[SimplifyLibCalls] Fix memchr expansion for constant strings.

[efriedma]

Summary

The C standard says "The memchr function locates the first occurrence of c (converted to an unsigned char)[...]". The expansion was missing the conversion to unsigned char.

Fixes https://bugs.llvm.org/show_bug.cgi?id=39041 .

[spatel]

Wouldn't it be more efficient to create 'and X, 255' directly?

[efriedma]

Emit "and" explicitly.

[spatel]

LGTM

lib/Transforms/Utils/SimplifyLibCalls.cpp

if (!CharC && !Str.empty() && isOnlyUsedInZeroEqualityComparison(CI)) {
     unsigned char Width = NextPowerOf2(std::max((unsigned char)7, Max));
 
     // Now build the bit field.
     APInt Bitfield(Width, 0);
     for (char C : Str)
       Bitfield.setBit((unsigned char)C);
     Value *BitfieldC = B.getInt(Bitfield);
 
-    // First check that the bit field access is within bounds.
+    // Adjust width of "C" to the bitfield width, then mask off the high bits.
     Value *C = B.CreateZExtOrTrunc(CI->getArgOperand(1), BitfieldC->getType());
+    C = B.CreateAnd(C, B.getIntN(Width, 0xFF));
+
+    // First check that the bit field access is within bounds.
     Value *Bounds = B.CreateICmp(ICmpInst::ICMP_ULT, C, B.getIntN(Width, Width),
                                  "memchr.bounds");
 
     // Create code that checks if the given bit is set in the field.
     Value *Shl = B.CreateShl(B.getIntN(Width, 1ULL), C);
     Value *Bits = B.CreateIsNotNull(B.CreateAnd(Shl, BitfieldC), "memchr.bits");
 
     // Finally merge both checks and cast to pointer type. The inttoptr

test/Transforms/InstCombine/memchr.ll

;
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @memchr(i8* %src, i32 0, i32 14)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test4(i32 %chr) {
 ; CHECK-LABEL: @test4(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 %chr, i32 14)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 [[CHR:%.*]], i32 14)
 ; CHECK-NEXT:    store i8* [[DST]], i8** @chp, align 4
 ; CHECK-NEXT:    ret void
 ;
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @memchr(i8* %src, i32 %chr, i32 14)
   store i8* %dst, i8** @chp
   ret void
 }
;
   %dst = call i8* @memchr(i8* %str, i32 119, i32 6)
   store i8* %dst, i8** @chp
   ret void
 }
 
 ; Check transformation memchr("\r\n", C, 2) != nullptr -> (C & 9216) != 0
 define i1 @test11(i32 %C) {
 ; CHECK-LABEL: @test11(
-; CHECK-NEXT:    [[TMP1:%.*]] = trunc i32 %C to i16
-; CHECK-NEXT:    [[MEMCHR_BOUNDS:%.*]] = icmp ult i16 [[TMP1]], 16
-; CHECK-NEXT:    [[TMP2:%.*]] = shl i16 1, [[TMP1]]
-; CHECK-NEXT:    [[TMP3:%.*]] = and i16 [[TMP2]], 9216
-; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp ne i16 [[TMP3]], 0
-; CHECK-NEXT:    [[MEMCHR:%.*]] = and i1 [[MEMCHR:%.*]].bounds, [[MEMCHR:%.*]].bits
+; CHECK-NEXT:    [[TMP1:%.*]] = trunc i32 [[C:%.*]] to i16
+; CHECK-NEXT:    [[TMP2:%.*]] = and i16 [[TMP1]], 255
+; CHECK-NEXT:    [[MEMCHR_BOUNDS:%.*]] = icmp ult i16 [[TMP2]], 16
+; CHECK-NEXT:    [[TMP3:%.*]] = shl i16 1, [[TMP2]]
+; CHECK-NEXT:    [[TMP4:%.*]] = and i16 [[TMP3]], 9216
+; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp ne i16 [[TMP4]], 0
+; CHECK-NEXT:    [[MEMCHR:%.*]] = and i1 [[MEMCHR_BOUNDS]], [[MEMCHR_BITS]]
 ; CHECK-NEXT:    ret i1 [[MEMCHR]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @newlines, i64 0, i64 0), i32 %C, i32 2)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 ; No 64 bits here
 define i1 @test12(i32 %C) {
 ; CHECK-LABEL: @test12(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i32 0, i32 0), i32 %C, i32 3)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i32 0, i32 0), i32 [[C:%.*]], i32 3)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i8* [[DST]], null
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i64 0, i64 0), i32 %C, i32 3)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 define i1 @test13(i32 %C) {
 ; CHECK-LABEL: @test13(
-; CHECK-NEXT:    [[MEMCHR_BOUNDS:%.*]] = icmp ult i32 %C, 32
-; CHECK-NEXT:    [[TMP1:%.*]] = shl i32 1, %C
-; CHECK-NEXT:    [[TMP2:%.*]] = and i32 [[TMP1]], -2147483647
-; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp ne i32 [[TMP2]], 0
-; CHECK-NEXT:    [[MEMCHR:%.*]] = and i1 [[MEMCHR:%.*]].bounds, [[MEMCHR:%.*]].bits
+; CHECK-NEXT:    [[TMP1:%.*]] = and i32 [[C:%.*]], 255
+; CHECK-NEXT:    [[MEMCHR_BOUNDS:%.*]] = icmp ult i32 [[TMP1]], 32
+; CHECK-NEXT:    [[TMP2:%.*]] = shl i32 1, [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = and i32 [[TMP2]], -2147483647
+; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK-NEXT:    [[MEMCHR:%.*]] = and i1 [[MEMCHR_BOUNDS]], [[MEMCHR_BITS]]
 ; CHECK-NEXT:    ret i1 [[MEMCHR]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @single, i64 0, i64 0), i32 %C, i32 2)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 define i1 @test14(i32 %C) {
 ; CHECK-LABEL: @test14(
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq i32 %C, 31
-; CHECK-NEXT:    ret i1 [[TMP1]]
+; CHECK-NEXT:    [[TMP1:%.*]] = and i32 [[C:%.*]], 255
+; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp eq i32 [[TMP1]], 31
+; CHECK-NEXT:    ret i1 [[MEMCHR_BITS]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @single, i64 0, i64 0), i32 %C, i32 1)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 define i1 @test15(i32 %C) {
 ; CHECK-LABEL: @test15(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i32 0, i32 0), i32 %C, i32 3)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i32 0, i32 0), i32 [[C:%.*]], i32 3)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i8* [[DST]], null
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i64 0, i64 0), i32 %C, i32 3)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 @s = internal constant [1 x i8] [i8 0], align 1
 define i8* @pr32124() {
 ; CHECK-LABEL: @pr32124(
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([1 x i8], [1 x i8]* @s, i32 0, i32 0)
 ;
   %res = tail call i8* @memchr(i8* getelementptr ([1 x i8], [1 x i8]* @s, i64 0, i64 0), i32 0, i32 1)
   ret i8* %res
 }

test/Transforms/InstCombine/strchr-1.ll

; CHECK: ret void
   store i8* %dst, i8** @chp
   ret void
 }
 
 ; Check transformation strchr("\r\n", C) != nullptr -> (C & 9217) != 0
 define i1 @test_simplify7(i32 %C) {
 ; CHECK-LABEL: @test_simplify7
 ; CHECK-NEXT: [[TRUNC:%.*]] = trunc i32 %C to i16
-; CHECK-NEXT: %memchr.bounds = icmp ult i16 [[TRUNC]], 16
-; CHECK-NEXT: [[SHL:%.*]] = shl i16 1, [[TRUNC]]
+; CHECK-NEXT: [[TRUNC_AND:%.*]] = and i16 [[TRUNC]], 255
+; CHECK-NEXT: %memchr.bounds = icmp ult i16 [[TRUNC_AND]], 16
+; CHECK-NEXT: [[SHL:%.*]] = shl i16 1, [[TRUNC_AND]]
 ; CHECK-NEXT: [[AND:%.*]] = and i16 [[SHL]], 9217
 ; CHECK-NEXT: %memchr.bits = icmp ne i16 [[AND]], 0
 ; CHECK-NEXT: %memchr1 = and i1 %memchr.bounds, %memchr.bits
 ; CHECK-NEXT: ret i1 %memchr1
 
   %dst = call i8* @strchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @newlines, i64 0, i64 0), i32 %C)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }