Android: adds stack symbolization utilities.

Android: adds stack symbolization utilities.

asan_symbolize.py is used for the logcat stacks using the special ASan format.
tombstones.py is used to manage tombstones files in devices.
symbolize.py is used for stack traces generated by base/debug/stack_trace_android.cc

BUG=234973
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=211134

[bulach, 2013-07-04 17:21:42]

ptal (depends on https://codereview.chromium.org/18715002/)

[bulach, 2013-07-05 09:05:35]

+scherkus for symbolize.py and its tests

[Isaac (away), 2013-07-05 23:36:27]

:-) Really looking forward to this and avoiding the following dialog:

Q: How do I symbolize chromium android crashs
A: We have a script!  First you checkout the private code...

[scherkus (not reviewing), 2013-07-08 21:53:48]

lgtm w/ nit on symbolize.py

don't forget to update CL description to mention what symbolize.py is used for
(it's used to symbolize traces generated by base/debug/stack_trace_android.cc --
it's currently broken but I'll fix that as soon as this lands)

https://codereview.chromium.org/18473004/diff/5001/build/android/tests/symbol...
File build/android/tests/symbolize/a.cc (right):

https://codereview.chromium.org/18473004/diff/5001/build/android/tests/symbol...
build/android/tests/symbolize/a.cc:5: class  A {
whoops look like I left an extra space between "class" and "A"

[scherkus (not reviewing), 2013-07-09 00:17:31]

actually not lgtm as symbolize.py is broken in its current form :\

https://codereview.chromium.org/18473004/diff/5001/build/android/symbolize.py
File build/android/symbolize.py (right):

https://codereview.chromium.org/18473004/diff/5001/build/android/symbolize.py...
build/android/symbolize.py:23: import symbol
FYI this isn't included in this CL

are we planning on upstreaming this module? if not, we'll have to look into
breaking the dependency by rewriting the missing bits :\

[bulach, 2013-07-09 08:26:49]

thanks scherkus!

addressed the nit, and sorry I didn't mention before:
the actual chain of patches is:
1) add third_party (which contains symbol.py):
https://codereview.chromium.org/18326020/

2) apply our optimizations:
https://codereview.chromium.org/18715002/

3) this patch :)

4) that downstream patch removing these utilities and updating callers to use
this new version.

they'll obviously land in order.. :)

I think symbolize.py is not called by any other script, but if it is, please let
me know and I'm happy to fix up any further dependency downstream..

fwiw, the very first patch is waiting for security review, once that unblocks,
the rest of the chain should move along.

https://codereview.chromium.org/18473004/diff/5001/build/android/tests/symbol...
File build/android/tests/symbolize/a.cc (right):

https://codereview.chromium.org/18473004/diff/5001/build/android/tests/symbol...
build/android/tests/symbolize/a.cc:5: class  A {
On 2013/07/08 21:53:48, scherkus wrote:
> whoops look like I left an extra space between "class" and "A"

Done.

[scherkus (not reviewing), 2013-07-09 15:29:16]

Thanks for the explanation! LGTM

> I think symbolize.py is not called by any other script, but if it is, please
let
> me know and I'm happy to fix up any further dependency downstream..

Correct -- it's a leaf node, if you will.

[commit-bot: I haz the power, 2013-07-11 12:26:27]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bulach@chromium.org/18473004/19001

[bulach, 2013-07-11 12:29:18]

scherkus: I just realized I had the .so binaries in patchset #3..
I removed them and I'm CQing patchset #4..

please let me know if you want me to add them. I think the instructions on how
to build them are sufficient, but anyways :)

(perhaps we could call "make" from symbolize_test.py..)

[commit-bot: I haz the power, 2013-07-11 16:18:14]

Change committed as 211134

[cjhopman, 2013-07-11 16:23:13]

On 2013/07/11 12:29:18, bulach wrote:
> scherkus: I just realized I had the .so binaries in patchset #3..
> I removed them and I'm CQing patchset #4..
> 
> please let me know if you want me to add them. I think the instructions on how
> to build them are sufficient, but anyways :)
> 
> (perhaps we could call "make" from symbolize_test.py..)

I think part of the problem is that the tests hardcode the address that they
expect symbols to be at in the binary.

[scherkus (not reviewing), 2013-07-11 22:06:00]

On 2013/07/11 16:23:13, cjhopman wrote:
> On 2013/07/11 12:29:18, bulach wrote:
> > scherkus: I just realized I had the .so binaries in patchset #3..
> > I removed them and I'm CQing patchset #4..
> > 
> > please let me know if you want me to add them. I think the instructions on
how
> > to build them are sufficient, but anyways :)
> > 
> > (perhaps we could call "make" from symbolize_test.py..)
> 
> I think part of the problem is that the tests hardcode the address that they
> expect symbols to be at in the binary.

Yep. CL out here: https://codereview.chromium.org/18755005/