CrashHandler for Windows.

CrashHandler for Windows.

Plus, print out assertion failures on Windows,
and some little tweaks to CrashHandler on other platforms for consistency.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/3f73e8c8d589e0d5a1f75327b4aa22c1e745732d

[bungeman-skia, 2014-06-18 22:28:28]

https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp
File tools/CrashHandler.cpp (right):

https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp#newco...
tools/CrashHandler.cpp:115: frame.AddrPC.Offset    = c->Eip;
CONTEXT is a completely different struct on x86 and AMD64, you'd need to use
Rip, Rsp, and Rbp/Rdi (not sure which one one is supposed to use these days) on
AMD64.

https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp#newco...
tools/CrashHandler.cpp:122: while (StackWalk64(sizeof(void*) == 4 ?
IMAGE_FILE_MACHINE_I386 : IMAGE_FILE_MACHINE_AMD64,
Because it's what we're actually after here, we should be checking for the
defines _AMD64_ and _X86_ here. They're the Windows way of doing things.

[mtklein, 2014-06-18 22:34:29]

On 2014/06/18 22:28:28, bungeman1 wrote:
> https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp
> File tools/CrashHandler.cpp (right):
> 
>
https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp#newco...
> tools/CrashHandler.cpp:115: frame.AddrPC.Offset    = c->Eip;
> CONTEXT is a completely different struct on x86 and AMD64, you'd need to use
> Rip, Rsp, and Rbp/Rdi (not sure which one one is supposed to use these days)
on
> AMD64.
> 
>
https://codereview.chromium.org/340523007/diff/1/tools/CrashHandler.cpp#newco...
> tools/CrashHandler.cpp:122: while (StackWalk64(sizeof(void*) == 4 ?
> IMAGE_FILE_MACHINE_I386 : IMAGE_FILE_MACHINE_AMD64,
> Because it's what we're actually after here, we should be checking for the
> defines _AMD64_ and _X86_ here. They're the Windows way of doing things.

Like this?

[bungeman-skia, 2014-06-18 22:54:02]

I'm sure there are ways to make this print out pretty things, but for now this
is certainly an improvement over what we currently have. lgtm with one nit. (So
long as it build and works and stuff.)

https://codereview.chromium.org/340523007/diff/60001/tools/CrashHandler.cpp
File tools/CrashHandler.cpp (right):

https://codereview.chromium.org/340523007/diff/60001/tools/CrashHandler.cpp#n...
tools/CrashHandler.cpp:124: #else
Can this be '#elif defined(_AMD64_)' and then below have an '#else' with an
'#error You Foo!!!' in it?

[mtklein, 2014-06-19 00:28:29]

Thanks for triggering the trybot.  Booyah.  Beautiful assert message and stack
trace!

https://codereview.chromium.org/340523007/diff/60001/tools/CrashHandler.cpp
File tools/CrashHandler.cpp (right):

https://codereview.chromium.org/340523007/diff/60001/tools/CrashHandler.cpp#n...
tools/CrashHandler.cpp:124: #else
On 2014/06/18 22:54:02, bungeman1 wrote:
> Can this be '#elif defined(_AMD64_)' and then below have an '#else' with an
> '#error You Foo!!!' in it?

Done.  Just did #if, #elif, #endif.  Future us will understand.

[mtklein, 2014-06-19 00:29:44]

+reed for include/.  

No public API changes: we're just making SkASSERT better on Windows by actually
printing something when it fails like we do on Linux, Mac and Android.

[mtklein, 2014-06-19 00:30:49]

On 2014/06/19 00:29:44, mtklein wrote:
> +reed for include/.  
> 
> No public API changes: we're just making SkASSERT better on Windows by
actually
> printing something when it fails like we do on Linux, Mac and Android.

You can see an example assertion failure and stack trace on the failing Test-
run on patch set 4.  That's what's breaking our bots right now.

[reed1, 2014-06-19 13:29:34]

lgtm

[mtklein, 2014-06-19 13:31:09]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-06-19 13:32:07]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/mtklein@chromium.org/340523007/70001

[commit-bot: I haz the power, 2014-06-19 14:42:04]

Change committed as 3f73e8c8d589e0d5a1f75327b4aa22c1e745732d