[llvm-commits] [llvm] r173342 - in /llvm/trunk: lib/Transforms/Utils/SimplifyCFG.cpp test/Transforms/SimplifyCFG/SpeculativeExec.ll

[Chandler Carruth]

On Fri, Jan 25, 2013 at 1:43 PM, Jakob Stoklund Olesen <stoklund at 2pi.dk>wrote:

>
> On Jan 25, 2013, at 12:00 PM, Chandler Carruth <chandlerc at gmail.com>
> wrote:
>
> The presence of a 'trunc' in the conditional basic block shouldn't be the
> thing that enables or disables this. All I wanted out of this patch was
> improved stability and consistency on x86.
>
>
> It basically sounds like you are already on your way down the path Duncan
> warned about.
>
> This is not the right way to do it.
>

I really am not...


Clearly this commit is more controversial than I had ever anticipated
(including when I mentioned it to Chris on IRC), so I'm happy to back it
out... It'll be a bit, as I can't just revert it at this point -- there
were other patches here that fixed somewhat serious bugs -- but I'll get to
it soon.


Anyways, I think there is a lot of confusion, miscommunication, and
misunderstanding in this thread (and the other on a related subject, but
which had *nothing* to do with my actual commits)... So I hope we can get
back to discussing the actual technical issues at hand here. Note that
while I've reverted this patch, SimplifyCFG still uses TTI, and fixing that
is a bigger task -- that was introduced with the lookup table optimization.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130125/c80b33bc/attachment.html>