Currently .drectve section contents are parsed after other sections are parsed.
That order may result in wrong results if other sections depend on command line
options in the directive section.
For example, if a weak symbol is defined using /alternatename option in the
directive section, we have to read it first and then read the text section
contents. Otherwise the weak symbol won't be defined.
This patch changes the order to fix the issue.
llvm-svn: 198071
There are many object files in the standard library who have empty .drective
sections. Parsing the empty string is not wrong but a waste.
llvm-svn: 198067
The main changes are in:
include/lld/Core/Reference.h
include/lld/ReaderWriter/Reader.h
Everything else is details to support the main change.
1) Registration based Readers
Previously, lld had a tangled interdependency with all the Readers. It would
have been impossible to make a streamlined linker (say for a JIT) which
just supported one file format and one architecture (no yaml, no archives, etc).
The old model also required a LinkingContext to read an object file, which
would have made .o inspection tools awkward.
The new model is that there is a global Registry object. You programmatically
register the Readers you want with the registry object. Whenever you need to
read/parse a file, you ask the registry to do it, and the registry tries each
registered reader.
For ease of use with the existing lld code base, there is one Registry
object inside the LinkingContext object.
2) Changing kind value to be a tuple
Beside Readers, the registry also keeps track of the mapping for Reference
Kind values to and from strings. Along with that, this patch also fixes
an ambiguity with the previous Reference::Kind values. The problem was that
we wanted to reuse existing relocation type values as Reference::Kind values.
But then how can the YAML write know how to convert a value to a string? The
fix is to change the 32-bit Reference::Kind into a tuple with an 8-bit namespace
(e.g. ELF, COFFF, etc), an 8-bit architecture (e.g. x86_64, PowerPC, etc), and
a 16-bit value. This tuple system allows conversion to and from strings with
no ambiguities.
llvm-svn: 197727
GroupedSectionsPass was a complicated pass. That pass's job was to reorder
atoms by section name, so that the atoms with the same section prefix will be
emitted consecutively to the executable. The pass added layout edges to atoms,
and let the layout pass to actually reorder them.
This patch simplifies the design by making GroupedSectionPass to directly
reorder atoms, rather than adding layout edges. This resembles ELF's
ArrayOrderPass.
This patch improves the performance of LLD; it used to take 7.1 seconds to
link LLD with LLD on my Macbook Pro, but it now takes 6.1 seconds.
llvm-svn: 196628
Atom ordinals are the indeces in a file. Currently the PECOFF reader assigns
ordinals for each section, so it's (incorrectly) assigning duplicate ordinals.
llvm-svn: 195852
According to the PE/COFF spec, a section with IMAGE_SCN_LNK_INFO should only
appear in an object file, and not allowed in an executable. So I believe
treating it as the same way as IMAGE_SCN_LNK_INFO is the right thing.
llvm-svn: 195692
This patch won't change LLD's behavior because it's a temporary file and
LLD does not use the file extension to determine file type. But using the
correct file extension is a good thing.
llvm-svn: 193211
We should dead-strip atoms only if they are created for COMDAT symbols. If we
remove non-COMDAT atoms from a binary, it will no longer be guaranteed that
the binary will work correctly.
In COFF, you can manipulate the order of section contents in the resulting
binary by section name. For example, if you have four sections
.data$unique_prefix_{a,b,c,d}, it's guaranteed that the contents of A, B, C,
and D will be consecutive in the resulting .data section in that order.
Thus, you can access B's and C's contents by incrementing a pointer pointing
to A until it reached to D. That's why we cannot dead-strip B or C even if
no one is directly referencing to them.
Some object files in the standard library actually use that technique.
llvm-svn: 193017
Changes :-
a) Functionality in InputGraph to insert Input elements at any position
b) Functionality in the Resolver to use nextFile
c) Move the functionality of assigning file ordinals to InputGraph
d) Changes all inputs to MemoryBuffers
e) Remove LinkerInput, InputFiles, ReaderArchive
llvm-svn: 192081
This patch inverts the return value of these functions, so that they return
"true" on success and "false" on failure. The meaning of boolean return value
was mixed in LLD; for example, InputGraph::validate() returns true on success.
With this patch they'll become consistent.
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D1748
llvm-svn: 191341
There was a bug that if a section has an alignment requirement and there are
multiple symbols at offset 0 in the section, only the last atom at offset 0
would be aligned properly. That bug would move only the last symbol to an
alignment boundary, leaving other symbols unaligned, although they should be at
the same location. That caused a mysterious SEGV error of the resultant
executable.
With this patch, we manage all symbols at the same location properly, rather
than keeping the last one.
llvm-svn: 190724
The compiler is allowed to add a linker option starting with -?<name> to
.drectve section. If the linker can interpret -<name>, it's processed as if
there's no question mark there. If not, such option is silently ignored.
This is a COFF's feature to allow the compiler to emit new linker options
while keeping compatibility with older linkers.
llvm-svn: 189897
Because of a bug, the last atom of each section contained a garbage at the
end of its data. In most cases the garbage is harmless but it could have cause
SEGV.
llvm-svn: 189572
We were creating undefined atoms for common symbols by mistake. That did not
lead to a link failure, for undefined atoms would be resolved by common symbols
in the same file, but that's a waste of resource.
llvm-svn: 189534
We scanned the symbol table twice; first to gather all regular symbols, and
second to process aux symbols. That's a bit inefficient and complicated. We
can instead cache aux symbols in the first pass, to eliminate the need of the
second pass.
llvm-svn: 189525
The COMDAT section is a section with a special attribute to tell the linker
whether the symbols in the section are allowed to be merged or not. This patch
add a function to interpret the COMDAT data and set "merge" attribute to the
atoms accordingly.
LLD supports multiple policies to merge atoms; atoms can be merged by name or
by content. COFF supports them, and in addition to that, it supports
choose-the-largest-atom policy, which LLD currently does not support. I simply
mapped it to merge-by-name attribute for now, but we eventually have to support
that policy in the core linker.
llvm-svn: 188025
Also change some local variable names: "ti" -> "context" and
"_targetInfo" -> "_context".
Differential Revision: http://llvm-reviews.chandlerc.com/D1301
llvm-svn: 187823
The aim of this patch is to reduce the dependency from COFFDefinedAtom
to COFF structs defined in llvm/Object/COFF.h. Currently many attributes
of the atom are computed in the atom. That provide a simple interface but
does not work well in some cases.
There are some cases that the same type atom is created from different
parts of a COFF file. One example is the BSS atom, which can be created
from the defined symbol in the .bss section or from the undefined symbol.
Computing attributes from different sources in the atom complicates the
code. We should compute it outside the atom.
In the next patch, I'll move more code from Atoms.h to ReaderCOFF.cpp.
llvm-svn: 187681
Summary:
The .drectve section contains linker command line options, and the linker is
expected to interpret them as if they were given via the command line. In this
patch, the command line parser in the driver is called from the object file
reader to parse the string.
I think this patch is important, because this is the first step towards mutable
TargetInfo. We had a discussion about that on llvm-commits mailing list before.
I haven't removed "const" from the function signature yet. Instead, I just use
cast to remove "const". This is a temporary aid for an experiment. If we don't
see any issue with this mutable TargetInfo appraoch, I'll change the function
signature, and rename the class LinkerContext from TargetInfo.
Reviewers: kledzik
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D1246
llvm-svn: 187677
For an invalid input we should not call report_fatal_error(), because
when LLD is used as a library, we don't want to kill the whole app
because of a malformed input.
llvm-svn: 187673
A instance of the class always represents a BSS atom, so we don't need
to look at the symbol or the section to retrieve its attributes.
llvm-svn: 187643
The BSS atom is similar to the regular defined atom, but it's different
in the sense that it does not have contents. Until now we assumed all the
defined atoms have its contents. That did not fit well to the BSS atom.
llvm-svn: 187453
Member functions to read the symbol table had too many parameters to propagate
all the temporary information from one to another. By storing the information
to data members, we can simplify the function signatures and improve the
readability.
llvm-svn: 187321
Some sections, such as with IMAGE_SCN_LNK_REMOVE attribute, is skipped
in the first pass. Such sections need to be skipped in the latter passes.
llvm-svn: 187281
Extract atom definitions as Atoms.h so that we can use them in other files.
Also applied clang-format to Atoms.h.
Reviewers: shankarke
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D995
llvm-svn: 184124
This is the first patch toward full DLL support. With this patch, lld can
read .lib file for a DLL.
Reviewers: Bigcheese
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D987
llvm-svn: 184101
Archive file in Windows has file extension of ".lib" but the file format is
in fact the same as Unix. It's an ar archive holding multiple .obj files.
The existing archive reader can read .lib files.
llvm-svn: 184036
Summary:
COFFReference class is defined to represent relocation information for
COFFDefinedAtom, as ELFReference for ELFDefinedAtom. ReaderCOFF can now
read relocation entries and create COFFReferences accordingly.
I need to make WriterPECOFF to handle the relocation references created by
the reader, but this patch is already big, so I think it's probably better
to get it reviewed now.
Reviewers: Bigcheese
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D976
llvm-svn: 183964
Architecture specific code should reside in architecture specific directory
not in Atom. Looks like there are no efforts being made at this moment to
support ARM, so let's remove it for now.
Reviewers: Bigcheese
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D959
llvm-svn: 183877
Split FileCOFF's constructor into mainly two private methods.
One method is responsible to iterate over symbol tables, and other
method is to atomize defined atoms. This is for readability and
no changes in functionality.
Reviewers: Bigcheese
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D940
llvm-svn: 183708
Add WinLinkDriver and connect it to the existing COFF reader. Remaining
parts are still stubs, so while it can now read a COFF file, it still
cannot link or output PE/COFF files yet.
Reviewers: Bigcheese
CC: llvm-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D865
llvm-svn: 182784
The major changes are:
1) LinkerOptions has been merged into TargetInfo
2) LinkerInvocation has been merged into Driver
3) Drivers no longer convert arguments into an intermediate (core) argument
list, but instead create a TargetInfo object and call setter methods on
it. This is only how in-process linking would work. That is, you can
programmatically set up a TargetInfo object which controls the linking.
4) Lots of tweaks to test suite to work with driver changes
5) Add the DarwinDriver
6) I heavily doxygen commented TargetInfo.h
Things to do after this patch is committed:
a) Consider renaming TargetInfo, given its new roll.
b) Consider pulling the list of input files out of TargetInfo. This will
enable in-process clients to create one TargetInfo the re-use it with
different input file lists.
c) Work out a way for Drivers to format the warnings and error done in
core linking.
llvm-svn: 178776
I really would have liked to split this patch up, but it would greatly
complicate the lld-core and lld drivers having to deal with both
{Reader,Writer}Option and TargetInfo.
llvm-svn: 173217
* Moves enum Scope from DefinedAtom.h to Atom.h
* Adds scope method to AbsoluteAtom class
* Updates YAML to print scope of AbsoluteAtoms
* Update Native Reader/Writer to handle this new, "attribute"
* Adds testcase
Reviewed and approved by Nick Kledzik
llvm-svn: 166189
now Reader and Writer subclasses for each file format. Each Reader and
Writer subclass defines an "options" class which controls how that Reader
or Writer operates.
llvm-svn: 157774
Rafael Espindola