<div dir="ltr">This change is causing ASan failures on the sanitizer bots: <a href="http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/21898/steps/check-clang%20asan/logs/stdio">http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/21898/steps/check-clang%20asan/logs/stdio</a><div><br></div><div>I've reverted it in r338904.</div></div><br><div class="gmail_quote"><div dir="ltr">On Fri, Aug 3, 2018 at 8:51 AM Scott Linder via cfe-commits <<a href="mailto:cfe-commits@lists.llvm.org">cfe-commits@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Author: scott.linder<br>
Date: Fri Aug 3 08:50:52 2018<br>
New Revision: 338899<br>
<br>
URL: <a href="http://llvm.org/viewvc/llvm-project?rev=338899&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project?rev=338899&view=rev</a><br>
Log:<br>
[OpenCL] Always emit alloca in entry block for enqueue_kernel builtin<br>
<br>
Ensures the statically sized alloca is not converted to DYNAMIC_STACKALLOC<br>
later because it is not in the entry block.<br>
<br>
Differential Revision: <a href="https://reviews.llvm.org/D50104" rel="noreferrer" target="_blank">https://reviews.llvm.org/D50104</a><br>
<br>
<br>
Added:<br>
cfe/trunk/test/CodeGenOpenCL/<a href="http://enqueue-kernel-non-entry-block.cl" rel="noreferrer" target="_blank">enqueue-kernel-non-entry-block.cl</a><br>
Modified:<br>
cfe/trunk/lib/CodeGen/CGBuiltin.cpp<br>
cfe/trunk/test/CodeGenOpenCL/<a href="http://cl20-device-side-enqueue.cl" rel="noreferrer" target="_blank">cl20-device-side-enqueue.cl</a><br>
<br>
Modified: cfe/trunk/lib/CodeGen/CGBuiltin.cpp<br>
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGBuiltin.cpp?rev=338899&r1=338898&r2=338899&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGBuiltin.cpp?rev=338899&r1=338898&r2=338899&view=diff</a><br>
==============================================================================<br>
--- cfe/trunk/lib/CodeGen/CGBuiltin.cpp (original)<br>
+++ cfe/trunk/lib/CodeGen/CGBuiltin.cpp Fri Aug 3 08:50:52 2018<br>
@@ -3338,23 +3338,29 @@ RValue CodeGenFunction::EmitBuiltinExpr(<br>
// Create a temporary array to hold the sizes of local pointer arguments<br>
// for the block. \p First is the position of the first size argument.<br>
auto CreateArrayForSizeVar = [=](unsigned First) {<br>
- auto *AT = llvm::ArrayType::get(SizeTy, NumArgs - First);<br>
- auto *Arr = Builder.CreateAlloca(AT);<br>
- llvm::Value *Ptr;<br>
+ llvm::APInt ArraySize(32, NumArgs - First);<br>
+ QualType SizeArrayTy = getContext().getConstantArrayType(<br>
+ getContext().getSizeType(), ArraySize, ArrayType::Normal,<br>
+ /*IndexTypeQuals=*/0);<br>
+ auto Tmp = CreateMemTemp(SizeArrayTy, "block_sizes");<br>
+ llvm::Value *TmpPtr = Tmp.getPointer();<br>
+ llvm::Value *TmpSize = EmitLifetimeStart(<br>
+ CGM.getDataLayout().getTypeAllocSize(Tmp.getElementType()), TmpPtr);<br>
+ llvm::Value *ElemPtr;<br>
// Each of the following arguments specifies the size of the corresponding<br>
// argument passed to the enqueued block.<br>
auto *Zero = llvm::ConstantInt::get(IntTy, 0);<br>
for (unsigned I = First; I < NumArgs; ++I) {<br>
auto *Index = llvm::ConstantInt::get(IntTy, I - First);<br>
- auto *GEP = Builder.CreateGEP(Arr, {Zero, Index});<br>
+ auto *GEP = Builder.CreateGEP(TmpPtr, {Zero, Index});<br>
if (I == First)<br>
- Ptr = GEP;<br>
+ ElemPtr = GEP;<br>
auto *V =<br>
Builder.CreateZExtOrTrunc(EmitScalarExpr(E->getArg(I)), SizeTy);<br>
Builder.CreateAlignedStore(<br>
V, GEP, CGM.getDataLayout().getPrefTypeAlignment(SizeTy));<br>
}<br>
- return Ptr;<br>
+ return std::tie(ElemPtr, TmpSize, TmpPtr);<br>
};<br>
<br>
// Could have events and/or varargs.<br>
@@ -3366,24 +3372,27 @@ RValue CodeGenFunction::EmitBuiltinExpr(<br>
llvm::Value *Kernel =<br>
Builder.CreatePointerCast(Info.Kernel, GenericVoidPtrTy);<br>
auto *Block = Builder.CreatePointerCast(Info.BlockArg, GenericVoidPtrTy);<br>
- auto *PtrToSizeArray = CreateArrayForSizeVar(4);<br>
+ llvm::Value *ElemPtr, *TmpSize, *TmpPtr;<br>
+ std::tie(ElemPtr, TmpSize, TmpPtr) = CreateArrayForSizeVar(4);<br>
<br>
// Create a vector of the arguments, as well as a constant value to<br>
// express to the runtime the number of variadic arguments.<br>
std::vector<llvm::Value *> Args = {<br>
Queue, Flags, Range,<br>
Kernel, Block, ConstantInt::get(IntTy, NumArgs - 4),<br>
- PtrToSizeArray};<br>
+ ElemPtr};<br>
std::vector<llvm::Type *> ArgTys = {<br>
- QueueTy, IntTy, RangeTy,<br>
- GenericVoidPtrTy, GenericVoidPtrTy, IntTy,<br>
- PtrToSizeArray->getType()};<br>
+ QueueTy, IntTy, RangeTy, GenericVoidPtrTy,<br>
+ GenericVoidPtrTy, IntTy, ElemPtr->getType()};<br>
<br>
llvm::FunctionType *FTy = llvm::FunctionType::get(<br>
Int32Ty, llvm::ArrayRef<llvm::Type *>(ArgTys), false);<br>
- return RValue::get(<br>
- Builder.CreateCall(CGM.CreateRuntimeFunction(FTy, Name),<br>
- llvm::ArrayRef<llvm::Value *>(Args)));<br>
+ auto Call =<br>
+ RValue::get(Builder.CreateCall(CGM.CreateRuntimeFunction(FTy, Name),<br>
+ llvm::ArrayRef<llvm::Value *>(Args)));<br>
+ if (TmpSize)<br>
+ EmitLifetimeEnd(TmpSize, TmpPtr);<br>
+ return Call;<br>
}<br>
// Any calls now have event arguments passed.<br>
if (NumArgs >= 7) {<br>
@@ -3430,15 +3439,19 @@ RValue CodeGenFunction::EmitBuiltinExpr(<br>
ArgTys.push_back(Int32Ty);<br>
Name = "__enqueue_kernel_events_varargs";<br>
<br>
- auto *PtrToSizeArray = CreateArrayForSizeVar(7);<br>
- Args.push_back(PtrToSizeArray);<br>
- ArgTys.push_back(PtrToSizeArray->getType());<br>
+ llvm::Value *ElemPtr, *TmpSize, *TmpPtr;<br>
+ std::tie(ElemPtr, TmpSize, TmpPtr) = CreateArrayForSizeVar(7);<br>
+ Args.push_back(ElemPtr);<br>
+ ArgTys.push_back(ElemPtr->getType());<br>
<br>
llvm::FunctionType *FTy = llvm::FunctionType::get(<br>
Int32Ty, llvm::ArrayRef<llvm::Type *>(ArgTys), false);<br>
- return RValue::get(<br>
- Builder.CreateCall(CGM.CreateRuntimeFunction(FTy, Name),<br>
- llvm::ArrayRef<llvm::Value *>(Args)));<br>
+ auto Call =<br>
+ RValue::get(Builder.CreateCall(CGM.CreateRuntimeFunction(FTy, Name),<br>
+ llvm::ArrayRef<llvm::Value *>(Args)));<br>
+ if (TmpSize)<br>
+ EmitLifetimeEnd(TmpSize, TmpPtr);<br>
+ return Call;<br>
}<br>
LLVM_FALLTHROUGH;<br>
}<br>
<br>
Modified: cfe/trunk/test/CodeGenOpenCL/<a href="http://cl20-device-side-enqueue.cl" rel="noreferrer" target="_blank">cl20-device-side-enqueue.cl</a><br>
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenOpenCL/cl20-device-side-enqueue.cl?rev=338899&r1=338898&r2=338899&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenOpenCL/cl20-device-side-enqueue.cl?rev=338899&r1=338898&r2=338899&view=diff</a><br>
==============================================================================<br>
--- cfe/trunk/test/CodeGenOpenCL/<a href="http://cl20-device-side-enqueue.cl" rel="noreferrer" target="_blank">cl20-device-side-enqueue.cl</a> (original)<br>
+++ cfe/trunk/test/CodeGenOpenCL/<a href="http://cl20-device-side-enqueue.cl" rel="noreferrer" target="_blank">cl20-device-side-enqueue.cl</a> Fri Aug 3 08:50:52 2018<br>
@@ -1,5 +1,6 @@<br>
// RUN: %clang_cc1 %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B32<br>
// RUN: %clang_cc1 %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B64<br>
+// RUN: %clang_cc1 %s -cl-std=CL2.0 -ffake-address-space-map -O1 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=CHECK-LIFETIMES<br>
<br>
#pragma OPENCL EXTENSION cl_khr_subgroups : enable<br>
<br>
@@ -46,8 +47,31 @@ kernel void device_side_enqueue(global i<br>
// COMMON: %event_wait_list2 = alloca [1 x %opencl.clk_event_t*]<br>
clk_event_t event_wait_list2[] = {clk_event};<br>
<br>
- // Emits block literal on stack and block kernel [[INVLK1]].<br>
// COMMON: [[NDR:%[a-z0-9]+]] = alloca %struct.ndrange_t, align 4<br>
+<br>
+ // B32: %[[BLOCK_SIZES1:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES1:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES1:.*]] = alloca [1 x i64]<br>
+ // B32: %[[BLOCK_SIZES2:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES2:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES2:.*]] = alloca [1 x i64]<br>
+ // B32: %[[BLOCK_SIZES3:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES3:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES3:.*]] = alloca [1 x i64]<br>
+ // B32: %[[BLOCK_SIZES4:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES4:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES4:.*]] = alloca [1 x i64]<br>
+ // B32: %[[BLOCK_SIZES5:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES5:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES5:.*]] = alloca [1 x i64]<br>
+ // B32: %[[BLOCK_SIZES6:.*]] = alloca [3 x i32]<br>
+ // B64: %[[BLOCK_SIZES6:.*]] = alloca [3 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES6:.*]] = alloca [3 x i64]<br>
+ // B32: %[[BLOCK_SIZES7:.*]] = alloca [1 x i32]<br>
+ // B64: %[[BLOCK_SIZES7:.*]] = alloca [1 x i64]<br>
+ // CHECK-LIFETIMES: %[[BLOCK_SIZES7:.*]] = alloca [1 x i64]<br>
+<br>
+ // Emits block literal on stack and block kernel [[INVLK1]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
// B32: [[BL:%[0-9]+]] = bitcast <{ i32, i32, i32 addrspace(1)*, i32, i32 addrspace(1)* }>* %block to void ()*<br>
@@ -73,7 +97,6 @@ kernel void device_side_enqueue(global i<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}}* addrspace(4)* [[WAIT_EVNT]], %opencl.clk_event_t{{.*}}* addrspace(4)* [[EVNT]],<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVLK2:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* [[BL_I8]])<br>
-<br>
enqueue_kernel(default_queue, flags, ndrange, 2, &event_wait_list, &clk_event,<br>
^(void) {<br>
a[i] = b[i];<br>
@@ -82,39 +105,46 @@ kernel void device_side_enqueue(global i<br>
// Emits global block literal [[BLG1]] and block kernel [[INVGK1]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 256, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 256, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES1]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES1]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES1]], i32 0, i32 0<br>
+ // B32: store i32 256, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES1]], i32 0, i32 0<br>
+ // B64: store i64 256, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_varargs(<br>
// COMMON-SAME: %opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{([0-9]+)?}},<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK1:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG1]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange,<br>
^(local void *p) {<br>
return;<br>
},<br>
256);<br>
+<br>
char c;<br>
// Emits global block literal [[BLG2]] and block kernel [[INVGK2]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES2]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES2]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES2]], i32 0, i32 0<br>
+ // B32: store i32 %{{.*}}, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES2]], i32 0, i32 0<br>
+ // B64: store i64 %{{.*}}, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_varargs(<br>
// COMMON-SAME: %opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{([0-9]+)?}},<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK2:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG2]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange,<br>
^(local void *p) {<br>
return;<br>
@@ -127,18 +157,21 @@ kernel void device_side_enqueue(global i<br>
// COMMON: [[AD:%arraydecay[0-9]*]] = getelementptr inbounds [1 x %opencl.clk_event_t*], [1 x %opencl.clk_event_t*]* %event_wait_list2, i32 0, i32 0<br>
// COMMON: [[WAIT_EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** [[AD]] to %opencl.clk_event_t{{.*}}* addrspace(4)*<br>
// COMMON: [[EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** %clk_event to %opencl.clk_event_t{{.*}}* addrspace(4)*<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 256, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 256, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES3]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES3]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_events_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES3]], i32 0, i32 0<br>
+ // B32: store i32 256, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES3]], i32 0, i32 0<br>
+ // B64: store i64 256, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_events_varargs<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}} [[WAIT_EVNT]], %opencl.clk_event_t{{.*}} [[EVNT]],<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK3:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG3]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange, 2, event_wait_list2, &clk_event,<br>
^(local void *p) {<br>
return;<br>
@@ -151,18 +184,21 @@ kernel void device_side_enqueue(global i<br>
// COMMON: [[AD:%arraydecay[0-9]*]] = getelementptr inbounds [1 x %opencl.clk_event_t*], [1 x %opencl.clk_event_t*]* %event_wait_list2, i32 0, i32 0<br>
// COMMON: [[WAIT_EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** [[AD]] to %opencl.clk_event_t{{.*}}* addrspace(4)*<br>
// COMMON: [[EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** %clk_event to %opencl.clk_event_t{{.*}}* addrspace(4)*<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES4]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES4]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_events_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES4]], i32 0, i32 0<br>
+ // B32: store i32 %{{.*}}, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES4]], i32 0, i32 0<br>
+ // B64: store i64 %{{.*}}, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_events_varargs<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}}* addrspace(4)* [[WAIT_EVNT]], %opencl.clk_event_t{{.*}}* addrspace(4)* [[EVNT]],<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK4:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG4]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange, 2, event_wait_list2, &clk_event,<br>
^(local void *p) {<br>
return;<br>
@@ -173,18 +209,21 @@ kernel void device_side_enqueue(global i<br>
// Emits global block literal [[BLG5]] and block kernel [[INVGK5]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES5]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES5]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES5]], i32 0, i32 0<br>
+ // B32: store i32 %{{.*}}, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES5]], i32 0, i32 0<br>
+ // B64: store i64 %{{.*}}, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_varargs<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{([0-9]+)?}},<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK5:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG5]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange,<br>
^(local void *p) {<br>
return;<br>
@@ -194,26 +233,29 @@ kernel void device_side_enqueue(global i<br>
// Emits global block literal [[BLG6]] and block kernel [[INVGK6]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
- // B32: %[[TMP:.*]] = alloca [3 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 1, i32* %[[TMP1]], align 4<br>
- // B32: %[[TMP2:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 1<br>
- // B32: store i32 2, i32* %[[TMP2]], align 4<br>
- // B32: %[[TMP3:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 2<br>
- // B32: store i32 4, i32* %[[TMP3]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [3 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 1, i64* %[[TMP1]], align 8<br>
- // B64: %[[TMP2:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 1<br>
- // B64: store i64 2, i64* %[[TMP2]], align 8<br>
- // B64: %[[TMP3:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 2<br>
- // B64: store i64 4, i64* %[[TMP3]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [3 x i64]* %[[BLOCK_SIZES6]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 24, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [3 x i64], [3 x i64]* %[[BLOCK_SIZES6]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 24, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [3 x i32], [3 x i32]* %[[BLOCK_SIZES6]], i32 0, i32 0<br>
+ // B32: store i32 1, i32* %[[TMP]], align 4<br>
+ // B32: %[[BLOCK_SIZES62:.*]] = getelementptr [3 x i32], [3 x i32]* %[[BLOCK_SIZES6]], i32 0, i32 1<br>
+ // B32: store i32 2, i32* %[[BLOCK_SIZES62]], align 4<br>
+ // B32: %[[BLOCK_SIZES63:.*]] = getelementptr [3 x i32], [3 x i32]* %[[BLOCK_SIZES6]], i32 0, i32 2<br>
+ // B32: store i32 4, i32* %[[BLOCK_SIZES63]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [3 x i64], [3 x i64]* %[[BLOCK_SIZES6]], i32 0, i32 0<br>
+ // B64: store i64 1, i64* %[[TMP]], align 8<br>
+ // B64: %[[BLOCK_SIZES62:.*]] = getelementptr [3 x i64], [3 x i64]* %[[BLOCK_SIZES6]], i32 0, i32 1<br>
+ // B64: store i64 2, i64* %[[BLOCK_SIZES62]], align 8<br>
+ // B64: %[[BLOCK_SIZES63:.*]] = getelementptr [3 x i64], [3 x i64]* %[[BLOCK_SIZES6]], i32 0, i32 2<br>
+ // B64: store i64 4, i64* %[[BLOCK_SIZES63]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_varargs<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{([0-9]+)?}},<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK6:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG6]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 3,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange,<br>
^(local void *p1, local void *p2, local void *p3) {<br>
return;<br>
@@ -223,18 +265,21 @@ kernel void device_side_enqueue(global i<br>
// Emits global block literal [[BLG7]] and block kernel [[INVGK7]].<br>
// COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t*, %opencl.queue_t** %default_queue<br>
// COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags<br>
- // B32: %[[TMP:.*]] = alloca [1 x i32]<br>
- // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0<br>
- // B32: store i32 0, i32* %[[TMP1]], align 4<br>
- // B64: %[[TMP:.*]] = alloca [1 x i64]<br>
- // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0<br>
- // B64: store i64 4294967296, i64* %[[TMP1]], align 8<br>
+ // CHECK-LIFETIMES: [[LIFETIME_PTR:%[0-9]+]] = bitcast [1 x i64]* %[[BLOCK_SIZES7]] to i8*<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // CHECK-LIFETIMES-NEXT: getelementptr inbounds [1 x i64], [1 x i64]* %[[BLOCK_SIZES7]], i64 0, i64 0<br>
+ // CHECK-LIFETIMES-LABEL: call i32 @__enqueue_kernel_varargs(<br>
+ // CHECK-LIFETIMES-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* nonnull [[LIFETIME_PTR]])<br>
+ // B32: %[[TMP:.*]] = getelementptr [1 x i32], [1 x i32]* %[[BLOCK_SIZES7]], i32 0, i32 0<br>
+ // B32: store i32 0, i32* %[[TMP]], align 4<br>
+ // B64: %[[TMP:.*]] = getelementptr [1 x i64], [1 x i64]* %[[BLOCK_SIZES7]], i32 0, i32 0<br>
+ // B64: store i64 4294967296, i64* %[[TMP]], align 8<br>
// COMMON-LABEL: call i32 @__enqueue_kernel_varargs<br>
// COMMON-SAME: (%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{([0-9]+)?}},<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8* bitcast ({{.*}} [[INVGK7:[^ ]+_kernel]] to i8*) to i8 addrspace(4)*),<br>
// COMMON-SAME: i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i32, i32 } addrspace(1)* [[BLG7]] to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1,<br>
- // B32-SAME: i32* %[[TMP1]])<br>
- // B64-SAME: i64* %[[TMP1]])<br>
+ // B32-SAME: i32* %[[TMP]])<br>
+ // B64-SAME: i64* %[[TMP]])<br>
enqueue_kernel(default_queue, flags, ndrange,<br>
^(local void *p) {<br>
return;<br>
<br>
Added: cfe/trunk/test/CodeGenOpenCL/<a href="http://enqueue-kernel-non-entry-block.cl" rel="noreferrer" target="_blank">enqueue-kernel-non-entry-block.cl</a><br>
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenOpenCL/enqueue-kernel-non-entry-block.cl?rev=338899&view=auto" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenOpenCL/enqueue-kernel-non-entry-block.cl?rev=338899&view=auto</a><br>
==============================================================================<br>
--- cfe/trunk/test/CodeGenOpenCL/<a href="http://enqueue-kernel-non-entry-block.cl" rel="noreferrer" target="_blank">enqueue-kernel-non-entry-block.cl</a> (added)<br>
+++ cfe/trunk/test/CodeGenOpenCL/<a href="http://enqueue-kernel-non-entry-block.cl" rel="noreferrer" target="_blank">enqueue-kernel-non-entry-block.cl</a> Fri Aug 3 08:50:52 2018<br>
@@ -0,0 +1,31 @@<br>
+// RUN: %clang_cc1 -cl-std=CL2.0 -O0 -emit-llvm -o - -triple amdgcn < %s | FileCheck %s --check-prefixes=COMMON,AMDGPU<br>
+// RUN: %clang_cc1 -cl-std=CL2.0 -O0 -emit-llvm -o - -triple "spir-unknown-unknown" < %s | FileCheck %s --check-prefixes=COMMON,SPIR32<br>
+// RUN: %clang_cc1 -cl-std=CL2.0 -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" < %s | FileCheck %s --check-prefixes=COMMON,SPIR64<br>
+// RUN: %clang_cc1 -cl-std=CL2.0 -O0 -debug-info-kind=limited -emit-llvm -o - -triple amdgcn < %s | FileCheck %s --check-prefixes=CHECK-DEBUG<br>
+<br>
+// Check that the enqueue_kernel array temporary is in the entry block to avoid<br>
+// a dynamic alloca<br>
+<br>
+typedef struct {int a;} ndrange_t;<br>
+<br>
+kernel void test(int i) {<br>
+// COMMON-LABEL: define {{.*}} void @test<br>
+// COMMON-LABEL: entry:<br>
+// AMDGPU: %block_sizes = alloca [1 x i64]<br>
+// SPIR32: %block_sizes = alloca [1 x i32]<br>
+// SPIR64: %block_sizes = alloca [1 x i64]<br>
+// COMMON-LABEL: if.then:<br>
+// COMMON-NOT: alloca<br>
+// CHECK-DEBUG: getelementptr {{.*}} %block_sizes, {{.*}} !dbg !34<br>
+// COMMON-LABEL: if.end<br>
+ queue_t default_queue;<br>
+ unsigned flags = 0;<br>
+ ndrange_t ndrange;<br>
+ if (i)<br>
+ enqueue_kernel(default_queue, flags, ndrange, ^(local void *a) { }, 32);<br>
+}<br>
+<br>
+// Check that the temporary is scoped to the `if`<br>
+<br>
+// CHECK-DEBUG: !32 = distinct !DILexicalBlock(scope: !7, file: !1, line: 24)<br>
+// CHECK-DEBUG: !34 = !DILocation(line: 25, scope: !32)<br>
<br>
<br>
_______________________________________________<br>
cfe-commits mailing list<br>
<a href="mailto:cfe-commits@lists.llvm.org" target="_blank">cfe-commits@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits</a><br>
</blockquote></div>