Symbol's NameSize is computed lazily. Currently, when we replace a symbol,
a cached length value can be discarded. This patch propagates that value.
Differential Revision: https://reviews.llvm.org/D62234
llvm-svn: 361364
Rather than report "undefined symbol: ", give more informative message
about the object file that defines the discarded section.
In particular, PR41133, if the section is a discarded COMDAT, print the
section group signature and the object file with the prevailing
definition. This is useful to track down some ODR issues.
We need to
* add `uint32_t DiscardedSecIdx` to Undefined for this feature.
* make ComdatGroups public and change its type to DenseMap<CachedHashStringRef, const InputFile *>
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D59649
llvm-svn: 361359
This reverts commit 7804dbddcc.
This change broke a bunch of tests of the WebAssembly waterfall.
Will hopefully reland with increased test coverage.
llvm-svn: 361273
--reproduce is a convenient option for debugging. If you invoke lld
with `--reproduce=repro.tar`, it creates `repro.tar` with all input
files and the command line options given to the linker, so that it is
very easy to run lld with the exact same inputs.
ELF and Windows lld have this option.
This patch add that option to lld/wasm.
Differential Revision: https://reviews.llvm.org/D62170
llvm-svn: 361244
Major refactor to better match the structure of the ELF linker.
- Split out relocation processing into scanRelocations
- Split out synthetic sections into their own classes.
Differential Revision: https://reviews.llvm.org/D61811
llvm-svn: 361233
For memory5.test, ld.bfd appears to ignore `. += 0x2000;`, so the test was testing
a wrong behavior. After deleting the code added in rLLD336335, we match ld.bfd and thus fix PR41357.
PR37836 (memory4.test) seems to have been fixed by another change.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D62177
llvm-svn: 361228
For a reference to a local symbol, ld.bfd and gold error if the symbol
is defined in a discarded section but accept it if the symbol is
undefined. This inconsistent behavior seems unnecessary for us (it
probably makes sense for them as they differentiate local/global
symbols, the error would mean more code).
Catch such errors. Symbol index 0 may be used by marker relocations,
e.g. R_*_NONE R_ARM_V4BX. Don't error on them.
The difference from D61563 (which caused msan failure) is we don't call
Sym.computeBinding() on local symbols - VersionId is uninitialized.
llvm-svn: 361213
Summary:
Valid section or chunk alignments are powers of 2 in the range [1,
8192]. These can be stored more canonically in log2 form to free up some
bits in Chunk. Combined with D61696, SectionChunk gets 8 bytes smaller.
Reviewers: ruiu, aganea
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61698
llvm-svn: 361206
This patch is a fix for https://bugs.llvm.org/show_bug.cgi?id=41804.
We try to solve the precedence of user-specified symbol ordering file and C3 ordering provided as call graph. It deals with two case:
(1) When both --symbol-ordering-file=<file> and --call-graph-order-file=<file> are present, whichever flag comes later will take precedence.
(2) When only --symbol-ordering-file=<file> is present, it takes precedence over implicit call graph (CGProfile) generated by CGProfilePass enabled in new pass manager.
llvm-svn: 361190
We currently sort dynamic relocations by (!is_relative,symbol_index).
Add r_offset as the third key. This makes `readelf -r` debugging easier
(relocations to the same symbol are ordered by r_offset).
Refactor the test combreloc.s (renamed from combrelocs.s) to check
R_X86_64_RELATIVE, and delete --expand-relocs.
The difference from the reverted D61477 is that we keep !is_relative as
the first key. In local dynamic TLS model, DTPMOD (e.g.
R_ARM_TLS_DTPMOD32 R_X86_64_DTPMOD and R_PPC{,64}_DTPMOD) may use 0 as
the symbol index.
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D62141
llvm-svn: 361164
This reverts commit r361144. It causes a use-of-uninitialized-value in
maybeReportUndefined at llvm/tools/lld/ELF/Relocations.cpp:682, as
detected by MemorySanitizer when local-undefined-symbol.s test is run.
llvm-svn: 361162
This reverts commit r361125. This linker change breaks shared libraries
in some subtle way on x86_64. (Specifically, gold segfaults when
loading the LLVMgold.so plugin linked with lldb with this patch.)
llvm-svn: 361150
For R_TLS:
1) Delete Sym.isTls() . The assembler ensures the symbol is STT_TLS.
If not (the input is broken), we would crash (dereferencing null Out::TlsPhdr).
2) Change Sym.isUndefWeak() to Sym.isUndefined(), otherwise with --noinhibit-exec
we would still evaluate the symbol and crash.
3) Return A if the symbol is undefined. This is PR40570.
The case is probably unrealistic but returning A matches R_ABS and the
behavior of several dynamic loaders.
R_NEG_TLS is obsoleted Sun TLS we don't fully support, but
R_RELAX_TLS_GD_TO_LE_NEG is still used by GD->LE relaxation (subl $var@tpoff,%eax).
They should add the addend. Unfortunately I can't test it as compilers don't seem to generate non-zero implicit addends.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D62098
llvm-svn: 361146
For a reference to a local symbol, ld.bfd and gold error if the symbol
is defined in a discarded section but accept it if the symbol is
undefined. This inconsistent behavior seems unnecessary for us (it
probably makes sense for them as they differentiate local/global
symbols, the error would mean more code).
Weaken the condition to getSymbol(Config->IsMips64EL) == 0 to catch such
errors. The symbol index can be 0 (e.g. R_*_NONE R_ARM_V4BX) and we shouldn't error on them.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D61563
llvm-svn: 361144
Fixes PR41692.
We currently sort dynamic relocations by (!is_relative,symbol_index).
Change it to (symbol_index,r_offset). We still place relative
relocations first because R_*_RELATIVE are the only dynamic relocations
with 0 symbol index (except on MIPS, which doesn't use DT_REL[A]COUNT
anyway).
This makes `readelf -r` debugging easier (relocations to the same symbol
are ordered by r_offset).
Refactor the test combreloc.s (renamed from combrelocs.s) to check
R_X86_64_RELATIVE, and delete --expand-relocs.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D61477
llvm-svn: 361125
This is a mechanical rewrite of replaceSymbol(A, B) to A->replace(B).
I also added a comment to Symbol::replace().
Technically this change is not necessary, but this change makes code a
bit more concise.
Differential Revision: https://reviews.llvm.org/D62117
llvm-svn: 361123
Otherwise, we may set IsPreemptible (e.g. --dynamic-list) then clear it
(in replaceCommonSymbols()).
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D62107
llvm-svn: 361122
This reverts D53906.
D53906 increased p_align of PT_TLS on ARM/AArch64 to 32/64 to make the
static TLS layout compatible with Android Bionic's ELF TLS. However,
this may cause glibc ARM/AArch64 programs to crash (see PR41527).
The faulty PT_TLS in the executable satisfies p_vaddr%p_align != 0. The
remainder is normally 0 but may be non-zero with the hack in place. The
problem is that we increase PT_TLS's p_align after OutputSections'
addresses are fixed (assignAddress()). It is possible that
p_vaddr%old_p_align = 0 while p_vaddr%new_p_align != 0.
For a thread local variable defined in the executable, lld computed TLS
offset (local exec) is different from glibc computed TLS offset from
another module (initial exec/generic dynamic). Note: PR41527 said the
bug affects initial exec but actually generic dynamic is affected as
well.
(glibc is correct in that it compute offsets that satisfy
`offset%p_align == p_vaddr%p_align`, which is a basic ELF requirement.
This hack appears to work on FreeBSD rtld, musl<=1.1.22, and Bionic, but
that is just because they (and lld) incorrectly compute offsets that
satisfy `offset%p_align = 0` instead.)
Android developers are fine to revert this patch, carry this patch in
their tree before figuring out a long-term solution (e.g. a dummy .tdata
with sh_addralign=64 sh_size={0,1} in crtbegin*.o files. The overhead is
now insignificant after D62059).
Reviewed By: rprichard, srhines
Differential Revision: https://reviews.llvm.org/D62055
llvm-svn: 361090
After D62059, we don't align p_memsz of PT_TLS to p_align. The
getRelocTargetVA formula should align it instead.
It becomes clear that R_NEG_TLS and R_TLS are opposite from each other.
In i386-tls-le-align.s, I put ret after call ___tls_get_addr@plt as
otherwise ld.bfd would reject the relaxation:
TLS transition from R_386_TLS_GD to R_386_TLS_LE_32 against `a' at 0x3 in section `.text' failed
llvm-svn: 361088
As Ryan Prichard pointed out, after D62059, the TP offset is incorrect.
Add x86-64-tls-le-align.s to check this. Better formulae for both
variants should take p_vaddr%p_align into account (offset%p_align =
p_vaddr%p_align is a basic ELF requirement), but I can't find a way to
test the behavior.
llvm-svn: 361084
On Elf*_Rel targets, for a relocation to a section symbol, an R_ABS is
added which will be used by relocateOne() to compute the implicit
addend.
Addends of R_*_NONE should be ignored, so don't emit an R_ABS.
This fixes crashes on X86 and ARM because their relocateOne() do not
handle R_*_NONE.
Reviewed By: peter.smith
Differential Revision: https://reviews.llvm.org/D62052
llvm-svn: 361036
The code was added in r252352, probably to address some layout issues.
Actually PT_TLS's p_memsz doesn't need to be aligned on either variant.
ld.bfd doesn't do that.
In case of larger alignment (e.g. 64 for Android Bionic on AArch64, see
D62055), this may make the overhead smaller.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D62059
llvm-svn: 361029
Libtool concludes that the linker doesn't support shared libraries,
unless this flag is listed in the output of --help.
Differential Revision: https://reviews.llvm.org/D62053
llvm-svn: 361017
When integrating PDB output in mingw targeting build systems, it
might be a lot of extra work to specify unique file names for
the pdb output. Therefore allow omitting the actual file name
and let it implicitly be the same name as the linker output, with
a pdb extension.
As the current form of the pdb option takes a separate parameter value,
e.g. "-pdb out.pdb", it is impractical to leave out the parameter value.
Therefore, introduce a second syntax for the option, with an equals
sign, like -pdb=out.pdb, where the value easily can be omitted.
The form -pdb= for requesting pdb files with an implicit name should
work fine, even though it looks a bit unconventional in that form.
Differential Revision: https://reviews.llvm.org/D62004
llvm-svn: 361014
Change R_{386,AARCH64}_NONE yaml2obj tests/icf10.test to use assembly
Add relocation-none-{arm,x86_64}.s.
Check the referenced section survives under --gc-sections.
Check -r copies R_X86_64_NONE R_AARCH64_NONE. (Elf*_Rel arches currently have a bug)
Delete the dtrace tests as they are covered by the R_X86_64_NONE test.
Reviewed By: ruiu
Differential Revision: https://reviews.llvm.org/D62051
llvm-svn: 361013
This patch implements a limited form of autolinking primarily designed to allow
either the --dependent-library compiler option, or "comment lib" pragmas (
https://docs.microsoft.com/en-us/cpp/preprocessor/comment-c-cpp?view=vs-2017) in
C/C++ e.g. #pragma comment(lib, "foo"), to cause an ELF linker to automatically
add the specified library to the link when processing the input file generated
by the compiler.
Currently this extension is unique to LLVM and LLD. However, care has been taken
to design this feature so that it could be supported by other ELF linkers.
The design goals were to provide:
- A simple linking model for developers to reason about.
- The ability to to override autolinking from the linker command line.
- Source code compatibility, where possible, with "comment lib" pragmas in other
environments (MSVC in particular).
Dependent library support is implemented differently for ELF platforms than on
the other platforms. Primarily this difference is that on ELF we pass the
dependent library specifiers directly to the linker without manipulating them.
This is in contrast to other platforms where they are mapped to a specific
linker option by the compiler. This difference is a result of the greater
variety of ELF linkers and the fact that ELF linkers tend to handle libraries in
a more complicated fashion than on other platforms. This forces us to defer
handling the specifiers to the linker.
In order to achieve a level of source code compatibility with other platforms
we have restricted this feature to work with libraries that meet the following
"reasonable" requirements:
1. There are no competing defined symbols in a given set of libraries, or
if they exist, the program owner doesn't care which is linked to their
program.
2. There may be circular dependencies between libraries.
The binary representation is a mergeable string section (SHF_MERGE,
SHF_STRINGS), called .deplibs, with custom type SHT_LLVM_DEPENDENT_LIBRARIES
(0x6fff4c04). The compiler forms this section by concatenating the arguments of
the "comment lib" pragmas and --dependent-library options in the order they are
encountered. Partial (-r, -Ur) links are handled by concatenating .deplibs
sections with the normal mergeable string section rules. As an example, #pragma
comment(lib, "foo") would result in:
.section ".deplibs","MS",@llvm_dependent_libraries,1
.asciz "foo"
For LTO, equivalent information to the contents of a the .deplibs section can be
retrieved by the LLD for bitcode input files.
LLD processes the dependent library specifiers in the following way:
1. Dependent libraries which are found from the specifiers in .deplibs sections
of relocatable object files are added when the linker decides to include that
file (which could itself be in a library) in the link. Dependent libraries
behave as if they were appended to the command line after all other options. As
a consequence the set of dependent libraries are searched last to resolve
symbols.
2. It is an error if a file cannot be found for a given specifier.
3. Any command line options in effect at the end of the command line parsing apply
to the dependent libraries, e.g. --whole-archive.
4. The linker tries to add a library or relocatable object file from each of the
strings in a .deplibs section by; first, handling the string as if it was
specified on the command line; second, by looking for the string in each of the
library search paths in turn; third, by looking for a lib<string>.a or
lib<string>.so (depending on the current mode of the linker) in each of the
library search paths.
5. A new command line option --no-dependent-libraries tells LLD to ignore the
dependent libraries.
Rationale for the above points:
1. Adding the dependent libraries last makes the process simple to understand
from a developers perspective. All linkers are able to implement this scheme.
2. Error-ing for libraries that are not found seems like better behavior than
failing the link during symbol resolution.
3. It seems useful for the user to be able to apply command line options which
will affect all of the dependent libraries. There is a potential problem of
surprise for developers, who might not realize that these options would apply
to these "invisible" input files; however, despite the potential for surprise,
this is easy for developers to reason about and gives developers the control
that they may require.
4. This algorithm takes into account all of the different ways that ELF linkers
find input files. The different search methods are tried by the linker in most
obvious to least obvious order.
5. I considered adding finer grained control over which dependent libraries were
ignored (e.g. MSVC has /nodefaultlib:<library>); however, I concluded that this
is not necessary: if finer control is required developers can fall back to using
the command line directly.
RFC thread: http://lists.llvm.org/pipermail/llvm-dev/2019-March/131004.html.
Differential Revision: https://reviews.llvm.org/D60274
llvm-svn: 360984
This is the last patch of the series of patches to make it possible to
resolve symbols without asking SymbolTable to do so.
The main point of this patch is the introduction of
`elf::resolveSymbol(Symbol *Old, Symbol *New)`. That function resolves
or merges given symbols by examining symbol types and call
replaceSymbol (which memcpy's New to Old) if necessary.
With the new function, we have now separated symbol resolution from
symbol lookup. If you already have a Symbol pointer, you can directly
resolve the symbol without asking SymbolTable to do that.
Now that the nice abstraction become available, I can start working on
performance improvement of the linker. As a starter, I'm thinking of
making --{start,end}-lib faster.
--{start,end}-lib is currently unnecessarily slow because it looks up
the symbol table twice for each symbol.
- The first hash table lookup/insertion occurs when we instantiate a
LazyObject file to insert LazyObject symbols.
- The second hash table lookup/insertion occurs when we create an
ObjFile from LazyObject file. That overwrites LazyObject symbols
with Defined symbols.
I think it is not too hard to see how we can now eliminate the second
hash table lookup. We can keep LazyObject symbols in Step 1, and then
call elf::resolveSymbol() to do Step 2.
Differential Revision: https://reviews.llvm.org/D61898
llvm-svn: 360975
The change broke some scenarios where debug information is still
needed, although MarkLive cannot see it, including the
Chromium/Android build. Reverting to unbreak that build.
llvm-svn: 360955
Previously these sections were being generated during their
constructors. This moves the work to finalizeContent, and also does
the same for the relocation sections because their contents depends
on the final layout too.
This change is part of a larger refactor to how we deal with synthetic
sections: https://reviews.llvm.org/D61811
Differential Revision: https://reviews.llvm.org/D61971
llvm-svn: 360941
Simon Atanasyan