[ConstantFold] Remove non-trivial gep-of-gep fold #93823
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
llvmbot
added
clang
|
@llvm/pr-subscribers-clang @llvm/pr-subscribers-llvm-ir Author: Nikita Popov (nikic) ChangesThis fold is subtly incorrect, because DL-unaware constant folding does not know the correct index type to use, and just performs the addition in the type that happens to already be there. This is incorrect, since sext(X)+sext(Y) is generally not the same as sext(X+Y). See the One could try to restrict the fold to cases where no overflow occurs, but I'm not bothering with that here, because the DL-aware constant folding will take care of this anyway. I've only kept the straightforward zero-index case, where we just concatenate two GEPs. Patch is 47.07 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/93823.diff 9 Files Affected:
diff --git a/clang/test/CodeGenCXX/2011-12-19-init-list-ctor.cpp b/clang/test/CodeGenCXX/2011-12-19-init-list-ctor.cpp
index 4033adc8f0390..6495fdebacf02 100644
--- a/clang/test/CodeGenCXX/2011-12-19-init-list-ctor.cpp
+++ b/clang/test/CodeGenCXX/2011-12-19-init-list-ctor.cpp
@@ -22,5 +22,5 @@ struct S {
// CHECK: call void @_ZN1AC1EPKc(ptr {{[^,]*}} getelementptr inbounds (%struct.S, ptr @arr, i32 0, i32 1), ptr noundef @.str)
// CHECK: store i32 1, ptr getelementptr inbounds (%struct.S, ptr @arr, i64 1)
// CHECK: call void @_ZN1AC1EPKc(ptr {{[^,]*}} getelementptr inbounds (%struct.S, ptr @arr, i64 1, i32 1), ptr noundef @.str.1)
-// CHECK: store i32 2, ptr getelementptr inbounds (%struct.S, ptr @arr, i64 2)
+// CHECK: store i32 2, ptr getelementptr inbounds (%struct.S, ptr getelementptr inbounds (%struct.S, ptr @arr, i64 1), i64 1)
// CHECK: call void @_ZN1AC1EPKc(ptr {{[^,]*}} getelementptr inbounds (%struct.S, ptr @arr, i64 2, i32 1), ptr noundef @.str.2)
diff --git a/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist-pr12086.cpp b/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist-pr12086.cpp
index 6fbe4c7fd17a7..4f4c88d9e96a5 100644
--- a/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist-pr12086.cpp
+++ b/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist-pr12086.cpp
@@ -83,8 +83,8 @@ std::initializer_list<std::initializer_list<int>> nested = {
// CHECK-DYNAMIC-BL: store i32 5, ptr @_ZGR6nested2_
// CHECK-DYNAMIC-BL: store i32 {{.*}}, ptr getelementptr inbounds (i32, ptr @_ZGR6nested2_, i64 1)
// CHECK-DYNAMIC-BL: store ptr @_ZGR6nested2_,
-// CHECK-DYNAMIC-BL: ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 2), align 8
-// CHECK-DYNAMIC-BL: store i64 2, ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 2, i32 1), align 8
+// CHECK-DYNAMIC-BL: ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 1), i64 1), align 8
+// CHECK-DYNAMIC-BL: store i64 2, ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 1), i64 1, i32 1), align 8
// CHECK-DYNAMIC-BL: store ptr @_ZGR6nested_,
// CHECK-DYNAMIC-BL: ptr @nested, align 8
// CHECK-DYNAMIC-BL: store i64 3, ptr getelementptr inbounds ({{.*}}, ptr @nested, i32 0, i32 1), align 8
@@ -123,9 +123,9 @@ std::initializer_list<std::initializer_list<int>> nested = {
// CHECK-DYNAMIC-BE: store i32 5, ptr @_ZGR6nested2_
// CHECK-DYNAMIC-BE: store i32 {{.*}}, ptr getelementptr inbounds (i32, ptr @_ZGR6nested2_, i64 1)
// CHECK-DYNAMIC-BE: store ptr @_ZGR6nested2_,
-// CHECK-DYNAMIC-BE: ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 2), align 8
+// CHECK-DYNAMIC-BE: ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 1), i64 1), align 8
// CHECK-DYNAMIC-BE: store ptr getelementptr inbounds ([2 x i32], ptr @_ZGR6nested2_, i64 0, i64 2),
-// CHECK-DYNAMIC-BE: ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 2, i32 1), align 8
+// CHECK-DYNAMIC-BE: ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr @_ZGR6nested_, i64 1), i64 1, i32 1), align 8
// CHECK-DYNAMIC-BE: store ptr @_ZGR6nested_,
// CHECK-DYNAMIC-BE: ptr @nested, align 8
// CHECK-DYNAMIC-BE: store ptr getelementptr inbounds ([3 x {{.*}}], ptr @_ZGR6nested_, i64 0, i64 3),
diff --git a/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist.cpp b/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist.cpp
index aa2f078a5fb0c..ac05910151a46 100644
--- a/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist.cpp
+++ b/clang/test/CodeGenCXX/cxx0x-initializer-stdinitializerlist.cpp
@@ -370,8 +370,8 @@ namespace partly_constant {
//
// Third init list.
// CHECK-NOT: @[[PARTLY_CONSTANT_THIRD]],
- // CHECK: store ptr {{.*}}@[[PARTLY_CONSTANT_THIRD]]{{.*}}, ptr getelementptr inbounds ({{.*}}, ptr {{.*}}@[[PARTLY_CONSTANT_INNER]]{{.*}}, i64 2)
- // CHECK: store i64 4, ptr getelementptr inbounds ({{.*}}, ptr {{.*}}@[[PARTLY_CONSTANT_INNER]]{{.*}}, i64 2, i32 1)
+ // CHECK: store ptr {{.*}}@[[PARTLY_CONSTANT_THIRD]]{{.*}}, ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr {{.*}}@[[PARTLY_CONSTANT_INNER]]{{.*}}, i64 1), i64 1)
+ // CHECK: store i64 4, ptr getelementptr inbounds ({{.*}}, ptr getelementptr inbounds ({{.*}}, ptr {{.*}}@[[PARTLY_CONSTANT_INNER]]{{.*}}, i64 1), i64 1, i32 1)
// CHECK-NOT: @[[PARTLY_CONSTANT_THIRD]],
//
// Outer init list.
diff --git a/clang/test/OpenMP/threadprivate_codegen.cpp b/clang/test/OpenMP/threadprivate_codegen.cpp
index b27783be829d6..7094decd41041 100644
--- a/clang/test/OpenMP/threadprivate_codegen.cpp
+++ b/clang/test/OpenMP/threadprivate_codegen.cpp
@@ -1260,8 +1260,8 @@ int foobar() {
// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// CHECK1-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]]
// CHECK1: invoke.cont2:
-// CHECK1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
-// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// CHECK1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
+// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// CHECK1-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]]
// CHECK1: invoke.cont3:
// CHECK1-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8
@@ -1273,8 +1273,8 @@ int foobar() {
// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// CHECK1-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]]
// CHECK1: invoke.cont8:
-// CHECK1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
-// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), i32 noundef 6)
+// CHECK1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
+// CHECK1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), i32 noundef 6)
// CHECK1-NEXT: to label [[INVOKE_CONT9:%.*]] unwind label [[LPAD6]]
// CHECK1: invoke.cont9:
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR3]]
@@ -1783,8 +1783,8 @@ int foobar() {
// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// CHECK2-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]]
// CHECK2: invoke.cont2:
-// CHECK2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
-// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// CHECK2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
+// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// CHECK2-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]]
// CHECK2: invoke.cont3:
// CHECK2-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8
@@ -1796,8 +1796,8 @@ int foobar() {
// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// CHECK2-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]]
// CHECK2: invoke.cont8:
-// CHECK2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
-// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), i32 noundef 6)
+// CHECK2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
+// CHECK2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), i32 noundef 6)
// CHECK2-NEXT: to label [[INVOKE_CONT9:%.*]] unwind label [[LPAD6]]
// CHECK2: invoke.cont9:
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR3]]
@@ -2458,8 +2458,8 @@ int foobar() {
// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// SIMD1-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]]
// SIMD1: invoke.cont2:
-// SIMD1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
-// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// SIMD1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
+// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// SIMD1-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]]
// SIMD1: invoke.cont3:
// SIMD1-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8
@@ -2471,8 +2471,8 @@ int foobar() {
// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// SIMD1-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]]
// SIMD1: invoke.cont8:
-// SIMD1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
-// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), i32 noundef 6)
+// SIMD1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
+// SIMD1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), i32 noundef 6)
// SIMD1-NEXT: to label [[INVOKE_CONT9:%.*]] unwind label [[LPAD6]]
// SIMD1: invoke.cont9:
// SIMD1-NEXT: [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR3]]
@@ -2922,8 +2922,8 @@ int foobar() {
// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// SIMD2-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]], !dbg [[DBG158:![0-9]+]]
// SIMD2: invoke.cont2:
-// SIMD2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8, !dbg [[DBG156]]
-// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// SIMD2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8, !dbg [[DBG156]]
+// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// SIMD2-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]], !dbg [[DBG159:![0-9]+]]
// SIMD2: invoke.cont3:
// SIMD2-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8, !dbg [[DBG154]]
@@ -2935,8 +2935,8 @@ int foobar() {
// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// SIMD2-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]], !dbg [[DBG162:![0-9]+]]
// SIMD2: invoke.cont8:
-// SIMD2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), ptr [[ARRAYINIT_ENDOFINIT5]], align 8, !dbg [[DBG160]]
-// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), i32 noundef 6)
+// SIMD2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT5]], align 8, !dbg [[DBG160]]
+// SIMD2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), i32 noundef 6)
// SIMD2-NEXT: to label [[INVOKE_CONT9:%.*]] unwind label [[LPAD6]], !dbg [[DBG163:![0-9]+]]
// SIMD2: invoke.cont9:
// SIMD2-NEXT: [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR3]], !dbg [[DBG164:![0-9]+]]
@@ -3451,8 +3451,8 @@ int foobar() {
// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// CHECK-TLS1-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]]
// CHECK-TLS1: invoke.cont2:
-// CHECK-TLS1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
-// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// CHECK-TLS1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
+// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// CHECK-TLS1-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]]
// CHECK-TLS1: invoke.cont3:
// CHECK-TLS1-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8
@@ -3464,8 +3464,8 @@ int foobar() {
// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// CHECK-TLS1-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]]
// CHECK-TLS1: invoke.cont8:
-// CHECK-TLS1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
-// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 2), i32 noundef 6)
+// CHECK-TLS1-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT5]], align 8
+// CHECK-TLS1-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i64 1), i32 noundef 6)
// CHECK-TLS1-NEXT: to label [[INVOKE_CONT9:%.*]] unwind label [[LPAD6]]
// CHECK-TLS1: invoke.cont9:
// CHECK-TLS1-NEXT: [[TMP0:%.*]] = call i32 @__cxa_thread_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR3]]
@@ -4178,8 +4178,8 @@ int foobar() {
// CHECK-TLS2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i32 noundef 2)
// CHECK-TLS2-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD]]
// CHECK-TLS2: invoke.cont2:
-// CHECK-TLS2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
-// CHECK-TLS2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 2), i32 noundef 3)
+// CHECK-TLS2-NEXT: store ptr getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), ptr [[ARRAYINIT_ENDOFINIT1]], align 8
+// CHECK-TLS2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([[STRUCT_S1]], ptr @arr_x, i64 1), i64 1), i32 noundef 3)
// CHECK-TLS2-NEXT: to label [[INVOKE_CONT3:%.*]] unwind label [[LPAD]]
// CHECK-TLS2: invoke.cont3:
// CHECK-TLS2-NEXT: store ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), ptr [[ARRAYINIT_ENDOFINIT]], align 8
@@ -4191,8 +4191,8 @@ int foobar() {
// CHECK-TLS2-NEXT: invoke void @_ZN2S1C1Ei(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S1]], ptr getelementptr inbounds ([3 x %struct.S1], ptr @arr_x, i64 1), i64 1), i32 noundef 5)
// CHECK-TLS2-NEXT: to label [[INVOKE_CONT8:%.*]] unwind label [[LPAD6]]
// CHECK-TLS2: invoke.con...
[truncated]
|
dtcxzyw
reviewed
May 30, 2024
dtcxzyw
added a commit
to dtcxzyw/llvm-opt-benchmark
that referenced
this pull request
May 30, 2024
It looks like we don't have an LLVM test, but it affects a few clang tests: I kind of expected more clang fallout, so I'd be open to just dropping this fold from the DL-unaware folder completely, even the basic case. |
aeubanks
approved these changes
May 30, 2024
This fold is subtly incorrect, because DL-unaware constant folding does not know the correct index type to use, and just performs the addition in the type that happens to already be there. This is incorrect, since sext(X)+sext(Y) is generally not the same as sext(X+Y). See the `@constexpr_gep_of_gep_with_narrow_type()` for a miscompile with the current implementation. One could try to restrict the fold to cases where no overflow occurs, but I'm not bothering with that here, because the DL-aware constant folding will take care of this anyway. I've only kept the straightforward zero-index case, where we just concatenate two GEPs.
nikic
added a commit
that referenced
this pull request
May 31, 2024
This reverts commit e1cc9e4. This causes some non-trivial text size increases in unoptimized builds for Bullet. Revert while I investigate.
nikic
added a commit
to nikic/llvm-project
that referenced
this pull request
May 31, 2024
This makes codegen for array initialization simpler in two ways: 1. Drop the zero-index GEP at the start, which is no longer needed with opaque pointers. 2. Emit GEPs directly to the correct element, instead of having a long chain of +1 GEPs. This is more canonical, and also avoids regressions in unoptimized builds from llvm#93823.
nikic
added a commit
that referenced
this pull request
Jun 10, 2024
This makes codegen for array initialization simpler in two ways: 1. Drop the zero-index GEP at the start, which is no longer needed with opaque pointers. 2. Emit GEPs directly to the correct element, instead of having a long chain of +1 GEPs. This is more canonical, and also avoids regressions in unoptimized builds from #93823.
nikic
added a commit
to nikic/llvm-project
that referenced
this pull request
Jun 10, 2024
This fold is subtly incorrect, because DL-unaware constant folding does not know the correct index type to use, and just performs the addition in the type that happens to already be there. This is incorrect, since sext(X)+sext(Y) is generally not the same as sext(X+Y). See the `@constexpr_gep_of_gep_with_narrow_type()` for a miscompile with the current implementation. One could try to restrict the fold to cases where no overflow occurs, but I'm not bothering with that here, because the DL-aware constant folding will take care of this anyway. I've only kept the straightforward zero-index case, where we just concatenate two GEPs.
nikic
added a commit
that referenced
this pull request
Jun 10, 2024
Reapply after #93956, which changed clang array initialization codegen to avoid size regressions for unoptimized builds. ----- This fold is subtly incorrect, because DL-unaware constant folding does not know the correct index type to use, and just performs the addition in the type that happens to already be there. This is incorrect, since sext(X)+sext(Y) is generally not the same as sext(X+Y). See the `@constexpr_gep_of_gep_with_narrow_type()` for a miscompile with the current implementation. One could try to restrict the fold to cases where no overflow occurs, but I'm not bothering with that here, because the DL-aware constant folding will take care of this anyway. I've only kept the straightforward zero-index case, where we just concatenate two GEPs.
nikic
added a commit
to nikic/llvm-project
that referenced
this pull request
Jun 11, 2024
This is a followup to llvm#93823 and drop the DataLayout-unaware GEP of GEP fold entirely. All cases are now left to the DataLayout-aware constant folder, which will fold everything to a single i8 GEP. We didn't have any test coverage for this fold in LLVM, but some Clang tests change.
nikic
added a commit
that referenced
this pull request
Jun 12, 2024
This is a followup to #93823 and drops the DataLayout-unaware GEP of GEP fold entirely. All cases are now left to the DataLayout-aware constant folder, which will fold everything to a single i8 GEP. We didn't have any test coverage for this fold in LLVM, but some Clang tests change.
Closed
nikic
added a commit
that referenced
this pull request
Jun 13, 2024
Reapplying without changes. The flang+openmp buildbot failure should be addressed by #94541. ----- This is a followup to #93823 and drops the DataLayout-unaware GEP of GEP fold entirely. All cases are now left to the DataLayout-aware constant folder, which will fold everything to a single i8 GEP. We didn't have any test coverage for this fold in LLVM, but some Clang tests change.
EthanLuisMcDonough
pushed a commit
to EthanLuisMcDonough/llvm-project
that referenced
this pull request
Aug 13, 2024
Reapplying without changes. The flang+openmp buildbot failure should be addressed by llvm#94541. ----- This is a followup to llvm#93823 and drops the DataLayout-unaware GEP of GEP fold entirely. All cases are now left to the DataLayout-aware constant folder, which will fold everything to a single i8 GEP. We didn't have any test coverage for this fold in LLVM, but some Clang tests change.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This fold is subtly incorrect, because DL-unaware constant folding does not know the correct index type to use, and just performs the addition in the type that happens to already be there. This is incorrect, since sext(X)+sext(Y) is generally not the same as sext(X+Y). See the
@constexpr_gep_of_gep_with_narrow_type()for a miscompile with the current implementation.One could try to restrict the fold to cases where no overflow occurs, but I'm not bothering with that here, because the DL-aware constant folding will take care of this anyway. I've only kept the straightforward zero-index case, where we just concatenate two GEPs.