This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][ARM] Optimized 64-bit multiplication for THUMB-1


Ping? Is the patch okay with an explict .arm directive added or some
other modification is required?

-Doug

在 2010年10月26日下午2:33,Doug Kwan (關振德) <dougkwan@google.com> 寫道:
> Hi,
>
>  I looked at the definition of the ARM_FUNC_START macro, the cases in
> which the macro does not force use of ARM mode are:
>
> - __thumb2__ is defined, the macro is defined but no .arm used.
> - __ARM_ARCH_6M__ is defined, the macro is not defined.
>
>  In both of the cases above, the code protected by the test is not
> assembled, so there is no problem observed.  I can add an .arm to be
> explicit like the attached patch.  Would that be better?
>
> -Doug
>
>
>
>
>
>
> 在 2010年10月25日下午4:36,Paul Brook <paul@codesourcery.com> 寫道:
>>> Hi Paul,
>>>
>>>    Thank you very much for your review and comments.  I have fixed the
>>> push/pop and use of 2-argument code in 32-bit code.  I am not quite
>>> sure what the problem in the __thumb2__ test is.  I built arm-eabi-gcc
>>> with arches armv4, armv5te, armv7-a and no-arch and all build was
>>> successful.  I did change the test so that forcing ARM mode is only
>>> done if:
>>
>> No. You're missing the point. ARM_FUNC_START does not force the use of ARM
>> mode.  See comments near the definition of that macro.
>>
>> Paul
>>
>


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]