Refactor some X86 Stackwalker bits into parent classes

I'm going to add stack scanning to the amd64 / arm stackwalkers, so this is some
refactoring to pull some bits that currently live on the X86 classes up to the
parent classes.

[jimb]

LGTM!  I love that moving the stack scanning function out to the base class and
making it a template required no code changes in the caller.

http://breakpad.appspot.com/205001/diff/1/4
File src/google_breakpad/processor/stackwalker.h (right):

http://breakpad.appspot.com/205001/diff/1/4#newcode120
Line 120: template<typename instruction_type>
Not sure, but I think type names that appear as template arguments should follow
the same naming conventions as other type names: StudlyCaps, no underscores.

http://breakpad.appspot.com/205001/diff/1/5
File src/processor/minidump_stackwalk.cc (right):

http://breakpad.appspot.com/205001/diff/1/5#newcode237
Line 237: const char *trust_name;
It might be nice to add a FrameTrustName static member function to the
StackFrame base class to do this.

http://breakpad.appspot.com/205001/diff/1/7
File src/processor/stackwalker_amd64_unittest.cc (right):

http://breakpad.appspot.com/205001/diff/1/7#newcode205
Line 205: EXPECT_EQ(StackFrame::FRAME_TRUST_CONTEXT, frame0->trust);
It's tested! Yay!

[Ted Mielczarek]

On 2010/09/30 17:42:57, jimb wrote:
> http://breakpad.appspot.com/205001/diff/1/4
> File src/google_breakpad/processor/stackwalker.h (right):
> 
> http://breakpad.appspot.com/205001/diff/1/4#newcode120
> Line 120: template<typename instruction_type>
> Not sure, but I think type names that appear as template arguments should
follow
> the same naming conventions as other type names: StudlyCaps, no underscores.

Done.

> http://breakpad.appspot.com/205001/diff/1/5
> File src/processor/minidump_stackwalk.cc (right):
> 
> http://breakpad.appspot.com/205001/diff/1/5#newcode237
> Line 237: const char *trust_name;
> It might be nice to add a FrameTrustName static member function to the
> StackFrame base class to do this.

I added a non-static StackFrame::trust_description method. Seemed like a more
natural fit, although probably a bit odd since it's a struct.