[llvm-commits] [LLVMdev] [RFC] "noclone" function attribute

[James Molloy]

Hi Krzysztof,

Yes, the spec is ambiguous. The Khronos member I spoke to believed the
spirit of the spec was that barriers were keyed by call path (i.e. PC
and call stack) rather than the PC alone.

Many implementations force-inline everything anyway, so this is kind of
the only interpretation of the spec that will work in real life :(

Cheers,

James

On Thu, 2012-12-06 at 16:50 +0000, Tim Northover wrote:
> > What begins to worry me a bit here, is that I'm no longer sure what "same
> > barrier" means.  Do we have a certainty exactly what problem we are trying
> > to solve?  Is a barrier "the same" only if it has been reached through the
> > same call chain from the moment the threads were created?
>
> The issue appears very similar to the old dynamic vs lexical scoping
> of variables. And being honest I have just as much distaste for for
> context-sensitive barriers as I do for dynamic scoping.
>
> But apparently there are good efficiency reasons for putting this
> burden on the programmer so it'll probably stick (in whatever form).
> :-(.
>
> Tim.
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>


-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium.  Thank you.