[cfe-commits] Plumbing through some MIPS options

[Carl Norum]

Here's the change for the -G flag.  I sent the mips-delay-filler one over to llvm-dev.

-- Carl


-------------- next part --------------
A non-text attachment was scrubbed...
Name: support-g-flag.patch
Type: application/octet-stream
Size: 622 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120817/9f18df0c/attachment.obj>
-------------- next part --------------



On Aug 17, 2012, at 4:26 PM, Carl Norum wrote:

> 
> Sure, easily done.
> 
> -- Carl
> 
> 
> On Aug 17, 2012, at 4:18 PM, Hatanaka, Akira wrote:
> 
>> Could you rename the option to "disable-mips-delay-filler" (false by default)?
>> 
>> ________________________________________
>> From: Carl Norum [carl at lytro.com]
>> Sent: Friday, August 17, 2012 4:00 PM
>> To: Hatanaka, Akira
>> Cc: Eli Friedman; cfe-commits
>> Subject: Re: [cfe-commits] Plumbing through some MIPS options
>> 
>> Yeah, -G is the GCC option.  We have it passed through right now in some -Xassembler flags.
>> 
>> I'll turn the delay filler on by default and change the other flag's name to -G.
>> 
>> -- Carl
>> 
>> 
>> On Aug 17, 2012, at 2:25 PM, Hatanaka, Akira wrote:
>> 
>>> I believe this "-G" option is it:
>>> 
>>> -G NUM                  allow referencing objects up to NUM bytes
>>>                            implicitly with the gp register [default 8]
>>> ________________________________________
>>> From: Eli Friedman [eli.friedman at gmail.com]
>>> Sent: Friday, August 17, 2012 2:20 PM
>>> To: Hatanaka, Akira
>>> Cc: Carl Norum; cfe-commits
>>> Subject: Re: [cfe-commits] Plumbing through some MIPS options
>>> 
>>> On Fri, Aug 17, 2012 at 2:16 PM, Hatanaka, Akira <ahatanaka at mips.com> wrote:
>>>> (These are my personal opinions)
>>>> 
>>>> Turning on "-enable-mips-delay-filler" by default (in MipsDelaySlotFiller.cpp) or turning it on when compiling with "-O3" is better. I don't think it should be exposed as a command line option. We have been planning to change this for a while, but haven't done so.
>>>> 
>>>> -mmips-short-section-threshold is fine. There is a similar option in gcc.
>>> 
>>> What's the name of the gcc option?  If there's precedent, we should
>>> probably follow it.
>>> 
>>> -Eli
>> 
>