diff --git a/lld/MachO/ICF.cpp b/lld/MachO/ICF.cpp --- a/lld/MachO/ICF.cpp +++ b/lld/MachO/ICF.cpp @@ -324,6 +324,7 @@ // parallelization. Therefore, we hash every InputSection here where we have // them all accessible as simple vectors. std::vector codeSections; + std::vector cfStringSections; // ICF can't fold functions with unwind info DenseSet functionsWithUnwindInfo = @@ -339,18 +340,30 @@ // ICF::segregate() uint64_t icfUniqueID = inputSections.size(); for (ConcatInputSection *isec : inputSections) { - bool isHashable = isCodeSection(isec) && !isec->shouldOmitFromOutput() && + bool isHashable = (isCodeSection(isec) || isCfStringSection(isec)) && + !isec->shouldOmitFromOutput() && !functionsWithUnwindInfo.contains(isec) && isec->isHashableForICF(); if (isHashable) { - codeSections.push_back(isec); + if (isCodeSection(isec)) + codeSections.push_back(isec); + else { + assert(isCfStringSection(isec)); + cfStringSections.push_back(isec); + } } else { isec->icfEqClass[0] = ++icfUniqueID; } } - parallelForEach(codeSections, + std::vector hashable(codeSections); + hashable.insert(hashable.end(), cfStringSections.begin(), + cfStringSections.end()); + parallelForEach(hashable, [](ConcatInputSection *isec) { isec->hashForICF(); }); // Now that every input section is either hashed or marked as unique, run the // segregation algorithm to detect foldable subsections. + // We dedup cfStringSections first since code sections may refer to them, but + // not vice-versa. + ICF(cfStringSections).run(); ICF(codeSections).run(); } diff --git a/lld/MachO/InputFiles.cpp b/lld/MachO/InputFiles.cpp --- a/lld/MachO/InputFiles.cpp +++ b/lld/MachO/InputFiles.cpp @@ -282,6 +282,16 @@ flags); } subsections.push_back({{0, isec}}); + } else if (name == section_names::cfString && + segname == segment_names::data) { + uint64_t literalSize = target->wordSize == 8 ? 32 : 16; + subsections.push_back({}); + SubsectionMap &subsecMap = subsections.back(); + for (uint64_t off = 0; off < data.size(); off += literalSize) + subsecMap.push_back( + {off, make(segname, name, this, + data.slice(off, literalSize), align, + flags)}); } else { auto *isec = make(segname, name, this, data, align, flags); @@ -596,22 +606,43 @@ } } - // Calculate symbol sizes and create subsections by splitting the sections - // along symbol boundaries. for (size_t i = 0; i < subsections.size(); ++i) { SubsectionMap &subsecMap = subsections[i]; if (subsecMap.empty()) continue; std::vector &symbolIndices = symbolsBySection[i]; - llvm::sort(symbolIndices, [&](uint32_t lhs, uint32_t rhs) { - return nList[lhs].n_value < nList[rhs].n_value; - }); uint64_t sectionAddr = sectionHeaders[i].addr; uint32_t sectionAlign = 1u << sectionHeaders[i].align; + InputSection *isec = subsecMap.back().isec; + // __cfstring has already been split into subsections during + // parseSections(), so we simply need to match Symbols to the corresponding + // subsection here. + if (isCfStringSection(isec)) { + for (size_t j = 0; j < symbolIndices.size(); ++j) { + uint32_t symIndex = symbolIndices[j]; + const NList &sym = nList[symIndex]; + StringRef name = strtab + sym.n_strx; + uint64_t symbolOffset = sym.n_value - sectionAddr; + InputSection *isec = findContainingSubsection(subsecMap, &symbolOffset); + if (symbolOffset != 0) { + error(toString(this) + ": __cfstring contains symbol " + name + + " at misaligned offset"); + continue; + } + symbols[symIndex] = createDefined(sym, name, isec, 0, isec->getSize()); + } + continue; + } + + // Calculate symbol sizes and create subsections by splitting the sections + // along symbol boundaries. // We populate subsecMap by repeatedly splitting the last (highest address) // subsection. + llvm::sort(symbolIndices, [&](uint32_t lhs, uint32_t rhs) { + return nList[lhs].n_value < nList[rhs].n_value; + }); SubsectionEntry subsecEntry = subsecMap.back(); for (size_t j = 0; j < symbolIndices.size(); ++j) { uint32_t symIndex = symbolIndices[j]; diff --git a/lld/MachO/InputSection.h b/lld/MachO/InputSection.h --- a/lld/MachO/InputSection.h +++ b/lld/MachO/InputSection.h @@ -247,6 +247,8 @@ bool isCodeSection(const InputSection *); +bool isCfStringSection(const InputSection *); + extern std::vector inputSections; namespace section_names { diff --git a/lld/MachO/InputSection.cpp b/lld/MachO/InputSection.cpp --- a/lld/MachO/InputSection.cpp +++ b/lld/MachO/InputSection.cpp @@ -230,6 +230,11 @@ return false; } +bool macho::isCfStringSection(const InputSection *isec) { + return isec->name == section_names::cfString && + isec->segname == segment_names::data; +} + std::string lld::toString(const InputSection *isec) { return (toString(isec->file) + ":(" + isec->name + ")").str(); } diff --git a/lld/test/MachO/Inputs/MacOSX.sdk/System/Library/Frameworks/CoreFoundation.framework/CoreFoundation.tbd b/lld/test/MachO/Inputs/MacOSX.sdk/System/Library/Frameworks/CoreFoundation.framework/CoreFoundation.tbd --- a/lld/test/MachO/Inputs/MacOSX.sdk/System/Library/Frameworks/CoreFoundation.framework/CoreFoundation.tbd +++ b/lld/test/MachO/Inputs/MacOSX.sdk/System/Library/Frameworks/CoreFoundation.framework/CoreFoundation.tbd @@ -7,7 +7,7 @@ compatibility-version: 150 exports: - archs: [ 'x86_64' ] - symbols: [ '__CFBigNumGetInt128' ] + symbols: [ __CFBigNumGetInt128, ___CFConstantStringClassReference ] objc-classes: [ NSObject ] objc-ivars: [ NSConstantArray._count ] objc-eh-types: [ NSException ] diff --git a/lld/test/MachO/cfstring-dedup.s b/lld/test/MachO/cfstring-dedup.s new file mode 100644 --- /dev/null +++ b/lld/test/MachO/cfstring-dedup.s @@ -0,0 +1,140 @@ +# REQUIRES: x86 +# RUN: rm -rf %t; split-file %s %t +# RUN: llvm-mc -filetype=obj -triple=x86_64-apple-darwin %t/foo1.s -o %t/foo1.o +# RUN: llvm-mc -filetype=obj -triple=x86_64-apple-darwin %t/foo2.s -o %t/foo2.o +# RUN: %lld -dylib --icf=all -framework CoreFoundation %t/foo1.o %t/foo2.o -o %t/foo +# RUN: llvm-objdump --macho --rebase --bind --syms -d %t/foo | FileCheck %s + +# CHECK: (__TEXT,__text) section +# CHECK-NEXT: _foo1: +# CHECK-NEXT: _foo2: +# CHECK-NEXT: movq _named_cfstring(%rip), %rax +# CHECK-NEXT: _foo1_utf16: +# CHECK-NEXT: movq [[#]](%rip), %rax +# CHECK-NEXT: _named_foo1: +# CHECK-NEXT: _named_foo2: +# CHECK-NEXT: movq _named_cfstring(%rip), %rax +# CHECK-NEXT: _foo2_utf16: +# CHECK-NEXT: movq [[#]](%rip), %rax + +# CHECK: SYMBOL TABLE: +# CHECK-DAG: [[#%.16x,FOO:]] g F __TEXT,__text _foo1 +# CHECK-DAG: [[#FOO]] g F __TEXT,__text _foo2 + +## Make sure we don't emit redundant bind / rebase opcodes for folded sections. +# CHECK: Rebase table: +# CHECK-NEXT: segment section address type +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer +# CHECK-EMPTY: +# CHECK-NEXT: Bind table: +# CHECK-NEXT: segment section address type addend dylib symbol +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer 0 CoreFoundation ___CFConstantStringClassReference +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer 0 CoreFoundation ___CFConstantStringClassReference +# CHECK-NEXT: __DATA_CONST __cfstring {{.*}} pointer 0 CoreFoundation ___CFConstantStringClassReference +# CHECK-EMPTY: + +#--- foo1.s +.cstring +L_.str: + .asciz "foo" + +.section __DATA,__cfstring +.p2align 3 +L__unnamed_cfstring_: + .quad ___CFConstantStringClassReference + .long 1992 ## utf-8 + .space 4 + .quad L_.str + .quad 3 ## strlen + +_named_cfstring: + .quad ___CFConstantStringClassReference + .long 1992 ## utf-8 + .space 4 + .quad L_.str + .quad 3 ## strlen + +.section __TEXT,__ustring +l_.str.2: + .short 102 ## f + .short 111 ## o + .short 111 ## o + .short 0 ## \0 + .short 0 ## \0 + +## FIXME: We should be able to deduplicate UTF-16 CFStrings too. +.section __DATA,__cfstring +.p2align 3 +L__unnamed_cfstring_.2: + .quad ___CFConstantStringClassReference + .long 2000 ## utf-16 + .space 4 + .quad l_.str.2 + .quad 4 ## strlen + +.text +.globl _foo1, _foo1_utf16, _named_foo1 +_foo1: + movq L__unnamed_cfstring_(%rip), %rax + +_foo1_utf16: + movq L__unnamed_cfstring_.2(%rip), %rax + +_named_foo1: + movq _named_cfstring(%rip), %rax + +.subsections_via_symbols + +#--- foo2.s +.cstring +L_.str: + .asciz "foo" + +.section __DATA,__cfstring +.p2align 3 +L__unnamed_cfstring_: + .quad ___CFConstantStringClassReference + .long 1992 ## utf-8 + .space 4 + .quad L_.str + .quad 3 ## strlen + +_named_cfstring: + .quad ___CFConstantStringClassReference + .long 1992 ## utf-8 + .space 4 + .quad L_.str + .quad 3 ## strlen + +.section __TEXT,__ustring + .p2align 1 +l_.str.2: + .short 102 ## f + .short 111 ## o + .short 111 ## o + .short 0 ## \0 + .short 0 ## \0 + +.section __DATA,__cfstring +.p2align 3 +L__unnamed_cfstring_.2: + .quad ___CFConstantStringClassReference + .long 2000 ## utf-16 + .space 4 + .quad l_.str.2 + .quad 4 ## strlen + +.text +.globl _foo2, _foo2_utf16, _named_foo2 +_foo2: + movq L__unnamed_cfstring_(%rip), %rax + +_foo2_utf16: + movq L__unnamed_cfstring_.2(%rip), %rax + +_named_foo2: + movq _named_cfstring(%rip), %rax + +.subsections_via_symbols diff --git a/lld/test/MachO/invalid/cfstring.s b/lld/test/MachO/invalid/cfstring.s new file mode 100644 --- /dev/null +++ b/lld/test/MachO/invalid/cfstring.s @@ -0,0 +1,19 @@ +# REQUIRES: x86 +# RUN: rm -rf %t; mkdir %t +# RUN: llvm-mc -filetype=obj -triple=x86_64-apple-darwin %s -o %t/test.o +# RUN: not %lld -dylib -framework CoreFoundation %t/test.o 2>&1 | FileCheck %s +# CHECK: error: {{.*}}test.o: __cfstring contains symbol _uh_oh at misaligned offset + +.cstring +L_.str: + .asciz "foo" + +.section __DATA,__cfstring +.p2align 3 +L__unnamed_cfstring_: + .quad ___CFConstantStringClassReference + .long 1992 ## utf-8 +_uh_oh: + .space 4 + .quad L_.str + .quad 3 ## strlen