[PATCH] Add a note to LangRef about static allocas.

Sean Silva chisophugis at gmail.com
Tue Jul 1 22:17:38 PDT 2014


On Tue, Jul 1, 2014 at 9:07 PM, Alp Toker <alp at nuanti.com> wrote:

>
> On 02/07/2014 05:48, Sean Silva wrote:
>
>
>>
>>
>> On Tue, Jul 1, 2014 at 6:41 PM, Matt Arsenault <Matthew.Arsenault at amd.com
>> <mailto:Matthew.Arsenault at amd.com>> wrote:
>>
>>     On 07/01/2014 05:31 PM, Reid Kleckner wrote:
>>
>>         Sounds good to me, go for it.  I'd wordsmith it a bit.  I
>>         don't think "is basically free" is important.  The important
>>         thing is that the stack adjustment is folded into the function
>>         prologue.
>>
>>     Both of these are very target specific details that don't really
>>     have anything to do with the IR instruction's semantics
>>
>>
>> I agree that it is target-specific, and that target-specific information
>> should be avoided in LangRef. However, this is a fairly important
>> distinction; for example, a frontend author should definitely be aware of
>> it.
>>
>> I'll wait for some more opinions.
>>
>
> Seems fine to add this, but perhaps drop the loud "Note" heading


There is already precedent to create sectional divisions for remarks that
don't fall into the "standard" headings of the instructions. E.g.
http://llvm.org/docs/LangRef.html#implementation
(gah that anchor name is really bad...)
Also there is precedent for removing them as needed e.g.
http://llvm.org/docs/LangRef.html#unreachable-instruction


> and qualify the statement with something like "a backend may fold" to
> address Matt's observation.
>

That sounds like a good thing to call out explicitly.

-- Sean Silva


>
> Alp.
>
>
>
>
>> -- Sean Silva
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>
> --
> http://www.nuanti.com
> the browser experts
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140701/d80f37e7/attachment.html>


More information about the llvm-commits mailing list