Added the base exploitability module for windows. This only adds the very bas...

Added the base exploitability module for windows. This only adds the very basic
exception type based analysis for now.

BUG=NONE
TEST=MinidumpProcessorTest.TestExploitilityEngine


Committed: http://code.google.com/p/google-breakpad/source/detail?r=698

[Cris Neckar]

This issue addresses the requested changes from
http://breakpad.appspot.com/188001/show

[mochalatte]

LGTM

http://breakpad.appspot.com/189001/diff/1/3
File src/google_breakpad/common/minidump_exception_win32.h (right):

http://breakpad.appspot.com/189001/diff/1/3#newcode108
Line 108: typedef enum {
it seems to me that the constants in this enum are actually set by the Windows
minidump writer but they aren't defined in the Windows SDK (or at least I could
not find them on MSDN) - can you add a clarifying comment indicating where you
got them from?

http://breakpad.appspot.com/189001/diff/1/6
File src/processor/exploitability_win.cc (right):

http://breakpad.appspot.com/189001/diff/1/6#newcode38
Line 38: #include "processor/exploitability_win.h"
move this guy up and into it's own section, according to style guide

[Ted Mielczarek]

http://breakpad.appspot.com/189001/diff/1/2
File Makefile.in (right):

http://breakpad.appspot.com/189001/diff/1/2#newcode1
Line 1: # Makefile.in generated by automake 1.11.1 from Makefile.am.
You actually need to edit Makefile.am. Makefile.in is generated from Makefile.am
by automake, so your changes will be lost if someone re-runs automake.

http://breakpad.appspot.com/189001/diff/1/9
File src/processor/minidump_processor_unittest.cc (right):

http://breakpad.appspot.com/189001/diff/1/9#newcode259
Line 259: // Test that exploitability module correctly fails to supply
Fix the comment here. Might also want to expound on why this test dump gets
classified as EXPLOITABILITY_HIGH.

Do you have plans to add more interesting tests here? I have a patch in my local
patch queue that adds a test to the windows client similar to the one I added
here:
http://code.google.com/p/google-breakpad/source/browse/trunk/src/client/linux...

So you could intentionally crash in interesting ways and verify the
exploitability ratings from the minidump that was produced. If you're interested
I can get the patch up for review.

[Cris Neckar]

Thanks guys. 

Ted if you have a patch you think might help with testing I would love to see
it. Currently the best way I can see (for testing windows dumps on a linux
build) is to included pre-generated dumps.

http://breakpad.appspot.com/189001/diff/1/2
File Makefile.in (right):

http://breakpad.appspot.com/189001/diff/1/2#newcode1
Line 1: # Makefile.in generated by automake 1.11.1 from Makefile.am.
On 2010/09/21 12:43:06, Ted Mielczarek wrote:
> You actually need to edit Makefile.am. Makefile.in is generated from
Makefile.am
> by automake, so your changes will be lost if someone re-runs automake.

Done.

http://breakpad.appspot.com/189001/diff/1/3
File src/google_breakpad/common/minidump_exception_win32.h (right):

http://breakpad.appspot.com/189001/diff/1/3#newcode108
Line 108: typedef enum {
On 2010/09/20 18:05:55, mochalatte wrote:
> it seems to me that the constants in this enum are actually set by the Windows
> minidump writer but they aren't defined in the Windows SDK (or at least I
could
> not find them on MSDN) - can you add a clarifying comment indicating where you
> got them from?

Done.

http://breakpad.appspot.com/189001/diff/1/6
File src/processor/exploitability_win.cc (right):

http://breakpad.appspot.com/189001/diff/1/6#newcode38
Line 38: #include "processor/exploitability_win.h"
On 2010/09/20 18:05:55, mochalatte wrote:
> move this guy up and into it's own section, according to style guide

Done.

http://breakpad.appspot.com/189001/diff/1/9
File src/processor/minidump_processor_unittest.cc (right):

http://breakpad.appspot.com/189001/diff/1/9#newcode259
Line 259: // Test that exploitability module correctly fails to supply
On 2010/09/21 12:43:06, Ted Mielczarek wrote:
> Fix the comment here. Might also want to expound on why this test dump gets
> classified as EXPLOITABILITY_HIGH.
> 
> Do you have plans to add more interesting tests here? I have a patch in my
local
> patch queue that adds a test to the windows client similar to the one I added
> here:
>
http://code.google.com/p/google-breakpad/source/browse/trunk/src/client/linux...
> 
> So you could intentionally crash in interesting ways and verify the
> exploitability ratings from the minidump that was produced. If you're
interested
> I can get the patch up for review.

Done.

Yeah I do plan to add tests once I get the rule sets more flushed out. I was
going to ask you guys what the preferred way of doing this. It seems like
committing a bunch of pregenerated minidump files is ugly but aside from that
the only option I see is building and testing on the platform you want to
generate dumps for.