report assertion info in minidump_stackwalk

Apparently I wrote this patch two years ago and forgot about it. Only one hunk
failed to apply, so I fixed that up, cleaned up a few things and it does what I
want. I've added an extra line of output to the minidump_stackwalk output, and
for the machine-readable output, I just made it output the assertion info (if
available) on the "Crash" line instead of "No Crash".

Mark, can you take a look?

This covers:
http://code.google.com/p/google-breakpad/issues/detail?id=170

[Mark Mentovai]

LG with these changes.

http://breakpad.appspot.com/45001/diff/1/9
File src/google_breakpad/processor/minidump.h (right):

http://breakpad.appspot.com/45001/diff/1/9#newcode676
Line 676: MDRawAssertionInfo   assertion_;
extra space on this line?

http://breakpad.appspot.com/45001/diff/1/4
File src/processor/minidump.cc (right):

http://breakpad.appspot.com/45001/diff/1/4#newcode191
Line 191: if (in_word == 0)
I don't think so.

The vector should be sized correctly already.  Are you hitting a case where it
isn't?

http://breakpad.appspot.com/45001/diff/1/4#newcode2773
Line 2773: 
This file uses two blank lines around these sections.  Same at line 2865.

http://breakpad.appspot.com/45001/diff/1/4#newcode2810
Line 2810: u_int32_t byte_length = sizeof(assertion_.expression);
Do you want to check that byte_length > 0?  I think that's a good idea.  Your
use of &expression_utf16[0] is invalid if you created that vector with
char_length 0.

Do you want to check that byte_length % 2 == 0?  Do we do that in the other
spots in this file where we read UTF-16 from the minidump?

http://breakpad.appspot.com/45001/diff/1/6
File src/processor/minidump_processor.cc (right):

http://breakpad.appspot.com/45001/diff/1/6#newcode89
Line 89: // this will just return an empty string if it doesn't exist
Sentence: capital T, period on the end.

http://breakpad.appspot.com/45001/diff/1/6#newcode1026
Line 1026: assertion_string = "Invalid Parameter passed to library function";
lowercase p on parameter

http://breakpad.appspot.com/45001/diff/1/5
File src/processor/minidump_stackwalk.cc (right):

http://breakpad.appspot.com/45001/diff/1/5#newcode428
Line 428: else {
Use ugly nested } else {.

[Mark Mentovai]

Sometimes, sizeofs read as strlens if that's what you're expecting...

Having this be a fixed-size array explains why you had the '\0' check in the
UTF-16-to-8 conversion function.  I think we should still account for that, to
avoid building up strings that have trailing '\0' characters.  But I don't think
that should happen as a '\0' check in the conversion function, which expects to
operate on vectors that are already properly-sized.  Instead, I think that the
caller should compute the length and size the vector properly before passing it
to the conversion function.

[Ted Mielczarek]

I wrote a UTF16codeunits function and used that, and added a check for
word_length == 0. Landed in r433.