Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Jacek Caban <jacek@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Also change return type to void since the function can no longer fail.
Signed-off-by: Jacek Caban <jacek@codeweavers.com>
Signed-off-by: Henri Verbeet <hverbeet@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Austin English <austinenglish@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Andrey Gusev <andrey.goosev@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Thomas Faber <thomas.faber@reactos.org>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Michael Stefaniuc <mstefani@redhat.de>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Ralf Habacker <ralf.habacker@freenet.de>
Signed-off-by: Marcus Meissner <marcus@jet.franken.de>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Austin English <austinenglish@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Android libraries will be handled differently.
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: André Hentschel <nerv@dawncrow.de>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Qian Hong <qhong@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

Signed-off-by: Qian Hong <qhong@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>

dbghelp: Provide a default value for DYLD_FALLBACK_LIBRARY_PATH, which is closer to how dyld behaves.

dbghelp: On Mac, get the wineloader path from the target or our own process before resorting to guessing.

This makes backtraces more reliably complete when WINELOADER isn't set
and the loader isn't in a typical location.

Also fixes a bug in UnDecorateSymbolName when undecorated_length == 0.

Images in the shared cache have their segments mapped non-contiguously.  We
don't know how to find where the non-__TEXT segments actually are.  Also, the
code assumes that the image is mapped contiguously in various places.  So, we
just ignore those segments.

Treating the non-contiguous segments as though they were contiguous caused
their apparent memory ranges to overlap, leading to addresses being attributed
to the wrong module, let alone symbol.

Such external debug info files are created by the dsymutil command.  Usually, it
creates a <name>.dSYM bundle directory but it can also produce a <name>.dwarf
flat file.  The code will find and use either.  It first checks if such a bundle
or file is adjacent to the module binary.  If it's not, it uses Spotlight to
search for a .dSYM bundle based on the binary's UUID, which is what gdb and lldb
do, too.  That way, it can find it anywhere on the system.

dbghelp: Don't parse the DWARF info from Mach-O files if we were requested to only provide public symbols.

This now works more like how the ELF and PE support works.

For backward compatibility with old, buggy GCC, Apple uses a different register
numbering scheme for the eh_frame section.

See, for reference, the comments near the top of this file from LLDB's source:
https://github.com/llvm-mirror/lldb/blob/release_36/source/Plugins/Process/Utility/RegisterContext_x86.h

At this point, this is just the .eh_frame information used for walking the stack.

It was set, but never read.

From outside of macho_module.c, it's only called with NULL.  When it's
called with a non-NULL parameter, it's just a thin wrapper around
macho_load_debug_info_from_map(), so the code can just call that
directly.

OS X uses a struct nlist_64 for 64-bit images, where the n_value field is a
64-bit unsigned integer.

These are from MSVC 2013.