

All good.  I'll keep it local for now.

>> Add support for softfp cross compilers to the GCC and GLIBC
>> configuration.  Needed for Ubuntu and other distros that are softfp.
>
> What about uClibc? How will it cope with softfp?

I'm not sure.  I've had a poke about in buildroot and it has support
for soft float and hard float but no softfp.  It seems to tie the FPU
into the calling convention like crosstool-NG currently does so hard
float == has VFP.

I can play with this more next week.

>> Signed-off-by: Michael Hope <michael.hope@linaro.org>
>>
>> diff -r a32156bd31c0 -r a31d097e28cd config/target.in
>> --- a/config/target.in        Sun Oct 16 17:51:42 2011 +0200
>> +++ b/config/target.in        Wed Oct 19 15:27:32 2011 +1300
>> @@ -271,6 +271,22 @@
>>        If your processor has no FPU, then you most probably want this, as it
>>        is faster than emulating the FPU in the kernel.
>>
>> +config ARCH_FLOAT_SOFTFP
>
> I'm a bit reluctant at adding an architecture-specific option to this
> generic file. Currently, all arch options are in the related arch file.
>
> However, I agreee that there is no easy way to nicely handle that with
> the current infrastructure... :-/
>
> *But* I recall a similar approach a few months back... It seemed that ARM
> is not the only architecture that support softfp. Seems PPC also uses it.
> So:
>
>> +    bool
>> +    prompt "softfp"
>> +    depends on ARCH_arm
>
>  - this arch-specific "depends on" should go away
>  - and either we keep the option un-protected, or we hide it behind
>   ARCH_SUPPORT_SOFTFP (or the like) which is set by archs that
>   support it.

OK.

>> +    help
>> +      Emit hardware floating point opcodes but use the software
>> +      floating point calling convention.
>> +
>> +      Architectures such as ARM use different registers for passing
>> +      floating point values depending on if they're in software mode
>> +      or hardware mode.  softfp emits FPU instructions but uses the
>> +      software FP calling convention allowing softfp code to
>> +      interoperate with legacy software only code.
>> +
>> +      If in doubt, use 'software' or 'hardware' mode instead.
>> +
>>  endchoice
>>
>>  config TARGET_CFLAGS
>> diff -r a32156bd31c0 -r a31d097e28cd scripts/build/libc/glibc-eglibc.sh-common
>> --- a/scripts/build/libc/glibc-eglibc.sh-common       Sun Oct 16 17:51:42 2011 +0200
>> +++ b/scripts/build/libc/glibc-eglibc.sh-common       Wed Oct 19 15:27:32 2011 +1300
>> @@ -132,9 +132,10 @@
>>          *) extra_config+=("--disable-shared");;
>>      esac
>>
>> -    case "${CT_ARCH_FLOAT_HW},${CT_ARCH_FLOAT_SW}" in
>> -        y,) extra_config+=("--with-fp");;
>> -        ,y) extra_config+=("--without-fp");;
>> +    case "${CT_ARCH_FLOAT_HW},${CT_ARCH_FLOAT_SW},${CT_ARCH_FLOAT_SOFTFP}" in
>> +        y,,) extra_config+=("--with-fp");;
>> +        ,y,) extra_config+=("--without-fp");;
>> +        ,,y) extra_config+=("--with-fp");;
>>      esac
>
> Argh!... This is starting to be unreadable... :-/
>
> config ARCH_FLOAT
>    string
>    default "hard"   if ARCH_FLOAT_HW
>    default "soft"   if ARCH_FLOAT_SW
>    default "softfp" if ARCH_FLOAT_SOFTFP
>
> Then:
>    case "${CT_ARCH_FLOAT}" in
>        hard)   ...;;
>        soft)   ...;;
>        softfp) ...;;
>    esac
>
> I'll do it.

OK.

>>      if [ "${CT_LIBC_DISABLE_VERSIONING}" = "y" ]; then
>> diff -r a32156bd31c0 -r a31d097e28cd scripts/functions
>> --- a/scripts/functions       Sun Oct 16 17:51:42 2011 +0200
>> +++ b/scripts/functions       Wed Oct 19 15:27:32 2011 +1300
>> @@ -984,6 +984,7 @@
>>      [ "${CT_ARCH_TUNE}"     ] && { CT_ARCH_TUNE_CFLAG="-mtune=${CT_ARCH_TUNE}";  CT_ARCH_WITH_TUNE="--with-tune=${CT_ARCH_TUNE}"; }
>>      [ "${CT_ARCH_FPU}"      ] && { CT_ARCH_FPU_CFLAG="-mfpu=${CT_ARCH_FPU}";     CT_ARCH_WITH_FPU="--with-fpu=${CT_ARCH_FPU}";    }
>>      [ "${CT_ARCH_FLOAT_SW}" ] && { CT_ARCH_FLOAT_CFLAG="-msoft-float";           CT_ARCH_WITH_FLOAT="--with-float=soft";          }
>> +    [ "${CT_ARCH_FLOAT_SOFTFP}" ] && { CT_ARCH_FLOAT_CFLAG="-mfloat-abi=softfp"; CT_ARCH_WITH_FLOAT="--with-float=softfp";        }
>
> And the last time this came up, it was pointed that CT_ARCH_FLOAT_HW did
> force neither -hard-float not --with-float=hard
>
> I'll look at it...

OK.

> Can we sit on this for now, and revisit after the release?

Sure.  I'll keep it locally.

-- Michael

--
For unsubscribe information see http://sourceware.org/lists.html#faq