llvm-svn: 194130

/defaultlib options can be specified implicitly via the .drectve section, and
it's pretty common that multiple object files add the same library, such as
user32.lib, to the input. We shouldn't add the same library multiple times.

llvm-svn: 194129

llvm-svn: 194128

We wrapped the linker internal file with a virtual archive file, so that the
linker internal file was linked only when it's actually used. This was to avoid
__ImageBase being included to the resulting executable. __ImageBase used to
occupy four bytes when emitted to executable.

And then it turned out that the implementation of __ImageBase was wrong -- it
shouldn't have been a regular atom but an absolute atom. Absolute atoms point
to some memory location, but they don't occupy disk space themselves. So it
wouldn't increase executable size (except the symbol table.) That means that
it's OK to link the linker internal file unconditionally.

So this patch does that, removing the wrapper archive file. Doing this
simplifies the code.

llvm-svn: 194127

msvcrt.lib contains "/disallowlib" command line option in its .drectve section.
I couldn't spot any documentation for the option. Ignore it for now so that we
can link the library without error.

llvm-svn: 194114

Errors in .drectve section were silently ignored. This patch fixes the issue.

llvm-svn: 194110

llvm-svn: 194089

llvm-svn: 194037

It was never transporting any value in addition to the error_code.

llvm-svn: 194028

Patch reviewed by Shankar Easwaran.

llvm-svn: 194010

MSVC 2012 raises an error in the lambda passed to vector::find_if, while it
seems valid code. Rewrote without high-order functions.

llvm-svn: 193975

n_desc field in MachO string table was not initialized. On Unix,
test/darwin/hello-world.objtxt did not fail because I think an nlist object
is always allocated to a fresh heap initialized with zeros. On Windows,
uninitialized fields are filled with 0xCC when compiled with /GZ. Because
of that the test was failing on Windows.

llvm-svn: 193909

Bugs that would be caught by this assertion would also be caught by
RoundTripYAMLPass test. We've enabled the pass for PECOFF, so we can remove
this.

llvm-svn: 193886

llvm-svn: 193885

llvm-svn: 193883

llvm-svn: 193882

llvm-svn: 193881

The data directory in the PE/COFF header consisted of list of data directory
atoms. This patch changes it -- now there's only one data directory entry that
contains former data directories. That's easier to handle in the writer as well
as to write to/read from YAML/Native files. The main purpose of this refactoring
is to enable RoundTrip tests for PE/COFF.

There's no functionality change.

llvm-svn: 193854

llvm-svn: 193797

This reverts commit r193479.

The atoms are already added to the file, so re-adding them caused the YAML
writer to write the same atoms twice. That made the YAML reader to fail with
"duplicate atom name" error.

This is not the only error we've got for RoundTripYAMLPass for PECOFF, so we
cannot enable the test yet. More fixes will come.

Differential Revision: http://llvm-reviews.chandlerc.com/D2069

llvm-svn: 193762

llvm-svn: 193712

llvm-svn: 193662

llvm-svn: 193646

Enable this for the following flavors

a) core
b) gnu
c) darwin

Its disabled for the flavor PECOFF. Convenient markers are added with FIXME
comments in the Driver that would be removed and code removed from each flavor.

llvm-svn: 193585

__ImageBase is an absolute symbol whose address is the same as the image base
address. What we did before this patch was to create __ImageBase symbol as a
symbol whose *contents* (not location) is the image base address, which is
clearly wrong.

llvm-svn: 193565

This change has no effect now, but will be needed to emit BSS atoms
to .bss section.

llvm-svn: 193556

llvm-svn: 193551

intended for debugging and diagnostic output), just inspect the spelling
to check for specific prefixes in drectve section flags.

In addition to being significantly cheaper and not relying on
a debugging interface, this also avoids creating a temporary string and
binding it to StringRef variable. We then went on to access it after the
memory had been deallocated.

This bug too was caught by ASan. I love ASan so much. =]

llvm-svn: 193487

llvm-svn: 193482

llvm-svn: 193481

llvm-svn: 193480

llvm-svn: 193479

llvm-svn: 193478

llvm-svn: 193477

On discussing this with Nick, it looks like the StubAtoms
that contain a lazyImmediate reference kind should be null
and the location needs to be fixed up later with some value
that is an offset into the __LINKEDIT segment.

The drawback is that it allows yaml files with references
that expect a target to be considered without one.

This results in bad yaml files that would need to be handled
in the YAML Reader.

Inorder to fix this, the Stub Atoms use a dummy target such
as itself.

llvm-svn: 193476

/merge:<from>=<to> option makes the linker to combine "from" section to "to"
section. This patch is to parse the option. The actual feature will be
implemented in a subsequent patch.

llvm-svn: 193454

llvm-svn: 193451

llvm-svn: 193446

We really need a test for the manifest file output, but because it depends
on external commands (CVTRES.EXE and RC.EXE), it's not very easy to write it.

llvm-svn: 193445

llvm-svn: 193444