<div dir="ltr"><div><a href="https://reviews.llvm.org/rG8cdcbcaa02e7" rel="noreferrer" target="_blank">https://reviews.llvm.org/rG8cdcbcaa02e7</a></div><div><a href="https://reviews.llvm.org/rG1e308452bf68" rel="noreferrer" target="_blank">https://reviews.llvm.org/rG1e308452bf68</a></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Feb 27, 2020 at 6:29 PM Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Sure. That sounds great. Thanks!</div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Feb 26, 2020 at 10:45 AM Sanjay Patel <<a href="mailto:spatel@rotateright.com" target="_blank">spatel@rotateright.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div>To be clear - the test is checking IR instructions, but it's checking -O1 IR for various targets. <br></div><div>So there must be different expectations per target...<br></div><div>But I just tried a test of turning everything  down to -O0, and it all passed except for the "fast-math" run for AArch64.</div><div>I can tweak that to not be so specific if that sounds like a reasonable solution.<br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Feb 26, 2020 at 1:05 PM Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto">I mean anything that's testing assembly output out of clang is less than ideal. There are some circumstances, but this doesn't seem like one of them. </div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Feb 26, 2020, 9:10 AM Sanjay Patel <<a href="mailto:spatel@rotateright.com" target="_blank">spatel@rotateright.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div>The test file dates back to:</div><div><a href="https://reviews.llvm.org/D5698" rel="noreferrer" target="_blank">https://reviews.llvm.org/D5698</a></div><div>...and I'm not familiar with _Complex enough to say how to fix this properly (seems like the check lines are already limited such that -O0 rather than -O1 would work?).</div><div><br> </div><div>But this file keeps wiggling unexpectedly, it's going to move again with <a href="https://reviews.llvm.org/D75130" rel="noreferrer noreferrer" target="_blank">https://reviews.llvm.org/D75130</a></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Feb 25, 2020 at 1:15 PM Eric Christopher <<a href="mailto:echristo@gmail.com" rel="noreferrer" target="_blank">echristo@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Is there any way to pull this test out of clang and as an opt test? What's it trying to test?<div><br></div><div>-eric</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Feb 25, 2020 at 6:15 AM Sanjay Patel via cfe-commits <<a href="mailto:cfe-commits@lists.llvm.org" rel="noreferrer" target="_blank">cfe-commits@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><br>
Author: Sanjay Patel<br>
Date: 2020-02-25T09:13:49-05:00<br>
New Revision: 83f4372f3a708ceaa800feff8b1bd92ae2c3be5f<br>
<br>
URL: <a href="https://github.com/llvm/llvm-project/commit/83f4372f3a708ceaa800feff8b1bd92ae2c3be5f" rel="noreferrer noreferrer" target="_blank">https://github.com/llvm/llvm-project/commit/83f4372f3a708ceaa800feff8b1bd92ae2c3be5f</a><br>
DIFF: <a href="https://github.com/llvm/llvm-project/commit/83f4372f3a708ceaa800feff8b1bd92ae2c3be5f.diff" rel="noreferrer noreferrer" target="_blank">https://github.com/llvm/llvm-project/commit/83f4372f3a708ceaa800feff8b1bd92ae2c3be5f.diff</a><br>
<br>
LOG: [CodeGen] fix clang test that runs the optimizer pipeline; NFC<br>
<br>
There's already a FIXME note on this file; it can break when the<br>
underlying LLVM behavior changes independently of anything in clang.<br>
<br>
Added: <br>
<br>
<br>
Modified: <br>
    clang/test/CodeGen/complex-math.c<br>
<br>
Removed: <br>
<br>
<br>
<br>
################################################################################<br>
diff  --git a/clang/test/CodeGen/complex-math.c b/clang/test/CodeGen/complex-math.c<br>
index e42418ad72c2..54dee473a364 100644<br>
--- a/clang/test/CodeGen/complex-math.c<br>
+++ b/clang/test/CodeGen/complex-math.c<br>
@@ -93,14 +93,15 @@ float _Complex mul_float_rc(float a, float _Complex b) {<br>
   // X86: ret<br>
   return a * b;<br>
 }<br>
+<br>
 float _Complex mul_float_cc(float _Complex a, float _Complex b) {<br>
   // X86-LABEL: @mul_float_cc(<br>
   // X86: %[[AC:[^ ]+]] = fmul<br>
   // X86: %[[BD:[^ ]+]] = fmul<br>
   // X86: %[[AD:[^ ]+]] = fmul<br>
   // X86: %[[BC:[^ ]+]] = fmul<br>
-  // X86: %[[RR:[^ ]+]] = fsub float %[[AC]], %[[BD]]<br>
-  // X86: %[[RI:[^ ]+]] = fadd float<br>
+  // X86: %[[RR:[^ ]+]] = fsub<br>
+  // X86: %[[RI:[^ ]+]] = fadd<br>
   // X86-DAG: %[[AD]]<br>
   // X86-DAG: ,<br>
   // X86-DAG: %[[BC]]<br>
<br>
<br>
<br>
_______________________________________________<br>
cfe-commits mailing list<br>
<a href="mailto:cfe-commits@lists.llvm.org" rel="noreferrer" target="_blank">cfe-commits@lists.llvm.org</a><br>
<a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits" rel="noreferrer noreferrer" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits</a><br>
</blockquote></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>