[LLVMdev] RFC: New EH representation for MSVC compatibility

[Reid Kleckner]

On Wed, May 20, 2015 at 12:04 PM, Joseph Tremoulet <jotrem at microsoft.com>
wrote:

>  Sure, I could do this as a follow-on change when LLILC is in a position
> to exercise it.  If you wanted to go ahead and use the "endcleanup from
> label %cleanup unwind label %nextaction" naming/syntax in the meantime,
> that would be cool.
>
>
>
> > If we're not going to use the construct in Clang
>
> Once the construct and the attendant code duplication optimization are in
> place, would there be any reason not to switch Clang over to use it?  I'd
> think that doing so would let you eschew the frameescape and get
> better-optimized code (even on the non-EH paths) in functions that use SEH
> __finally.
>

Yeah, once LLVM supports entercleanup/endcleanup then we could definitely
use it for SEH __finally.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20150520/4dc98d4e/attachment.html>