[cfe-dev] noinlining in clang/llvm

[Halfdan Ingvarsson via cfe-dev]

My understanding is that the portion of the talk you've linked to, is in
the context of profiling hints (e.g. if profiling detects just a single
callsite for a specific function, always inline its code, that sort of
thing).

The noinline attribute works just fine.

E.g:  https://godbolt.org/g/frM8J7

- ½

On 2 October 2017 at 23:44, Hayden Livingston via cfe-dev <
cfe-dev at lists.llvm.org> wrote:

> Chandler Carruth at a CppCon/NWCPP talk said that Clang will not
> respect the noinline gcc attribute (or clang attribute).
>
> Sometimes it is needed for semantic correctness that a method not be
> inlined (e.g. for GC lifetime visibility etc.)
>
> Is this true?
>
> Video: https://youtu.be/9RFYVbHRoLg?t=3822
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20171003/fd03281d/attachment.html>