aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorYonghong Song <yhs@fb.com>2019-10-08 18:23:17 +0000
committerYonghong Song <yhs@fb.com>2019-10-08 18:23:17 +0000
commit2eb3835e1cb4b13450a7a1c2bbe5e96904567612 (patch)
treed35eec42361442d33123f144af87bc5fabdc38d7
parentbd6eec4f5f95b5659e0bd30688b00290856b905a (diff)
[BPF] do compile-once run-everywhere relocation for bitfields
A bpf specific clang intrinsic is introduced: u32 __builtin_preserve_field_info(member_access, info_kind) Depending on info_kind, different information will be returned to the program. A relocation is also recorded for this builtin so that bpf loader can patch the instruction on the target host. This clang intrinsic is used to get certain information to facilitate struct/union member relocations. The offset relocation is extended by 4 bytes to include relocation kind. Currently supported relocation kinds are enum { FIELD_BYTE_OFFSET = 0, FIELD_BYTE_SIZE, FIELD_EXISTENCE, FIELD_SIGNEDNESS, FIELD_LSHIFT_U64, FIELD_RSHIFT_U64, }; for __builtin_preserve_field_info. The old access offset relocation is covered by FIELD_BYTE_OFFSET = 0. An example: struct s { int a; int b1:9; int b2:4; }; enum { FIELD_BYTE_OFFSET = 0, FIELD_BYTE_SIZE, FIELD_EXISTENCE, FIELD_SIGNEDNESS, FIELD_LSHIFT_U64, FIELD_RSHIFT_U64, }; void bpf_probe_read(void *, unsigned, const void *); int field_read(struct s *arg) { unsigned long long ull = 0; unsigned offset = __builtin_preserve_field_info(arg->b2, FIELD_BYTE_OFFSET); unsigned size = __builtin_preserve_field_info(arg->b2, FIELD_BYTE_SIZE); #ifdef USE_PROBE_READ bpf_probe_read(&ull, size, (const void *)arg + offset); unsigned lshift = __builtin_preserve_field_info(arg->b2, FIELD_LSHIFT_U64); #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ lshift = lshift + (size << 3) - 64; #endif #else switch(size) { case 1: ull = *(unsigned char *)((void *)arg + offset); break; case 2: ull = *(unsigned short *)((void *)arg + offset); break; case 4: ull = *(unsigned int *)((void *)arg + offset); break; case 8: ull = *(unsigned long long *)((void *)arg + offset); break; } unsigned lshift = __builtin_preserve_field_info(arg->b2, FIELD_LSHIFT_U64); #endif ull <<= lshift; if (__builtin_preserve_field_info(arg->b2, FIELD_SIGNEDNESS)) return (long long)ull >> __builtin_preserve_field_info(arg->b2, FIELD_RSHIFT_U64); return ull >> __builtin_preserve_field_info(arg->b2, FIELD_RSHIFT_U64); } There is a minor overhead for bpf_probe_read() on big endian. The code and relocation generated for field_read where bpf_probe_read() is used to access argument data on little endian mode: r3 = r1 r1 = 0 r1 = 4 <=== relocation (FIELD_BYTE_OFFSET) r3 += r1 r1 = r10 r1 += -8 r2 = 4 <=== relocation (FIELD_BYTE_SIZE) call bpf_probe_read r2 = 51 <=== relocation (FIELD_LSHIFT_U64) r1 = *(u64 *)(r10 - 8) r1 <<= r2 r2 = 60 <=== relocation (FIELD_RSHIFT_U64) r0 = r1 r0 >>= r2 r3 = 1 <=== relocation (FIELD_SIGNEDNESS) if r3 == 0 goto LBB0_2 r1 s>>= r2 r0 = r1 LBB0_2: exit Compare to the above code between relocations FIELD_LSHIFT_U64 and FIELD_LSHIFT_U64, the code with big endian mode has four more instructions. r1 = 41 <=== relocation (FIELD_LSHIFT_U64) r6 += r1 r6 += -64 r6 <<= 32 r6 >>= 32 r1 = *(u64 *)(r10 - 8) r1 <<= r6 r2 = 60 <=== relocation (FIELD_RSHIFT_U64) The code and relocation generated when using direct load. r2 = 0 r3 = 4 r4 = 4 if r4 s> 3 goto LBB0_3 if r4 == 1 goto LBB0_5 if r4 == 2 goto LBB0_6 goto LBB0_9 LBB0_6: # %sw.bb1 r1 += r3 r2 = *(u16 *)(r1 + 0) goto LBB0_9 LBB0_3: # %entry if r4 == 4 goto LBB0_7 if r4 == 8 goto LBB0_8 goto LBB0_9 LBB0_8: # %sw.bb9 r1 += r3 r2 = *(u64 *)(r1 + 0) goto LBB0_9 LBB0_5: # %sw.bb r1 += r3 r2 = *(u8 *)(r1 + 0) goto LBB0_9 LBB0_7: # %sw.bb5 r1 += r3 r2 = *(u32 *)(r1 + 0) LBB0_9: # %sw.epilog r1 = 51 r2 <<= r1 r1 = 60 r0 = r2 r0 >>= r1 r3 = 1 if r3 == 0 goto LBB0_11 r2 s>>= r1 r0 = r2 LBB0_11: # %sw.epilog exit Considering verifier is able to do limited constant propogation following branches. The following is the code actually traversed. r2 = 0 r3 = 4 <=== relocation r4 = 4 <=== relocation if r4 s> 3 goto LBB0_3 LBB0_3: # %entry if r4 == 4 goto LBB0_7 LBB0_7: # %sw.bb5 r1 += r3 r2 = *(u32 *)(r1 + 0) LBB0_9: # %sw.epilog r1 = 51 <=== relocation r2 <<= r1 r1 = 60 <=== relocation r0 = r2 r0 >>= r1 r3 = 1 if r3 == 0 goto LBB0_11 r2 s>>= r1 r0 = r2 LBB0_11: # %sw.epilog exit For native load case, the load size is calculated to be the same as the size of load width LLVM otherwise used to load the value which is then used to extract the bitfield value. Differential Revision: https://reviews.llvm.org/D67980 git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@374099 91177308-0d34-0410-b5e6-96231b3b80d8
-rw-r--r--include/clang/Basic/BuiltinsBPF.def24
-rw-r--r--include/clang/Basic/DiagnosticSemaKinds.td5
-rw-r--r--include/clang/Basic/TargetBuiltins.h10
-rw-r--r--include/clang/Sema/Sema.h1
-rw-r--r--include/clang/module.modulemap1
-rw-r--r--lib/Basic/Targets/BPF.cpp12
-rw-r--r--lib/Basic/Targets/BPF.h4
-rw-r--r--lib/CodeGen/CGBuiltin.cpp34
-rw-r--r--lib/CodeGen/CGExpr.cpp16
-rw-r--r--lib/CodeGen/CodeGenFunction.h1
-rw-r--r--lib/Sema/SemaChecking.cpp39
-rw-r--r--test/CodeGen/builtins-bpf-preserve-field-info-1.c35
-rw-r--r--test/CodeGen/builtins-bpf-preserve-field-info-2.c26
-rw-r--r--test/Sema/builtins-bpf.c48
14 files changed, 252 insertions, 4 deletions
diff --git a/include/clang/Basic/BuiltinsBPF.def b/include/clang/Basic/BuiltinsBPF.def
new file mode 100644
index 0000000000..bd96b9ef53
--- /dev/null
+++ b/include/clang/Basic/BuiltinsBPF.def
@@ -0,0 +1,24 @@
+//===--- BuiltinsBPF.def - BPF Builtin function database --------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines the BPF-specific builtin function database. Users of
+// this file must define the BUILTIN macro to make use of this information.
+//
+//===----------------------------------------------------------------------===//
+
+// The format of this database matches clang/Basic/Builtins.def.
+
+#if defined(BUILTIN) && !defined(TARGET_BUILTIN)
+# define TARGET_BUILTIN(ID, TYPE, ATTRS, FEATURE) BUILTIN(ID, TYPE, ATTRS)
+#endif
+
+// Get record field information.
+TARGET_BUILTIN(__builtin_preserve_field_info, "Ui.", "t", "")
+
+#undef BUILTIN
+#undef TARGET_BUILTIN
diff --git a/include/clang/Basic/DiagnosticSemaKinds.td b/include/clang/Basic/DiagnosticSemaKinds.td
index 23cedcab27..d0d1a7d4ea 100644
--- a/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/include/clang/Basic/DiagnosticSemaKinds.td
@@ -9954,6 +9954,11 @@ def err_builtin_launder_invalid_arg : Error<
"%select{non-pointer|function pointer|void pointer}0 argument to "
"'__builtin_launder' is not allowed">;
+def err_preserve_field_info_not_field : Error<
+ "__builtin_preserve_field_info argument %0 not a field access">;
+def err_preserve_field_info_not_const: Error<
+ "__builtin_preserve_field_info argument %0 not a constant">;
+
def err_bit_cast_non_trivially_copyable : Error<
"__builtin_bit_cast %select{source|destination}0 type must be trivially copyable">;
def err_bit_cast_type_size_mismatch : Error<
diff --git a/include/clang/Basic/TargetBuiltins.h b/include/clang/Basic/TargetBuiltins.h
index 50262fa310..0e2f0753b0 100644
--- a/include/clang/Basic/TargetBuiltins.h
+++ b/include/clang/Basic/TargetBuiltins.h
@@ -52,6 +52,16 @@ namespace clang {
};
}
+ /// BPF builtins
+ namespace BPF {
+ enum {
+ LastTIBuiltin = clang::Builtin::FirstTSBuiltin - 1,
+ #define BUILTIN(ID, TYPE, ATTRS) BI##ID,
+ #include "clang/Basic/BuiltinsBPF.def"
+ LastTSBuiltin
+ };
+ }
+
/// PPC builtins
namespace PPC {
enum {
diff --git a/include/clang/Sema/Sema.h b/include/clang/Sema/Sema.h
index 28180ed6d1..bb05a08e54 100644
--- a/include/clang/Sema/Sema.h
+++ b/include/clang/Sema/Sema.h
@@ -11056,6 +11056,7 @@ private:
bool CheckARMBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall);
bool CheckAArch64BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall);
+ bool CheckBPFBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall);
bool CheckHexagonBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall);
bool CheckHexagonBuiltinCpu(unsigned BuiltinID, CallExpr *TheCall);
bool CheckHexagonBuiltinArgument(unsigned BuiltinID, CallExpr *TheCall);
diff --git a/include/clang/module.modulemap b/include/clang/module.modulemap
index f7bf482fbb..0f9b7bb908 100644
--- a/include/clang/module.modulemap
+++ b/include/clang/module.modulemap
@@ -34,6 +34,7 @@ module Clang_Basic {
textual header "Basic/BuiltinsAArch64.def"
textual header "Basic/BuiltinsAMDGPU.def"
textual header "Basic/BuiltinsARM.def"
+ textual header "Basic/BuiltinsBPF.def"
textual header "Basic/Builtins.def"
textual header "Basic/BuiltinsHexagon.def"
textual header "Basic/BuiltinsLe64.def"
diff --git a/lib/Basic/Targets/BPF.cpp b/lib/Basic/Targets/BPF.cpp
index 0cf55a58a9..2fe2450b9a 100644
--- a/lib/Basic/Targets/BPF.cpp
+++ b/lib/Basic/Targets/BPF.cpp
@@ -13,11 +13,18 @@
#include "BPF.h"
#include "Targets.h"
#include "clang/Basic/MacroBuilder.h"
+#include "clang/Basic/TargetBuiltins.h"
#include "llvm/ADT/StringRef.h"
using namespace clang;
using namespace clang::targets;
+const Builtin::Info BPFTargetInfo::BuiltinInfo[] = {
+#define BUILTIN(ID, TYPE, ATTRS) \
+ {#ID, TYPE, ATTRS, nullptr, ALL_LANGUAGES, nullptr},
+#include "clang/Basic/BuiltinsBPF.def"
+};
+
void BPFTargetInfo::getTargetDefines(const LangOptions &Opts,
MacroBuilder &Builder) const {
Builder.defineMacro("__bpf__");
@@ -34,3 +41,8 @@ bool BPFTargetInfo::isValidCPUName(StringRef Name) const {
void BPFTargetInfo::fillValidCPUList(SmallVectorImpl<StringRef> &Values) const {
Values.append(std::begin(ValidCPUNames), std::end(ValidCPUNames));
}
+
+ArrayRef<Builtin::Info> BPFTargetInfo::getTargetBuiltins() const {
+ return llvm::makeArrayRef(BuiltinInfo, clang::BPF::LastTSBuiltin -
+ Builtin::FirstTSBuiltin);
+}
diff --git a/lib/Basic/Targets/BPF.h b/lib/Basic/Targets/BPF.h
index 79abd8828a..117f81430b 100644
--- a/lib/Basic/Targets/BPF.h
+++ b/lib/Basic/Targets/BPF.h
@@ -22,6 +22,8 @@ namespace clang {
namespace targets {
class LLVM_LIBRARY_VISIBILITY BPFTargetInfo : public TargetInfo {
+ static const Builtin::Info BuiltinInfo[];
+
public:
BPFTargetInfo(const llvm::Triple &Triple, const TargetOptions &)
: TargetInfo(Triple) {
@@ -54,7 +56,7 @@ public:
Features[Name] = Enabled;
}
- ArrayRef<Builtin::Info> getTargetBuiltins() const override { return None; }
+ ArrayRef<Builtin::Info> getTargetBuiltins() const override;
const char *getClobbers() const override { return ""; }
diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index 905db49747..d5389d74cb 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -4242,6 +4242,9 @@ static Value *EmitTargetArchBuiltinExpr(CodeGenFunction *CGF,
case llvm::Triple::aarch64:
case llvm::Triple::aarch64_be:
return CGF->EmitAArch64BuiltinExpr(BuiltinID, E, Arch);
+ case llvm::Triple::bpfeb:
+ case llvm::Triple::bpfel:
+ return CGF->EmitBPFBuiltinExpr(BuiltinID, E);
case llvm::Triple::x86:
case llvm::Triple::x86_64:
return CGF->EmitX86BuiltinExpr(BuiltinID, E);
@@ -9300,6 +9303,37 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
}
}
+Value *CodeGenFunction::EmitBPFBuiltinExpr(unsigned BuiltinID,
+ const CallExpr *E) {
+ assert(BuiltinID == BPF::BI__builtin_preserve_field_info &&
+ "unexpected ARM builtin");
+
+ const Expr *Arg = E->getArg(0);
+ bool IsBitField = Arg->IgnoreParens()->getObjectKind() == OK_BitField;
+
+ if (!getDebugInfo()) {
+ CGM.Error(E->getExprLoc(), "using builtin_preserve_field_info() without -g");
+ return IsBitField ? EmitLValue(Arg).getBitFieldPointer()
+ : EmitLValue(Arg).getPointer();
+ }
+
+ // Enable underlying preserve_*_access_index() generation.
+ bool OldIsInPreservedAIRegion = IsInPreservedAIRegion;
+ IsInPreservedAIRegion = true;
+ Value *FieldAddr = IsBitField ? EmitLValue(Arg).getBitFieldPointer()
+ : EmitLValue(Arg).getPointer();
+ IsInPreservedAIRegion = OldIsInPreservedAIRegion;
+
+ ConstantInt *C = cast<ConstantInt>(EmitScalarExpr(E->getArg(1)));
+ Value *InfoKind = ConstantInt::get(Int64Ty, C->getSExtValue());
+
+ // Built the IR for the preserve_field_info intrinsic.
+ llvm::Function *FnGetFieldInfo = llvm::Intrinsic::getDeclaration(
+ &CGM.getModule(), llvm::Intrinsic::bpf_preserve_field_info,
+ {FieldAddr->getType()});
+ return Builder.CreateCall(FnGetFieldInfo, {FieldAddr, InfoKind});
+}
+
llvm::Value *CodeGenFunction::
BuildVector(ArrayRef<llvm::Value*> Ops) {
assert((Ops.size() & (Ops.size() - 1)) == 0 &&
diff --git a/lib/CodeGen/CGExpr.cpp b/lib/CodeGen/CGExpr.cpp
index 1ea02677b4..2bd1b0ba7e 100644
--- a/lib/CodeGen/CGExpr.cpp
+++ b/lib/CodeGen/CGExpr.cpp
@@ -3990,9 +3990,19 @@ LValue CodeGenFunction::EmitLValueForField(LValue base,
const CGBitFieldInfo &Info = RL.getBitFieldInfo(field);
Address Addr = base.getAddress();
unsigned Idx = RL.getLLVMFieldNo(field);
- if (Idx != 0)
- // For structs, we GEP to the field that the record layout suggests.
- Addr = Builder.CreateStructGEP(Addr, Idx, field->getName());
+ if (!IsInPreservedAIRegion) {
+ if (Idx != 0)
+ // For structs, we GEP to the field that the record layout suggests.
+ Addr = Builder.CreateStructGEP(Addr, Idx, field->getName());
+ } else {
+ const RecordDecl *rec = field->getParent();
+ llvm::DIType *DbgInfo = getDebugInfo()->getOrCreateRecordType(
+ getContext().getRecordType(rec), rec->getLocation());
+ Addr = Builder.CreatePreserveStructAccessIndex(Addr, Idx,
+ getDebugInfoFIndex(rec, field->getFieldIndex()),
+ DbgInfo);
+ }
+
// Get the access type.
llvm::Type *FieldIntTy =
llvm::Type::getIntNTy(getLLVMContext(), Info.StorageSize);
diff --git a/lib/CodeGen/CodeGenFunction.h b/lib/CodeGen/CodeGenFunction.h
index 02007ab28a..ef16cd3800 100644
--- a/lib/CodeGen/CodeGenFunction.h
+++ b/lib/CodeGen/CodeGenFunction.h
@@ -3760,6 +3760,7 @@ public:
llvm::Value *vectorWrapScalar16(llvm::Value *Op);
llvm::Value *EmitAArch64BuiltinExpr(unsigned BuiltinID, const CallExpr *E,
llvm::Triple::ArchType Arch);
+ llvm::Value *EmitBPFBuiltinExpr(unsigned BuiltinID, const CallExpr *E);
llvm::Value *BuildVector(ArrayRef<llvm::Value*> Ops);
llvm::Value *EmitX86BuiltinExpr(unsigned BuiltinID, const CallExpr *E);
diff --git a/lib/Sema/SemaChecking.cpp b/lib/Sema/SemaChecking.cpp
index de8e1ef87a..07d3648dc9 100644
--- a/lib/Sema/SemaChecking.cpp
+++ b/lib/Sema/SemaChecking.cpp
@@ -1540,6 +1540,11 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID,
if (CheckAArch64BuiltinFunctionCall(BuiltinID, TheCall))
return ExprError();
break;
+ case llvm::Triple::bpfeb:
+ case llvm::Triple::bpfel:
+ if (CheckBPFBuiltinFunctionCall(BuiltinID, TheCall))
+ return ExprError();
+ break;
case llvm::Triple::hexagon:
if (CheckHexagonBuiltinFunctionCall(BuiltinID, TheCall))
return ExprError();
@@ -1940,6 +1945,40 @@ bool Sema::CheckAArch64BuiltinFunctionCall(unsigned BuiltinID,
return SemaBuiltinConstantArgRange(TheCall, i, l, u + l);
}
+bool Sema::CheckBPFBuiltinFunctionCall(unsigned BuiltinID,
+ CallExpr *TheCall) {
+ assert(BuiltinID == BPF::BI__builtin_preserve_field_info &&
+ "unexpected ARM builtin");
+
+ if (checkArgCount(*this, TheCall, 2))
+ return true;
+
+ // The first argument needs to be a record field access.
+ // If it is an array element access, we delay decision
+ // to BPF backend to check whether the access is a
+ // field access or not.
+ Expr *Arg = TheCall->getArg(0);
+ if (Arg->getType()->getAsPlaceholderType() ||
+ (Arg->IgnoreParens()->getObjectKind() != OK_BitField &&
+ !dyn_cast<MemberExpr>(Arg->IgnoreParens()) &&
+ !dyn_cast<ArraySubscriptExpr>(Arg->IgnoreParens()))) {
+ Diag(Arg->getBeginLoc(), diag::err_preserve_field_info_not_field)
+ << 1 << Arg->getSourceRange();
+ return true;
+ }
+
+ // The second argument needs to be a constant int
+ llvm::APSInt Value;
+ if (!TheCall->getArg(1)->isIntegerConstantExpr(Value, Context)) {
+ Diag(Arg->getBeginLoc(), diag::err_preserve_field_info_not_const)
+ << 2 << Arg->getSourceRange();
+ return true;
+ }
+
+ TheCall->setType(Context.UnsignedIntTy);
+ return false;
+}
+
bool Sema::CheckHexagonBuiltinCpu(unsigned BuiltinID, CallExpr *TheCall) {
struct BuiltinAndString {
unsigned BuiltinID;
diff --git a/test/CodeGen/builtins-bpf-preserve-field-info-1.c b/test/CodeGen/builtins-bpf-preserve-field-info-1.c
new file mode 100644
index 0000000000..a244bd1a18
--- /dev/null
+++ b/test/CodeGen/builtins-bpf-preserve-field-info-1.c
@@ -0,0 +1,35 @@
+// REQUIRES: bpf-registered-target
+// RUN: %clang -target bpf -emit-llvm -S -g %s -o - | FileCheck %s
+
+#define _(x, y) (__builtin_preserve_field_info((x), (y)))
+
+struct s1 {
+ char a;
+ char b:2;
+};
+
+union u1 {
+ char a;
+ char b:2;
+};
+
+unsigned unit1(struct s1 *arg) {
+ return _(arg->a, 10) + _(arg->b, 10);
+}
+// CHECK: define dso_local i32 @unit1
+// CHECK: call i8* @llvm.preserve.struct.access.index.p0i8.p0s_struct.s1s(%struct.s1* %{{[0-9a-z]+}}, i32 0, i32 0), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+// CHECK: call i8* @llvm.preserve.struct.access.index.p0i8.p0s_struct.s1s(%struct.s1* %{{[0-9a-z]+}}, i32 1, i32 1), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+
+unsigned unit2(union u1 *arg) {
+ return _(arg->a, 10) + _(arg->b, 10);
+}
+// CHECK: define dso_local i32 @unit2
+// CHECK: call %union.u1* @llvm.preserve.union.access.index.p0s_union.u1s.p0s_union.u1s(%union.u1* %{{[0-9a-z]+}}, i32 0), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[UNION_U1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+// CHECK: call i8* @llvm.preserve.struct.access.index.p0i8.p0s_union.u1s(%union.u1* %{{[0-9a-z]+}}, i32 0, i32 1), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[UNION_U1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+
+// CHECK: ![[STRUCT_S1]] = distinct !DICompositeType(tag: DW_TAG_structure_type, name: "s1"
+// CHECK: ![[UNION_U1]] = distinct !DICompositeType(tag: DW_TAG_union_type, name: "u1"
diff --git a/test/CodeGen/builtins-bpf-preserve-field-info-2.c b/test/CodeGen/builtins-bpf-preserve-field-info-2.c
new file mode 100644
index 0000000000..f50c9974d7
--- /dev/null
+++ b/test/CodeGen/builtins-bpf-preserve-field-info-2.c
@@ -0,0 +1,26 @@
+// REQUIRES: bpf-registered-target
+// RUN: %clang -target bpf -emit-llvm -S -g %s -o - | FileCheck %s
+
+#define _(x, y) (__builtin_preserve_field_info((x), (y)))
+
+struct s1 {
+ char a;
+ char b:2;
+};
+struct s2 {
+ struct s1 s;
+};
+
+unsigned unit1(struct s2 *arg) {
+ return _(arg->s.a, 10) + _(arg->s.b, 10);
+}
+// CHECK: define dso_local i32 @unit1
+// CHECK: call %struct.s1* @llvm.preserve.struct.access.index.p0s_struct.s1s.p0s_struct.s2s(%struct.s2* %{{[0-9a-z]+}}, i32 0, i32 0), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S2:[0-9]+]]
+// CHECK: call i8* @llvm.preserve.struct.access.index.p0i8.p0s_struct.s1s(%struct.s1* %{{[0-9a-z]+}}, i32 0, i32 0), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+// CHECK: call %struct.s1* @llvm.preserve.struct.access.index.p0s_struct.s1s.p0s_struct.s2s(%struct.s2* %{{[0-9a-z]+}}, i32 0, i32 0), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S2:[0-9]+]]
+// CHECK: call i8* @llvm.preserve.struct.access.index.p0i8.p0s_struct.s1s(%struct.s1* %{{[0-9a-z]+}}, i32 1, i32 1), !dbg !{{[0-9]+}}, !llvm.preserve.access.index ![[STRUCT_S1:[0-9]+]]
+// CHECK: call i32 @llvm.bpf.preserve.field.info.p0i8(i8* %{{[0-9a-z]+}}, i64 10), !dbg !{{[0-9]+}}
+
+// CHECK: ![[STRUCT_S2]] = distinct !DICompositeType(tag: DW_TAG_structure_type, name: "s2"
+// CHECK: ![[STRUCT_S1]] = distinct !DICompositeType(tag: DW_TAG_structure_type, name: "s1"
diff --git a/test/Sema/builtins-bpf.c b/test/Sema/builtins-bpf.c
new file mode 100644
index 0000000000..8df96976f1
--- /dev/null
+++ b/test/Sema/builtins-bpf.c
@@ -0,0 +1,48 @@
+// RUN: %clang_cc1 -x c -triple bpf-pc-linux-gnu -dwarf-version=4 -fsyntax-only -verify %s
+
+struct s { int a; int b[4]; int c:1; };
+union u { int a; int b[4]; int c:1; };
+
+unsigned invalid1(const int *arg) {
+ return __builtin_preserve_field_info(arg, 1); // expected-error {{__builtin_preserve_field_info argument 1 not a field access}}
+}
+
+unsigned invalid2(const int *arg) {
+ return __builtin_preserve_field_info(*arg, 1); // expected-error {{__builtin_preserve_field_info argument 1 not a field access}}
+}
+
+void *invalid3(struct s *arg) {
+ return __builtin_preserve_field_info(arg->a, 1); // expected-warning {{incompatible integer to pointer conversion returning 'unsigned int' from a function with result type 'void *'}}
+}
+
+unsigned valid4(struct s *arg) {
+ return __builtin_preserve_field_info(arg->b[1], 1);
+}
+
+unsigned valid5(union u *arg) {
+ return __builtin_preserve_field_info(arg->b[2], 1);
+}
+
+unsigned valid6(struct s *arg) {
+ return __builtin_preserve_field_info(arg->a, 1);
+}
+
+unsigned valid7(struct s *arg) {
+ return __builtin_preserve_field_info(arg->c, 1ULL);
+}
+
+unsigned valid8(union u *arg) {
+ return __builtin_preserve_field_info(arg->a, 1);
+}
+
+unsigned valid9(union u *arg) {
+ return __builtin_preserve_field_info(arg->c, 'a');
+}
+
+unsigned invalid10(struct s *arg) {
+ return __builtin_preserve_field_info(arg->a, arg); // expected-error {{__builtin_preserve_field_info argument 2 not a constant}}
+}
+
+unsigned invalid11(struct s *arg, int info_kind) {
+ return __builtin_preserve_field_info(arg->a, info_kind); // expected-error {{__builtin_preserve_field_info argument 2 not a constant}}
+}