[llvm-dev] RFC: We should stop merging allocas in the inliner
Duncan P. N. Exon Smith via llvm-dev
llvm-dev at lists.llvm.org
Mon Aug 1 19:16:08 PDT 2016
> On 2016-Aug-01, at 19:14, Chandler Carruth <chandlerc at gmail.com> wrote:
> On Mon, Aug 1, 2016 at 5:51 PM Duncan P. N. Exon Smith via llvm-dev <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>> wrote:
> I just want to note that there's currently a miscompile in C code with lifetime markers:
> http://lists.llvm.org/pipermail/cfe-dev/2016-July/050066.html <http://lists.llvm.org/pipermail/cfe-dev/2016-July/050066.html>
> Depending on how we fix that, we could end up being more pessimistic about when markers are emitted. I suggest we fix that before benchmarking this change.
> That shouldn't impact this change.
> The inliner will emit *inlining* based lifetime markers all on its own for any static allocas that the frontend doesn't annotate. So we could turn off all lifetime markers in Clang and the merging here should still not be necessary to get the stack savings.
> (And I have now verified that yes, the inliner inserts these markers...)
Ah, thanks, I'd misunderstood up-thread somehow. In that case I agree that it's independent.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the llvm-dev