[llvm] [InstCombine] Improve folding of `icmp pred (and X, Mask/~Mask), Y)` (PR #81562)

via llvm-commits llvm-commits at lists.llvm.org
Thu Feb 29 09:56:05 PST 2024


https://github.com/goldsteinn updated https://github.com/llvm/llvm-project/pull/81562

>From f24f920145f296998b16ec21b02dbf1f9c58646a Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Wed, 13 Sep 2023 13:45:49 -0500
Subject: [PATCH 1/3] [InstCombine] Add tests for expanding
 `foldICmpWithLowBitMaskedVal`; NFC

Differential Revision: https://reviews.llvm.org/D159057
---
 .../InstCombine/icmp-and-lowbit-mask.ll       | 662 ++++++++++++++++++
 1 file changed, 662 insertions(+)
 create mode 100644 llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll

diff --git a/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
new file mode 100644
index 00000000000000..89f59eac60f803
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
@@ -0,0 +1,662 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=instcombine -S | FileCheck %s
+
+declare void @use.i8(i8)
+declare void @use.i16(i16)
+define i1 @src_is_mask_zext(i16 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_zext(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[M_IN:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = zext i8 [[M_IN]] to i16
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = lshr i8 -1, %y
+  %mask = zext i8 %m_in to i16
+
+  %and = and i16 %x, %mask
+  %r = icmp eq i16 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_zext_fail_not_mask(i16 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_zext_fail_not_mask(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[M_IN:%.*]] = lshr i8 -2, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = zext i8 [[M_IN]] to i16
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = lshr i8 -2, %y
+  %mask = zext i8 %m_in to i16
+
+  %and = and i16 %x, %mask
+  %r = icmp eq i16 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_sext(i16 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_sext(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[TMP1:%.*]] = ashr i8 -32, [[Y:%.*]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = sext i8 [[TMP1]] to i16
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = lshr i8 31, %y
+  %mask = sext i8 %m_in to i16
+  %notmask = xor i16 %mask, -1
+
+  %and = and i16 %notmask, %x
+  %r = icmp eq i16 %and, 0
+  ret i1 %r
+}
+
+define i1 @src_is_mask_sext_fail_multiuse(i16 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_sext_fail_multiuse(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 122
+; CHECK-NEXT:    [[M_IN:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i8 [[M_IN]], -1
+; CHECK-NEXT:    [[NOTMASK:%.*]] = sext i8 [[TMP1]] to i16
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    call void @use.i16(i16 [[AND]])
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = lshr i8 -1, %y
+  %mask = sext i8 %m_in to i16
+  %notmask = xor i16 %mask, -1
+
+  %and = and i16 %notmask, %x
+  call void @use.i16(i16 %and)
+  %r = icmp eq i16 %and, 0
+  ret i1 %r
+}
+
+define i1 @src_is_mask_and(i8 %x_in, i8 %y, i8 %z) {
+; CHECK-LABEL: @src_is_mask_and(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[MY:%.*]] = lshr i8 7, [[Y:%.*]]
+; CHECK-NEXT:    [[MZ:%.*]] = lshr i8 -1, [[Z:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[MY]], [[MZ]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %my = ashr i8 7, %y
+  %mz = lshr i8 -1, %z
+  %mask = and i8 %my, %mz
+
+  %and = and i8 %x, %mask
+  %r = icmp eq i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_and_fail_mixed(i8 %x_in, i8 %y, i8 %z) {
+; CHECK-LABEL: @src_is_mask_and_fail_mixed(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[MY:%.*]] = ashr i8 -8, [[Y:%.*]]
+; CHECK-NEXT:    [[MZ:%.*]] = lshr i8 -1, [[Z:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[MY]], [[MZ]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %my = ashr i8 -8, %y
+  %mz = lshr i8 -1, %z
+  %mask = and i8 %my, %mz
+
+  %and = and i8 %x, %mask
+  %r = icmp eq i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_or(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_or(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[MY:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[MY]], 7
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %my = lshr i8 -1, %y
+  %mask = and i8 %my, 7
+
+  %and = and i8 %mask, %x
+  %r = icmp eq i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_xor(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_xor(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[MASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %mask = xor i8 %y, %y_m1
+  %and = and i8 %x, %mask
+  %r = icmp ne i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_xor_fail_notmask(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_xor_fail_notmask(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[TMP1:%.*]] = sub i8 0, [[Y:%.*]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = xor i8 [[TMP1]], [[Y]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %mask = xor i8 %y, %y_m1
+  %notmask = xor i8 %mask, -1
+  %and = and i8 %x, %notmask
+  %r = icmp ne i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_select(i8 %x_in, i8 %y, i1 %cond) {
+; CHECK-LABEL: @src_is_mask_select(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = select i1 %cond, i8 %ymask, i8 15
+
+  %and = and i8 %mask, %x
+  %r = icmp ne i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_select_fail_wrong_pattern(i8 %x_in, i8 %y, i1 %cond, i8 %z) {
+; CHECK-LABEL: @src_is_mask_select_fail_wrong_pattern(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[Z:%.*]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = select i1 %cond, i8 %ymask, i8 15
+
+  %and = and i8 %mask, %x
+  %r = icmp ne i8 %and, %z
+  ret i1 %r
+}
+
+define i1 @src_is_mask_shl_lshr(i8 %x_in, i8 %y, i1 %cond) {
+; CHECK-LABEL: @src_is_mask_shl_lshr(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 122
+; CHECK-NEXT:    [[TMP1:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = xor i8 [[TMP1]], -1
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %m_shl = shl i8 -1, %y
+  %mask = lshr i8 %m_shl, %y
+  %notmask = xor i8 %mask, -1
+
+  %and = and i8 %x, %notmask
+  %r = icmp ne i8 0, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_shl_lshr_fail_not_allones(i8 %x_in, i8 %y, i1 %cond) {
+; CHECK-LABEL: @src_is_mask_shl_lshr_fail_not_allones(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[TMP1:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[TMP1]], -2
+; CHECK-NEXT:    [[NOTMASK:%.*]] = xor i8 [[MASK]], -1
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %m_shl = shl i8 -2, %y
+  %mask = lshr i8 %m_shl, %y
+  %notmask = xor i8 %mask, -1
+
+  %and = and i8 %x, %notmask
+  %r = icmp ne i8 0, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_lshr(i8 %x_in, i8 %y, i8 %z, i1 %cond) {
+; CHECK-LABEL: @src_is_mask_lshr(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[SMASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
+; CHECK-NEXT:    [[MASK:%.*]] = lshr i8 [[SMASK]], [[Z:%.*]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %smask = select i1 %cond, i8 %ymask, i8 15
+  %mask = lshr i8 %smask, %z
+  %and = and i8 %mask, %x
+  %r = icmp ne i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_ashr(i8 %x_in, i8 %y, i8 %z, i1 %cond) {
+; CHECK-LABEL: @src_is_mask_ashr(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[SMASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
+; CHECK-NEXT:    [[MASK:%.*]] = ashr i8 [[SMASK]], [[Z:%.*]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %smask = select i1 %cond, i8 %ymask, i8 15
+  %mask = ashr i8 %smask, %z
+  %and = and i8 %x, %mask
+  %r = icmp ult i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_p2_m1(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_p2_m1(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[P2ORZ:%.*]] = shl i8 2, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = add i8 [[P2ORZ]], -1
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %p2orz = shl i8 2, %y
+  %mask = add i8 %p2orz, -1
+  %and = and i8 %mask, %x
+  %r = icmp ult i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_umax(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_umax(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.umax.i8(i8 [[YMASK]], i8 3)
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = call i8 @llvm.umax.i8(i8 %ymask, i8 3)
+
+  %and = and i8 %x, %mask
+  %r = icmp ugt i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_umin(i8 %x_in, i8 %y, i8 %z) {
+; CHECK-LABEL: @src_is_mask_umin(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[ZMASK:%.*]] = lshr i8 15, [[Z:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.umin.i8(i8 [[YMASK]], i8 [[ZMASK]])
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %zmask = lshr i8 15, %z
+  %mask = call i8 @llvm.umin.i8(i8 %ymask, i8 %zmask)
+
+  %and = and i8 %mask, %x
+  %r = icmp ugt i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_umin_fail_mismatch(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_umin_fail_mismatch(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.umin.i8(i8 [[YMASK]], i8 -32)
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = call i8 @llvm.umin.i8(i8 %ymask, i8 -32)
+
+  %and = and i8 %mask, %x
+  %r = icmp ugt i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_smax(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_smax(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.smax.i8(i8 [[YMASK]], i8 -1)
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = call i8 @llvm.smax.i8(i8 %ymask, i8 -1)
+
+  %and = and i8 %x, %mask
+  %r = icmp uge i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_smin(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_smin(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.smin.i8(i8 [[YMASK]], i8 0)
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %ymask = xor i8 %y, %y_m1
+  %mask = call i8 @llvm.smin.i8(i8 %ymask, i8 0)
+
+  %and = and i8 %mask, %x
+  %r = icmp uge i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_bitreverse_not_mask(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_mask_bitreverse_not_mask(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[NMASK:%.*]] = shl nsw i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[NMASK]])
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %nmask = shl i8 -1, %y
+  %mask = call i8 @llvm.bitreverse.i8(i8 %nmask)
+
+  %and = and i8 %x, %mask
+  %r = icmp ule i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_sext(i16 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_notmask_sext(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[M_IN:%.*]] = shl i8 -8, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i8 [[M_IN]], -1
+; CHECK-NEXT:    [[MASK:%.*]] = sext i8 [[TMP1]] to i16
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = shl i8 -8, %y
+  %nmask = sext i8 %m_in to i16
+  %mask = xor i16 %nmask, -1
+  %and = and i16 %mask, %x
+  %r = icmp ule i16 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_shl(i8 %x_in, i8 %y, i1 %cond) {
+; CHECK-LABEL: @src_is_notmask_shl(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 122
+; CHECK-NEXT:    [[NMASK:%.*]] = shl nsw i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i8 [[NMASK]], -1
+; CHECK-NEXT:    [[NOTMASK0:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[TMP1]])
+; CHECK-NEXT:    [[NOTMASK:%.*]] = select i1 [[COND:%.*]], i8 [[NOTMASK0]], i8 -8
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %nmask = shl i8 -1, %y
+  %mask = call i8 @llvm.bitreverse.i8(i8 %nmask)
+  %notmask0 = xor i8 %mask, -1
+  %notmask = select i1 %cond, i8 %notmask0, i8 -8
+  %and = and i8 %x, %notmask
+  %r = icmp eq i8 %and, 0
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_shl_fail_multiuse_invert(i8 %x_in, i8 %y, i1 %cond) {
+; CHECK-LABEL: @src_is_notmask_shl_fail_multiuse_invert(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 122
+; CHECK-NEXT:    [[NMASK:%.*]] = shl nsw i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i8 [[NMASK]], -1
+; CHECK-NEXT:    [[NOTMASK0:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[TMP1]])
+; CHECK-NEXT:    [[NOTMASK:%.*]] = select i1 [[COND:%.*]], i8 [[NOTMASK0]], i8 -8
+; CHECK-NEXT:    call void @use.i8(i8 [[NOTMASK]])
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %nmask = shl i8 -1, %y
+  %mask = call i8 @llvm.bitreverse.i8(i8 %nmask)
+  %notmask0 = xor i8 %mask, -1
+  %notmask = select i1 %cond, i8 %notmask0, i8 -8
+  call void @use.i8(i8 %notmask)
+  %and = and i8 %x, %notmask
+  %r = icmp eq i8 %and, 0
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_lshr_shl(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_notmask_lshr_shl(
+; CHECK-NEXT:    [[TMP1:%.*]] = shl nsw i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP2:%.*]] = xor i8 [[X_IN:%.*]], -124
+; CHECK-NEXT:    [[R:%.*]] = icmp uge i8 [[TMP2]], [[TMP1]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %mask_shr = lshr i8 -1, %y
+  %nmask = shl i8 %mask_shr, %y
+  %mask = xor i8 %nmask, -1
+  %and = and i8 %mask, %x
+  %r = icmp eq i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_lshr_shl_fail_mismatch_shifts(i8 %x_in, i8 %y, i8 %z) {
+; CHECK-LABEL: @src_is_notmask_lshr_shl_fail_mismatch_shifts(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[MASK_SHR:%.*]] = lshr i8 -1, [[Y:%.*]]
+; CHECK-NEXT:    [[NMASK:%.*]] = shl i8 [[MASK_SHR]], [[Z:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = xor i8 [[NMASK]], -1
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %mask_shr = lshr i8 -1, %y
+  %nmask = shl i8 %mask_shr, %z
+  %mask = xor i8 %nmask, -1
+  %and = and i8 %mask, %x
+  %r = icmp eq i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_ashr(i16 %x_in, i8 %y, i16 %z) {
+; CHECK-LABEL: @src_is_notmask_ashr(
+; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[M_IN:%.*]] = shl i8 -32, [[Y:%.*]]
+; CHECK-NEXT:    [[NMASK:%.*]] = sext i8 [[M_IN]] to i16
+; CHECK-NEXT:    [[NMASK_SHR:%.*]] = ashr i16 [[NMASK]], [[Z:%.*]]
+; CHECK-NEXT:    [[MASK:%.*]] = xor i16 [[NMASK_SHR]], -1
+; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i16 %x_in, 123
+  %m_in = shl i8 -32, %y
+  %nmask = sext i8 %m_in to i16
+  %nmask_shr = ashr i16 %nmask, %z
+  %mask = xor i16 %nmask_shr, -1
+  %and = and i16 %x, %mask
+  %r = icmp eq i16 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_neg_p2(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_notmask_neg_p2(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[TMP1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[TMP2:%.*]] = xor i8 [[Y]], -1
+; CHECK-NEXT:    [[TMP3:%.*]] = and i8 [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[TMP3]])
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[NOTMASK]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %ny = sub i8 0, %y
+  %p2 = and i8 %ny, %y
+  %nmask = sub i8 0, %p2
+  %mask = call i8 @llvm.bitreverse.i8(i8 %nmask)
+  %notmask = xor i8 %mask, -1
+  %and = and i8 %notmask, %x
+  %r = icmp eq i8 0, %and
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_neg_p2_fail_not_invertable(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_notmask_neg_p2_fail_not_invertable(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[NY:%.*]] = sub i8 0, [[Y:%.*]]
+; CHECK-NEXT:    [[P2:%.*]] = and i8 [[NY]], [[Y]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = sub i8 0, [[P2]]
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], 0
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %ny = sub i8 0, %y
+  %p2 = and i8 %ny, %y
+  %notmask = sub i8 0, %p2
+  %and = and i8 %notmask, %x
+  %r = icmp eq i8 0, %and
+  ret i1 %r
+}
+
+define i1 @src_is_notmask_xor_fail(i8 %x_in, i8 %y) {
+; CHECK-LABEL: @src_is_notmask_xor_fail(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[TMP1:%.*]] = sub i8 0, [[Y:%.*]]
+; CHECK-NEXT:    [[NOTMASK_REV:%.*]] = xor i8 [[TMP1]], [[Y]]
+; CHECK-NEXT:    [[NOTMASK:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[NOTMASK_REV]])
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
+; CHECK-NEXT:    [[R:%.*]] = icmp slt i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %y_m1 = add i8 %y, -1
+  %mask = xor i8 %y, %y_m1
+  %notmask_rev = xor i8 %mask, -1
+  %notmask = call i8 @llvm.bitreverse.i8(i8 %notmask_rev)
+  %and = and i8 %x, %notmask
+  %r = icmp slt i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_const_slt(i8 %x_in) {
+; CHECK-LABEL: @src_is_mask_const_slt(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], 7
+; CHECK-NEXT:    [[R:%.*]] = icmp slt i8 [[X]], [[AND]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %and = and i8 %x, 7
+  %r = icmp slt i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_const_sgt(i8 %x_in) {
+; CHECK-LABEL: @src_is_mask_const_sgt(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt i8 [[X]], 7
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %and = and i8 %x, 7
+  %r = icmp sgt i8 %x, %and
+  ret i1 %r
+}
+
+define i1 @src_is_mask_const_sle(i8 %x_in) {
+; CHECK-LABEL: @src_is_mask_const_sle(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], 31
+; CHECK-NEXT:    [[R:%.*]] = icmp sle i8 [[AND]], [[X]]
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %and = and i8 %x, 31
+  %r = icmp sle i8 %and, %x
+  ret i1 %r
+}
+
+define i1 @src_is_mask_const_sge(i8 %x_in) {
+; CHECK-LABEL: @src_is_mask_const_sge(
+; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
+; CHECK-NEXT:    [[R:%.*]] = icmp slt i8 [[X]], 32
+; CHECK-NEXT:    ret i1 [[R]]
+;
+  %x = xor i8 %x_in, 123
+  %and = and i8 %x, 31
+  %r = icmp sge i8 %and, %x
+  ret i1 %r
+}

>From c65043e51495d9ab455511363c929aaf9bf13d7e Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Wed, 13 Sep 2023 13:45:52 -0500
Subject: [PATCH 2/3] [InstCombine] Improve mask detection in
 `foldICmpWithLowBitMaskedVal`

Make recursive matcher that is able to detect a lot more patterns.
Proofs for all supported patterns: https://alive2.llvm.org/ce/z/fSQ3nZ

Differential Revision: https://reviews.llvm.org/D159058
---
 llvm/include/llvm/IR/PatternMatch.h           |  25 ++++
 .../InstCombine/InstCombineCompares.cpp       | 135 ++++++++++++++++--
 ...nt-low-bit-mask-and-icmp-eq-to-icmp-ule.ll |   3 +-
 ...nt-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll |   3 +-
 ...t-low-bit-mask-and-icmp-sge-to-icmp-sle.ll |   3 +-
 ...t-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll |   3 +-
 ...t-low-bit-mask-and-icmp-sle-to-icmp-sle.ll |   3 +-
 ...t-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll |   3 +-
 ...t-low-bit-mask-and-icmp-uge-to-icmp-ule.ll |   3 +-
 ...t-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll |   3 +-
 ...t-low-bit-mask-and-icmp-ule-to-icmp-ule.ll |   3 +-
 ...t-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll |   3 +-
 .../InstCombine/icmp-and-lowbit-mask.ll       |  53 +++----
 llvm/unittests/IR/PatternMatch.cpp            |  22 +++
 14 files changed, 195 insertions(+), 70 deletions(-)

diff --git a/llvm/include/llvm/IR/PatternMatch.h b/llvm/include/llvm/IR/PatternMatch.h
index fed552414298ad..487ae170210de9 100644
--- a/llvm/include/llvm/IR/PatternMatch.h
+++ b/llvm/include/llvm/IR/PatternMatch.h
@@ -564,6 +564,19 @@ inline api_pred_ty<is_negated_power2> m_NegatedPower2(const APInt *&V) {
   return V;
 }
 
+struct is_negated_power2_or_zero {
+  bool isValue(const APInt &C) { return !C || C.isNegatedPowerOf2(); }
+};
+/// Match a integer or vector negated power-of-2.
+/// For vectors, this includes constants with undefined elements.
+inline cst_pred_ty<is_negated_power2_or_zero> m_NegatedPower2OrZero() {
+  return cst_pred_ty<is_negated_power2_or_zero>();
+}
+inline api_pred_ty<is_negated_power2_or_zero>
+m_NegatedPower2OrZero(const APInt *&V) {
+  return V;
+}
+
 struct is_power2_or_zero {
   bool isValue(const APInt &C) { return !C || C.isPowerOf2(); }
 };
@@ -595,6 +608,18 @@ inline cst_pred_ty<is_lowbit_mask> m_LowBitMask() {
 }
 inline api_pred_ty<is_lowbit_mask> m_LowBitMask(const APInt *&V) { return V; }
 
+struct is_lowbit_mask_or_zero {
+  bool isValue(const APInt &C) { return !C || C.isMask(); }
+};
+/// Match an integer or vector with only the low bit(s) set.
+/// For vectors, this includes constants with undefined elements.
+inline cst_pred_ty<is_lowbit_mask_or_zero> m_LowBitMaskOrZero() {
+  return cst_pred_ty<is_lowbit_mask_or_zero>();
+}
+inline api_pred_ty<is_lowbit_mask_or_zero> m_LowBitMaskOrZero(const APInt *&V) {
+  return V;
+}
+
 struct icmp_pred_with_threshold {
   ICmpInst::Predicate Pred;
   const APInt *Thr;
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 49e597171b1c6f..9b133aa4d6ea66 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -4068,6 +4068,95 @@ Instruction *InstCombinerImpl::foldSelectICmp(ICmpInst::Predicate Pred,
   return nullptr;
 }
 
+// Returns whether V is a Mask ((X + 1) & X == 0) or ~Mask (-Pow2OrZero)
+static bool isMaskOrZero(const Value *V, bool Not, const SimplifyQuery &Q,
+                         unsigned Depth = 0) {
+  if (Not ? match(V, m_NegatedPower2OrZero()) : match(V, m_LowBitMaskOrZero()))
+    return true;
+  if (V->getType()->getScalarSizeInBits() == 1)
+    return true;
+  if (Depth++ >= MaxAnalysisRecursionDepth)
+    return false;
+  Value *X;
+  const Operator *I = dyn_cast<Operator>(V);
+  if (I == nullptr)
+    return false;
+  switch (I->getOpcode()) {
+  case Instruction::ZExt:
+    // ZExt(Mask) is a Mask.
+    return !Not && isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::SExt:
+    // SExt(Mask) is a Mask.
+    // SExt(~Mask) is a ~Mask.
+    return isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::And:
+  case Instruction::Or:
+    // Mask0 | Mask1 is a Mask.
+    // Mask0 & Mask1 is a Mask.
+    // ~Mask0 | ~Mask1 is a ~Mask.
+    // ~Mask0 & ~Mask1 is a ~Mask.
+    return isMaskOrZero(I->getOperand(1), Not, Q, Depth) &&
+           isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::Xor:
+    if (match(V, m_Not(m_Value(X))))
+      return isMaskOrZero(X, !Not, Q, Depth);
+
+    // (X ^ (X - 1)) is a Mask
+    return !Not &&
+           match(V, m_c_Xor(m_Value(X), m_Add(m_Deferred(X), m_AllOnes())));
+  case Instruction::Select:
+    // c ? Mask0 : Mask1 is a Mask.
+    return isMaskOrZero(I->getOperand(1), Not, Q, Depth) &&
+           isMaskOrZero(I->getOperand(2), Not, Q, Depth);
+  case Instruction::Shl:
+    // (~Mask) << X is a ~Mask.
+    return Not && isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::LShr:
+    // Mask >> X is a Mask.
+    return !Not && isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::AShr:
+    // Mask s>> X is a Mask.
+    // ~Mask s>> X is a ~Mask.
+    return isMaskOrZero(I->getOperand(0), Not, Q, Depth);
+  case Instruction::Add:
+    // Pow2 - 1 is a Mask.
+    if (!Not && match(I->getOperand(1), m_AllOnes()))
+      return isKnownToBeAPowerOfTwo(I->getOperand(0), Q.DL, /*OrZero*/ true,
+                                    Depth, Q.AC, Q.CxtI, Q.DT);
+    break;
+  case Instruction::Sub:
+    // -Pow2 is a ~Mask.
+    if (Not && match(I->getOperand(0), m_Zero()))
+      return isKnownToBeAPowerOfTwo(I->getOperand(1), Q.DL, /*OrZero*/ true,
+                                    Depth, Q.AC, Q.CxtI, Q.DT);
+    break;
+  case Instruction::Call: {
+    if (auto *II = dyn_cast<IntrinsicInst>(I)) {
+      switch (II->getIntrinsicID()) {
+        // min/max(Mask0, Mask1) is a Mask.
+        // min/max(~Mask0, ~Mask1) is a ~Mask.
+      case Intrinsic::umax:
+      case Intrinsic::smax:
+      case Intrinsic::umin:
+      case Intrinsic::smin:
+        return isMaskOrZero(II->getArgOperand(1), Not, Q, Depth) &&
+               isMaskOrZero(II->getArgOperand(0), Not, Q, Depth);
+
+        // In the context of masks, bitreverse(Mask) == ~Mask
+      case Intrinsic::bitreverse:
+        return isMaskOrZero(II->getArgOperand(0), !Not, Q, Depth);
+      default:
+        break;
+      }
+    }
+    break;
+  }
+  default:
+    break;
+  }
+  return false;
+}
+
 /// Some comparisons can be simplified.
 /// In this case, we are looking for comparisons that look like
 /// a check for a lossy truncation.
@@ -4081,21 +4170,35 @@ Instruction *InstCombinerImpl::foldSelectICmp(ICmpInst::Predicate Pred,
 /// The Mask can be a constant, too.
 /// For some predicates, the operands are commutative.
 /// For others, x can only be on a specific side.
-static Value *foldICmpWithLowBitMaskedVal(ICmpInst &I,
-                                          InstCombiner::BuilderTy &Builder) {
+static Value *foldICmpWithLowBitMaskedVal(ICmpInst &I, const SimplifyQuery &Q,
+                                          InstCombiner &IC) {
+
+  Value *X, *M;
+  ICmpInst::Predicate Pred = I.getPredicate();
   ICmpInst::Predicate SrcPred;
-  Value *X, *M, *Y;
-  auto m_VariableMask = m_CombineOr(
-      m_CombineOr(m_Not(m_Shl(m_AllOnes(), m_Value())),
-                  m_Add(m_Shl(m_One(), m_Value()), m_AllOnes())),
-      m_CombineOr(m_LShr(m_AllOnes(), m_Value()),
-                  m_LShr(m_Shl(m_AllOnes(), m_Value(Y)), m_Deferred(Y))));
-  auto m_Mask = m_CombineOr(m_VariableMask, m_LowBitMask());
-  if (!match(&I, m_c_ICmp(SrcPred,
-                          m_c_And(m_CombineAnd(m_Mask, m_Value(M)), m_Value(X)),
-                          m_Deferred(X))))
-    return nullptr;
+  bool NeedsNot = false;
 
+  auto CheckMask = [&](Value *V, bool Not) {
+    if (!ICmpInst::isSigned(Pred))
+      return isMaskOrZero(V, Not, Q);
+    return Not ? match(V, m_NegatedPower2OrZero())
+               : match(V, m_LowBitMaskOrZero());
+  };
+
+  auto TryMatch = [&](unsigned OpNo) {
+    SrcPred = Pred;
+    if (match(I.getOperand(OpNo),
+              m_c_And(m_Specific(I.getOperand(1 - OpNo)), m_Value(M)))) {
+      X = I.getOperand(1 - OpNo);
+      if (OpNo)
+        SrcPred = ICmpInst::getSwappedPredicate(Pred);
+      return CheckMask(M, /*Not*/ false);
+    }
+    return false;
+  };
+
+  if (!TryMatch(0) && !TryMatch(1))
+    return nullptr;
   ICmpInst::Predicate DstPred;
   switch (SrcPred) {
   case ICmpInst::Predicate::ICMP_EQ:
@@ -4163,7 +4266,9 @@ static Value *foldICmpWithLowBitMaskedVal(ICmpInst &I,
     M = Constant::replaceUndefsWith(VecC, SafeReplacementConstant);
   }
 
-  return Builder.CreateICmp(DstPred, X, M);
+  if (NeedsNot)
+    M = IC.Builder.CreateNot(M);
+  return IC.Builder.CreateICmp(DstPred, X, M);
 }
 
 /// Some comparisons can be simplified.
@@ -5080,7 +5185,7 @@ Instruction *InstCombinerImpl::foldICmpBinOp(ICmpInst &I,
   if (Value *V = foldMultiplicationOverflowCheck(I))
     return replaceInstUsesWith(I, V);
 
-  if (Value *V = foldICmpWithLowBitMaskedVal(I, Builder))
+  if (Value *V = foldICmpWithLowBitMaskedVal(I, Q, *this))
     return replaceInstUsesWith(I, V);
 
   if (Instruction *R = foldICmpAndXX(I, Q, *this))
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-eq-to-icmp-ule.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-eq-to-icmp-ule.ll
index a957fb2d088ef4..5b7a99d53c308c 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-eq-to-icmp-ule.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-eq-to-icmp-ule.ll
@@ -62,8 +62,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase0(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X:%.*]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp eq <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ult <2 x i8> [[X:%.*]], <i8 4, i8 1>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
index 57361cdf38977c..160d968b9ac4c7 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ne-to-icmp-ugt.ll
@@ -62,8 +62,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase0(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X:%.*]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp ne <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ugt <2 x i8> [[X:%.*]], <i8 3, i8 0>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sge-to-icmp-sle.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sge-to-icmp-sle.ll
index 0dfc9f51baf9c2..60921042d52435 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sge-to-icmp-sle.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sge-to-icmp-sle.ll
@@ -50,8 +50,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase(
-; CHECK-NEXT:    [[TMP0:%.*]] = and <2 x i8> [[X:%.*]], <i8 3, i8 0>
-; CHECK-NEXT:    [[RET:%.*]] = icmp sge <2 x i8> [[TMP0]], [[X]]
+; CHECK-NEXT:    [[RET:%.*]] = icmp slt <2 x i8> [[X:%.*]], <i8 4, i8 1>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
index e0893ce4cf2ecb..6345e70d7220e2 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sgt-to-icmp-sgt.ll
@@ -63,8 +63,7 @@ define <2 x i1> @p2_vec_nonsplat() {
 define <2 x i1> @p2_vec_nonsplat_edgecase() {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase(
 ; CHECK-NEXT:    [[X:%.*]] = call <2 x i8> @gen2x8()
-; CHECK-NEXT:    [[TMP0:%.*]] = and <2 x i8> [[X]], <i8 3, i8 0>
-; CHECK-NEXT:    [[RET:%.*]] = icmp sgt <2 x i8> [[X]], [[TMP0]]
+; CHECK-NEXT:    [[RET:%.*]] = icmp sgt <2 x i8> [[X]], <i8 3, i8 0>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %x = call <2 x i8> @gen2x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sle-to-icmp-sle.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sle-to-icmp-sle.ll
index 81887a39091573..b7aec53fed6760 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sle-to-icmp-sle.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-sle-to-icmp-sle.ll
@@ -63,8 +63,7 @@ define <2 x i1> @p2_vec_nonsplat() {
 define <2 x i1> @p2_vec_nonsplat_edgecase() {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase(
 ; CHECK-NEXT:    [[X:%.*]] = call <2 x i8> @gen2x8()
-; CHECK-NEXT:    [[TMP0:%.*]] = and <2 x i8> [[X]], <i8 3, i8 0>
-; CHECK-NEXT:    [[RET:%.*]] = icmp sle <2 x i8> [[X]], [[TMP0]]
+; CHECK-NEXT:    [[RET:%.*]] = icmp slt <2 x i8> [[X]], <i8 4, i8 1>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %x = call <2 x i8> @gen2x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
index 8ce8687f198446..56661d335c4f60 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-slt-to-icmp-sgt.ll
@@ -50,8 +50,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase(
-; CHECK-NEXT:    [[TMP0:%.*]] = and <2 x i8> [[X:%.*]], <i8 3, i8 0>
-; CHECK-NEXT:    [[RET:%.*]] = icmp slt <2 x i8> [[TMP0]], [[X]]
+; CHECK-NEXT:    [[RET:%.*]] = icmp sgt <2 x i8> [[X:%.*]], <i8 3, i8 0>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-uge-to-icmp-ule.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-uge-to-icmp-ule.ll
index ff09e255185b5a..a93e8f779435fc 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-uge-to-icmp-ule.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-uge-to-icmp-ule.ll
@@ -62,8 +62,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase0(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X:%.*]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp eq <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ult <2 x i8> [[X:%.*]], <i8 4, i8 1>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
index 4ad04710fd7bb9..73ea4d456d2462 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ugt-to-icmp-ugt.ll
@@ -75,8 +75,7 @@ define <2 x i1> @p2_vec_nonsplat() {
 define <2 x i1> @p2_vec_nonsplat_edgecase0() {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
 ; CHECK-NEXT:    [[X:%.*]] = call <2 x i8> @gen2x8()
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp ne <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ugt <2 x i8> [[X]], <i8 3, i8 0>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %x = call <2 x i8> @gen2x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ule-to-icmp-ule.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ule-to-icmp-ule.ll
index 8e513dcbf4ef3a..53886b5f2dc9c3 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ule-to-icmp-ule.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ule-to-icmp-ule.ll
@@ -75,8 +75,7 @@ define <2 x i1> @p2_vec_nonsplat() {
 define <2 x i1> @p2_vec_nonsplat_edgecase0() {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
 ; CHECK-NEXT:    [[X:%.*]] = call <2 x i8> @gen2x8()
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp eq <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ult <2 x i8> [[X]], <i8 4, i8 1>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %x = call <2 x i8> @gen2x8()
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
index d02ecf6965e878..d66be571008c2f 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-constant-low-bit-mask-and-icmp-ult-to-icmp-ugt.ll
@@ -62,8 +62,7 @@ define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
 
 define <2 x i1> @p2_vec_nonsplat_edgecase0(<2 x i8> %x) {
 ; CHECK-LABEL: @p2_vec_nonsplat_edgecase0(
-; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X:%.*]], <i8 -4, i8 -1>
-; CHECK-NEXT:    [[RET:%.*]] = icmp ne <2 x i8> [[TMP1]], zeroinitializer
+; CHECK-NEXT:    [[RET:%.*]] = icmp ugt <2 x i8> [[X:%.*]], <i8 3, i8 0>
 ; CHECK-NEXT:    ret <2 x i1> [[RET]]
 ;
   %tmp0 = and <2 x i8> %x, <i8 3, i8 0>
diff --git a/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
index 89f59eac60f803..4a8339fd5e1bc0 100644
--- a/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
@@ -8,8 +8,7 @@ define i1 @src_is_mask_zext(i16 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[M_IN:%.*]] = lshr i8 -1, [[Y:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = zext i8 [[M_IN]] to i16
-; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i16 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i16 %x_in, 123
@@ -86,8 +85,7 @@ define i1 @src_is_mask_and(i8 %x_in, i8 %y, i8 %z) {
 ; CHECK-NEXT:    [[MY:%.*]] = lshr i8 7, [[Y:%.*]]
 ; CHECK-NEXT:    [[MZ:%.*]] = lshr i8 -1, [[Z:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[MY]], [[MZ]]
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[X]], [[AND]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -125,8 +123,7 @@ define i1 @src_is_mask_or(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[MY:%.*]] = lshr i8 -1, [[Y:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[MY]], 7
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[X]], [[AND]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -143,8 +140,7 @@ define i1 @src_is_mask_xor(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
 ; CHECK-NEXT:    [[MASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -179,8 +175,7 @@ define i1 @src_is_mask_select(i8 %x_in, i8 %y, i1 %cond) {
 ; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[MASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -259,8 +254,7 @@ define i1 @src_is_mask_lshr(i8 %x_in, i8 %y, i8 %z, i1 %cond) {
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[SMASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
 ; CHECK-NEXT:    [[MASK:%.*]] = lshr i8 [[SMASK]], [[Z:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[X]], [[AND]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -280,8 +274,7 @@ define i1 @src_is_mask_ashr(i8 %x_in, i8 %y, i8 %z, i1 %cond) {
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[SMASK:%.*]] = select i1 [[COND:%.*]], i8 [[YMASK]], i8 15
 ; CHECK-NEXT:    [[MASK:%.*]] = ashr i8 [[SMASK]], [[Z:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -299,8 +292,7 @@ define i1 @src_is_mask_p2_m1(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[P2ORZ:%.*]] = shl i8 2, [[Y:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = add i8 [[P2ORZ]], -1
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -317,8 +309,7 @@ define i1 @src_is_mask_umax(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.umax.i8(i8 [[YMASK]], i8 3)
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -338,8 +329,7 @@ define i1 @src_is_mask_umin(i8 %x_in, i8 %y, i8 %z) {
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[ZMASK:%.*]] = lshr i8 15, [[Z:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.umin.i8(i8 [[YMASK]], i8 [[ZMASK]])
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -379,8 +369,7 @@ define i1 @src_is_mask_smax(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.smax.i8(i8 [[YMASK]], i8 -1)
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -399,8 +388,7 @@ define i1 @src_is_mask_smin(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[Y_M1:%.*]] = add i8 [[Y:%.*]], -1
 ; CHECK-NEXT:    [[YMASK:%.*]] = xor i8 [[Y_M1]], [[Y]]
 ; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.smin.i8(i8 [[YMASK]], i8 0)
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[MASK]], [[X]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -418,8 +406,7 @@ define i1 @src_is_mask_bitreverse_not_mask(i8 %x_in, i8 %y) {
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[NMASK:%.*]] = shl nsw i8 -1, [[Y:%.*]]
 ; CHECK-NEXT:    [[MASK:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[NMASK]])
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], [[X]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[MASK]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -433,12 +420,10 @@ define i1 @src_is_mask_bitreverse_not_mask(i8 %x_in, i8 %y) {
 
 define i1 @src_is_notmask_sext(i16 %x_in, i8 %y) {
 ; CHECK-LABEL: @src_is_notmask_sext(
-; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[M_IN:%.*]] = shl i8 -8, [[Y:%.*]]
-; CHECK-NEXT:    [[TMP1:%.*]] = xor i8 [[M_IN]], -1
-; CHECK-NEXT:    [[MASK:%.*]] = sext i8 [[TMP1]] to i16
-; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], [[X]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i16 [[X_IN:%.*]], -124
+; CHECK-NEXT:    [[TMP2:%.*]] = sext i8 [[M_IN]] to i16
+; CHECK-NEXT:    [[R:%.*]] = icmp uge i16 [[TMP1]], [[TMP2]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i16 %x_in, 123
@@ -531,13 +516,11 @@ define i1 @src_is_notmask_lshr_shl_fail_mismatch_shifts(i8 %x_in, i8 %y, i8 %z)
 
 define i1 @src_is_notmask_ashr(i16 %x_in, i8 %y, i16 %z) {
 ; CHECK-LABEL: @src_is_notmask_ashr(
-; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
 ; CHECK-NEXT:    [[M_IN:%.*]] = shl i8 -32, [[Y:%.*]]
 ; CHECK-NEXT:    [[NMASK:%.*]] = sext i8 [[M_IN]] to i16
 ; CHECK-NEXT:    [[NMASK_SHR:%.*]] = ashr i16 [[NMASK]], [[Z:%.*]]
-; CHECK-NEXT:    [[MASK:%.*]] = xor i16 [[NMASK_SHR]], -1
-; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[MASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[X]], [[AND]]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor i16 [[X_IN:%.*]], -124
+; CHECK-NEXT:    [[R:%.*]] = icmp uge i16 [[TMP1]], [[NMASK_SHR]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i16 %x_in, 123
diff --git a/llvm/unittests/IR/PatternMatch.cpp b/llvm/unittests/IR/PatternMatch.cpp
index 883149c686b42a..533a30bfba45dd 100644
--- a/llvm/unittests/IR/PatternMatch.cpp
+++ b/llvm/unittests/IR/PatternMatch.cpp
@@ -579,17 +579,39 @@ TEST_F(PatternMatchTest, Power2) {
   EXPECT_TRUE(m_Power2().match(C128));
   EXPECT_FALSE(m_Power2().match(CNeg128));
 
+  EXPECT_TRUE(m_Power2OrZero().match(C128));
+  EXPECT_FALSE(m_Power2OrZero().match(CNeg128));
+
   EXPECT_FALSE(m_NegatedPower2().match(C128));
   EXPECT_TRUE(m_NegatedPower2().match(CNeg128));
 
+  EXPECT_FALSE(m_NegatedPower2OrZero().match(C128));
+  EXPECT_TRUE(m_NegatedPower2OrZero().match(CNeg128));
+
   Value *CIntMin = IRB.getInt64(APSInt::getSignedMinValue(64).getSExtValue());
   Value *CNegIntMin = ConstantExpr::getNeg(cast<Constant>(CIntMin));
 
   EXPECT_TRUE(m_Power2().match(CIntMin));
   EXPECT_TRUE(m_Power2().match(CNegIntMin));
 
+  EXPECT_TRUE(m_Power2OrZero().match(CIntMin));
+  EXPECT_TRUE(m_Power2OrZero().match(CNegIntMin));
+
   EXPECT_TRUE(m_NegatedPower2().match(CIntMin));
   EXPECT_TRUE(m_NegatedPower2().match(CNegIntMin));
+
+  EXPECT_TRUE(m_NegatedPower2OrZero().match(CIntMin));
+  EXPECT_TRUE(m_NegatedPower2OrZero().match(CNegIntMin));
+
+  Value *CZero = IRB.getInt64(0);
+
+  EXPECT_FALSE(m_Power2().match(CZero));
+
+  EXPECT_TRUE(m_Power2OrZero().match(CZero));
+
+  EXPECT_FALSE(m_NegatedPower2().match(CZero));
+
+  EXPECT_TRUE(m_NegatedPower2OrZero().match(CZero));
 }
 
 TEST_F(PatternMatchTest, Not) {

>From 194152ed20dde5864305caa6be7a049aff0d3b61 Mon Sep 17 00:00:00 2001
From: Noah Goldstein <goldstein.w.n at gmail.com>
Date: Wed, 13 Sep 2023 13:45:55 -0500
Subject: [PATCH 3/3] [InstCombine] Recognize `(icmp eq/ne (and X, ~Mask), 0)`
 pattern in `foldICmpWithLowBitMaskedVal`

`(icmp eq/ne (and X, ~Mask), 0)` is equivilent to `(icmp eq/ne (and X,
Mask), X` and we sometimes generate the former pattern intentionally
to reduce number of uses of `X`.
Proof: https://alive2.llvm.org/ce/z/3u-usC

Differential Revision: https://reviews.llvm.org/D159329
---
 .../InstCombine/InstCombineCompares.cpp       | 11 ++++++++++
 .../InstCombine/icmp-and-lowbit-mask.ll       | 20 ++++++++-----------
 .../InstCombine/lshr-and-negC-icmpeq-zero.ll  |  9 +++------
 .../lshr-and-signbit-icmpeq-zero.ll           |  9 +++------
 .../InstCombine/shl-and-negC-icmpeq-zero.ll   |  9 +++------
 .../shl-and-signbit-icmpeq-zero.ll            |  9 +++------
 6 files changed, 31 insertions(+), 36 deletions(-)

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 9b133aa4d6ea66..363f0ec57496b6 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -4162,6 +4162,7 @@ static bool isMaskOrZero(const Value *V, bool Not, const SimplifyQuery &Q,
 /// a check for a lossy truncation.
 /// Folds:
 ///   icmp SrcPred (x & Mask), x    to    icmp DstPred x, Mask
+///   icmp eq/ne (x & ~Mask), 0     to    icmp DstPred x, Mask
 /// Where Mask is some pattern that produces all-ones in low bits:
 ///    (-1 >> y)
 ///    ((-1 << y) >> y)     <- non-canonical, has extra uses
@@ -4194,6 +4195,16 @@ static Value *foldICmpWithLowBitMaskedVal(ICmpInst &I, const SimplifyQuery &Q,
         SrcPred = ICmpInst::getSwappedPredicate(Pred);
       return CheckMask(M, /*Not*/ false);
     }
+    if (OpNo == 1 && match(I.getOperand(1), m_Zero()) &&
+        ICmpInst::isEquality(Pred) &&
+        match(I.getOperand(0), m_OneUse(m_And(m_Value(X), m_Value(M))))) {
+      NeedsNot = true;
+      if (IC.isFreeToInvert(X, X->hasOneUse()) && CheckMask(X, /*Not*/ true)) {
+        std::swap(X, M);
+        return true;
+      }
+      return IC.isFreeToInvert(M, M->hasOneUse()) && CheckMask(M, /*Not*/ true);
+    }
     return false;
   };
 
diff --git a/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
index 4a8339fd5e1bc0..640a95b0561602 100644
--- a/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-and-lowbit-mask.ll
@@ -41,10 +41,9 @@ define i1 @src_is_mask_zext_fail_not_mask(i16 %x_in, i8 %y) {
 define i1 @src_is_mask_sext(i16 %x_in, i8 %y) {
 ; CHECK-LABEL: @src_is_mask_sext(
 ; CHECK-NEXT:    [[X:%.*]] = xor i16 [[X_IN:%.*]], 123
-; CHECK-NEXT:    [[TMP1:%.*]] = ashr i8 -32, [[Y:%.*]]
-; CHECK-NEXT:    [[NOTMASK:%.*]] = sext i8 [[TMP1]] to i16
-; CHECK-NEXT:    [[AND:%.*]] = and i16 [[X]], [[NOTMASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i16 [[AND]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = lshr i8 31, [[Y:%.*]]
+; CHECK-NEXT:    [[TMP2:%.*]] = zext nneg i8 [[TMP1]] to i16
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i16 [[X]], [[TMP2]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i16 %x_in, 123
@@ -212,9 +211,7 @@ define i1 @src_is_mask_shl_lshr(i8 %x_in, i8 %y, i1 %cond) {
 ; CHECK-LABEL: @src_is_mask_shl_lshr(
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 122
 ; CHECK-NEXT:    [[TMP1:%.*]] = lshr i8 -1, [[Y:%.*]]
-; CHECK-NEXT:    [[NOTMASK:%.*]] = xor i8 [[TMP1]], -1
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp ne i8 [[AND]], 0
+; CHECK-NEXT:    [[R:%.*]] = icmp ugt i8 [[X]], [[TMP1]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
@@ -558,11 +555,10 @@ define i1 @src_is_notmask_neg_p2(i8 %x_in, i8 %y) {
 define i1 @src_is_notmask_neg_p2_fail_not_invertable(i8 %x_in, i8 %y) {
 ; CHECK-LABEL: @src_is_notmask_neg_p2_fail_not_invertable(
 ; CHECK-NEXT:    [[X:%.*]] = xor i8 [[X_IN:%.*]], 123
-; CHECK-NEXT:    [[NY:%.*]] = sub i8 0, [[Y:%.*]]
-; CHECK-NEXT:    [[P2:%.*]] = and i8 [[NY]], [[Y]]
-; CHECK-NEXT:    [[NOTMASK:%.*]] = sub i8 0, [[P2]]
-; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[NOTMASK]]
-; CHECK-NEXT:    [[R:%.*]] = icmp eq i8 [[AND]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add i8 [[Y:%.*]], -1
+; CHECK-NEXT:    [[TMP2:%.*]] = xor i8 [[Y]], -1
+; CHECK-NEXT:    [[TMP3:%.*]] = and i8 [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[R:%.*]] = icmp ule i8 [[X]], [[TMP3]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %x = xor i8 %x_in, 123
diff --git a/llvm/test/Transforms/InstCombine/lshr-and-negC-icmpeq-zero.ll b/llvm/test/Transforms/InstCombine/lshr-and-negC-icmpeq-zero.ll
index 79aef3a5406cfb..847a7940bad8c7 100644
--- a/llvm/test/Transforms/InstCombine/lshr-and-negC-icmpeq-zero.ll
+++ b/llvm/test/Transforms/InstCombine/lshr-and-negC-icmpeq-zero.ll
@@ -84,8 +84,7 @@ define <4 x i1> @vec_4xi32_lshr_and_negC_eq(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_lshr_and_negC_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_lshr_and_negC_eq_undef1(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -8, i32 undef, i32 -8, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], zeroinitializer
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[LSHR]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
@@ -97,8 +96,7 @@ define <4 x i1> @vec_lshr_and_negC_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_lshr_and_negC_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_lshr_and_negC_eq_undef2(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -8, i32 -8, i32 -8, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[LSHR]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
@@ -110,8 +108,7 @@ define <4 x i1> @vec_lshr_and_negC_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_lshr_and_negC_eq_undef3(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_lshr_and_negC_eq_undef3(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -8, i32 -8, i32 undef, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[LSHR]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
diff --git a/llvm/test/Transforms/InstCombine/lshr-and-signbit-icmpeq-zero.ll b/llvm/test/Transforms/InstCombine/lshr-and-signbit-icmpeq-zero.ll
index 5335a4736896d7..39f4e58b25dc84 100644
--- a/llvm/test/Transforms/InstCombine/lshr-and-signbit-icmpeq-zero.ll
+++ b/llvm/test/Transforms/InstCombine/lshr-and-signbit-icmpeq-zero.ll
@@ -84,8 +84,7 @@ define <4 x i1> @vec_4xi32_lshr_and_signbit_eq(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_4xi32_lshr_and_signbit_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_lshr_and_signbit_eq_undef1(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -2147483648, i32 undef, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], zeroinitializer
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[LSHR]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
@@ -97,8 +96,7 @@ define <4 x i1> @vec_4xi32_lshr_and_signbit_eq_undef1(<4 x i32> %x, <4 x i32> %y
 define <4 x i1> @vec_4xi32_lshr_and_signbit_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_lshr_and_signbit_eq_undef2(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 undef, i32 0, i32 0, i32 0>
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[LSHR]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
@@ -110,8 +108,7 @@ define <4 x i1> @vec_4xi32_lshr_and_signbit_eq_undef2(<4 x i32> %x, <4 x i32> %y
 define <4 x i1> @vec_4xi32_lshr_and_signbit_eq_undef3(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_lshr_and_signbit_eq_undef3(
 ; CHECK-NEXT:    [[LSHR:%.*]] = lshr <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[LSHR]], <i32 -2147483648, i32 undef, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[LSHR]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %lshr = lshr <4 x i32> %x, %y
diff --git a/llvm/test/Transforms/InstCombine/shl-and-negC-icmpeq-zero.ll b/llvm/test/Transforms/InstCombine/shl-and-negC-icmpeq-zero.ll
index d8e7fe2e2a2c16..406dc72f2646e5 100644
--- a/llvm/test/Transforms/InstCombine/shl-and-negC-icmpeq-zero.ll
+++ b/llvm/test/Transforms/InstCombine/shl-and-negC-icmpeq-zero.ll
@@ -84,8 +84,7 @@ define <4 x i1> @vec_4xi32_shl_and_negC_eq(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_shl_and_negC_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_shl_and_negC_eq_undef1(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -8, i32 undef, i32 -8, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], zeroinitializer
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[SHL]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y
@@ -97,8 +96,7 @@ define <4 x i1> @vec_shl_and_negC_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_shl_and_negC_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_shl_and_negC_eq_undef2(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -8, i32 -8, i32 -8, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[SHL]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y
@@ -110,8 +108,7 @@ define <4 x i1> @vec_shl_and_negC_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_shl_and_negC_eq_undef3(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_shl_and_negC_eq_undef3(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -8, i32 -8, i32 undef, i32 -8>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp ult <4 x i32> [[SHL]], <i32 8, i32 8, i32 8, i32 8>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y
diff --git a/llvm/test/Transforms/InstCombine/shl-and-signbit-icmpeq-zero.ll b/llvm/test/Transforms/InstCombine/shl-and-signbit-icmpeq-zero.ll
index 42b755f51a9713..4c2c876e3925bf 100644
--- a/llvm/test/Transforms/InstCombine/shl-and-signbit-icmpeq-zero.ll
+++ b/llvm/test/Transforms/InstCombine/shl-and-signbit-icmpeq-zero.ll
@@ -84,8 +84,7 @@ define <4 x i1> @vec_4xi32_shl_and_signbit_eq(<4 x i32> %x, <4 x i32> %y) {
 define <4 x i1> @vec_4xi32_shl_and_signbit_eq_undef1(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_shl_and_signbit_eq_undef1(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -2147483648, i32 undef, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], zeroinitializer
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[SHL]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y
@@ -97,8 +96,7 @@ define <4 x i1> @vec_4xi32_shl_and_signbit_eq_undef1(<4 x i32> %x, <4 x i32> %y)
 define <4 x i1> @vec_4xi32_shl_and_signbit_eq_undef2(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_shl_and_signbit_eq_undef2(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 undef, i32 0, i32 0, i32 0>
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[SHL]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y
@@ -110,8 +108,7 @@ define <4 x i1> @vec_4xi32_shl_and_signbit_eq_undef2(<4 x i32> %x, <4 x i32> %y)
 define <4 x i1> @vec_4xi32_shl_and_signbit_eq_undef3(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @vec_4xi32_shl_and_signbit_eq_undef3(
 ; CHECK-NEXT:    [[SHL:%.*]] = shl <4 x i32> [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT:    [[AND:%.*]] = and <4 x i32> [[SHL]], <i32 -2147483648, i32 undef, i32 -2147483648, i32 -2147483648>
-; CHECK-NEXT:    [[R:%.*]] = icmp eq <4 x i32> [[AND]], <i32 0, i32 0, i32 0, i32 undef>
+; CHECK-NEXT:    [[R:%.*]] = icmp sgt <4 x i32> [[SHL]], <i32 -1, i32 -1, i32 -1, i32 -1>
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %shl = shl <4 x i32> %x, %y



More information about the llvm-commits mailing list