[d8] Fix tsan bugs

[d8] Fix tsan bugs

script_executed and last_run are read/written by multiple threads. Also
externalized_shared_contents_ is modified by multiple threads.

BUG=4306
R=jarin@chromium.org
LOG=n

Committed: https://crrev.com/053b8434f00d5e0de9bea884dbe3fa2cdeaaab83
Cr-Commit-Position: refs/heads/master@{#29912}

[Jarin, 2015-07-23 06:49:14]

lgtm with bunch of questions.

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc#newcode338
src/d8.cc:338: options.set_script_executed(true);
It looks like the set_script_executed could be calculated from the command line
arguments: script_executed is true if there is an argument not prefixed with '-'
or if there is an '-e' argument. Perhaps we could just do a walk over the
arguments (basically cut-down version of SourceGroup::Execute) rather than the
Atomic hackery?

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc#newcode2554
src/d8.cc:2554: options.set_last_run(i == stress_runs - 1);
I am a bit confused about how this can be run in parallel with anything. I
thought that everything should be stopped at this point, no?

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc#newcode2564
src/d8.cc:2564: options.set_last_run(i == stress_runs - 1);
Same here.

[binji, 2015-07-24 23:13:08]

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc
File src/d8.cc (right):

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc#newcode338
src/d8.cc:338: options.set_script_executed(true);
On 2015/07/23 at 06:49:14, jarin wrote:
> It looks like the set_script_executed could be calculated from the command
line arguments: script_executed is true if there is an argument not prefixed
with '-' or if there is an '-e' argument. Perhaps we could just do a walk over
the arguments (basically cut-down version of SourceGroup::Execute) rather than
the Atomic hackery?

Done.

https://codereview.chromium.org/1252623003/diff/20001/src/d8.cc#newcode2554
src/d8.cc:2554: options.set_last_run(i == stress_runs - 1);
On 2015/07/23 at 06:49:14, jarin wrote:
> I am a bit confused about how this can be run in parallel with anything. I
thought that everything should be stopped at this point, no?

No, the SourceGroup spawns a thread per isolate that persists across RunMain.
But I refactored my change so that there is no need for last_run anymore.

[Jarin, 2015-07-27 04:01:44]

lgtm. Love it!

[binji, 2015-07-28 21:47:14]

The CQ bit was checked by binji@chromium.org

[commit-bot: I haz the power, 2015-07-28 21:47:23]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1252623003/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1252623003/40001

[commit-bot: I haz the power, 2015-07-30 08:20:40]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-07-30 08:20:58]

Patchset 3 (id:??) landed as
https://crrev.com/053b8434f00d5e0de9bea884dbe3fa2cdeaaab83
Cr-Commit-Position: refs/heads/master@{#29912}