Skip to content

[CIR] Allow use different Int types together in Vec Shift Op #141111

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

Merged
merged 5 commits into from
May 28, 2025

Conversation

AmrDeveloper
Copy link
Member

Update the verification of ShiftOp for Vector to allow performing shift op between signed and unsigned integers, similar to LLVM IR

Issue #136487

@llvmbot llvmbot added clang Clang issues not falling into any other category ClangIR Anything related to the ClangIR project labels May 22, 2025
@llvmbot
Copy link
Member

llvmbot commented May 22, 2025

@llvm/pr-subscribers-clangir

Author: Amr Hesham (AmrDeveloper)

Changes

Update the verification of ShiftOp for Vector to allow performing shift op between signed and unsigned integers, similar to LLVM IR

Issue #136487


Full diff: https://github.com/llvm/llvm-project/pull/141111.diff

2 Files Affected:

  • (modified) clang/lib/CIR/Dialect/IR/CIRDialect.cpp (+2-2)
  • (modified) clang/test/CIR/CodeGen/vector.cpp (+58-12)
diff --git a/clang/lib/CIR/Dialect/IR/CIRDialect.cpp b/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
index 36dcbc6a4be4a..a49a6c4b42de0 100644
--- a/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
+++ b/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
@@ -1427,12 +1427,12 @@ OpFoldResult cir::SelectOp::fold(FoldAdaptor adaptor) {
 //===----------------------------------------------------------------------===//
 LogicalResult cir::ShiftOp::verify() {
   mlir::Operation *op = getOperation();
-  mlir::Type resType = getResult().getType();
   const bool isOp0Vec = mlir::isa<cir::VectorType>(op->getOperand(0).getType());
   const bool isOp1Vec = mlir::isa<cir::VectorType>(op->getOperand(1).getType());
   if (isOp0Vec != isOp1Vec)
     return emitOpError() << "input types cannot be one vector and one scalar";
-  if (isOp1Vec && op->getOperand(1).getType() != resType) {
+
+  if (isOp1Vec && !mlir::isa<cir::VectorType>(getResult().getType())) {
     return emitOpError() << "shift amount must have the type of the result "
                          << "if it is vector shift";
   }
diff --git a/clang/test/CIR/CodeGen/vector.cpp b/clang/test/CIR/CodeGen/vector.cpp
index 2ad5b247cd1b0..63b2ea46d7b13 100644
--- a/clang/test/CIR/CodeGen/vector.cpp
+++ b/clang/test/CIR/CodeGen/vector.cpp
@@ -6,6 +6,7 @@
 // RUN: FileCheck --input-file=%t.ll %s -check-prefix=OGCG
 
 typedef int vi4 __attribute__((vector_size(16)));
+typedef unsigned int uvi4 __attribute__((vector_size(16)));
 typedef double vd2 __attribute__((vector_size(16)));
 typedef long long vll2 __attribute__((vector_size(16)));
 
@@ -394,19 +395,9 @@ void foo9() {
 // CIR: %[[VEC_B:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["b", init]
 // CIR: %[[SHL_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shl", init]
 // CIR: %[[SHR_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shr", init]
-// CIR: %[[CONST_1:.*]] = cir.const #cir.int<1> : !s32i
-// CIR: %[[CONST_2:.*]] = cir.const #cir.int<2> : !s32i
-// CIR: %[[CONST_3:.*]] = cir.const #cir.int<3> : !s32i
-// CIR: %[[CONST_4:.*]] = cir.const #cir.int<4> : !s32i
-// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%[[CONST_1]], %[[CONST_2]], %[[CONST_3]], %[[CONST_4]] :
-// CIR-SAME: !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} :  !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
 // CIR: cir.store %[[VEC_A_VAL]], %[[VEC_A]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
-// CIR: %[[CONST_5:.*]] = cir.const #cir.int<5> : !s32i
-// CIR: %[[CONST_6:.*]] = cir.const #cir.int<6> : !s32i
-// CIR: %[[CONST_7:.*]] = cir.const #cir.int<7> : !s32i
-// CIR: %[[CONST_8:.*]] = cir.const #cir.int<8> : !s32i
-// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%[[CONST_5]], %[[CONST_6]], %[[CONST_7]], %[[CONST_8]] :
-// CIR-SAME: !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
 // CIR: cir.store %[[VEC_B_VAL]], %[[VEC_B]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
 // CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
 // CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
@@ -447,6 +438,61 @@ void foo9() {
 // OGCG: %[[SHR:.*]] = ashr <4 x i32> %[[TMP_A]], %[[TMP_B]]
 // OGCG: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
 
+void foo10() {
+  vi4 a = {1, 2, 3, 4};
+  uvi4 b = {5u, 6u, 7u, 8u};
+
+  vi4 shl = a << b;
+  uvi4 shr = b >> a;
+}
+
+// CIR: %[[VEC_A:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["a", init]
+// CIR: %[[VEC_B:.*]] = cir.alloca !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>, ["b", init]
+// CIR: %[[SHL_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shl", init]
+// CIR: %[[SHR_RES:.*]] = cir.alloca !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>, ["shr", init]
+// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: cir.store %[[VEC_A_VAL]], %[[VEC_A]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} :  !u32i, !u32i, !u32i, !u32i) : !cir.vector<4 x !u32i>
+// CIR: cir.store %[[VEC_B_VAL]], %[[VEC_B]] : !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR: %[[SHL:.*]] = cir.shift(left, %[[TMP_A]] : !cir.vector<4 x !s32i>, %[[TMP_B]] : !cir.vector<4 x !u32i>) -> !cir.vector<4 x !s32i>
+// CIR: cir.store %[[SHL]], %[[SHL_RES]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR: %[[SHR:.*]] = cir.shift(right, %[[TMP_B]] : !cir.vector<4 x !u32i>, %[[TMP_A]] : !cir.vector<4 x !s32i>) -> !cir.vector<4 x !u32i>
+// CIR: cir.store %[[SHR]], %[[SHR_RES]] : !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>
+
+// LLVM: %[[VEC_A:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[VEC_B:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[SHL_RES:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[SHR_RES:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: store <4 x i32> <i32 1, i32 2, i32 3, i32 4>, ptr %[[VEC_A]], align 16
+// LLVM: store <4 x i32> <i32 5, i32 6, i32 7, i32 8>, ptr %[[VEC_B]], align 16
+// LLVM: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// LLVM: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// LLVM: %[[SHL:.*]] = shl <4 x i32> %[[TMP_A]], %[[TMP_B]]
+// LLVM: store <4 x i32> %[[SHL]], ptr %[[SHL_RES]], align 16
+// LLVM: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// LLVM: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// LLVM: %[[SHR:.*]] = lshr <4 x i32> %[[TMP_B]], %[[TMP_A]]
+// LLVM: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
+
+// OGCG: %[[VEC_A:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[VEC_B:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[SHL_RES:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[SHR_RES:.*]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> <i32 1, i32 2, i32 3, i32 4>, ptr %[[VEC_A]], align 16
+// OGCG: store <4 x i32> <i32 5, i32 6, i32 7, i32 8>, ptr %[[VEC_B]], align 16
+// OGCG: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// OGCG: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// OGCG: %[[SHL:.*]] = shl <4 x i32> %[[TMP_A]], %[[TMP_B]]
+// OGCG: store <4 x i32> %[[SHL]], ptr %[[SHL_RES]], align 16
+// OGCG: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// OGCG: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// OGCG: %[[SHR:.*]] = lshr <4 x i32> %[[TMP_B]], %[[TMP_A]]
+// OGCG: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
+
 void foo11() {
   vi4 a = {1, 2, 3, 4};
   vi4 b = {5, 6, 7, 8};

@llvmbot
Copy link
Member

llvmbot commented May 22, 2025

@llvm/pr-subscribers-clang

Author: Amr Hesham (AmrDeveloper)

Changes

Update the verification of ShiftOp for Vector to allow performing shift op between signed and unsigned integers, similar to LLVM IR

Issue #136487


Full diff: https://github.com/llvm/llvm-project/pull/141111.diff

2 Files Affected:

  • (modified) clang/lib/CIR/Dialect/IR/CIRDialect.cpp (+2-2)
  • (modified) clang/test/CIR/CodeGen/vector.cpp (+58-12)
diff --git a/clang/lib/CIR/Dialect/IR/CIRDialect.cpp b/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
index 36dcbc6a4be4a..a49a6c4b42de0 100644
--- a/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
+++ b/clang/lib/CIR/Dialect/IR/CIRDialect.cpp
@@ -1427,12 +1427,12 @@ OpFoldResult cir::SelectOp::fold(FoldAdaptor adaptor) {
 //===----------------------------------------------------------------------===//
 LogicalResult cir::ShiftOp::verify() {
   mlir::Operation *op = getOperation();
-  mlir::Type resType = getResult().getType();
   const bool isOp0Vec = mlir::isa<cir::VectorType>(op->getOperand(0).getType());
   const bool isOp1Vec = mlir::isa<cir::VectorType>(op->getOperand(1).getType());
   if (isOp0Vec != isOp1Vec)
     return emitOpError() << "input types cannot be one vector and one scalar";
-  if (isOp1Vec && op->getOperand(1).getType() != resType) {
+
+  if (isOp1Vec && !mlir::isa<cir::VectorType>(getResult().getType())) {
     return emitOpError() << "shift amount must have the type of the result "
                          << "if it is vector shift";
   }
diff --git a/clang/test/CIR/CodeGen/vector.cpp b/clang/test/CIR/CodeGen/vector.cpp
index 2ad5b247cd1b0..63b2ea46d7b13 100644
--- a/clang/test/CIR/CodeGen/vector.cpp
+++ b/clang/test/CIR/CodeGen/vector.cpp
@@ -6,6 +6,7 @@
 // RUN: FileCheck --input-file=%t.ll %s -check-prefix=OGCG
 
 typedef int vi4 __attribute__((vector_size(16)));
+typedef unsigned int uvi4 __attribute__((vector_size(16)));
 typedef double vd2 __attribute__((vector_size(16)));
 typedef long long vll2 __attribute__((vector_size(16)));
 
@@ -394,19 +395,9 @@ void foo9() {
 // CIR: %[[VEC_B:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["b", init]
 // CIR: %[[SHL_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shl", init]
 // CIR: %[[SHR_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shr", init]
-// CIR: %[[CONST_1:.*]] = cir.const #cir.int<1> : !s32i
-// CIR: %[[CONST_2:.*]] = cir.const #cir.int<2> : !s32i
-// CIR: %[[CONST_3:.*]] = cir.const #cir.int<3> : !s32i
-// CIR: %[[CONST_4:.*]] = cir.const #cir.int<4> : !s32i
-// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%[[CONST_1]], %[[CONST_2]], %[[CONST_3]], %[[CONST_4]] :
-// CIR-SAME: !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} :  !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
 // CIR: cir.store %[[VEC_A_VAL]], %[[VEC_A]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
-// CIR: %[[CONST_5:.*]] = cir.const #cir.int<5> : !s32i
-// CIR: %[[CONST_6:.*]] = cir.const #cir.int<6> : !s32i
-// CIR: %[[CONST_7:.*]] = cir.const #cir.int<7> : !s32i
-// CIR: %[[CONST_8:.*]] = cir.const #cir.int<8> : !s32i
-// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%[[CONST_5]], %[[CONST_6]], %[[CONST_7]], %[[CONST_8]] :
-// CIR-SAME: !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
 // CIR: cir.store %[[VEC_B_VAL]], %[[VEC_B]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
 // CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
 // CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
@@ -447,6 +438,61 @@ void foo9() {
 // OGCG: %[[SHR:.*]] = ashr <4 x i32> %[[TMP_A]], %[[TMP_B]]
 // OGCG: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
 
+void foo10() {
+  vi4 a = {1, 2, 3, 4};
+  uvi4 b = {5u, 6u, 7u, 8u};
+
+  vi4 shl = a << b;
+  uvi4 shr = b >> a;
+}
+
+// CIR: %[[VEC_A:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["a", init]
+// CIR: %[[VEC_B:.*]] = cir.alloca !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>, ["b", init]
+// CIR: %[[SHL_RES:.*]] = cir.alloca !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>, ["shl", init]
+// CIR: %[[SHR_RES:.*]] = cir.alloca !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>, ["shr", init]
+// CIR: %[[VEC_A_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : !s32i, !s32i, !s32i, !s32i) : !cir.vector<4 x !s32i>
+// CIR: cir.store %[[VEC_A_VAL]], %[[VEC_A]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR: %[[VEC_B_VAL:.*]] = cir.vec.create(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} :  !u32i, !u32i, !u32i, !u32i) : !cir.vector<4 x !u32i>
+// CIR: cir.store %[[VEC_B_VAL]], %[[VEC_B]] : !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>
+// CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR: %[[SHL:.*]] = cir.shift(left, %[[TMP_A]] : !cir.vector<4 x !s32i>, %[[TMP_B]] : !cir.vector<4 x !u32i>) -> !cir.vector<4 x !s32i>
+// CIR: cir.store %[[SHL]], %[[SHL_RES]] : !cir.vector<4 x !s32i>, !cir.ptr<!cir.vector<4 x !s32i>>
+// CIR: %[[TMP_B:.*]] = cir.load %[[VEC_B]] : !cir.ptr<!cir.vector<4 x !u32i>>, !cir.vector<4 x !u32i>
+// CIR: %[[TMP_A:.*]] = cir.load %[[VEC_A]] : !cir.ptr<!cir.vector<4 x !s32i>>, !cir.vector<4 x !s32i>
+// CIR: %[[SHR:.*]] = cir.shift(right, %[[TMP_B]] : !cir.vector<4 x !u32i>, %[[TMP_A]] : !cir.vector<4 x !s32i>) -> !cir.vector<4 x !u32i>
+// CIR: cir.store %[[SHR]], %[[SHR_RES]] : !cir.vector<4 x !u32i>, !cir.ptr<!cir.vector<4 x !u32i>>
+
+// LLVM: %[[VEC_A:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[VEC_B:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[SHL_RES:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: %[[SHR_RES:.*]] = alloca <4 x i32>, i64 1, align 16
+// LLVM: store <4 x i32> <i32 1, i32 2, i32 3, i32 4>, ptr %[[VEC_A]], align 16
+// LLVM: store <4 x i32> <i32 5, i32 6, i32 7, i32 8>, ptr %[[VEC_B]], align 16
+// LLVM: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// LLVM: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// LLVM: %[[SHL:.*]] = shl <4 x i32> %[[TMP_A]], %[[TMP_B]]
+// LLVM: store <4 x i32> %[[SHL]], ptr %[[SHL_RES]], align 16
+// LLVM: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// LLVM: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// LLVM: %[[SHR:.*]] = lshr <4 x i32> %[[TMP_B]], %[[TMP_A]]
+// LLVM: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
+
+// OGCG: %[[VEC_A:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[VEC_B:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[SHL_RES:.*]] = alloca <4 x i32>, align 16
+// OGCG: %[[SHR_RES:.*]] = alloca <4 x i32>, align 16
+// OGCG: store <4 x i32> <i32 1, i32 2, i32 3, i32 4>, ptr %[[VEC_A]], align 16
+// OGCG: store <4 x i32> <i32 5, i32 6, i32 7, i32 8>, ptr %[[VEC_B]], align 16
+// OGCG: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// OGCG: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// OGCG: %[[SHL:.*]] = shl <4 x i32> %[[TMP_A]], %[[TMP_B]]
+// OGCG: store <4 x i32> %[[SHL]], ptr %[[SHL_RES]], align 16
+// OGCG: %[[TMP_B:.*]] = load <4 x i32>, ptr %[[VEC_B]], align 16
+// OGCG: %[[TMP_A:.*]] = load <4 x i32>, ptr %[[VEC_A]], align 16
+// OGCG: %[[SHR:.*]] = lshr <4 x i32> %[[TMP_B]], %[[TMP_A]]
+// OGCG: store <4 x i32> %[[SHR]], ptr %[[SHR_RES]], align 16
+
 void foo11() {
   vi4 a = {1, 2, 3, 4};
   vi4 b = {5, 6, 7, 8};

xlauko
xlauko approved these changes May 22, 2025
Comment on lines 1436 to 1437
return emitOpError() << "shift amount must have the type of the result "
<< "if it is vector shift";
Copy link
Contributor

@xlauko xlauko May 22, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The message still states that result type needs to match amount type, what is not true now.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you, message updated, and I added a test for it

@xlauko xlauko self-requested a review May 22, 2025 19:06
Copy link
Contributor

@xlauko xlauko left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

if (isOp1Vec && op->getOperand(1).getType() != resType) {
return emitOpError() << "shift amount must have the type of the result "

if (isOp1Vec && !mlir::isa<cir::VectorType>(getResult().getType())) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is the requirement that the operands and the result must have the same size elements enforced somewhere else?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I tested a case with a different size, and there is no guard for that case. I will add a requirement for that

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I added requirements for element type, width, and vector size because now we don't compare the vector types, I think if we will need this more than one time we can create a function to compare vectors without comparing element type isSigned

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If there is no other comment, I can merge it @andykaylor

@AmrDeveloper AmrDeveloper force-pushed the cir_improve_shift_op_verify branch from 23936ed to 413f73e Compare May 23, 2025 20:31
Copy link
Member

@bcardosolopes bcardosolopes left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@AmrDeveloper AmrDeveloper force-pushed the cir_improve_shift_op_verify branch from 413f73e to e23fdd2 Compare May 24, 2025 11:42
bcardosolopes pushed a commit to llvm/clangir that referenced this pull request May 27, 2025
Copy link
Contributor

@andykaylor andykaylor left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

@AmrDeveloper AmrDeveloper merged commit 6e5f9bb into llvm:main May 28, 2025
11 checks passed
sivan-shani pushed a commit to sivan-shani/llvm-project that referenced this pull request Jun 3, 2025
…1111)

Update the verification of ShiftOp for Vector to allow performing shift
op between signed and unsigned integers, similar to LLVM IR

Issue llvm#136487
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
clang Clang issues not falling into any other category ClangIR Anything related to the ClangIR project
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants