Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis closed this revision. eugenis added a comment. http://llvm.org/viewvc/llvm-project?rev=259716=rev Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis added inline comments. Comment at: lib/CodeGen/CGClass.cpp:2608 @@ -2607,3 +2607,3 @@ auto TypeId = CGM.CreateCfiIdForTypeMetadata(MD); if (CGM.getCodeGenOpts().SanitizeCfiCrossDso && TypeId) { EmitCfiSlowPathCheck(M, BitSetTest, TypeId, CastedVTable, StaticData); samsonov wrote: > Can we rewrite this as if-elseif-else block now? even better, with 2 early returns. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis updated this revision to Diff 46807. Repository: rL LLVM http://reviews.llvm.org/D16823 Files: lib/CodeGen/CGClass.cpp lib/CodeGen/CGExpr.cpp lib/CodeGen/CodeGenModule.cpp lib/CodeGen/CodeGenModule.h test/CodeGen/cfi-check-fail.c test/CodeGenCXX/cfi-cast.cpp test/CodeGenCXX/cfi-vcall.cpp Index: test/CodeGenCXX/cfi-vcall.cpp === --- test/CodeGenCXX/cfi-vcall.cpp +++ test/CodeGenCXX/cfi-vcall.cpp @@ -1,5 +1,5 @@ -// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=NDIAG %s -// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=DIAG --check-prefix=DIAG-ABORT %s +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=ITANIUM-NDIAG --check-prefix=NDIAG %s +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=ITANIUM-DIAG --check-prefix=DIAG --check-prefix=DIAG-ABORT %s // RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-recover=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=DIAG --check-prefix=DIAG-RECOVER %s // RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=MS --check-prefix=NDIAG %s @@ -55,23 +55,25 @@ // DIAG: @[[SRC:.*]] = private unnamed_addr constant [{{.*}} x i8] c"{{.*}}cfi-vcall.cpp\00", align 1 // DIAG: @[[TYPE:.*]] = private unnamed_addr constant { i16, i16, [4 x i8] } { i16 -1, i16 0, [4 x i8] c"'A'\00" } -// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+21]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } +// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+23]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } // ITANIUM: define void @_Z2afP1A // MS: define void @"\01?af@@YAXPEAUA@@@Z" void af(A *a) { // ITANIUM: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"_ZTS1A") // MS: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"?AUA@@") + // DIAG-NEXT: [[VTVALID0:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT]], metadata !"all-vtables") // CHECK-NEXT: br i1 [[P]], label %[[CONTBB:[^ ,]*]], label %[[TRAPBB:[^ ,]*]] // CHECK-NEXT: {{^$}} // CHECK: [[TRAPBB]] // NDIAG-NEXT: call void @llvm.trap() // NDIAG-NEXT: unreachable // DIAG-NEXT: [[VTINT:%[^ ]*]] = ptrtoint i8* [[VT]] to i64 - // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-NEXT: [[VTVALID:%[^ ]*]] = zext i1 [[VTVALID0]] to i64 + // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-ABORT-NEXT: unreachable - // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-RECOVER-NEXT: br label %[[CONTBB]] // CHECK: [[CONTBB]] @@ -157,32 +159,47 @@ } -// Check for the expected number of elements (9 or 15 respectively). -// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){8}]]} -// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// Check for the expected number of elements (15 or 23 respectively). +// MS-NDIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){9}]]} +// MS-DIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){15}]]} +// ITANIUM-NDIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// ITANIUM-DIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){23}]]} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTV1A, i64 16} +// ITANIUM-DIAG-DAG: !{!"all-vtables", [3 x i8*]* @_ZTV1A, i64 16} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} +// ITANIUM-DIAG-DAG: !{!"all-vtables", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} +// ITANIUM-DIAG-DAG: !{!"all-vtables", [9 x i8*]*
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
samsonov accepted this revision. samsonov added a reviewer: samsonov. samsonov added a comment. This revision is now accepted and ready to land. LGTM Comment at: lib/CodeGen/CGClass.cpp:2608 @@ -2607,3 +2607,3 @@ auto TypeId = CGM.CreateCfiIdForTypeMetadata(MD); if (CGM.getCodeGenOpts().SanitizeCfiCrossDso && TypeId) { EmitCfiSlowPathCheck(M, BitSetTest, TypeId, CastedVTable, StaticData); Can we rewrite this as if-elseif-else block now? Comment at: lib/CodeGen/CGExpr.cpp:2642 @@ +2641,3 @@ + llvm::MDString::get(CGM.getLLVMContext(), "all-vtables")); + llvm::Value *ValidVtable = Builder.CreateZExt( + Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::bitset_test), eugenis wrote: > samsonov wrote: > > This is almost the same as EmitVTablePtrCheck, but with ZExt? Is the > > difference intentional/important? Is it possible to extract this logic > > (getting "all-vtables" metadata and running bitset test) to a function? > Not important. Zext makes the test a bit simpler. > Extracting these two lines to a function is surely possible, but is it worth > it? > Probably not, as long as we have the test coverage. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis created this revision. eugenis added reviewers: pcc, kcc. eugenis added a subscriber: cfe-commits. eugenis set the repository for this revision to rL LLVM. Avoid crashing when printing diagnostics for vtable-related CFI errors. In diagnostic mode, the frontend does an additional check of the vtable pointer against the set of all known vtable addresses and lets the runtime handler know if it is safe to inspect the vtable. Repository: rL LLVM http://reviews.llvm.org/D16823 Files: lib/CodeGen/CGClass.cpp lib/CodeGen/CGExpr.cpp lib/CodeGen/CodeGenFunction.h lib/CodeGen/CodeGenModule.cpp test/CodeGen/cfi-check-fail.c test/CodeGenCXX/cfi-cast.cpp test/CodeGenCXX/cfi-vcall.cpp Index: test/CodeGenCXX/cfi-vcall.cpp === --- test/CodeGenCXX/cfi-vcall.cpp +++ test/CodeGenCXX/cfi-vcall.cpp @@ -55,7 +55,7 @@ // DIAG: @[[SRC:.*]] = private unnamed_addr constant [{{.*}} x i8] c"{{.*}}cfi-vcall.cpp\00", align 1 // DIAG: @[[TYPE:.*]] = private unnamed_addr constant { i16, i16, [4 x i8] } { i16 -1, i16 0, [4 x i8] c"'A'\00" } -// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+21]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } +// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+23]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } // ITANIUM: define void @_Z2afP1A // MS: define void @"\01?af@@YAXPEAUA@@@Z" @@ -68,10 +68,12 @@ // CHECK: [[TRAPBB]] // NDIAG-NEXT: call void @llvm.trap() // NDIAG-NEXT: unreachable + // DIAG-NEXT: [[VTVALID0:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT]], metadata !"all-vtables") // DIAG-NEXT: [[VTINT:%[^ ]*]] = ptrtoint i8* [[VT]] to i64 - // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-NEXT: [[VTVALID:%[^ ]*]] = zext i1 [[VTVALID0]] to i64 + // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-ABORT-NEXT: unreachable - // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-RECOVER-NEXT: br label %[[CONTBB]] // CHECK: [[CONTBB]] @@ -157,32 +159,45 @@ } -// Check for the expected number of elements (9 or 15 respectively). -// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){8}]]} -// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// Check for the expected number of elements (15 or 23 respectively). +// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){15}]]} +// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){23}]]} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTV1A, i64 16} +// ITANIUM-DAG: !{!"all-vtables", [3 x i8*]* @_ZTV1A, i64 16} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} +// ITANIUM-DAG: !{!"all-vtables", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} // ITANIUM-DAG: !{!"_ZTS1C", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1C", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 88} +// ITANIUM-DAG: !{!"all-vtables", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 88} // ITANIUM-DAG: !{![[DTYPE]], [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTV1B, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [7 x i8*]* @_ZTV1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTV1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [5 x i8*]* @_ZTV1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1C", [5 x i8*]* @_ZTV1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTVZ3foovE2FA, i64 16} // ITANIUM-DAG: !{!{{[0-9]+}}, [3 x i8*]* @_ZTVZ3foovE2FA, i64 16} // MS-DAG: !{!"?AUA@@", [2 x i8*]* @[[VTA]], i64 8} +// MS-DAG: !{!"all-vtables", [2 x i8*]* @[[VTA]], i64 8} // MS-DAG: !{!"?AUB@@", [3 x i8*]* @[[VTB]], i64 8} +// MS-DAG: !{!"all-vtables", [3 x i8*]* @[[VTB]], i64 8} // MS-DAG: !{!"?AUA@@", [2 x i8*]* @[[VTAinB]], i64 8} +// MS-DAG: !{!"all-vtables", [2 x i8*]*
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis added inline comments. Comment at: lib/CodeGen/CGExpr.cpp:2494 @@ +2493,3 @@ + llvm::Value *ValidVtable = nullptr; + if (CheckAndAppendValidVtable) { +llvm::Value *AllVtables = llvm::MetadataAsValue::get( samsonov wrote: > This is really ugly. Why are you not passing it down in DynamicArgs? Is it > performance penalty you don't want to pay if the check will not succeed? How > large will it be? Yes, I want this code to be on the failing side of the check. This would cost about the same as the check itself, so I suspect it could double the overhead. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
samsonov added a subscriber: samsonov. Comment at: lib/CodeGen/CGExpr.cpp:2494 @@ +2493,3 @@ + llvm::Value *ValidVtable = nullptr; + if (CheckAndAppendValidVtable) { +llvm::Value *AllVtables = llvm::MetadataAsValue::get( This is really ugly. Why are you not passing it down in DynamicArgs? Is it performance penalty you don't want to pay if the check will not succeed? How large will it be? Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis added inline comments. Comment at: lib/CodeGen/CGExpr.cpp:2494 @@ +2493,3 @@ + llvm::Value *ValidVtable = nullptr; + if (CheckAndAppendValidVtable) { +llvm::Value *AllVtables = llvm::MetadataAsValue::get( pcc wrote: > eugenis wrote: > > samsonov wrote: > > > This is really ugly. Why are you not passing it down in DynamicArgs? Is > > > it performance penalty you don't want to pay if the check will not > > > succeed? How large will it be? > > Yes, I want this code to be on the failing side of the check. > > This would cost about the same as the check itself, so I suspect it could > > double the overhead. > > > I would just emit the call unconditionally. We don't care too much about the > performance in non-trapping mode, and if it becomes a problem in practice we > can see if we can have the optimizer move the call into the conditional block > (which I suspect it already knows how to do). I care about performance in non-trapping mode. Doing this change would not make the code any less ugly. For example, EmitCheck may not use the argument if the check has -fsanitize-trap behaviour, in which case we get an unused llvm.bitset.test call that affects some of the clang tests. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
pcc added inline comments. Comment at: lib/CodeGen/CGExpr.cpp:2494 @@ +2493,3 @@ + llvm::Value *ValidVtable = nullptr; + if (CheckAndAppendValidVtable) { +llvm::Value *AllVtables = llvm::MetadataAsValue::get( eugenis wrote: > samsonov wrote: > > This is really ugly. Why are you not passing it down in DynamicArgs? Is it > > performance penalty you don't want to pay if the check will not succeed? > > How large will it be? > Yes, I want this code to be on the failing side of the check. > This would cost about the same as the check itself, so I suspect it could > double the overhead. > I would just emit the call unconditionally. We don't care too much about the performance in non-trapping mode, and if it becomes a problem in practice we can see if we can have the optimizer move the call into the conditional block (which I suspect it already knows how to do). Comment at: lib/CodeGen/CodeGenModule.cpp:4053 @@ +4052,3 @@ + + if (!CodeGenOpts.SanitizeTrap.has(SanitizerKind::CFIVCall) || + !CodeGenOpts.SanitizeTrap.has(SanitizerKind::CFINVCall) || This conditional doesn't look right. It should be something like ```if (sanitize.has(this) && !sanitizetrap.has(this)) || (sanitize.has(that) && !sanitizetrap.has(that)) || ... ``` But that's sufficiently ugly that I wonder if we should just do this unconditionally. It shouldn't make a difference to the generated code either way. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
samsonov added inline comments. Comment at: lib/CodeGen/CGExpr.cpp:2642 @@ +2641,3 @@ + llvm::MDString::get(CGM.getLLVMContext(), "all-vtables")); + llvm::Value *ValidVtable = Builder.CreateZExt( + Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::bitset_test), This is almost the same as EmitVTablePtrCheck, but with ZExt? Is the difference intentional/important? Is it possible to extract this logic (getting "all-vtables" metadata and running bitset test) to a function? Comment at: lib/CodeGen/CodeGenModule.cpp:4034 @@ +4033,3 @@ + // is not in the trapping mode. + return ((LangOpts.Sanitize.has(SanitizerKind::CFIVCall) && + !CodeGenOpts.SanitizeTrap.has(SanitizerKind::CFIVCall)) || Hm, can you write this as a loop? Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis updated this revision to Diff 46718. eugenis added a comment. Moved bitset.text call outside. LLVM is smart enough to sink it along the cold branch, so performance should not suffer. Repository: rL LLVM http://reviews.llvm.org/D16823 Files: lib/CodeGen/CGClass.cpp lib/CodeGen/CGExpr.cpp lib/CodeGen/CodeGenModule.cpp test/CodeGen/cfi-check-fail.c test/CodeGenCXX/cfi-cast.cpp test/CodeGenCXX/cfi-vcall.cpp Index: test/CodeGenCXX/cfi-vcall.cpp === --- test/CodeGenCXX/cfi-vcall.cpp +++ test/CodeGenCXX/cfi-vcall.cpp @@ -55,23 +55,25 @@ // DIAG: @[[SRC:.*]] = private unnamed_addr constant [{{.*}} x i8] c"{{.*}}cfi-vcall.cpp\00", align 1 // DIAG: @[[TYPE:.*]] = private unnamed_addr constant { i16, i16, [4 x i8] } { i16 -1, i16 0, [4 x i8] c"'A'\00" } -// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+21]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } +// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+23]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } // ITANIUM: define void @_Z2afP1A // MS: define void @"\01?af@@YAXPEAUA@@@Z" void af(A *a) { // ITANIUM: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"_ZTS1A") // MS: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"?AUA@@") + // DIAG-NEXT: [[VTVALID0:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT]], metadata !"all-vtables") // CHECK-NEXT: br i1 [[P]], label %[[CONTBB:[^ ,]*]], label %[[TRAPBB:[^ ,]*]] // CHECK-NEXT: {{^$}} // CHECK: [[TRAPBB]] // NDIAG-NEXT: call void @llvm.trap() // NDIAG-NEXT: unreachable // DIAG-NEXT: [[VTINT:%[^ ]*]] = ptrtoint i8* [[VT]] to i64 - // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-NEXT: [[VTVALID:%[^ ]*]] = zext i1 [[VTVALID0]] to i64 + // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-ABORT-NEXT: unreachable - // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-RECOVER-NEXT: br label %[[CONTBB]] // CHECK: [[CONTBB]] @@ -157,32 +159,45 @@ } -// Check for the expected number of elements (9 or 15 respectively). -// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){8}]]} -// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// Check for the expected number of elements (15 or 23 respectively). +// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){15}]]} +// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){23}]]} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTV1A, i64 16} +// ITANIUM-DAG: !{!"all-vtables", [3 x i8*]* @_ZTV1A, i64 16} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} +// ITANIUM-DAG: !{!"all-vtables", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} // ITANIUM-DAG: !{!"_ZTS1C", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1C", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 88} +// ITANIUM-DAG: !{!"all-vtables", [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 88} // ITANIUM-DAG: !{![[DTYPE]], [12 x i8*]* @_ZTVN12_GLOBAL__N_11DE, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTV1B, i64 32} +// ITANIUM-DAG: !{!"all-vtables", [7 x i8*]* @_ZTV1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTV1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [5 x i8*]* @_ZTV1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1C", [5 x i8*]* @_ZTV1C, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTVZ3foovE2FA, i64 16} // ITANIUM-DAG: !{!{{[0-9]+}}, [3 x i8*]* @_ZTVZ3foovE2FA, i64 16} // MS-DAG: !{!"?AUA@@", [2 x i8*]* @[[VTA]], i64 8} +// MS-DAG: !{!"all-vtables", [2 x i8*]* @[[VTA]], i64 8} // MS-DAG: !{!"?AUB@@", [3 x i8*]* @[[VTB]], i64 8} +// MS-DAG: !{!"all-vtables", [3 x i8*]* @[[VTB]], i64 8} // MS-DAG: !{!"?AUA@@", [2 x i8*]* @[[VTAinB]], i64 8} +// MS-DAG:
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis updated this revision to Diff 46723. eugenis marked an inline comment as done. Repository: rL LLVM http://reviews.llvm.org/D16823 Files: lib/CodeGen/CGClass.cpp lib/CodeGen/CGExpr.cpp lib/CodeGen/CodeGenModule.cpp lib/CodeGen/CodeGenModule.h test/CodeGen/cfi-check-fail.c test/CodeGenCXX/cfi-cast.cpp test/CodeGenCXX/cfi-vcall.cpp Index: test/CodeGenCXX/cfi-vcall.cpp === --- test/CodeGenCXX/cfi-vcall.cpp +++ test/CodeGenCXX/cfi-vcall.cpp @@ -1,5 +1,5 @@ -// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=NDIAG %s -// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=DIAG --check-prefix=DIAG-ABORT %s +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=ITANIUM-NDIAG --check-prefix=NDIAG %s +// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=ITANIUM-DIAG --check-prefix=DIAG --check-prefix=DIAG-ABORT %s // RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-vcall -fsanitize-recover=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=ITANIUM --check-prefix=DIAG --check-prefix=DIAG-RECOVER %s // RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -fsanitize=cfi-vcall -fsanitize-trap=cfi-vcall -emit-llvm -o - %s | FileCheck --check-prefix=CHECK --check-prefix=MS --check-prefix=NDIAG %s @@ -55,23 +55,25 @@ // DIAG: @[[SRC:.*]] = private unnamed_addr constant [{{.*}} x i8] c"{{.*}}cfi-vcall.cpp\00", align 1 // DIAG: @[[TYPE:.*]] = private unnamed_addr constant { i16, i16, [4 x i8] } { i16 -1, i16 0, [4 x i8] c"'A'\00" } -// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+21]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } +// DIAG: @[[BADTYPESTATIC:.*]] = private unnamed_addr global { i8, { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }* } { i8 0, { [{{.*}} x i8]*, i32, i32 } { [{{.*}} x i8]* @[[SRC]], i32 [[@LINE+23]], i32 3 }, { i16, i16, [4 x i8] }* @[[TYPE]] } // ITANIUM: define void @_Z2afP1A // MS: define void @"\01?af@@YAXPEAUA@@@Z" void af(A *a) { // ITANIUM: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"_ZTS1A") // MS: [[P:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT:%[^ ]*]], metadata !"?AUA@@") + // DIAG-NEXT: [[VTVALID0:%[^ ]*]] = call i1 @llvm.bitset.test(i8* [[VT]], metadata !"all-vtables") // CHECK-NEXT: br i1 [[P]], label %[[CONTBB:[^ ,]*]], label %[[TRAPBB:[^ ,]*]] // CHECK-NEXT: {{^$}} // CHECK: [[TRAPBB]] // NDIAG-NEXT: call void @llvm.trap() // NDIAG-NEXT: unreachable // DIAG-NEXT: [[VTINT:%[^ ]*]] = ptrtoint i8* [[VT]] to i64 - // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-NEXT: [[VTVALID:%[^ ]*]] = zext i1 [[VTVALID0]] to i64 + // DIAG-ABORT-NEXT: call void @__ubsan_handle_cfi_check_fail_abort(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-ABORT-NEXT: unreachable - // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]]) + // DIAG-RECOVER-NEXT: call void @__ubsan_handle_cfi_check_fail(i8* getelementptr inbounds ({{.*}} @[[BADTYPESTATIC]], i32 0, i32 0), i64 [[VTINT]], i64 [[VTVALID]]) // DIAG-RECOVER-NEXT: br label %[[CONTBB]] // CHECK: [[CONTBB]] @@ -157,32 +159,47 @@ } -// Check for the expected number of elements (9 or 15 respectively). -// MS: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){8}]]} -// ITANIUM: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// Check for the expected number of elements (15 or 23 respectively). +// MS-NDIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){9}]]} +// MS-DIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){15}]]} +// ITANIUM-NDIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){14}]]} +// ITANIUM-DIAG: !llvm.bitsets = !{[[X:[^,]*(,[^,]*){23}]]} // ITANIUM-DAG: !{!"_ZTS1A", [3 x i8*]* @_ZTV1A, i64 16} +// ITANIUM-DIAG-DAG: !{!"all-vtables", [3 x i8*]* @_ZTV1A, i64 16} // ITANIUM-DAG: !{!"_ZTS1A", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} +// ITANIUM-DIAG-DAG: !{!"all-vtables", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1B", [7 x i8*]* @_ZTCN12_GLOBAL__N_11DE0_1B, i64 32} // ITANIUM-DAG: !{!"_ZTS1A", [9 x i8*]* @_ZTCN12_GLOBAL__N_11DE8_1C, i64 64} +// ITANIUM-DIAG-DAG:
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis added inline comments. Comment at: lib/CodeGen/CodeGenModule.cpp:4053 @@ +4052,3 @@ + + if (!CodeGenOpts.SanitizeTrap.has(SanitizerKind::CFIVCall) || + !CodeGenOpts.SanitizeTrap.has(SanitizerKind::CFINVCall) || I don't like emitting all these bitset entries if they are not needed. Fixed. Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D16823: [cfi] Safe handling of unaddressable vtable pointers (clang).
eugenis added inline comments. Comment at: lib/CodeGen/CGExpr.cpp:2642 @@ +2641,3 @@ + llvm::MDString::get(CGM.getLLVMContext(), "all-vtables")); + llvm::Value *ValidVtable = Builder.CreateZExt( + Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::bitset_test), samsonov wrote: > This is almost the same as EmitVTablePtrCheck, but with ZExt? Is the > difference intentional/important? Is it possible to extract this logic > (getting "all-vtables" metadata and running bitset test) to a function? Not important. Zext makes the test a bit simpler. Extracting these two lines to a function is surely possible, but is it worth it? Repository: rL LLVM http://reviews.llvm.org/D16823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits