We should not take in account a type of "source" symbol. Cross mode jump
adjustment is requred when target symbol and relocation belong to
different (regular/microMIPS) instruction sets.
llvm-svn: 231639
In the resolver, we maintain a list of undefined symbols, and when we
visit an archive file, we check that file if undefined symbols can be
resolved using files in the archive. The archive file class provides
find() function to lookup a symbol.
Previously, we call find() for each undefined symbols. Archive files
may be visited multiple times if they are in a --start-group and
--end-group. If we visit a file M times and if we have N undefined
symbols, find() is called M*N times. I found that that is one of the
most significant bottlenecks in LLD when linking a large executable.
find() is not a very cheap operation because it looks up a hash table
for a given string. And a string, or a symbol name, can be pretty long
if you are dealing with C++ symbols.
We can eliminate the bottleneck.
Calling find() with the same symbol multiple times is a waste. If a
result of looking up a symbol is "not found", it stays "not found"
forever because the symbol simply doesn't exist in the archive.
Thus, we should call find() only for newly-added undefined symbols.
This optimization makes O(M*N) O(N).
In this patch, all undefined symbols are added to a vector. For each
archive/shared library file, we maintain a start position P. All
symbols [0, P) are already searched. [P, end of the vector) are not
searched yet. For each file, we scan the vector only once.
This patch changes the order in which undefined symbols are looked for.
Previously, we iterated over the result of _symbolTable.undefines().
Now we iterate over the new vector. This is a benign change but caused
differences in output if remaining undefines exist. This is why some
tests are updated.
The performance improvement of this patch seems sometimes significant.
Previously, linking chrome.dll on my workstation (Xeon 2.4GHz 8 cores)
took about 70 seconds. Now it takes (only?) 30 seconds!
http://reviews.llvm.org/D8091
llvm-svn: 231434
The round-trip passes were introduced in r193300. The intention of
the change was to make sure that LLD is capable of reading end
writing such file formats.
But that turned out to be yet another over-designed stuff that had
been slowing down everyday development.
The passes ran after the core linker and before the writer. If you
had an additional piece of information that needs to be passed from
front-end to the writer, you had to invent a way to save the data to
YAML/Native. These passes forced us to do that even if that data
was not needed to be represented neither in an object file nor in
an executable/DSO. It doesn't make sense. We don't need these passes.
http://reviews.llvm.org/D7480
llvm-svn: 230069
Previously we applied the LayoutPass to order atoms and then
apply elf::ArrayOrderPass to sort them again. The first pass is
basically supposed to sort atoms in the normal fashion (which
is to sort symbols in the same order as the input files).
The second pass sorts atoms in {init,fini}_array.<priority> by
priority.
The problem is that the LayoutPass is overkill. It analyzes
references between atoms to make a decision how to sort them.
It's slow, hard to understand, and above all, it doesn't seem
that we need its feature for ELF in the first place.
This patch remove the LayoutPass from ELF pass list. Now all
reordering is done in elf::OrderPass. That pass sorts atoms by
{init,fini}_array, and if they are not in the special section,
they are ordered as the same order as they appear in the command
line. The new code is far easier to understand, faster, and
still able to create valid executables.
Unlike the previous layout pass, elf::OrderPass doesn't count
any attributes of an atom (e.g. permissions) except its
position. It's OK because the writer takes care of them if we
have to.
This patch changes the order of final output, although that's
benign. Tests are updated.
http://reviews.llvm.org/D7278
llvm-svn: 227666
That kind of reference was used only in ELFFile, and the use of
that reference there didn't seem to make sense. All test still
pass (after adjusting symbol names) without that code. LLD is
still be able to link LLD and Clang. Looks like we just don't
need this.
http://reviews.llvm.org/D7189
llvm-svn: 227259
This is initial patch to support MIPS64 object files linking.
The patch just makes some classes more generalized, and rejects
attempts to interlinking O32 and N64 ABI object files.
I try to reuse the current MIPS target related classes as much as
possible because O32 and N64 MIPS ABI are tightly related and share
almost the same set of relocations, GOT, flags etc.
llvm-svn: 227058
The previous default behavior of LLD is --as-needed. LLD linked
against a DSO only if the DSO file was actually used to link an
executable (i.e. at least one symbol was resolved using the shared
library file.)
In this patch I added a boolean flag to FileNode for --as-needed.
I also added an accessor to DSO name to shared library file class.
llvm-svn: 226274
If a regular symbol has microMIPS-bit we need to stop linking. Now the
LLD does not check the `applyRelocation` return value and continues
linking anyway. As a temporary workaround use the `llvm_unreachable`
call to stop the linker.
llvm-svn: 224831
The LLD output in the YAML mode depends on LLD_RUN_ROUNDTRIP_TEST
environment variable. Do not check unimportant YAML items like section-name.
llvm-svn: 224830
Some targets like microMIPS and ARM Thumb use the last bit of a symbol's
value to mark 'compressed' code. This patch adds new virtual function
`DynamicTable::getAtomVirtualAddress` which allows to adjust a symbol's
value before using it in a dynamic table tags like DT_INIT / DT_FINI.
llvm-svn: 223963
To find an AtomLayout object for the given symbol I replace the
`Layout::findAtomAddrByName` method by `Layout::findAtomLayoutByName` method.
llvm-svn: 223359
The AtomSections were improperly merging sections from various input files. This
patch fixes the problem, with an updated test that was provided by Simon.
Thanks to Simon Atanasyan for catching this issue.
llvm-svn: 222982
MIPS ELF symbols might contain some additional MIPS-specific flags
in the st_other field besides visibility ones. These flags indicate
code properties like microMIPS / MIPS16 encoding, position independent
code etc. We need to transfer the flags from input objects to the
output linked file to write them into the symbol table, adjust symbols
addresses etc.
I add new attribute CodeModel to the DefinedAtom class to hold target
specific flag and to get over YAML/Native format conversion barrier.
Other architectures/targets can extend CodeModel enumeration by their
own flags.
MIPS specific part of this patch adds support for STO_MIPS_MICROMIPS
flag. This flag marks microMIPS symbols. Such symbol should:
a) Has STO_MIPS_MICROMIPS in the corresponding .symtab record.
b) Has adjusted (odd) address in the corresponding .symtab
and .dynsym records.
llvm-svn: 221864
The dynamic table was creating the entry DT_FINI_ARRAY{SZ} even when there was
no .fini_array section. The entries should be creating in the dynamic section
only if there are sections .init_array/.fini_array in the output.
Fixes the tests that checked for errroneous outputs.
llvm-svn: 221588
1. The path checks ELF header flags to prevent linking of incompatible files.
For example we do not allow to link files with different ABI, -mnan
flags, some combination of target CPU etc.
2. The patch merge ELF header flags from input object files to put their
combination to the generated file. For example, if some input files
have EF_MIPS_NOREORDER flag we need to put this flag to the output
file header.
I use the `parseFile()` (not `canParse()`) method because in case of
recognition of incorrect input flags combination we should show detailed
error message and stop the linking process and should not try to use
another `Reader`.
llvm-svn: 221439
The commit of the .ent/.end implementation will change the result of the
relocation evaluations (because of a new section and additional relocations)
which will lead to a failure if the .ent/.end directives are present in this
test.
We don't really need the .ent/.end directives in this test so let's just remove
them to preserve the current output.
llvm-svn: 215534
Simon Atanasyan