Hi Galina,<br><br>Apologies, I am between jobs at the moment and do not have access to any machine that can build LLVM. I will have access to one on Monday evening when I start with Google. <br><br>In the meantime perhaps you could commit the minimal change to make the bot green and I could check it / handle the fallout on Monday?<br><br>James<br><div class="gmail_quote"><div dir="ltr">On Wed, 7 Jun 2017 at 03:15, Galina Kistanova <<a href="mailto:gkistanova@gmail.com">gkistanova@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div><div><div>Hello James.<br><br>Here is the bug:</div></div></div></div></div><div dir="ltr"><div><div><div><div><br><br>+ std::stable_sort(<br>
+ Candidates.begin(), Candidates.end(),<br>
+ [](const SinkingInstructionCandidate &A,<br>
+ const SinkingInstructionCandidate &B) { return A >= B; });<br><br></div></div></div></div></div><div dir="ltr"><div><div><div><div></div>This comparison operator is invalid, because it does not implement strict weak ordering. With this logic, if 2 candidates A and B have the same cost, the comparison operator will return true, but must return false.<br><br></div>This fix would fix the <a href="http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win" target="_blank">http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win</a> bot and <a href="https://bugs.llvm.org/show_bug.cgi?id=33210" target="_blank">https://bugs.llvm.org/show_bug.cgi?id=33210</a>.<br><br></div><div>Could you take care of this, please?<br></div><div><br></div>Thanks<br><br></div></div><div dir="ltr">Galina<br><div><div><br></div></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Jun 2, 2017 at 12:18 PM, Galina Kistanova <span dir="ltr"><<a href="mailto:gkistanova@gmail.com" target="_blank">gkistanova@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">The <a href="http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win" target="_blank">http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win</a> bot builds the debug configuration now.<br><br><div><div><div><div>James, could you fix the GVNSink.cpp, line 749, please?<br></div><span><br>DEBUG(dbgs() << " -- Sinking candidates:\n"; for (auto &C<br></span> : Candidates) dbgs()<br> << " " << C << "\n";);<br></div><div>It triggers a crash.<br><a href="http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/2891/steps/test-check-all/logs/stdio" target="_blank">http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/2891/steps/test-check-all/logs/stdio</a><br><br></div></div>Thanks<span class="m_-9028625266688623605HOEnZb"><font color="#888888"><br><br></font></span></div><span class="m_-9028625266688623605HOEnZb"><font color="#888888">Galina<br><br></font></span></div><div class="m_-9028625266688623605HOEnZb"><div class="m_-9028625266688623605h5"><div class="gmail_extra"><br><div class="gmail_quote">On Tue, May 30, 2017 at 12:14 PM, Davide Italiano <span dir="ltr"><<a href="mailto:davide@freebsd.org" target="_blank">davide@freebsd.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span>On Tue, May 30, 2017 at 11:34 AM, Galina Kistanova <<a href="mailto:gkistanova@gmail.com" target="_blank">gkistanova@gmail.com</a>> wrote:<br>
> I'll see if I could find few spare CPU circles on one of the MSVC builder.<br>
><br>
> What targets we are after here? Just LLVM or LLVM+Clang?<br>
> If this is LLVM only, I might be able to reconfigure one of the existing<br>
> builders to build debug instead of the release.<br>
><br>
<br>
</span>I think the crash was triggered by LLVM, but having LLVM + clang would<br>
be better.<br>
<br>
--<br>
Davide<br>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>
</blockquote></div>