This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran 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: Gcc stage 3 ends January 14th


On Fri, Jan 12, 2018 at 3:50 PM, Paul Richard Thomas
<paul.richard.thomas@gmail.com> wrote:
> Hi Thomas,
>
> I could adapt the patch for PR34640 to add the extra fields; two more
> for the breakup of the dtype and at least one more for flags
> (associate, allocated, contiguous, etc.) and...? I will take a look at
> the F2018 C descriptor and remind myself of the necessary fields.

There's https://gcc.gnu.org/wiki/ArrayDescriptorUpdate . One thing to
ponder about is whether the attribute and type fields should be char
(saves a tiny bit of space) or int (avoids partial regs and length
prefix changes on x86,  slowdowns or other issues on subword access on
some RISC archs (e.g. RISC-V compressed extension doesn't have subword
load/store instructions)).

>
> Cheers
>
> Paul
>
>
> On 11 January 2018 at 23:13, Thomas Koenig <tkoenig@netcologne.de> wrote:
>> For those who don't read the gcc mailing list:
>>
>> https://gcc.gnu.org/ml/gcc/2018-01/msg00033.html
>>
>> Question: How should we deal with the ABI breakage that
>> we already caused, and which we could potentially
>> exploit in time for gcc 8?
>>
>> Or should we just resign ourselves that we are going to
>> break the ABI again for gcc 9, or whenever we feel like
>> increasing our dimensions to 15 in accordance to F2008?
>>
>> Regards
>>
>>         Thomas
>
>
>
> --
> "If you can't explain it simply, you don't understand it well enough"
> - Albert Einstein



-- 
Janne Blomqvist


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