Use stdint types everywhere.

It's always sort of bothered me that we use the nonstandard u_intN_t types
from sys/types.h in Breakpad. Visual C++ 2010 ships a stdint.h, so it seems
reasonable to switch to just using stdint types everywhere. This patch is
mostly mechanically generated, using:
find ./src -name "*.cc" -o -name "*.h" | xargs perl -pi -e 's/u_int/uint/g'

The only manual changes were in:
breakpad_types.h,
src/processor/minidump.cc
src/processor/static_map_iterator.h
src/processor/simple_serializer.h
src/processor/disassembler_x86.h
src/client/windows/crash_generation/minidump_generator.h

I preserved the existing typedefs for pre-MSVC2010, and also added a
preprocessor define, BREAKPAD_CUSTOM_STDINT_H, for consumers using
pre-MSVC2010 who already have a stdint.h solution. Mozilla has one
in-tree since we haven't dropped support for older MSVC versions yet.

I built and ran the tests on Linux/x86-64, Windows, and Mac.

[Mark Mentovai]

Thanks for fixing the u_ error.

http://breakpad.appspot.com/535002/diff/1/src/common/mac/MachIPC.h
File src/common/mac/MachIPC.h (right):

http://breakpad.appspot.com/535002/diff/1/src/common/mac/MachIPC.h#newcode213
src/common/mac/MachIPC.h:213: uint8_t     data[1];     // actual size limited by
sizeof(MachMessage)
Fix alignment.

http://breakpad.appspot.com/535002/diff/1/src/common/mac/MachIPC.h#newcode226
src/common/mac/MachIPC.h:226: uint8_t           padding[1024]; // descriptors
and data may be embedded here
Me too.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/breakpad...
File src/google_breakpad/common/breakpad_types.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/breakpad...
src/google_breakpad/common/breakpad_types.h:74: } uint128_t;
Seems like this is going to collide with something someday.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
File src/google_breakpad/common/minidump_cpu_arm.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
src/google_breakpad/common/minidump_cpu_arm.h:80: uint64_t	fpscr;      /* FPU
status register */
Get rid of the tabs.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
File src/google_breakpad/common/minidump_cpu_ppc.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
src/google_breakpad/common/minidump_cpu_ppc.h:134: uint32_t             vrsave; 
/* Vector save */
Looks like these were supposed to line up with float_save.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
File src/google_breakpad/common/minidump_cpu_ppc64.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
src/google_breakpad/common/minidump_cpu_ppc64.h:106: uint64_t            
vrsave;  /* Vector save */
Looks like these were also supposed to line up with float_save.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
File src/google_breakpad/common/minidump_cpu_sparc.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
src/google_breakpad/common/minidump_cpu_sparc.h:85:
uint64_t	regs[MD_FLOATINGSAVEAREA_SPARC_FPR_COUNT];
Get rid of the tabs.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
File src/google_breakpad/common/minidump_format.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/minidump...
src/google_breakpad/common/minidump_format.h:237: uint32_t data_size;
Alignment, all throughout this file.

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/processor/memor...
File src/google_breakpad/processor/memory_region.h (right):

http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/processor/memor...
src/google_breakpad/processor/memory_region.h:66: virtual bool
GetMemoryAtAddress(uint64_t address, uint8_t*  value) const =0;
We can adopt the correct |= 0| form now (with the space).

http://breakpad.appspot.com/535002/diff/1/src/processor/basic_source_line_res...
File src/processor/basic_source_line_resolver.cc (right):

http://breakpad.appspot.com/535002/diff/1/src/processor/basic_source_line_res...
src/processor/basic_source_line_resolver.cc:303: uint64_t address    =
strtoull(tokens[0], NULL, 16);
Alignment, throughout this file.

http://breakpad.appspot.com/535002/diff/1/src/processor/disassembler_x86.h
File src/processor/disassembler_x86.h (right):

http://breakpad.appspot.com/535002/diff/1/src/processor/disassembler_x86.h#ne...
src/processor/disassembler_x86.h:40: #include <sys/types.h>
Did you want <sys/types.h> or breakpad_types.h?

http://breakpad.appspot.com/535002/diff/1/src/processor/windows_frame_info.h
File src/processor/windows_frame_info.h (right):

http://breakpad.appspot.com/535002/diff/1/src/processor/windows_frame_info.h#...
src/processor/windows_frame_info.h:131: uint32_t prolog_size         = 
strtoul(tokens[3],  NULL, 16);
Alignment.

[Ted Mielczarek]

Sorry about the alignment screwup, I somehow failed to consider that when
patching this.

On 2013/03/04 20:24:46, Mark Mentovai wrote:
>
http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/breakpad...
> File src/google_breakpad/common/breakpad_types.h (right):
> 
>
http://breakpad.appspot.com/535002/diff/1/src/google_breakpad/common/breakpad...
> src/google_breakpad/common/breakpad_types.h:74: } uint128_t;
> Seems like this is going to collide with something someday.

Agreed, I thought about that while patching this. Do you have a suggestion for a
replacement name, or are you okay with kicking that can down the road? (It'd be
a pretty simple search-and-replace like most of this patch if it becomes a
problem.)

>
http://breakpad.appspot.com/535002/diff/1/src/processor/disassembler_x86.h#ne...
> src/processor/disassembler_x86.h:40: #include <sys/types.h>
> Did you want <sys/types.h> or breakpad_types.h?

This needs to be here to replace the inclusion from breakpad_types.h, because
otherwise it gets included transitively through libdis.h, but inside namespace
libdis {}, which declares a bunch of types inside that namespace and breaks
things in unistd.h. We probably should fix libdis.h, but this seemed like the
easiest band-aid.

I fixed everything else, new patch in a minute.

[Mark Mentovai]

ted.mielczarek@gmail.com wrote:

> Agreed, I thought about that while patching this. Do you have a
> suggestion for a replacement name, or are you okay with kicking that can
> down the road? (It'd be a pretty simple search-and-replace like most of
> this patch if it becomes a problem.)
>

How about UInt128 or uint128_struct?

[ted_mielczarek.org]

On 3/5/2013 11:39 AM, Mark Mentovai wrote:
> ted.mielczarek@gmail.com <mailto:ted.mielczarek@gmail.com> wrote:
>
>     Agreed, I thought about that while patching this. Do you have a
>     suggestion for a replacement name, or are you okay with kicking
>     that can
>     down the road? (It'd be a pretty simple search-and-replace like
>     most of
>     this patch if it becomes a problem.)
>
>
> How about UInt128 or uint128_struct?
The latter sounds reasonable to me and is roughly in-line with the other
type names. Will update the patch in a bit.

[Ted Mielczarek]

https://breakpad.appspot.com/535002/diff/18001/src/google_breakpad/common/min...
File src/google_breakpad/common/minidump_cpu_ppc.h (right):

https://breakpad.appspot.com/535002/diff/18001/src/google_breakpad/common/min...
src/google_breakpad/common/minidump_cpu_ppc.h:100: uint32_t       save_vrvalid; 
/* Identifies which vector registers are saved*/
I know the lack of a trailing space here probably drives you nuts, but I didn't
want to spill onto a second line just to get that space in there. If it keeps
you up at night I'll fix it. :)

[Mark Mentovai]

LGTM

https://breakpad.appspot.com/535002/diff/18001/src/google_breakpad/common/min...
File src/google_breakpad/common/minidump_cpu_ppc.h (right):

https://breakpad.appspot.com/535002/diff/18001/src/google_breakpad/common/min...
src/google_breakpad/common/minidump_cpu_ppc.h:100: uint32_t       save_vrvalid; 
/* Identifies which vector registers are saved*/
Ted Mielczarek wrote:
> I know the lack of a trailing space here probably drives you nuts, but I
didn't
> want to spill onto a second line just to get that space in there. If it keeps
> you up at night I'll fix it. :)

So pick a shorter way of saying it, like “indicates” instead of “identifies.”

https://breakpad.appspot.com/535002/diff/18001/src/processor/windows_frame_in...
File src/processor/windows_frame_info.h (right):

https://breakpad.appspot.com/535002/diff/18001/src/processor/windows_frame_in...
src/processor/windows_frame_info.h:134: uint32_t saved_register_size  = 
strtoul(tokens[6],  NULL, 16);
You’ve got one more space on the left side of the = than you need in this
section.