win: Write memory map info as MINIDUMP_MEMORY_INFO[_LIST]

win: Write memory map info as MINIDUMP_MEMORY_INFO[_LIST]

Makes !vprot work in windbg, e.g.

0:000> !vprot 0x970000
BaseAddress:       00970000
AllocationBase:    00970000
AllocationProtect: 00000004  PAGE_READWRITE
RegionSize:        00001000
State:             00001000  MEM_COMMIT
Protect:           00000001  PAGE_NOACCESS
Type:              00020000  MEM_PRIVATE

...

0:000> !vprot 0x97a000
BaseAddress:       0097a000
AllocationBase:    00970000
AllocationProtect: 00000004  PAGE_READWRITE
RegionSize:        00001000
State:             00001000  MEM_COMMIT
Protect:           00000140  PAGE_EXECUTE_READWRITE + PAGE_GUARD
Type:              00020000  MEM_PRIVATE


Follows https://codereview.chromium.org/1377133006.

R=mark@chromium.org
BUG=crashpad:20, crashpad:46
Committed: https://chromium.googlesource.com/crashpad/crashpad/+/019a0cec8b8e0fd5f47c564772aa7a7d88921b79

[scottmg, 2015-10-07 22:33:27]

This is the "other one" that I mentioned. Needs
https://codereview.chromium.org/1377133006 first though.

[Mark Mentovai, 2015-10-12 14:34:47]

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/dbghelp.h
File compat/non_win/dbghelp.h (right):

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/dbghelp...
compat/non_win/dbghelp.h:865: //! `MEM_COMMIT`, `MEM_FREE`, or `MEM_RESERVE`.
Since we have our own versions of these now, this should become the ugly

\ref MEM_x MEM_COMMIT, \ref MEM_x MEM_FREE, or \ref MEM_x MEM_RESERVE.

Similar in the comments for the Type field below and the AllocationProtect field
above.

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h
File compat/non_win/winnt.h (right):

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h...
compat/non_win/winnt.h:182: //! \anchor PAGE_*
PAGE_x, because a * won’t work here. Same on line 201.

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h...
compat/non_win/winnt.h:188: #define PAGE_NOACCESS 0x01
Meh on the leading zeroes, which none of the other constants have. Just 0x1,
0x2, ….

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.cc:49: total_size_ =
sizeof(MINIDUMP_MEMORY_INFO_LIST) +
You don’t need total_size_ as a member, you can compute it in SizeOfObject().
That’s what the other list writers do.

It should be sizeof(memory_info_list_) + items_.size() * sizeof(items_[0]). Use
sizeof with variables instead of types when possible.

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.h (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:65: MINIDUMP_MEMORY_INFO_LIST
memory_info_list_;
Most of the list writers would put _base on the end of this name
(memory_info_list_base_)

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer_test.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer_test.cc:30: const MinidumpStreamType
kBogusStreamType =
This never comes into play, you can get rid of everything related to it
including the expected_streams argument.

https://codereview.chromium.org/1379873005/diff/360001/minidump/test/minidump...
File minidump/test/minidump_writable_test_util.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/test/minidump...
minidump/test/minidump_writable_test_util.cc:186: enum : size_t { kElementSize =
sizeof(MINIDUMP_MEMORY_INFO) };
Not mandated by the Minidump format, but since there isn’t anything like
MINIDUMP_MEMORY_INFO2 for now, this is OK.

https://codereview.chromium.org/1379873005/diff/360001/snapshot/test/test_mem...
File snapshot/test/test_memory_map_region_snapshot.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/snapshot/test/test_mem...
snapshot/test/test_memory_map_region_snapshot.cc:21: memset(&memory_info_, 0,
sizeof(memory_info_));
Use the initializer form (memory_info()) instead of a memset().

[Mark Mentovai, 2015-10-12 14:47:30]

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h
File compat/non_win/winnt.h (right):

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h...
compat/non_win/winnt.h:182: //! \anchor PAGE_*
Mark Mentovai wrote:
> PAGE_x, because a * won’t work here. Same on line 201.

Note: this applies only to \anchor. \name is pretty free-form, the * is correct
there.

[scottmg, 2015-10-13 19:51:29]

Thanks

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/dbghelp.h
File compat/non_win/dbghelp.h (right):

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/dbghelp...
compat/non_win/dbghelp.h:865: //! `MEM_COMMIT`, `MEM_FREE`, or `MEM_RESERVE`.
On 2015/10/12 14:34:46, Mark Mentovai wrote:
> Since we have our own versions of these now, this should become the ugly
> 
> \ref MEM_x MEM_COMMIT, \ref MEM_x MEM_FREE, or \ref MEM_x MEM_RESERVE.
> 
> Similar in the comments for the Type field below and the AllocationProtect
field
> above.

Done.

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h
File compat/non_win/winnt.h (right):

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h...
compat/non_win/winnt.h:182: //! \anchor PAGE_*
On 2015/10/12 14:34:46, Mark Mentovai wrote:
> PAGE_x, because a * won’t work here. Same on line 201.

Done.

https://codereview.chromium.org/1379873005/diff/360001/compat/non_win/winnt.h...
compat/non_win/winnt.h:188: #define PAGE_NOACCESS 0x01
On 2015/10/12 14:34:46, Mark Mentovai wrote:
> Meh on the leading zeroes, which none of the other constants have. Just 0x1,
> 0x2, ….

Done.

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.cc:49: total_size_ =
sizeof(MINIDUMP_MEMORY_INFO_LIST) +
On 2015/10/12 14:34:46, Mark Mentovai wrote:
> You don’t need total_size_ as a member, you can compute it in SizeOfObject().
> That’s what the other list writers do.
> 
> It should be sizeof(memory_info_list_) + items_.size() * sizeof(items_[0]).
Use
> sizeof with variables instead of types when possible.

Done.

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.h (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:65: MINIDUMP_MEMORY_INFO_LIST
memory_info_list_;
On 2015/10/12 14:34:47, Mark Mentovai wrote:
> Most of the list writers would put _base on the end of this name
> (memory_info_list_base_)

Done.

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer_test.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/minidump_memo...
minidump/minidump_memory_info_writer_test.cc:30: const MinidumpStreamType
kBogusStreamType =
On 2015/10/12 14:34:47, Mark Mentovai wrote:
> This never comes into play, you can get rid of everything related to it
> including the expected_streams argument.

Done.

https://codereview.chromium.org/1379873005/diff/360001/minidump/test/minidump...
File minidump/test/minidump_writable_test_util.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/minidump/test/minidump...
minidump/test/minidump_writable_test_util.cc:186: enum : size_t { kElementSize =
sizeof(MINIDUMP_MEMORY_INFO) };
On 2015/10/12 14:34:47, Mark Mentovai wrote:
> Not mandated by the Minidump format, but since there isn’t anything like
> MINIDUMP_MEMORY_INFO2 for now, this is OK.

Good point, I was only looking at it from one side.

https://codereview.chromium.org/1379873005/diff/360001/snapshot/test/test_mem...
File snapshot/test/test_memory_map_region_snapshot.cc (right):

https://codereview.chromium.org/1379873005/diff/360001/snapshot/test/test_mem...
snapshot/test/test_memory_map_region_snapshot.cc:21: memset(&memory_info_, 0,
sizeof(memory_info_));
On 2015/10/12 14:34:47, Mark Mentovai wrote:
> Use the initializer form (memory_info()) instead of a memset().

Done.

[Mark Mentovai, 2015-10-13 20:09:29]

LGTM

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.h (right):

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:25: #include "base/memory/scoped_ptr.h"
Unused

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:28: #include
"util/stdlib/pointer_container.h"
Unused.

[scottmg, 2015-10-13 20:15:32]

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
File minidump/minidump_memory_info_writer.h (right):

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:25: #include "base/memory/scoped_ptr.h"
On 2015/10/13 20:09:29, Mark Mentovai wrote:
> Unused

Done.

https://codereview.chromium.org/1379873005/diff/400001/minidump/minidump_memo...
minidump/minidump_memory_info_writer.h:28: #include
"util/stdlib/pointer_container.h"
On 2015/10/13 20:09:29, Mark Mentovai wrote:
> Unused.

Done.

[scottmg, 2015-10-13 20:15:49]

Committed patchset #22 (id:420001) manually as
019a0cec8b8e0fd5f47c564772aa7a7d88921b79 (presubmit successful).