Add crashpad_info, MachOImageAnnotationsReader, and its test

Add crashpad_info, MachOImageAnnotationsReader, and its test.

TEST=snapshot_test MachOImageAnnotationsReader.*
R=rsesek@chromium.org

Committed: https://chromium.googlesource.com/crashpad/crashpad/+/b43f510a529e2dbdc9da79345f26ee4043a3b7dd

BUG=

[Mark Mentovai, 2014-10-14 21:36:58]

mark@chromium.org changed reviewers:
+ rsesek@chromium.org

[Robert Sesek, 2014-10-15 20:25:21]

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
File util/mac/mach_o_image_annotations_reader.cc (right):

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:90: ->ReadCStringSizeLimited(
nit: break after ( instead of -> ?

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:126:
->ReadCStringSizeLimited(error_string_address, 1024, &message)) {
Globalize and use the kMaxMessageSize from above.

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:169: struct Entry {
Why not use SimpleStringDictionary::Entry here?

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:173: const size_t kNumEntries = 64;
Similarly, SimpleStringDictionary::num_entries.

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
File util/mac/mach_o_image_annotations_reader.h (right):

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.h:63: std::vector<std::string> Vector()
const;
What if there are errors during the reading process?

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.h:67: std::map<std::string,
std::string> SimpleMap() const;
Why SimpleMap() instead of just Map() to match Vector()?

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
File util/mac/mach_o_image_annotations_reader_test.cc (right):

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader_test.cc:151: expected_annotation) {
nit: indent +4

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader_test.cc:169: kExpectedAnnotation) {
+4

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader_test.cc:253:
simple_annotations->SetKeyValue("#TEST# pad", "crash");
Maybe also add a key/value pair that doesn't have the same length.

[Mark Mentovai, 2014-10-15 21:20:21]

Most feedback addressed.

What do you think about the layering problem?

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
File util/mac/mach_o_image_annotations_reader.cc (right):

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:126:
->ReadCStringSizeLimited(error_string_address, 1024, &message)) {
Robert Sesek wrote:
> Globalize and use the kMaxMessageSize from above.

But they are different sizes. We could reasonably raise the one above to 2048 or
16384 or remove it entirely, but this one should stay 1024, because that’s how
it’s declared in dyld.

https://codereview.chromium.org/651283003/diff/20001/util/mac/mach_o_image_an...
util/mac/mach_o_image_annotations_reader.cc:169: struct Entry {
Robert Sesek wrote:
> Why not use SimpleStringDictionary::Entry here?

Layering violation…util shouldn’t depend on client.

Should we move SimpleStringDictionary into util? I thought about it, but it’s
really only useful in client, except for the constants. I would never expect any
code outside of client to use the SimpleStringDictionary implementation.

[Mark Mentovai, 2014-10-17 16:20:31]

Patchset #5 (id:80001) has been deleted

[Mark Mentovai, 2014-10-17 16:22:14]

Updated as discussed to resolve the layering problem.

This now depends on https://codereview.chromium.org/666483002/, which moves all
of the reader bits from ProcessReader and everything that uses a ProcessReader
from util into snapshot. MachOImageAnnotationsReader is now part of snapshot
too. snapshot can depend on client for structure definitions and constants.

[Robert Sesek, 2014-10-17 17:17:23]

LGTM. Glad we could sort out that layering issue.

[Mark Mentovai, 2014-10-17 17:38:22]

Me too.

[Mark Mentovai, 2014-10-17 17:47:11]

Committed patchset #6 (id:120001) manually as
b43f510a529e2dbdc9da79345f26ee4043a3b7dd (presubmit successful).