Skip to content

[Hexagon] Shuffle patterns to vdeal + vpack #146122

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

quic-akaryaki
Copy link
Contributor

Lowering shuffle patterns to vdeal + vpack caused an assertion because the vdeal parameter value is negative but an unsigned one was expected.

Lowering shuffle patterns to vdeal + vpack caused an assertion because
the vdeal parameter value is negative but an unsigned one was expected.
@llvmbot
Copy link
Member

llvmbot commented Jun 27, 2025

@llvm/pr-subscribers-backend-hexagon

Author: Alexey Karyakin (quic-akaryaki)

Changes

Lowering shuffle patterns to vdeal + vpack caused an assertion because the vdeal parameter value is negative but an unsigned one was expected.


Full diff: https://github.com/llvm/llvm-project/pull/146122.diff

2 Files Affected:

  • (modified) llvm/lib/Target/Hexagon/HexagonISelDAGToDAGHVX.cpp (+9-3)
  • (added) llvm/test/CodeGen/Hexagon/hvx-vdeal-vpack.ll (+17)
diff --git a/llvm/lib/Target/Hexagon/HexagonISelDAGToDAGHVX.cpp b/llvm/lib/Target/Hexagon/HexagonISelDAGToDAGHVX.cpp
index 52fa678a600e9..48494c0380638 100644
--- a/llvm/lib/Target/Hexagon/HexagonISelDAGToDAGHVX.cpp
+++ b/llvm/lib/Target/Hexagon/HexagonISelDAGToDAGHVX.cpp
@@ -958,7 +958,8 @@ namespace llvm {
     void select(SDNode *ISelN);
     void materialize(const ResultStack &Results);
 
-    SDValue getConst32(int Val, const SDLoc &dl);
+    SDValue getConst32(unsigned Val, const SDLoc &dl);
+    SDValue getSignedConst32(int Val, const SDLoc &dl);
     SDValue getVectorConstant(ArrayRef<uint8_t> Data, const SDLoc &dl);
 
     enum : unsigned {
@@ -2145,7 +2146,8 @@ OpRef HvxSelector::contracting(ShuffleMask SM, OpRef Va, OpRef Vb,
     for (int i = 0, e = std::size(Opcodes); i != e; ++i) {
       auto [Size, Odd] = Packs[i];
       if (same(SM.Mask, shuffles::mask(shuffles::vdeal, HwLen, Size, Odd))) {
-        Results.push(Hexagon::A2_tfrsi, MVT::i32, {getConst32(-2 * Size, dl)});
+        Results.push(Hexagon::A2_tfrsi, MVT::i32,
+                     {getSignedConst32(-2 * Size, dl)});
         Results.push(Hexagon::V6_vdealvdd, PairTy, {Vb, Va, OpRef::res(-1)});
         auto vdeal = OpRef::res(Results.top());
         Results.push(Opcodes[i], SingleTy,
@@ -2545,10 +2547,14 @@ OpRef HvxSelector::butterfly(ShuffleMask SM, OpRef Va, ResultStack &Results) {
   return OpRef::fail();
 }
 
-SDValue HvxSelector::getConst32(int Val, const SDLoc &dl) {
+SDValue HvxSelector::getConst32(unsigned Val, const SDLoc &dl) {
   return DAG.getTargetConstant(Val, dl, MVT::i32);
 }
 
+SDValue HvxSelector::getSignedConst32(int Val, const SDLoc &dl) {
+  return DAG.getSignedTargetConstant(Val, dl, MVT::i32);
+}
+
 SDValue HvxSelector::getVectorConstant(ArrayRef<uint8_t> Data,
                                        const SDLoc &dl) {
   SmallVector<SDValue, 128> Elems;
diff --git a/llvm/test/CodeGen/Hexagon/hvx-vdeal-vpack.ll b/llvm/test/CodeGen/Hexagon/hvx-vdeal-vpack.ll
new file mode 100644
index 0000000000000..29bb462004943
--- /dev/null
+++ b/llvm/test/CodeGen/Hexagon/hvx-vdeal-vpack.ll
@@ -0,0 +1,17 @@
+; RUN: llc -march=hexagon -O2  < %s | FileCheck %s
+
+;; Test that a vdeal + vpack can be lowered.
+
+; CHECK: v[[#V1:]]:[[#V0:]] = vdeal(v[[#]],v[[#]],r[[#]])
+; CHECK: v[[#]].h = vpacke(v[[#V1]].w,v[[#V0]].w)
+
+target datalayout = "e-m:e-p:32:32:32-a:0-n16:32-i64:64:64-i32:32:32-i16:16:16-i1:8:8-f32:32:32-f64:64:64-v32:32:32-v64:64:64-v512:512:512-v1024:1024:1024-v2048:2048:2048"
+target triple = "hexagon"
+
+define dso_local <64 x i16> @f(<128 x i16> noundef %index) local_unnamed_addr #1 {
+entry:
+  %b = shufflevector <128 x i16> %index, <128 x i16> poison, <64 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60, i32 64, i32 68, i32 72, i32 76, i32 80, i32 84, i32 88, i32 92, i32 96, i32 100, i32 104, i32 108, i32 112, i32 116, i32 120, i32 124, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30, i32 34, i32 38, i32 42, i32 46, i32 50, i32 54, i32 58, i32 62, i32 66, i32 70, i32 74, i32 78, i32 82, i32 86, i32 90, i32 94, i32 98, i32 102, i32 106, i32 110, i32 114, i32 118, i32 122, i32 126>
+  ret <64 x i16> %b
+}
+
+attributes #1 = { mustprogress nofree norecurse nosync nounwind willreturn "target-cpu"="hexagonv75" "target-features"="+hvx-length128b,+hvx" }

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants