Skip to content

Commit

Permalink
[InstSimplify] Add some basic simplifications for llvm.ptrmask
Browse files Browse the repository at this point in the history
Mostly the same as `and`. We also have a check for a useless
`llvm.ptrmask` if the ptr is already known aligned.

Differential Revision: https://reviews.llvm.org/D156633
  • Loading branch information
goldsteinn committed Nov 2, 2023
1 parent a37861d commit 8c2fcf5
Show file tree
Hide file tree
Showing 4 changed files with 63 additions and 53 deletions.
38 changes: 38 additions & 0 deletions llvm/lib/Analysis/InstructionSimplify.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -6412,6 +6412,44 @@ static Value *simplifyBinaryIntrinsic(Function *F, Value *Op0, Value *Op1,
return Constant::getNullValue(ReturnType);
break;
}
case Intrinsic::ptrmask: {
if (isa<PoisonValue>(Op0) || isa<PoisonValue>(Op1))
return PoisonValue::get(Op0->getType());

// NOTE: We can't apply this simplifications based on the value of Op1
// because we need to preserve provenance.
if (Q.isUndefValue(Op0) || match(Op0, m_Zero()))
return Constant::getNullValue(Op0->getType());

assert(Op1->getType()->getScalarSizeInBits() ==
Q.DL.getIndexTypeSizeInBits(Op0->getType()) &&
"Invalid mask width");
// If index-width (mask size) is less than pointer-size then mask is
// 1-extended.
if (match(Op1, m_PtrToInt(m_Specific(Op0))))
return Op0;

// NOTE: We may have attributes associated with the return value of the
// llvm.ptrmask intrinsic that will be lost when we just return the
// operand. We should try to preserve them.
if (match(Op1, m_AllOnes()) || Q.isUndefValue(Op1))
return Op0;

Constant *C;
if (match(Op1, m_ImmConstant(C))) {
KnownBits PtrKnown = computeKnownBits(Op0, /*Depth=*/0, Q);
// See if we only masking off bits we know are already zero due to
// alignment.
APInt IrrelevantPtrBits =
PtrKnown.Zero.zextOrTrunc(C->getType()->getScalarSizeInBits());
C = ConstantFoldBinaryOpOperands(
Instruction::Or, C, ConstantInt::get(C->getType(), IrrelevantPtrBits),
Q.DL);
if (C != nullptr && C->isAllOnesValue())
return Op0;
}
break;
}
case Intrinsic::smax:
case Intrinsic::smin:
case Intrinsic::umax:
Expand Down
5 changes: 2 additions & 3 deletions llvm/test/Transforms/InferAddressSpaces/AMDGPU/ptrmask.ll
Original file line number Diff line number Diff line change
Expand Up @@ -320,8 +320,7 @@ define i8 @ptrmask_cast_local_to_flat_const_mask_fffffffffffffffe(ptr addrspace(

define i8 @ptrmask_cast_local_to_flat_const_mask_ffffffffffffffff(ptr addrspace(3) %src.ptr) {
; CHECK-LABEL: @ptrmask_cast_local_to_flat_const_mask_ffffffffffffffff(
; CHECK-NEXT: [[TMP1:%.*]] = call ptr addrspace(3) @llvm.ptrmask.p3.i32(ptr addrspace(3) [[SRC_PTR:%.*]], i32 -1)
; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr addrspace(3) [[TMP1]], align 1
; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr addrspace(3) [[SRC_PTR:%.*]], align 1
; CHECK-NEXT: ret i8 [[LOAD]]
;
%cast = addrspacecast ptr addrspace(3) %src.ptr to ptr
Expand All @@ -333,7 +332,7 @@ define i8 @ptrmask_cast_local_to_flat_const_mask_ffffffffffffffff(ptr addrspace(
; Make sure non-constant masks can also be handled.
define i8 @ptrmask_cast_local_to_flat_load_range_mask(ptr addrspace(3) %src.ptr, ptr addrspace(1) %mask.ptr) {
; CHECK-LABEL: @ptrmask_cast_local_to_flat_load_range_mask(
; CHECK-NEXT: [[LOAD_MASK:%.*]] = load i64, ptr addrspace(1) [[MASK_PTR:%.*]], align 8, !range !0
; CHECK-NEXT: [[LOAD_MASK:%.*]] = load i64, ptr addrspace(1) [[MASK_PTR:%.*]], align 8, !range [[RNG0:![0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[LOAD_MASK]] to i32
; CHECK-NEXT: [[TMP2:%.*]] = call ptr addrspace(3) @llvm.ptrmask.p3.i32(ptr addrspace(3) [[SRC_PTR:%.*]], i32 [[TMP1]])
; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr addrspace(3) [[TMP2]], align 1
Expand Down
8 changes: 2 additions & 6 deletions llvm/test/Transforms/InstCombine/align-addr.ll
Original file line number Diff line number Diff line change
Expand Up @@ -181,11 +181,9 @@ define <16 x i8> @ptrmask_align8_ptr_align1(ptr align 1 %ptr) {
}

; Underlying alignment already the same as forced alignment by ptrmask
; TODO: Should be able to drop the ptrmask
define <16 x i8> @ptrmask_align8_ptr_align8(ptr align 8 %ptr) {
; CHECK-LABEL: @ptrmask_align8_ptr_align8(
; CHECK-NEXT: [[ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[PTR:%.*]], i64 -8)
; CHECK-NEXT: [[LOAD:%.*]] = load <16 x i8>, ptr [[ALIGNED]], align 1
; CHECK-NEXT: [[LOAD:%.*]] = load <16 x i8>, ptr [[PTR:%.*]], align 1
; CHECK-NEXT: ret <16 x i8> [[LOAD]]
;
%aligned = call ptr @llvm.ptrmask.p0.i64(ptr %ptr, i64 -8)
Expand All @@ -194,11 +192,9 @@ define <16 x i8> @ptrmask_align8_ptr_align8(ptr align 8 %ptr) {
}

; Underlying alignment greater than alignment forced by ptrmask
; TODO: Should be able to drop the ptrmask
define <16 x i8> @ptrmask_align8_ptr_align16(ptr align 16 %ptr) {
; CHECK-LABEL: @ptrmask_align8_ptr_align16(
; CHECK-NEXT: [[ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[PTR:%.*]], i64 -8)
; CHECK-NEXT: [[LOAD:%.*]] = load <16 x i8>, ptr [[ALIGNED]], align 1
; CHECK-NEXT: [[LOAD:%.*]] = load <16 x i8>, ptr [[PTR:%.*]], align 1
; CHECK-NEXT: ret <16 x i8> [[LOAD]]
;
%aligned = call ptr @llvm.ptrmask.p0.i64(ptr %ptr, i64 -8)
Expand Down
65 changes: 21 additions & 44 deletions llvm/test/Transforms/InstSimplify/ptrmask.ll
Original file line number Diff line number Diff line change
Expand Up @@ -12,8 +12,7 @@ declare <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr>, <2 x i64>)
define ptr @ptrmask_simplify_poison_mask(ptr %p) {
; CHECK-LABEL: define ptr @ptrmask_simplify_poison_mask
; CHECK-SAME: (ptr [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[P]], i64 poison)
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr poison
;
%r = call ptr @llvm.ptrmask.p0.i64(ptr %p, i64 poison)
ret ptr %r
Expand All @@ -22,8 +21,7 @@ define ptr @ptrmask_simplify_poison_mask(ptr %p) {
define <2 x ptr addrspace(1) > @ptrmask_simplify_poison_mask_vec(<2 x ptr addrspace(1) > %p) {
; CHECK-LABEL: define <2 x ptr addrspace(1)> @ptrmask_simplify_poison_mask_vec
; CHECK-SAME: (<2 x ptr addrspace(1)> [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr addrspace(1)> @llvm.ptrmask.v2p1.v2i32(<2 x ptr addrspace(1)> [[P]], <2 x i32> poison)
; CHECK-NEXT: ret <2 x ptr addrspace(1)> [[R]]
; CHECK-NEXT: ret <2 x ptr addrspace(1)> poison
;
%r = call <2 x ptr addrspace(1) > @llvm.ptrmask.v2p1.v2i32(<2 x ptr addrspace(1) > %p, <2 x i32> poison)
ret <2 x ptr addrspace(1) > %r
Expand All @@ -42,8 +40,7 @@ define <2 x ptr addrspace(1) > @ptrmask_simplify_poison_and_zero_i32_vec_fail(<2
define <2 x ptr> @ptrmask_simplify_undef_and_ones_vec(<2 x ptr> %p) {
; CHECK-LABEL: define <2 x ptr> @ptrmask_simplify_undef_and_ones_vec
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 undef, i64 -1>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[P]]
;
%r = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 undef, i64 -1>)
ret <2 x ptr> %r
Expand All @@ -52,8 +49,7 @@ define <2 x ptr> @ptrmask_simplify_undef_and_ones_vec(<2 x ptr> %p) {
define <2 x ptr> @ptrmask_simplify_poison_and_ones_vec(<2 x ptr> %p) {
; CHECK-LABEL: define <2 x ptr> @ptrmask_simplify_poison_and_ones_vec
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 poison, i64 -1>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[P]]
;
%r = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 poison, i64 -1>)
ret <2 x ptr> %r
Expand All @@ -62,8 +58,7 @@ define <2 x ptr> @ptrmask_simplify_poison_and_ones_vec(<2 x ptr> %p) {
define <2 x ptr> @ptrmask_simplify_ones_vec(<2 x ptr> %p) {
; CHECK-LABEL: define <2 x ptr> @ptrmask_simplify_ones_vec
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 -1, i64 -1>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[P]]
;
%r = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 -1, i64 -1>)
ret <2 x ptr> %r
Expand All @@ -72,8 +67,7 @@ define <2 x ptr> @ptrmask_simplify_ones_vec(<2 x ptr> %p) {
define <2 x ptr addrspace(1) > @ptrmask_simplify_ones_i32_vec(<2 x ptr addrspace(1) > %p) {
; CHECK-LABEL: define <2 x ptr addrspace(1)> @ptrmask_simplify_ones_i32_vec
; CHECK-SAME: (<2 x ptr addrspace(1)> [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr addrspace(1)> @llvm.ptrmask.v2p1.v2i32(<2 x ptr addrspace(1)> [[P]], <2 x i32> <i32 -1, i32 -1>)
; CHECK-NEXT: ret <2 x ptr addrspace(1)> [[R]]
; CHECK-NEXT: ret <2 x ptr addrspace(1)> [[P]]
;
%r = call <2 x ptr addrspace(1) > @llvm.ptrmask.v2p1.v2i32(<2 x ptr addrspace(1) > %p, <2 x i32> <i32 -1, i32 -1>)
ret <2 x ptr addrspace(1) > %r
Expand All @@ -82,8 +76,7 @@ define <2 x ptr addrspace(1) > @ptrmask_simplify_ones_i32_vec(<2 x ptr addrspace
define ptr addrspace(1) @ptrmask_simplify_undef_mask(ptr addrspace(1) %p) {
; CHECK-LABEL: define ptr addrspace(1) @ptrmask_simplify_undef_mask
; CHECK-SAME: (ptr addrspace(1) [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) [[P]], i32 undef)
; CHECK-NEXT: ret ptr addrspace(1) [[R]]
; CHECK-NEXT: ret ptr addrspace(1) [[P]]
;
%r = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) %p, i32 undef)
ret ptr addrspace(1) %r
Expand All @@ -102,8 +95,7 @@ define ptr @ptrmask_simplify_0_mask(ptr %p) {
define ptr @ptrmask_simplify_1s_mask(ptr %p) {
; CHECK-LABEL: define ptr @ptrmask_simplify_1s_mask
; CHECK-SAME: (ptr [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[P]], i64 -1)
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr [[P]]
;
%r = call ptr @llvm.ptrmask.p0.i64(ptr %p, i64 -1)
ret ptr %r
Expand All @@ -112,8 +104,7 @@ define ptr @ptrmask_simplify_1s_mask(ptr %p) {
define ptr addrspace(1) @ptrmask_simplify_1s_mask_i32(ptr addrspace(1) %p) {
; CHECK-LABEL: define ptr addrspace(1) @ptrmask_simplify_1s_mask_i32
; CHECK-SAME: (ptr addrspace(1) [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) [[P]], i32 -1)
; CHECK-NEXT: ret ptr addrspace(1) [[R]]
; CHECK-NEXT: ret ptr addrspace(1) [[P]]
;
%r = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) %p, i32 -1)
ret ptr addrspace(1) %r
Expand All @@ -122,8 +113,7 @@ define ptr addrspace(1) @ptrmask_simplify_1s_mask_i32(ptr addrspace(1) %p) {
define ptr @ptrmask_simplify_poison_ptr(i64 %m) {
; CHECK-LABEL: define ptr @ptrmask_simplify_poison_ptr
; CHECK-SAME: (i64 [[M:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr poison, i64 [[M]])
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr poison
;
%r = call ptr @llvm.ptrmask.p0.i64(ptr poison, i64 %m)
ret ptr %r
Expand All @@ -132,8 +122,7 @@ define ptr @ptrmask_simplify_poison_ptr(i64 %m) {
define ptr addrspace(1) @ptrmask_simplify_undef_ptr(i32 %m) {
; CHECK-LABEL: define ptr addrspace(1) @ptrmask_simplify_undef_ptr
; CHECK-SAME: (i32 [[M:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) undef, i32 [[M]])
; CHECK-NEXT: ret ptr addrspace(1) [[R]]
; CHECK-NEXT: ret ptr addrspace(1) null
;
%r = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) undef, i32 %m)
ret ptr addrspace(1) %r
Expand All @@ -142,8 +131,7 @@ define ptr addrspace(1) @ptrmask_simplify_undef_ptr(i32 %m) {
define ptr @ptrmask_simplify_null_ptr(i64 %m) {
; CHECK-LABEL: define ptr @ptrmask_simplify_null_ptr
; CHECK-SAME: (i64 [[M:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr null, i64 [[M]])
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr null
;
%r = call ptr @llvm.ptrmask.p0.i64(ptr null, i64 %m)
ret ptr %r
Expand All @@ -152,9 +140,7 @@ define ptr @ptrmask_simplify_null_ptr(i64 %m) {
define ptr @ptrmask_simplify_ptrmask(ptr %p) {
; CHECK-LABEL: define ptr @ptrmask_simplify_ptrmask
; CHECK-SAME: (ptr [[P:%.*]]) {
; CHECK-NEXT: [[M:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[P]], i64 [[M]])
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr [[P]]
;
%m = ptrtoint ptr %p to i64
%r = call ptr @llvm.ptrmask.p0.i64(ptr %p, i64 %m)
Expand All @@ -164,9 +150,7 @@ define ptr @ptrmask_simplify_ptrmask(ptr %p) {
define ptr addrspace(1) @ptrmask_simplify_ptrmask_i32(ptr addrspace(1) %p) {
; CHECK-LABEL: define ptr addrspace(1) @ptrmask_simplify_ptrmask_i32
; CHECK-SAME: (ptr addrspace(1) [[P:%.*]]) {
; CHECK-NEXT: [[M:%.*]] = ptrtoint ptr addrspace(1) [[P]] to i32
; CHECK-NEXT: [[R:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) [[P]], i32 [[M]])
; CHECK-NEXT: ret ptr addrspace(1) [[R]]
; CHECK-NEXT: ret ptr addrspace(1) [[P]]
;
%m = ptrtoint ptr addrspace(1) %p to i32
%r = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) %p, i32 %m)
Expand All @@ -176,8 +160,7 @@ define ptr addrspace(1) @ptrmask_simplify_ptrmask_i32(ptr addrspace(1) %p) {
define ptr @ptrmask_simplify_aligned_unused(ptr align 64 %p) {
; CHECK-LABEL: define ptr @ptrmask_simplify_aligned_unused
; CHECK-SAME: (ptr align 64 [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[P]], i64 -64)
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr [[P]]
;
%r = call ptr @llvm.ptrmask.p0.i64(ptr %p, i64 -64)
ret ptr %r
Expand Down Expand Up @@ -206,8 +189,7 @@ define <2 x ptr> @ptrmask_simplify_aligned_unused_vec_todo(<2 x ptr> align 128 %
define ptr addrspace(1) @ptrmask_simplify_aligned_unused_i32(ptr addrspace(1) align 64 %p) {
; CHECK-LABEL: define ptr addrspace(1) @ptrmask_simplify_aligned_unused_i32
; CHECK-SAME: (ptr addrspace(1) align 64 [[P:%.*]]) {
; CHECK-NEXT: [[R:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) [[P]], i32 -64)
; CHECK-NEXT: ret ptr addrspace(1) [[R]]
; CHECK-NEXT: ret ptr addrspace(1) [[P]]
;
%r = call ptr addrspace(1) @llvm.ptrmask.p1.i32(ptr addrspace(1) %p, i32 -64)
ret ptr addrspace(1) %r
Expand All @@ -218,8 +200,7 @@ define ptr @ptrmask_simplify_known_unused(ptr %p) {
; CHECK-SAME: (ptr [[P:%.*]]) {
; CHECK-NEXT: [[PM0:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[P]], i64 -64)
; CHECK-NEXT: [[PGEP:%.*]] = getelementptr i8, ptr [[PM0]], i64 32
; CHECK-NEXT: [[R:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[PGEP]], i64 -32)
; CHECK-NEXT: ret ptr [[R]]
; CHECK-NEXT: ret ptr [[PGEP]]
;
%pm0 = call ptr @llvm.ptrmask.p0.i64(ptr %p, i64 -64)
%pgep = getelementptr i8, ptr %pm0, i64 32
Expand All @@ -232,8 +213,7 @@ define <2 x ptr> @ptrmask_simplify_known_unused_vec(<2 x ptr> %p) {
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[PM0:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 -64, i64 -64>)
; CHECK-NEXT: [[PGEP:%.*]] = getelementptr i8, <2 x ptr> [[PM0]], <2 x i64> <i64 32, i64 32>
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[PGEP]], <2 x i64> <i64 -32, i64 -32>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[PGEP]]
;
%pm0 = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 -64, i64 -64>)
%pgep = getelementptr i8, <2 x ptr> %pm0, <2 x i64> <i64 32, i64 32>
Expand All @@ -246,8 +226,7 @@ define <2 x ptr> @ptrmask_simplify_known_unused_vec2(<2 x ptr> %p) {
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[PM0:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 -64, i64 -64>)
; CHECK-NEXT: [[PGEP:%.*]] = getelementptr i8, <2 x ptr> [[PM0]], <2 x i64> <i64 32, i64 32>
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[PGEP]], <2 x i64> <i64 -32, i64 -16>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[PGEP]]
;
%pm0 = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 -64, i64 -64>)
%pgep = getelementptr i8, <2 x ptr> %pm0, <2 x i64> <i64 32, i64 32>
Expand All @@ -260,8 +239,7 @@ define <2 x ptr> @ptrmask_simplify_known_unused_vec3(<2 x ptr> %p) {
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[PM0:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 -64, i64 -128>)
; CHECK-NEXT: [[PGEP:%.*]] = getelementptr i8, <2 x ptr> [[PM0]], <2 x i64> <i64 32, i64 32>
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[PGEP]], <2 x i64> <i64 -32, i64 -32>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[PGEP]]
;
%pm0 = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 -64, i64 -128>)
%pgep = getelementptr i8, <2 x ptr> %pm0, <2 x i64> <i64 32, i64 32>
Expand All @@ -274,8 +252,7 @@ define <2 x ptr> @ptrmask_simplify_known_unused_vec4(<2 x ptr> %p) {
; CHECK-SAME: (<2 x ptr> [[P:%.*]]) {
; CHECK-NEXT: [[PM0:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[P]], <2 x i64> <i64 -64, i64 -128>)
; CHECK-NEXT: [[PGEP:%.*]] = getelementptr i8, <2 x ptr> [[PM0]], <2 x i64> <i64 32, i64 64>
; CHECK-NEXT: [[R:%.*]] = call <2 x ptr> @llvm.ptrmask.v2p0.v2i64(<2 x ptr> [[PGEP]], <2 x i64> <i64 -32, i64 -32>)
; CHECK-NEXT: ret <2 x ptr> [[R]]
; CHECK-NEXT: ret <2 x ptr> [[PGEP]]
;
%pm0 = call <2 x ptr> @llvm.ptrmask.v2p1.v2i64(<2 x ptr> %p, <2 x i64> <i64 -64, i64 -128>)
%pgep = getelementptr i8, <2 x ptr> %pm0, <2 x i64> <i64 32, i64 64>
Expand Down

0 comments on commit 8c2fcf5

Please sign in to comment.