Fix some crash bugs in Stackwalker implementations when resolver is NULL

Apparently these have existed since this code was written. I just ran into them
while writing a one-off tool that didn't bother to use a resolver.

[Mark Mentovai]

LGTM.

http://breakpad.appspot.com/257001/diff/1/6
File src/processor/stackwalker_arm_unittest.cc (right):

http://breakpad.appspot.com/257001/diff/1/6#newcode148
Line 148: EXPECT_TRUE(memcmp(&raw_context, &frame->context, sizeof(raw_context))
== 0);
EXPECT_EQ instead?

http://breakpad.appspot.com/257001/diff/1/8
File src/processor/stackwalker_x86_unittest.cc (right):

http://breakpad.appspot.com/257001/diff/1/8#newcode149
Line 149: EXPECT_TRUE(memcmp(&raw_context, &frame->context, sizeof(raw_context))
== 0);
EXPECT_EQ?

[Ted Mielczarek]

http://breakpad.appspot.com/257001/diff/1/6
File src/processor/stackwalker_arm_unittest.cc (right):

http://breakpad.appspot.com/257001/diff/1/6#newcode148
Line 148: EXPECT_TRUE(memcmp(&raw_context, &frame->context, sizeof(raw_context))
== 0);
On 2011/01/24 19:15:44, Mark Mentovai wrote:
> EXPECT_EQ instead?

I copy/pasted these from elsewhere in the file. I'll fix them all.