From e9b4278ca7a492d3710aeae324250189379e6d30 Mon Sep 17 00:00:00 2001 From: serge-sans-paille Date: Thu, 13 Oct 2022 10:26:41 +0200 Subject: [PATCH] [lto] Do not try to internalize symbols with escaped name Because of LLVM mangling escape sequence (through '\01' prefix), it is possible for a single symbols two have two different IR representations. For instance, consider @symbol and @"\01_symbol". On OSX, because of the system mangling rules, these two IR names point are converted in the same final symbol upon linkage. LTO doesn't model this behavior, which may result in symbols being incorrectly internalized (if all reference use the escaping sequence while the definition doesn't). The proper approach is probably to use the mangled name to compute GUID to avoid the dual representation, but we can also avoid discarding symbols that are bound to two different IR names. This is an approximation, but it's less intrusive on the codebase. Fix #57864 Differential Revision: https://reviews.llvm.org/D135710 --- llvm/lib/LTO/LTO.cpp | 16 +++++++ .../LTO/X86/hidden-escaped-symbols-alt.ll | 41 ++++++++++++++++++ llvm/test/LTO/X86/hidden-escaped-symbols.ll | 41 ++++++++++++++++++ .../ThinLTO/X86/hidden-escaped-symbols-alt.ll | 42 +++++++++++++++++++ .../ThinLTO/X86/hidden-escaped-symbols.ll | 42 +++++++++++++++++++ 5 files changed, 182 insertions(+) create mode 100644 llvm/test/LTO/X86/hidden-escaped-symbols-alt.ll create mode 100644 llvm/test/LTO/X86/hidden-escaped-symbols.ll create mode 100644 llvm/test/ThinLTO/X86/hidden-escaped-symbols-alt.ll create mode 100644 llvm/test/ThinLTO/X86/hidden-escaped-symbols.ll diff --git a/llvm/lib/LTO/LTO.cpp b/llvm/lib/LTO/LTO.cpp index 0f78c4ebd5ca..93aed72b3d15 100644 --- a/llvm/lib/LTO/LTO.cpp +++ b/llvm/lib/LTO/LTO.cpp @@ -565,6 +565,22 @@ void LTO::addModuleToGlobalRes(ArrayRef Syms, GlobalRes.IRName = std::string(Sym.getIRName()); } + // In rare occasion, the symbol used to initialize GlobalRes has a different + // IRName from the inspected Symbol. This can happen on macOS + iOS, when a + // symbol is referenced through its mangled name, say @"\01_symbol" while + // the IRName is @symbol (the prefix underscore comes from MachO mangling). + // In that case, we have the same actual Symbol that can get two different + // GUID, leading to some invalid internalization. Workaround this by marking + // the GlobalRes external. + + // FIXME: instead of this check, it would be desirable to compute GUIDs + // based on mangled name, but this requires an access to the Target Triple + // and would be relatively invasive on the codebase. + if (GlobalRes.IRName != Sym.getIRName()) { + GlobalRes.Partition = GlobalResolution::External; + GlobalRes.VisibleOutsideSummary = true; + } + // Set the partition to external if we know it is re-defined by the linker // with -defsym or -wrap options, used elsewhere, e.g. it is visible to a // regular object, is referenced from llvm.compiler.used/llvm.used, or was diff --git a/llvm/test/LTO/X86/hidden-escaped-symbols-alt.ll b/llvm/test/LTO/X86/hidden-escaped-symbols-alt.ll new file mode 100644 index 000000000000..ca8e7d8eb2b2 --- /dev/null +++ b/llvm/test/LTO/X86/hidden-escaped-symbols-alt.ll @@ -0,0 +1,41 @@ +; Check interaction between LTO and LLVM mangling escape char, see #57864. + +; RUN: split-file %s %t +; RUN: opt %t/hide-me.ll -o %t/hide-me.bc +; RUN: opt %t/ref.ll -o %t/ref.bc +; RUN: llvm-lto2 run \ +; RUN: -r %t/hide-me.bc,_hide_me,p \ +; RUN: -r %t/ref.bc,_main,plx \ +; RUN: -r %t/ref.bc,_hide_me,l \ +; RUN: --select-save-temps=precodegen \ +; RUN: -o %t/out \ +; RUN: %t/hide-me.bc %t/ref.bc +; RUN: llvm-dis %t/out.0.5.precodegen.bc -o - | FileCheck %s + + +;--- hide-me.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@"\01_hide_me" = hidden local_unnamed_addr global i8 8, align 1 + +;--- ref.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@hide_me = external local_unnamed_addr global i8 + +define i8 @main() { + %1 = load i8, ptr @hide_me, align 1 + ret i8 %1 +} + + +; CHECK: @"\01_hide_me" = hidden local_unnamed_addr global i8 8, align 1 +; CHECK: @hide_me = external dso_local local_unnamed_addr global i8 + +; CHECK: define dso_local i8 @main() local_unnamed_addr #0 { +; CHECK: %1 = load i8, ptr @hide_me, align 1 +; CHECK: ret i8 %1 +; CHECK: } + diff --git a/llvm/test/LTO/X86/hidden-escaped-symbols.ll b/llvm/test/LTO/X86/hidden-escaped-symbols.ll new file mode 100644 index 000000000000..aec617e25d3d --- /dev/null +++ b/llvm/test/LTO/X86/hidden-escaped-symbols.ll @@ -0,0 +1,41 @@ +; Check interaction between LTO and LLVM mangling escape char, see #57864. + +; RUN: split-file %s %t +; RUN: opt %t/hide-me.ll -o %t/hide-me.bc +; RUN: opt %t/ref.ll -o %t/ref.bc +; RUN: llvm-lto2 run \ +; RUN: -r %t/hide-me.bc,_hide_me,p \ +; RUN: -r %t/ref.bc,_main,plx \ +; RUN: -r %t/ref.bc,_hide_me,l \ +; RUN: --select-save-temps=precodegen \ +; RUN: -o %t/out \ +; RUN: %t/hide-me.bc %t/ref.bc +; RUN: llvm-dis %t/out.0.5.precodegen.bc -o - | FileCheck %s + + +;--- hide-me.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@hide_me = hidden local_unnamed_addr global i8 8, align 1 + +;--- ref.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@"\01_hide_me" = external local_unnamed_addr global i8 + +define i8 @main() { + %1 = load i8, ptr @"\01_hide_me", align 1 + ret i8 %1 +} + + +; CHECK: @hide_me = hidden local_unnamed_addr global i8 8, align 1 +; CHECK: @"\01_hide_me" = external dso_local local_unnamed_addr global i8 + +; CHECK: define dso_local i8 @main() local_unnamed_addr #0 { +; CHECK: %1 = load i8, ptr @"\01_hide_me", align 1 +; CHECK: ret i8 %1 +; CHECK: } + diff --git a/llvm/test/ThinLTO/X86/hidden-escaped-symbols-alt.ll b/llvm/test/ThinLTO/X86/hidden-escaped-symbols-alt.ll new file mode 100644 index 000000000000..dadd1d434256 --- /dev/null +++ b/llvm/test/ThinLTO/X86/hidden-escaped-symbols-alt.ll @@ -0,0 +1,42 @@ +; Check interaction between LTO and LLVM mangling escape char, see #57864. + +; RUN: split-file %s %t +; RUN: opt -module-summary %t/hide-me.ll -o %t/hide-me.bc +; RUN: opt -module-summary %t/ref.ll -o %t/ref.bc +; RUN: llvm-lto2 run \ +; RUN: -r %t/hide-me.bc,_hide_me,p \ +; RUN: -r %t/ref.bc,_main,plx \ +; RUN: -r %t/ref.bc,_hide_me,l \ +; RUN: --select-save-temps=precodegen \ +; RUN: -o %t/out \ +; RUN: %t/hide-me.bc %t/ref.bc +; RUN: llvm-dis %t/out.1.5.precodegen.bc -o - | FileCheck --check-prefix=CHECK-HIDE %s +; RUN: llvm-dis %t/out.2.5.precodegen.bc -o - | FileCheck --check-prefix=CHECK-REF %s + + +;--- hide-me.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@"\01_hide_me" = hidden local_unnamed_addr global i8 8, align 1 + +;--- ref.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@hide_me = external local_unnamed_addr global i8 + +define i8 @main() { + %1 = load i8, ptr @hide_me, align 1 + ret i8 %1 +} + + +; CHECK-HIDE: @"\01_hide_me" = hidden local_unnamed_addr global i8 8, align 1 + +; CHECK-REF: @hide_me = external local_unnamed_addr global i8 +; CHECK-REF: define dso_local i8 @main() local_unnamed_addr #0 { +; CHECK-REF: %1 = load i8, ptr @hide_me, align 1 +; CHECK-REF: ret i8 %1 +; CHECK-REF: } + diff --git a/llvm/test/ThinLTO/X86/hidden-escaped-symbols.ll b/llvm/test/ThinLTO/X86/hidden-escaped-symbols.ll new file mode 100644 index 000000000000..8d0e22f0fd22 --- /dev/null +++ b/llvm/test/ThinLTO/X86/hidden-escaped-symbols.ll @@ -0,0 +1,42 @@ +; Check interaction between LTO and LLVM mangling escape char, see #57864. + +; RUN: split-file %s %t +; RUN: opt -module-summary %t/hide-me.ll -o %t/hide-me.bc +; RUN: opt -module-summary %t/ref.ll -o %t/ref.bc +; RUN: llvm-lto2 run \ +; RUN: -r %t/hide-me.bc,_hide_me,p \ +; RUN: -r %t/ref.bc,_main,plx \ +; RUN: -r %t/ref.bc,_hide_me,l \ +; RUN: --select-save-temps=precodegen \ +; RUN: -o %t/out \ +; RUN: %t/hide-me.bc %t/ref.bc +; RUN: llvm-dis %t/out.1.5.precodegen.bc -o - | FileCheck --check-prefix=CHECK-HIDE %s +; RUN: llvm-dis %t/out.2.5.precodegen.bc -o - | FileCheck --check-prefix=CHECK-REF %s + + +;--- hide-me.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@hide_me = hidden local_unnamed_addr global i8 8, align 1 + +;--- ref.ll +target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.7.0" + +@"\01_hide_me" = external local_unnamed_addr global i8 + +define i8 @main() { + %1 = load i8, ptr @"\01_hide_me", align 1 + ret i8 %1 +} + + +; CHECK-HIDE: @hide_me = hidden local_unnamed_addr global i8 8, align 1 + +; CHECK-REF: @"\01_hide_me" = external local_unnamed_addr global i8 +; CHECK-REF: define dso_local i8 @main() local_unnamed_addr #0 { +; CHECK-REF: %1 = load i8, ptr @"\01_hide_me", align 1 +; CHECK-REF: ret i8 %1 +; CHECK-REF: } + -- 2.38.1.1.g6d9df9d320