[llvm-branch-commits] [llvm] 2b20c3f - [SelectionDAG] Fix problematic call to EVT::changeVectorElementType().
Tobias Hieta via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Tue Aug 15 02:09:21 PDT 2023
Author: Paul Walker
Date: 2023-08-15T11:07:47+02:00
New Revision: 2b20c3fe19637e4ad842decce552d5ad95c199b7
URL: https://github.com/llvm/llvm-project/commit/2b20c3fe19637e4ad842decce552d5ad95c199b7
DIFF: https://github.com/llvm/llvm-project/commit/2b20c3fe19637e4ad842decce552d5ad95c199b7.diff
LOG: [SelectionDAG] Fix problematic call to EVT::changeVectorElementType().
The function changeVectorElementType assumes MVT input types will
result in MVT output types. There's no gurantee this is possible
during early code generation and so this patch converts an instance
used during initial DAG construction to instead explicitly create a
new EVT.
NOTE: I could have added more MVTs, but that seemed unscalable as
you can either have MVTs with 100% element count coverage or 100%
bitwidth coverage, but not both.
Differential Revision: https://reviews.llvm.org/D157392
(cherry picked from commit b7e6e568b4f28124060b868e433f36af18c510db)
Added:
Modified:
llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
llvm/test/CodeGen/AArch64/active_lane_mask.ll
Removed:
################################################################################
diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
index bd4fa40b41c809..566881466a1a4a 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
@@ -7396,7 +7396,8 @@ void SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I,
}
SDValue TripCount = getValue(I.getOperand(1));
- auto VecTy = CCVT.changeVectorElementType(ElementVT);
+ EVT VecTy = EVT::getVectorVT(*DAG.getContext(), ElementVT,
+ CCVT.getVectorElementCount());
SDValue VectorIndex = DAG.getSplat(VecTy, sdl, Index);
SDValue VectorTripCount = DAG.getSplat(VecTy, sdl, TripCount);
diff --git a/llvm/test/CodeGen/AArch64/active_lane_mask.ll b/llvm/test/CodeGen/AArch64/active_lane_mask.ll
index b1d5e2c4dca349..8de5edc7b017a0 100644
--- a/llvm/test/CodeGen/AArch64/active_lane_mask.ll
+++ b/llvm/test/CodeGen/AArch64/active_lane_mask.ll
@@ -315,6 +315,16 @@ define <vscale x 32 x i1> @lane_mask_nxv32i1_i8(i8 %index, i8 %TC) {
ret <vscale x 32 x i1> %active.lane.mask
}
+; UTC_ARGS: --disable
+; This test exists to protect against a compiler crash caused by an attempt to
+; convert (via changeVectorElementType) an MVT into an EVT, which is impossible.
+; The test's output is large and not relevant so check lines have been disabled.
+define <vscale x 64 x i1> @lane_mask_nxv64i1_i64(i64 %index, i64 %TC) {
+; CHECK-LABEL: lane_mask_nxv64i1_i64:
+ %active.lane.mask = call <vscale x 64 x i1> @llvm.get.active.lane.mask.nxv64i1.i64(i64 %index, i64 %TC)
+ ret <vscale x 64 x i1> %active.lane.mask
+}
+; UTC_ARGS: --enable
; == Fixed width ==
@@ -409,9 +419,9 @@ define <2 x i1> @lane_mask_v2i1_i64(i64 %index, i64 %TC) {
define <16 x i1> @lane_mask_v16i1_i8(i8 %index, i8 %TC) {
; CHECK-LABEL: lane_mask_v16i1_i8:
; CHECK: // %bb.0:
-; CHECK-NEXT: adrp x8, .LCPI23_0
+; CHECK-NEXT: adrp x8, .LCPI24_0
; CHECK-NEXT: dup v1.16b, w0
-; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI23_0]
+; CHECK-NEXT: ldr q0, [x8, :lo12:.LCPI24_0]
; CHECK-NEXT: uqadd v0.16b, v1.16b, v0.16b
; CHECK-NEXT: dup v1.16b, w1
; CHECK-NEXT: cmhi v0.16b, v1.16b, v0.16b
@@ -423,10 +433,10 @@ define <16 x i1> @lane_mask_v16i1_i8(i8 %index, i8 %TC) {
define <8 x i1> @lane_mask_v8i1_i8(i8 %index, i8 %TC) {
; CHECK-LABEL: lane_mask_v8i1_i8:
; CHECK: // %bb.0:
-; CHECK-NEXT: adrp x8, .LCPI24_0
+; CHECK-NEXT: adrp x8, .LCPI25_0
; CHECK-NEXT: dup v0.8b, w0
; CHECK-NEXT: dup v2.8b, w1
-; CHECK-NEXT: ldr d1, [x8, :lo12:.LCPI24_0]
+; CHECK-NEXT: ldr d1, [x8, :lo12:.LCPI25_0]
; CHECK-NEXT: uqadd v0.8b, v0.8b, v1.8b
; CHECK-NEXT: cmhi v0.8b, v2.8b, v0.8b
; CHECK-NEXT: ret
@@ -437,11 +447,11 @@ define <8 x i1> @lane_mask_v8i1_i8(i8 %index, i8 %TC) {
define <4 x i1> @lane_mask_v4i1_i8(i8 %index, i8 %TC) {
; CHECK-LABEL: lane_mask_v4i1_i8:
; CHECK: // %bb.0:
-; CHECK-NEXT: adrp x8, .LCPI25_0
+; CHECK-NEXT: adrp x8, .LCPI26_0
; CHECK-NEXT: dup v0.4h, w0
; CHECK-NEXT: movi d2, #0xff00ff00ff00ff
; CHECK-NEXT: dup v3.4h, w1
-; CHECK-NEXT: ldr d1, [x8, :lo12:.LCPI25_0]
+; CHECK-NEXT: ldr d1, [x8, :lo12:.LCPI26_0]
; CHECK-NEXT: bic v0.4h, #255, lsl #8
; CHECK-NEXT: bic v3.4h, #255, lsl #8
; CHECK-NEXT: add v0.4h, v0.4h, v1.4h
@@ -455,11 +465,11 @@ define <4 x i1> @lane_mask_v4i1_i8(i8 %index, i8 %TC) {
define <2 x i1> @lane_mask_v2i1_i8(i8 %index, i8 %TC) {
; CHECK-LABEL: lane_mask_v2i1_i8:
; CHECK: // %bb.0:
-; CHECK-NEXT: adrp x8, .LCPI26_0
+; CHECK-NEXT: adrp x8, .LCPI27_0
; CHECK-NEXT: movi d0, #0x0000ff000000ff
; CHECK-NEXT: dup v1.2s, w0
; CHECK-NEXT: dup v3.2s, w1
-; CHECK-NEXT: ldr d2, [x8, :lo12:.LCPI26_0]
+; CHECK-NEXT: ldr d2, [x8, :lo12:.LCPI27_0]
; CHECK-NEXT: and v1.8b, v1.8b, v0.8b
; CHECK-NEXT: add v1.2s, v1.2s, v2.2s
; CHECK-NEXT: umin v1.2s, v1.2s, v0.2s
@@ -483,7 +493,7 @@ entry:
define <vscale x 4 x i1> @lane_mask_nxv4i1_imm5() {
; CHECK-LABEL: lane_mask_nxv4i1_imm5:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov w8, #5
+; CHECK-NEXT: mov w8, #5 // =0x5
; CHECK-NEXT: whilelo p0.s, xzr, x8
; CHECK-NEXT: ret
entry:
@@ -504,7 +514,7 @@ entry:
define <vscale x 16 x i1> @lane_mask_nxv16i1_imm10() {
; CHECK-LABEL: lane_mask_nxv16i1_imm10:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov w8, #10
+; CHECK-NEXT: mov w8, #10 // =0xa
; CHECK-NEXT: whilelo p0.b, xzr, x8
; CHECK-NEXT: ret
entry:
@@ -529,6 +539,7 @@ declare <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i32(i32, i32)
declare <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i32(i32, i32)
declare <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i32(i32, i32)
+declare <vscale x 64 x i1> @llvm.get.active.lane.mask.nxv64i1.i64(i64, i64)
declare <vscale x 32 x i1> @llvm.get.active.lane.mask.nxv32i1.i64(i64, i64)
declare <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64, i64)
declare <vscale x 8 x i1> @llvm.get.active.lane.mask.nxv8i1.i64(i64, i64)
More information about the llvm-branch-commits
mailing list