[ValueTracking] determine sign of 0.0 from select when matching min/max FP
In PR39475:
https://bugs.llvm.org/show_bug.cgi?id=39475
..we may fail to recognize/simplify fabs() in some cases because we do not
canonicalize fcmp with a -0.0 operand.
Adding that canonicalization can cause regressions on min/max FP tests, so
that's this patch: for the purpose of determining whether something is min/max,
let the value returned by the select determine how we treat a 0.0 operand in the fcmp.
This patch doesn't actually change the -0.0 to +0.0. It just changes the analysis, so
we don't fail to recognize equivalent min/max patterns that only differ in the
signbit of 0.0.
Differential Revision: https://reviews.llvm.org/D54001
llvm-svn: 346097
diff --git a/llvm/unittests/Analysis/ValueTrackingTest.cpp b/llvm/unittests/Analysis/ValueTrackingTest.cpp
index b13948a..c4adde4 100644
--- a/llvm/unittests/Analysis/ValueTrackingTest.cpp
+++ b/llvm/unittests/Analysis/ValueTrackingTest.cpp
@@ -156,8 +156,8 @@
" %A = select i1 %1, float 0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_NAN, true});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero2) {
@@ -167,8 +167,8 @@
" %A = select i1 %1, float 0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_NAN, false});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero3) {
@@ -178,8 +178,8 @@
" %A = select i1 %1, float -0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_NAN, true});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero4) {
@@ -189,8 +189,8 @@
" %A = select i1 %1, float -0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_NAN, false});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero5) {
@@ -200,8 +200,8 @@
" %A = select i1 %1, float %a, float 0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_OTHER, false});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero6) {
@@ -211,8 +211,8 @@
" %A = select i1 %1, float %a, float 0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_OTHER, true});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero7) {
@@ -222,8 +222,8 @@
" %A = select i1 %1, float %a, float -0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_OTHER, false});
}
TEST_F(MatchSelectPatternTest, FMinMismatchConstantZero8) {
@@ -233,8 +233,8 @@
" %A = select i1 %1, float %a, float -0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMINNUM, SPNB_RETURNS_OTHER, true});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero1) {
@@ -244,8 +244,8 @@
" %A = select i1 %1, float 0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_NAN, true});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero2) {
@@ -255,8 +255,8 @@
" %A = select i1 %1, float 0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_NAN, false});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero3) {
@@ -266,8 +266,8 @@
" %A = select i1 %1, float -0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_NAN, true});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero4) {
@@ -277,8 +277,8 @@
" %A = select i1 %1, float -0.0, float %a\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_NAN, false});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero5) {
@@ -288,8 +288,8 @@
" %A = select i1 %1, float %a, float 0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_OTHER, false});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero6) {
@@ -299,8 +299,8 @@
" %A = select i1 %1, float %a, float 0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_OTHER, true});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero7) {
@@ -310,8 +310,8 @@
" %A = select i1 %1, float %a, float -0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
- expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_OTHER, false});
}
TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZero8) {
@@ -321,7 +321,29 @@
" %A = select i1 %1, float %a, float -0.0\n"
" ret float %A\n"
"}\n");
- // FIXME: The sign of zero doesn't matter in fcmp.
+ // The sign of zero doesn't matter in fcmp.
+ expectPattern({SPF_FMAXNUM, SPNB_RETURNS_OTHER, true});
+}
+
+TEST_F(MatchSelectPatternTest, FMinMismatchConstantZeroVecUndef) {
+ parseAssembly(
+ "define <2 x float> @test(<2 x float> %a) {\n"
+ " %1 = fcmp ogt <2 x float> %a, <float -0.0, float -0.0>\n"
+ " %A = select <2 x i1> %1, <2 x float> <float undef, float 0.0>, <2 x float> %a\n"
+ " ret <2 x float> %A\n"
+ "}\n");
+ // An undef in a vector constant can not be back-propagated for this analysis.
+ expectPattern({SPF_UNKNOWN, SPNB_NA, false});
+}
+
+TEST_F(MatchSelectPatternTest, FMaxMismatchConstantZeroVecUndef) {
+ parseAssembly(
+ "define <2 x float> @test(<2 x float> %a) {\n"
+ " %1 = fcmp ogt <2 x float> %a, zeroinitializer\n"
+ " %A = select <2 x i1> %1, <2 x float> %a, <2 x float> <float -0.0, float undef>\n"
+ " ret <2 x float> %A\n"
+ "}\n");
+ // An undef in a vector constant can not be back-propagated for this analysis.
expectPattern({SPF_UNKNOWN, SPNB_NA, false});
}