<div dir="ltr"><div>This is just documenting the current status. If we choose not to transform these in InstCombine, then we should have a record of that limitation that we acknowledge is a good thing. This way someone won't come along in the future and wonder if the transform doesn't exist intentionally or it's just an oversight. I can add some test comments once we have a conclusion in D33172.<br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, May 15, 2017 at 6:23 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 class="">On Mon, May 15, 2017 at 4:59 PM, Sanjay Patel via llvm-commits<br>
<<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>> wrote:<br>
> Author: spatel<br>
> Date: Mon May 15 18:59:28 2017<br>
> New Revision: 303133<br>
><br>
> URL: <a href="http://llvm.org/viewvc/llvm-project?rev=303133&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-<wbr>project?rev=303133&view=rev</a><br>
> Log:<br>
> [InstCombine] add tests for PR32791; NFC<br>
><br>
<br>
</span>I'm slightly confused by these tests being committed inside<br>
InstCombine & labeled as InstCombine.<br>
We seem to not have an agreement yet on whether this transformation<br>
should live, so I don't see the reason for committing these<br>
beforehand.<br>
<div class="HOEnZb"><div class="h5"><br>
> Modified:<br>
> llvm/trunk/test/Transforms/<wbr>InstCombine/logical-select.ll<br>
><br>
> Modified: llvm/trunk/test/Transforms/<wbr>InstCombine/logical-select.ll<br>
> URL: <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/logical-select.ll?rev=303133&r1=303132&r2=303133&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-<wbr>project/llvm/trunk/test/<wbr>Transforms/InstCombine/<wbr>logical-select.ll?rev=303133&<wbr>r1=303132&r2=303133&view=diff</a><br>
> ==============================<wbr>==============================<wbr>==================<br>
> --- llvm/trunk/test/Transforms/<wbr>InstCombine/logical-select.ll (original)<br>
> +++ llvm/trunk/test/Transforms/<wbr>InstCombine/logical-select.ll Mon May 15 18:59:28 2017<br>
> @@ -62,6 +62,76 @@ define i32 @poo(i32 %a, i32 %b, i32 %c,<br>
> ret i32 %t3<br>
> }<br>
><br>
> +; PR32791 - <a href="https://bugs.llvm.org//show_bug.cgi?id=32791" rel="noreferrer" target="_blank">https://bugs.llvm.org//show_<wbr>bug.cgi?id=32791</a><br>
> +; Fold two selects with inverted predicates and zero operands.<br>
> +define i32 @fold_inverted_icmp_preds(i32 %a, i32 %b, i32 %c, i32 %d) {<br>
> +; CHECK-LABEL: @fold_inverted_icmp_preds(<br>
> +; CHECK-NEXT: [[CMP1:%.*]] = icmp slt i32 %a, %b<br>
> +; CHECK-NEXT: [[SEL1:%.*]] = select i1 [[CMP1]], i32 %c, i32 0<br>
> +; CHECK-NEXT: [[CMP2:%.*]] = icmp sge i32 %a, %b<br>
> +; CHECK-NEXT: [[SEL2:%.*]] = select i1 [[CMP2]], i32 %d, i32 0<br>
> +; CHECK-NEXT: [[OR:%.*]] = or i32 [[SEL1]], [[SEL2]]<br>
> +; CHECK-NEXT: ret i32 [[OR]]<br>
> +;<br>
> + %cmp1 = icmp slt i32 %a, %b<br>
> + %sel1 = select i1 %cmp1, i32 %c, i32 0<br>
> + %cmp2 = icmp sge i32 %a, %b<br>
> + %sel2 = select i1 %cmp2, i32 %d, i32 0<br>
> + %or = or i32 %sel1, %sel2<br>
> + ret i32 %or<br>
> +}<br>
> +<br>
> +define i32 @fold_inverted_icmp_preds_<wbr>reverse(i32 %a, i32 %b, i32 %c, i32 %d) {<br>
> +; CHECK-LABEL: @fold_inverted_icmp_preds_<wbr>reverse(<br>
> +; CHECK-NEXT: [[CMP1:%.*]] = icmp slt i32 %a, %b<br>
> +; CHECK-NEXT: [[SEL1:%.*]] = select i1 [[CMP1]], i32 0, i32 %c<br>
> +; CHECK-NEXT: [[CMP2:%.*]] = icmp sge i32 %a, %b<br>
> +; CHECK-NEXT: [[SEL2:%.*]] = select i1 [[CMP2]], i32 0, i32 %d<br>
> +; CHECK-NEXT: [[OR:%.*]] = or i32 [[SEL1]], [[SEL2]]<br>
> +; CHECK-NEXT: ret i32 [[OR]]<br>
> +;<br>
> + %cmp1 = icmp slt i32 %a, %b<br>
> + %sel1 = select i1 %cmp1, i32 0, i32 %c<br>
> + %cmp2 = icmp sge i32 %a, %b<br>
> + %sel2 = select i1 %cmp2, i32 0, i32 %d<br>
> + %or = or i32 %sel1, %sel2<br>
> + ret i32 %or<br>
> +}<br>
> +<br>
> +define i32 @fold_inverted_fcmp_preds(<wbr>float %a, float %b, i32 %c, i32 %d) {<br>
> +; CHECK-LABEL: @fold_inverted_fcmp_preds(<br>
> +; CHECK-NEXT: [[CMP1:%.*]] = fcmp olt float %a, %b<br>
> +; CHECK-NEXT: [[SEL1:%.*]] = select i1 [[CMP1]], i32 %c, i32 0<br>
> +; CHECK-NEXT: [[CMP2:%.*]] = fcmp uge float %a, %b<br>
> +; CHECK-NEXT: [[SEL2:%.*]] = select i1 [[CMP2]], i32 %d, i32 0<br>
> +; CHECK-NEXT: [[OR:%.*]] = or i32 [[SEL1]], [[SEL2]]<br>
> +; CHECK-NEXT: ret i32 [[OR]]<br>
> +;<br>
> + %cmp1 = fcmp olt float %a, %b<br>
> + %sel1 = select i1 %cmp1, i32 %c, i32 0<br>
> + %cmp2 = fcmp uge float %a, %b<br>
> + %sel2 = select i1 %cmp2, i32 %d, i32 0<br>
> + %or = or i32 %sel1, %sel2<br>
> + ret i32 %or<br>
> +}<br>
> +<br>
> +define <2 x i32> @fold_inverted_icmp_vector_<wbr>preds(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c, <2 x i32> %d) {<br>
> +; CHECK-LABEL: @fold_inverted_icmp_vector_<wbr>preds(<br>
> +; CHECK-NEXT: [[CMP1:%.*]] = icmp ne <2 x i32> %a, %b<br>
> +; CHECK-NEXT: [[SEL1:%.*]] = select <2 x i1> [[CMP1]], <2 x i32> %c, <2 x i32> zeroinitializer<br>
> +; CHECK-NEXT: [[CMP2:%.*]] = icmp eq <2 x i32> %a, %b<br>
> +; CHECK-NEXT: [[SEL2:%.*]] = select <2 x i1> [[CMP2]], <2 x i32> %d, <2 x i32> zeroinitializer<br>
> +; CHECK-NEXT: [[OR:%.*]] = or <2 x i32> [[SEL1]], [[SEL2]]<br>
> +; CHECK-NEXT: ret <2 x i32> [[OR]]<br>
> +;<br>
> + %cmp1 = icmp ne <2 x i32> %a, %b<br>
> + %sel1 = select <2 x i1> %cmp1, <2 x i32> %c, <2 x i32> <i32 0, i32 0><br>
> + %cmp2 = icmp eq <2 x i32> %a, %b<br>
> + %sel2 = select <2 x i1> %cmp2, <2 x i32> %d, <2 x i32> <i32 0, i32 0><br>
> + %or = or <2 x i32> %sel1, %sel2<br>
> + ret <2 x i32> %or<br>
> +}<br>
> +<br>
> define i32 @par(i32 %a, i32 %b, i32 %c, i32 %d) {<br>
> ; CHECK-LABEL: @par(<br>
> ; CHECK-NEXT: [[T0:%.*]] = icmp slt i32 %a, %b<br>
><br>
><br>
> ______________________________<wbr>_________________<br>
> llvm-commits mailing list<br>
> <a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
> <a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/llvm-commits</a><br>
<br>
</div></div><span class="HOEnZb"><font color="#888888">--<br>
Davide<br>
<br>
"There are no solved problems; there are only problems that are more<br>
or less solved" -- Henri Poincare<br>
</font></span></blockquote></div><br></div>