[InstCombine] [NFC] Add tests for peeking through unsigned FP casts for zero-equality compares (PR36682)

[lebedev.ri]

Summary

This pattern came up in PR36682 / D44390
https://bugs.llvm.org/show_bug.cgi?id=36682
https://reviews.llvm.org/D44390
https://godbolt.org/g/oKvT5H

Looking at the IR pattern in question, as per alive-nj, for all the type combinations i checked
(input: i16, i32, i64; intermediate: half/i16, float/i32, double/i64)
for the following icmp comparisons the uitofp+bitcast can be dropped:

• eq 0
• ne 0

I did not check vectors, but i'm guessing it's the same there.

alive-eq.txt3 KBDownload

Thus all these cases are in the testcase (along with the vector variant with additional undef element in the middle).
There are no negative patterns here (unless alive-nj lied/is broken), all of these should be optimized.

Generated with

bitcast.cpp3 KBDownload

[spatel]

LGTM - although I think it would be more useful and more compact to have 1 or 2 vector tests at the end with a zeroinitializer constant followed by 1 or 2 vector tests with an undef element in the constant. This will show that we already do support vector types; we're just lacking the last piece of the puzzle - discounting undef elements (and I'll try to fix that soon).

And as I said before, if we verify that at least one case works with vectors, then it's a good bet that all of these cases work with vectors.

[lebedev.ri]

Reduced amounts of test patterns, while improving test coverage:

• one single input type - i32
• still three sizes it's cast to - half/i16, float/i32, double/i64 - so we do have cast/upcast/downcast cases
• scalar, vector case with undef element, and added vector without undef Since the <i32 0, i32 0> is canonicalized to zeroinitializer, i did not add explicit test with zeroinitializer.

I could selectively manually prune some more testcases, if needed.

llvm/trunk/test/Transforms/InstCombine/cast-unsigned-icmp-eqcmp-0.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+; target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+; This is related to https://bugs.llvm.org/show_bug.cgi?id=36682
+
+; FIXME: *all* of these are true tests.
+; In *all* of these, uitofp and bitcast should be instcombine'd out.
+
+define i1 @i32_cast_cmp_eq_int_0_uitofp_float(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_float(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to float
+; CHECK-NEXT:    [[B:%.*]] = bitcast float [[F]] to i32
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to float
+  %b = bitcast float %f to i32
+  %cmp = icmp eq i32 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_eq_int_0_uitofp_float_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_float_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x float>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x float> [[F]] to <2 x i32>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <2 x i32> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x float>
+  %b = bitcast <2 x float> %f to <2 x i32>
+  %cmp = icmp eq <2 x i32> %b, <i32 0, i32 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_eq_int_0_uitofp_float_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_float_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x float>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x float> [[F]] to <3 x i32>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <3 x i32> [[B]], <i32 0, i32 undef, i32 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x float>
+  %b = bitcast <3 x float> %f to <3 x i32>
+  %cmp = icmp eq <3 x i32> %b, <i32 0, i32 undef, i32 0>
+  ret <3 x i1> %cmp
+}
+
+define i1 @i32_cast_cmp_ne_int_0_uitofp_float(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_float(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to float
+; CHECK-NEXT:    [[B:%.*]] = bitcast float [[F]] to i32
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to float
+  %b = bitcast float %f to i32
+  %cmp = icmp ne i32 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_ne_int_0_uitofp_float_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_float_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x float>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x float> [[F]] to <2 x i32>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <2 x i32> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x float>
+  %b = bitcast <2 x float> %f to <2 x i32>
+  %cmp = icmp ne <2 x i32> %b, <i32 0, i32 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_ne_int_0_uitofp_float_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_float_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x float>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x float> [[F]] to <3 x i32>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <3 x i32> [[B]], <i32 0, i32 undef, i32 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x float>
+  %b = bitcast <3 x float> %f to <3 x i32>
+  %cmp = icmp ne <3 x i32> %b, <i32 0, i32 undef, i32 0>
+  ret <3 x i1> %cmp
+}
+
+define i1 @i32_cast_cmp_eq_int_0_uitofp_double(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_double(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to double
+; CHECK-NEXT:    [[B:%.*]] = bitcast double [[F]] to i64
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to double
+  %b = bitcast double %f to i64
+  %cmp = icmp eq i64 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_eq_int_0_uitofp_double_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_double_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x double>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x double> [[F]] to <2 x i64>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <2 x i64> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x double>
+  %b = bitcast <2 x double> %f to <2 x i64>
+  %cmp = icmp eq <2 x i64> %b, <i64 0, i64 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_eq_int_0_uitofp_double_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_double_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x double>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x double> [[F]] to <3 x i64>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <3 x i64> [[B]], <i64 0, i64 undef, i64 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x double>
+  %b = bitcast <3 x double> %f to <3 x i64>
+  %cmp = icmp eq <3 x i64> %b, <i64 0, i64 undef, i64 0>
+  ret <3 x i1> %cmp
+}
+
+define i1 @i32_cast_cmp_ne_int_0_uitofp_double(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_double(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to double
+; CHECK-NEXT:    [[B:%.*]] = bitcast double [[F]] to i64
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i64 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to double
+  %b = bitcast double %f to i64
+  %cmp = icmp ne i64 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_ne_int_0_uitofp_double_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_double_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x double>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x double> [[F]] to <2 x i64>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <2 x i64> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x double>
+  %b = bitcast <2 x double> %f to <2 x i64>
+  %cmp = icmp ne <2 x i64> %b, <i64 0, i64 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_ne_int_0_uitofp_double_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_double_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x double>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x double> [[F]] to <3 x i64>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <3 x i64> [[B]], <i64 0, i64 undef, i64 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x double>
+  %b = bitcast <3 x double> %f to <3 x i64>
+  %cmp = icmp ne <3 x i64> %b, <i64 0, i64 undef, i64 0>
+  ret <3 x i1> %cmp
+}
+
+define i1 @i32_cast_cmp_eq_int_0_uitofp_half(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_half(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to half
+; CHECK-NEXT:    [[B:%.*]] = bitcast half [[F]] to i16
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i16 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to half
+  %b = bitcast half %f to i16
+  %cmp = icmp eq i16 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_eq_int_0_uitofp_half_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_half_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x half>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x half> [[F]] to <2 x i16>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <2 x i16> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x half>
+  %b = bitcast <2 x half> %f to <2 x i16>
+  %cmp = icmp eq <2 x i16> %b, <i16 0, i16 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_eq_int_0_uitofp_half_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_eq_int_0_uitofp_half_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x half>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x half> [[F]] to <3 x i16>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq <3 x i16> [[B]], <i16 0, i16 undef, i16 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x half>
+  %b = bitcast <3 x half> %f to <3 x i16>
+  %cmp = icmp eq <3 x i16> %b, <i16 0, i16 undef, i16 0>
+  ret <3 x i1> %cmp
+}
+
+define i1 @i32_cast_cmp_ne_int_0_uitofp_half(i32 %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_half(
+; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[I:%.*]] to half
+; CHECK-NEXT:    [[B:%.*]] = bitcast half [[F]] to i16
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i16 [[B]], 0
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %f = uitofp i32 %i to half
+  %b = bitcast half %f to i16
+  %cmp = icmp ne i16 %b, 0
+  ret i1 %cmp
+}
+
+define <2 x i1> @i32_cast_cmp_ne_int_0_uitofp_half_vec(<2 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_half_vec(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <2 x i32> [[I:%.*]] to <2 x half>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <2 x half> [[F]] to <2 x i16>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <2 x i16> [[B]], zeroinitializer
+; CHECK-NEXT:    ret <2 x i1> [[CMP]]
+;
+  %f = uitofp <2 x i32> %i to  <2 x half>
+  %b = bitcast <2 x half> %f to <2 x i16>
+  %cmp = icmp ne <2 x i16> %b, <i16 0, i16 0>
+  ret <2 x i1> %cmp
+}
+
+define <3 x i1> @i32_cast_cmp_ne_int_0_uitofp_half_vec_undef(<3 x i32> %i) {
+; CHECK-LABEL: @i32_cast_cmp_ne_int_0_uitofp_half_vec_undef(
+; CHECK-NEXT:    [[F:%.*]] = uitofp <3 x i32> [[I:%.*]] to <3 x half>
+; CHECK-NEXT:    [[B:%.*]] = bitcast <3 x half> [[F]] to <3 x i16>
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne <3 x i16> [[B]], <i16 0, i16 undef, i16 0>
+; CHECK-NEXT:    ret <3 x i1> [[CMP]]
+;
+  %f = uitofp <3 x i32> %i to <3 x half>
+  %b = bitcast <3 x half> %f to <3 x i16>
+  %cmp = icmp ne <3 x i16> %b, <i16 0, i16 undef, i16 0>
+  ret <3 x i1> %cmp
+}