aboutsummaryrefslogtreecommitdiffstats
path: root/test/Transforms/DeadStoreElimination
diff options
context:
space:
mode:
Diffstat (limited to 'test/Transforms/DeadStoreElimination')
-rw-r--r--test/Transforms/DeadStoreElimination/2011-09-06-EndOfFunction.ll27
-rw-r--r--test/Transforms/DeadStoreElimination/2011-09-06-MemCpy.ll85
-rw-r--r--test/Transforms/DeadStoreElimination/atomic.ll107
-rw-r--r--test/Transforms/DeadStoreElimination/simple.ll8
4 files changed, 223 insertions, 4 deletions
diff --git a/test/Transforms/DeadStoreElimination/2011-09-06-EndOfFunction.ll b/test/Transforms/DeadStoreElimination/2011-09-06-EndOfFunction.ll
new file mode 100644
index 0000000..c5cc101
--- /dev/null
+++ b/test/Transforms/DeadStoreElimination/2011-09-06-EndOfFunction.ll
@@ -0,0 +1,27 @@
+; RUN: opt -dse -S < %s | FileCheck %s
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-apple-darwin"
+
+%"class.std::auto_ptr" = type { i32* }
+
+; CHECK: @_Z3foov
+define void @_Z3foov(%"class.std::auto_ptr"* noalias nocapture sret %agg.result) uwtable ssp {
+_ZNSt8auto_ptrIiED1Ev.exit:
+ %temp.lvalue = alloca %"class.std::auto_ptr", align 8
+ call void @_Z3barv(%"class.std::auto_ptr"* sret %temp.lvalue)
+ %_M_ptr.i.i = getelementptr inbounds %"class.std::auto_ptr"* %temp.lvalue, i64 0, i32 0
+ %tmp.i.i = load i32** %_M_ptr.i.i, align 8, !tbaa !0
+; CHECK-NOT: store i32* null
+ store i32* null, i32** %_M_ptr.i.i, align 8, !tbaa !0
+ %_M_ptr.i.i4 = getelementptr inbounds %"class.std::auto_ptr"* %agg.result, i64 0, i32 0
+ store i32* %tmp.i.i, i32** %_M_ptr.i.i4, align 8, !tbaa !0
+; CHECK: ret void
+ ret void
+}
+
+declare void @_Z3barv(%"class.std::auto_ptr"* sret)
+
+!0 = metadata !{metadata !"any pointer", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA", null}
diff --git a/test/Transforms/DeadStoreElimination/2011-09-06-MemCpy.ll b/test/Transforms/DeadStoreElimination/2011-09-06-MemCpy.ll
new file mode 100644
index 0000000..22b8786
--- /dev/null
+++ b/test/Transforms/DeadStoreElimination/2011-09-06-MemCpy.ll
@@ -0,0 +1,85 @@
+; RUN: opt -dse -S < %s | FileCheck %s
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-f128:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+%struct.pair.162 = type { %struct.BasicBlock*, i32, [4 x i8] }
+%struct.BasicBlock = type { %struct.Value, %struct.ilist_node.24, %struct.iplist.22, %struct.Function* }
+%struct.Value = type { i32 (...)**, i8, i8, i16, %struct.Type*, %struct.Use*, %struct.StringMapEntry* }
+%struct.Type = type { %struct.LLVMContext*, i8, [3 x i8], i32, {}* }
+%struct.LLVMContext = type { %struct.LLVMContextImpl* }
+%struct.LLVMContextImpl = type opaque
+%struct.Use = type { %struct.Value*, %struct.Use*, %struct.PointerIntPair }
+%struct.PointerIntPair = type { i64 }
+%struct.StringMapEntry = type opaque
+%struct.ilist_node.24 = type { %struct.ilist_half_node.23, %struct.BasicBlock* }
+%struct.ilist_half_node.23 = type { %struct.BasicBlock* }
+%struct.iplist.22 = type { %struct.ilist_traits.21, %struct.Instruction* }
+%struct.ilist_traits.21 = type { %struct.ilist_half_node.25 }
+%struct.ilist_half_node.25 = type { %struct.Instruction* }
+%struct.Instruction = type { [52 x i8], %struct.ilist_node.26, %struct.BasicBlock*, %struct.DebugLoc }
+%struct.ilist_node.26 = type { %struct.ilist_half_node.25, %struct.Instruction* }
+%struct.DebugLoc = type { i32, i32 }
+%struct.Function = type { %struct.GlobalValue, %struct.ilist_node.14, %struct.iplist.4, %struct.iplist, %struct.ValueSymbolTable*, %struct.AttrListPtr }
+%struct.GlobalValue = type <{ [52 x i8], [4 x i8], %struct.Module*, i8, i16, [5 x i8], %struct.basic_string }>
+%struct.Module = type { %struct.LLVMContext*, %struct.iplist.20, %struct.iplist.16, %struct.iplist.12, %struct.vector.2, %struct.ilist, %struct.basic_string, %struct.ValueSymbolTable*, %struct.OwningPtr, %struct.basic_string, %struct.basic_string, %struct.basic_string, i8* }
+%struct.iplist.20 = type { %struct.ilist_traits.19, %struct.GlobalVariable* }
+%struct.ilist_traits.19 = type { %struct.ilist_node.18 }
+%struct.ilist_node.18 = type { %struct.ilist_half_node.17, %struct.GlobalVariable* }
+%struct.ilist_half_node.17 = type { %struct.GlobalVariable* }
+%struct.GlobalVariable = type { %struct.GlobalValue, %struct.ilist_node.18, i8, [7 x i8] }
+%struct.iplist.16 = type { %struct.ilist_traits.15, %struct.Function* }
+%struct.ilist_traits.15 = type { %struct.ilist_node.14 }
+%struct.ilist_node.14 = type { %struct.ilist_half_node.13, %struct.Function* }
+%struct.ilist_half_node.13 = type { %struct.Function* }
+%struct.iplist.12 = type { %struct.ilist_traits.11, %struct.GlobalAlias* }
+%struct.ilist_traits.11 = type { %struct.ilist_node.10 }
+%struct.ilist_node.10 = type { %struct.ilist_half_node.9, %struct.GlobalAlias* }
+%struct.ilist_half_node.9 = type { %struct.GlobalAlias* }
+%struct.GlobalAlias = type { %struct.GlobalValue, %struct.ilist_node.10 }
+%struct.vector.2 = type { %struct._Vector_base.1 }
+%struct._Vector_base.1 = type { %struct._Vector_impl.0 }
+%struct._Vector_impl.0 = type { %struct.basic_string*, %struct.basic_string*, %struct.basic_string* }
+%struct.basic_string = type { %struct._Alloc_hider }
+%struct._Alloc_hider = type { i8* }
+%struct.ilist = type { %struct.iplist.8 }
+%struct.iplist.8 = type { %struct.ilist_traits.7, %struct.NamedMDNode* }
+%struct.ilist_traits.7 = type { %struct.ilist_node.6 }
+%struct.ilist_node.6 = type { %struct.ilist_half_node.5, %struct.NamedMDNode* }
+%struct.ilist_half_node.5 = type { %struct.NamedMDNode* }
+%struct.NamedMDNode = type { %struct.ilist_node.6, %struct.basic_string, %struct.Module*, i8* }
+%struct.ValueSymbolTable = type opaque
+%struct.OwningPtr = type { %struct.GVMaterializer* }
+%struct.GVMaterializer = type opaque
+%struct.iplist.4 = type { %struct.ilist_traits.3, %struct.BasicBlock* }
+%struct.ilist_traits.3 = type { %struct.ilist_half_node.23 }
+%struct.iplist = type { %struct.ilist_traits, %struct.Argument* }
+%struct.ilist_traits = type { %struct.ilist_half_node }
+%struct.ilist_half_node = type { %struct.Argument* }
+%struct.Argument = type { %struct.Value, %struct.ilist_node, %struct.Function* }
+%struct.ilist_node = type { %struct.ilist_half_node, %struct.Argument* }
+%struct.AttrListPtr = type { %struct.AttributeListImpl* }
+%struct.AttributeListImpl = type opaque
+
+declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
+
+; CHECK: _ZSt9iter_swapIPSt4pairIPN4llvm10BasicBlockEjES5_EvT_T0_
+; CHECK: store
+; CHECK: ret void
+define void @_ZSt9iter_swapIPSt4pairIPN4llvm10BasicBlockEjES5_EvT_T0_(%struct.pair.162* %__a, %struct.pair.162* %__b) nounwind uwtable inlinehint {
+entry:
+ %memtmp = alloca %struct.pair.162, align 8
+ %0 = getelementptr inbounds %struct.pair.162* %memtmp, i64 0, i32 0
+ %1 = getelementptr inbounds %struct.pair.162* %__a, i64 0, i32 0
+ %2 = load %struct.BasicBlock** %1, align 8
+ store %struct.BasicBlock* %2, %struct.BasicBlock** %0, align 8
+ %3 = getelementptr inbounds %struct.pair.162* %memtmp, i64 0, i32 1
+ %4 = getelementptr inbounds %struct.pair.162* %__a, i64 0, i32 1
+ %5 = load i32* %4, align 4
+ store i32 %5, i32* %3, align 8
+ %6 = bitcast %struct.pair.162* %__a to i8*
+ %7 = bitcast %struct.pair.162* %__b to i8*
+ call void @llvm.memcpy.p0i8.p0i8.i64(i8* %6, i8* %7, i64 12, i32 1, i1 false)
+ %8 = bitcast %struct.pair.162* %memtmp to i8*
+ call void @llvm.memcpy.p0i8.p0i8.i64(i8* %7, i8* %8, i64 12, i32 1, i1 false)
+ ret void
+}
diff --git a/test/Transforms/DeadStoreElimination/atomic.ll b/test/Transforms/DeadStoreElimination/atomic.ll
new file mode 100644
index 0000000..2e84298
--- /dev/null
+++ b/test/Transforms/DeadStoreElimination/atomic.ll
@@ -0,0 +1,107 @@
+; RUN: opt -basicaa -dse -S < %s | FileCheck %s
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-apple-macosx10.7.0"
+
+; Sanity tests for atomic stores.
+; Note that it turns out essentially every transformation DSE does is legal on
+; atomic ops, just some transformations are not allowed across them.
+
+@x = common global i32 0, align 4
+@y = common global i32 0, align 4
+
+declare void @randomop(i32*)
+
+; DSE across unordered store (allowed)
+define void @test1() nounwind uwtable ssp {
+; CHECK: test1
+; CHECK-NOT: store i32 0
+; CHECK: store i32 1
+entry:
+ store i32 0, i32* @x
+ store atomic i32 0, i32* @y unordered, align 4
+ store i32 1, i32* @x
+ ret void
+}
+
+; DSE across seq_cst load (allowed in theory; not implemented ATM)
+define i32 @test2() nounwind uwtable ssp {
+; CHECK: test2
+; CHECK: store i32 0
+; CHECK: store i32 1
+entry:
+ store i32 0, i32* @x
+ %x = load atomic i32* @y seq_cst, align 4
+ store i32 1, i32* @x
+ ret i32 %x
+}
+
+; DSE across seq_cst store (store before atomic store must not be removed)
+define void @test3() nounwind uwtable ssp {
+; CHECK: test3
+; CHECK: store i32
+; CHECK: store atomic i32 2
+entry:
+ store i32 0, i32* @x
+ store atomic i32 2, i32* @y seq_cst, align 4
+ store i32 1, i32* @x
+ ret void
+}
+
+; DSE remove unordered store (allowed)
+define void @test4() nounwind uwtable ssp {
+; CHECK: test4
+; CHECK-NOT: store atomic
+; CHECK: store i32 1
+entry:
+ store atomic i32 0, i32* @x unordered, align 4
+ store i32 1, i32* @x
+ ret void
+}
+
+; DSE unordered store overwriting non-atomic store (allowed)
+define void @test5() nounwind uwtable ssp {
+; CHECK: test5
+; CHECK: store atomic i32 1
+entry:
+ store i32 0, i32* @x
+ store atomic i32 1, i32* @x unordered, align 4
+ ret void
+}
+
+; DSE no-op unordered atomic store (allowed)
+define void @test6() nounwind uwtable ssp {
+; CHECK: test6
+; CHECK-NOT: store
+; CHECK: ret void
+entry:
+ %x = load atomic i32* @x unordered, align 4
+ store atomic i32 %x, i32* @x unordered, align 4
+ ret void
+}
+
+; DSE seq_cst store (be conservative; DSE doesn't have infrastructure
+; to reason about atomic operations).
+define void @test7() nounwind uwtable ssp {
+; CHECK: test7
+; CHECK: store atomic
+entry:
+ %a = alloca i32
+ store atomic i32 0, i32* %a seq_cst, align 4
+ ret void
+}
+
+; DSE and seq_cst load (be conservative; DSE doesn't have infrastructure
+; to reason about atomic operations).
+define i32 @test8() nounwind uwtable ssp {
+; CHECK: test8
+; CHECK: store
+; CHECK: load atomic
+entry:
+ %a = alloca i32
+ call void @randomop(i32* %a)
+ store i32 0, i32* %a, align 4
+ %x = load atomic i32* @x seq_cst, align 4
+ ret i32 %x
+}
+
diff --git a/test/Transforms/DeadStoreElimination/simple.ll b/test/Transforms/DeadStoreElimination/simple.ll
index 5f143fc..ec2f157 100644
--- a/test/Transforms/DeadStoreElimination/simple.ll
+++ b/test/Transforms/DeadStoreElimination/simple.ll
@@ -42,20 +42,20 @@ define i32 @test3(i32* %g_addr) nounwind {
define void @test4(i32* %Q) {
%a = load i32* %Q
- volatile store i32 %a, i32* %Q
+ store volatile i32 %a, i32* %Q
ret void
; CHECK: @test4
; CHECK-NEXT: load i32
-; CHECK-NEXT: volatile store
+; CHECK-NEXT: store volatile
; CHECK-NEXT: ret void
}
define void @test5(i32* %Q) {
- %a = volatile load i32* %Q
+ %a = load volatile i32* %Q
store i32 %a, i32* %Q
ret void
; CHECK: @test5
-; CHECK-NEXT: volatile load
+; CHECK-NEXT: load volatile
; CHECK-NEXT: ret void
}