roll clang 159409:160702

roll clang 159409:160702

* Lets -Wunused-private-field find more stuff
* Fixes an ASan problem on precise
* Improves the plugin's handling of inner classes (off by default,
  can be enabled with |-Xclang -plugin-arg-find-bad-constructs
  -Xclang check-inner-classes|)

Abandoned because of http://llvm.org/PR13475

BUG=139346
TBR=hans

[Nico, 2012-07-27 23:34:26]

(Hans and I discussed the roll over IRC, hence the TBR. Hans, please take a look
anyway.)

[commit-bot: I haz the power, 2012-07-27 23:51:01]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/10833055/2004

[commit-bot: I haz the power, 2012-07-28 00:46:16]

Try job failure for 10833055-2004 (retry) on linux_clang for step "compile"
(clobber build).
It's a second try, previously, step "compile" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_clan...

[Nico, 2012-07-28 00:49:34]

mark / rsesek, can one of you take a look at the process_util_unittest.cc
change?

[Mark Mentovai, 2012-07-28 02:09:25]

LGTM. Questions:

Would volatile have worked?

What’s clang’s new smartness doing here?

https://chromiumcodereview.appspot.com/10833055/diff/3002/base/process_util_u...
File base/process_util_unittest.cc (right):

https://chromiumcodereview.appspot.com/10833055/diff/3002/base/process_util_u...
base/process_util_unittest.cc:507: asm("" : "=r" (buf));  // Prevent clang from
being too smart.
OK

[Nico, 2012-07-28 02:11:47]

I tried just "volatile buf[3]". It'd complain that free() doesn't take a
volatile pointer, and when i casted it to void*, clang elided the free() call
again.

[Mark Mentovai, 2012-07-28 02:15:00]

LGTM