Changes in Symbols.cpp

e980f16d - [ELF] Move whyExtract/backwardReferences from LinkerDriver to Ctx. NFC

Thu, 30 Jun 2022 00:34:30 +0000

[ELF] Move whyExtract/backwardReferences from LinkerDriver to Ctx. NFCCtx was recently added as a more suitable place for such singletons. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

b862bcbf - [ELF] Move SymbolUnion assertions to source file

Tue, 19 Apr 2022 20:42:05 +0000

[ELF] Move SymbolUnion assertions to source fileOtherwise they fires for every single file which includes the header,which is very noisy when building.Reviewed By: MaskRay, peter.smithDifferential Revision: https://reviews.llvm.org/D124041 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

7c7702b3 - [ELF] Move section assignment from initializeSymbols to postParse

Wed, 16 Mar 2022 02:24:41 +0000

[ELF] Move section assignment from initializeSymbols to postParsehttps://discourse.llvm.org/t/parallel-input-file-parsing/60164initializeSymbols currently sets Defined::section and handles non-prevailingCOMDAT groups. Move the code to the parallel postParse to reduce work from thesingle-threading code path and make parallel section initialization infeasible.Postpone reporting duplicate symbol errors so that the messages have thesection information. (`Defined::section` is assigned in postParse and anotherthread may not have the information).* duplicated-synthetic-sym.s: BinaryFile duplicate definition (very rare) now has no section information* comdat-binding: `%t/w.o %t/g.o` leads to an undesired undefined symbol. This is not ideal but we report a diagnostic to inform that this is unsupported. (See release note)* comdat-discarded-lazy.s: %tdef.o is unextracted. The new behavior (discarded section error) makes more sense* i386-comdat.s: switched to a better approach working around .gnu.linkonce.t.__x86.get_pc_thunk.bx in glibc<2.32 for x86-32. Drop the ancient no-longer-relevant workaround for __i686.get_pc_thunk.bxDepends on D120640Differential Revision: https://reviews.llvm.org/D120626 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

9b61fff0 - Revert D120626 "[ELF] Move section assignment from initializeSymbols to postParse"

Wed, 16 Mar 2022 02:00:54 +0000

Revert D120626 "[ELF] Move section assignment from initializeSymbols to postParse"This reverts commit c30e6447c0225f675773d07f2b6d4e3c2b962155.It exposed brittle support for __x86.get_pc_thunk.bx.Need to think a bit how to support __x86.get_pc_thunk.bx. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

c1d4c677 - [ELF] Suppress duplicate symbol error for __x86.get_pc_thunk.bx

Wed, 16 Mar 2022 00:20:29 +0000

[ELF] Suppress duplicate symbol error for __x86.get_pc_thunk.bx List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

c30e6447 - [ELF] Move section assignment from initializeSymbols to postParse

Mon, 14 Mar 2022 21:13:41 +0000

[ELF] Move section assignment from initializeSymbols to postParsehttps://discourse.llvm.org/t/parallel-input-file-parsing/60164initializeSymbols currently sets Defined::section and handles non-prevailingCOMDAT groups. Move the code to the parallel postParse to reduce work from thesingle-threading code path and make parallel section initialization infeasible.Postpone reporting duplicate symbol errors so that the messages have thesection information. (`Defined::section` is assigned in postParse and anotherthread may not have the information).* duplicated-synthetic-sym.s: BinaryFile duplicate definition (very rare) now has no section information* comdat-binding: `%t/w.o %t/g.o` leads to an undesired undefined symbol. This is not ideal but we report a diagnostic to inform that this is unsupported. (See release note)* comdat-discarded-lazy.s: %tdef.o is unextracted. The new behavior (discarded section error) makes more senseDepends on D120640Reviewed By: peter.smithDifferential Revision: https://reviews.llvm.org/D120626 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

c7cf960d - [ELF] Set the priority of STB_GNU_UNIQUE the same as STB_WEAK

Mon, 14 Mar 2022 19:00:15 +0000

[ELF] Set the priority of STB_GNU_UNIQUE the same as STB_WEAKIn GCC -fgnu-unique output, STB_GNU_UNIQUE symbols are always defined relativeto a section in a COMDAT group. Currently `other` cannot be STB_GNU_UNIQUE forvalid input, so this patch is NFC.If we switch to the model that ignores COMDAT resolution when performing symbolresolution (D120626), this will fix bogus `relocation refers to a symbol in adiscarded section` errors when mixing -fno-gnu-unique objects with -fgnu-uniqueobjects.Differential Revision: https://reviews.llvm.org/D120640 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

b07ef4d5 - [ELF] Rename Symbol::compare to shouldReplace. NFC

Mon, 28 Feb 2022 18:25:21 +0000

[ELF] Rename Symbol::compare to shouldReplace. NFCThe return value is not a boolean instead of a tri-state.Suggested by Peter Smith in D120640. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

d14d8664 - [ELF] Change global variable backwardReferences to a LinkerDriver member variable. NFC

Sun, 27 Feb 2022 20:33:28 +0000

[ELF] Change global variable backwardReferences to a LinkerDriver member variable. NFCSimilar to whyExtract. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

7fd3849b - [ELF] Move --print-archive-stats= and --why-extract= beside --warn-backrefs report

Sun, 27 Feb 2022 20:23:09 +0000

[ELF] Move --print-archive-stats= and --why-extract= beside --warn-backrefs reportSo that early errors don't suppress their output. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

19e37a74 - [ELF] Update comment. NFC

Thu, 24 Feb 2022 22:09:00 +0000

[ELF] Update comment. NFC List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

6d943408 - [ELF] Simplify resolveDefined and resolveCommon

Thu, 24 Feb 2022 22:08:06 +0000

[ELF] Simplify resolveDefined and resolveCommonThis is NFC for valid input (COMMON symbols cannot be weak or versioned). List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

15617cdb - [ELF] Simplify --fortran-common. NFC

Thu, 24 Feb 2022 20:21:40 +0000

[ELF] Simplify --fortran-common. NFC List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

4129890d - [ELF] De-template Symbol::resolveLazy. NFC

Thu, 24 Feb 2022 20:20:05 +0000

[ELF] De-template Symbol::resolveLazy. NFC List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

8ca46bba - [ELF] Move isUsedInRegularObj assignment from ctor to call sites. NFC

Thu, 24 Feb 2022 05:32:50 +0000

[ELF] Move isUsedInRegularObj assignment from ctor to call sites. NFCThis removes the tricky`isUsedInRegularObj(!file || file->kind() == InputFile::ObjKind)`and the copy from `Symbol::mergeProperties`. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

88d66f6e - [ELF] Move duplicate symbol check after input file parsing

Tue, 22 Feb 2022 18:07:58 +0000

[ELF] Move duplicate symbol check after input file parsinghttps://discourse.llvm.org/t/parallel-input-file-parsing/60164To decouple symbol initialization and section initialization, `Defined::section`assignment should be postponed after input file parsing. To avoid spuriousduplicate definition error due to two definitions in COMDAT groups of the samesignature, we should postpone the duplicate symbol check.The function is called postScan instead of a more specific name likecheckDuplicateSymbols, because we may merge Symbol::mergeProperties intopostScan. It is placed after compileBitcodeFiles to apply to ET_REL filesproduced by LTO. This causes minor diagnostic regressionfor skipLinkedOutput configurations: ld.lld --thinlto-index-only a.bc b.o(bitcode definition prevails) won't detect duplicate symbol error. I think thisis an acceptable compromise. The important cases where (a) both files arebitcode or (b) --thinlto-index-only is unused are still detected.Reviewed By: ikudrinDifferential Revision: https://reviews.llvm.org/D119908 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

467e1b3a - [ELF] reportDuplicate: change Symbol * to const Symbol &. NFC

Tue, 15 Feb 2022 19:18:31 +0000

[ELF] reportDuplicate: change Symbol * to const Symbol &. NFC List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

3d854240 - [ELF] Parse archives as --start-lib object files

Tue, 15 Feb 2022 17:38:00 +0000

[ELF] Parse archives as --start-lib object fileshttps://maskray.me/blog/2022-01-16-archives-and-start-libFor every definition in an extracted archive member, we intern the symbol twice,once for the archive index entry, once for the .o symbol table after extraction.This is inefficient.Symbols in a --start-lib ObjFile/BitcodeFile are only interned once because theresult is cached in symbols[i].Just handle an archive using the --start-lib code path. We can therefore removeArchiveFile and LazyArchive. For many projects, archive member extraction ratiois high and it is a net performance win. Linking a Release build of clang is1.01x as fast.Note: --start-lib scans symbols in the same order that llvm-ar adds them to theindex, so in the common case the semantics should be identical. If the archivesymbol table was created in a different order, or is incomplete, this strategymay have different semantics. Such cases are considered user error.The `is neither ET_REL nor LLVM bitcode` error is changed to a warning.Previously an archive may have such members without a diagnostic. Using awarning prevents breakage.* For some tests, the diagnostics get improved where we did not consider the archive member name: `b.a:` => `b.a(b.o):`.* `no-obj.s`: the link is now allowed, matching GNU ld* `archive-no-index.s`: the `is neither ET_REL nor LLVM bitcode` diagnostic is demoted to a warning.* `incompatible.s`: even when an archive is unextracted, we may report an "incompatible with" error.---I recently decreased sizeof(SymbolUnion) by 8 and decreased memory usage quite abit, so retaining `symbols` for un-extracted archive members should not cause amemory usage problem.Reviewed By: peter.smithDifferential Revision: https://reviews.llvm.org/D119074 List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

27bb7990 - [ELF] Clean up headers. NFC

Tue, 08 Feb 2022 05:53:34 +0000

[ELF] Clean up headers. NFC List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp

977a1a52 - [ELF] Symbol::replace: use the old nameData/nameSize. NFC

Sun, 06 Feb 2022 00:34:02 +0000

[ELF] Symbol::replace: use the old nameData/nameSize. NFCCurrently `this->getName() == newSym.getName()`.By keeping the old nameData/nameSize, newSym's nameData/nameSize will beignored. The call sites can avoid calling getName().printTraceSymbol needs to take the symbol name since `other`'s name is empty. List of files: /llvm-project-15.0.7/lld/ELF/Symbols.cpp