Revert of [profiler] Fix attribution for the top-most interpreted frame.

Revert of [profiler] Fix attribution for the top-most interpreted frame. (patchset #3 id:40001 of https://codereview.chromium.org/2667253004/ )

Reason for revert:
Flaky crashes on mac asan:
https://build.chromium.org/p/client.v8/builders/V8%20Mac64%20ASAN/builds/10739

Original issue's description:
> [profiler] Fix attribution for the top-most interpreted frame.
> 
> Before this change, we attributed samples for the top-most interpreter frame to the second-topmost frame if we were in a bytecode handler with elided frame. With this change we try to detect that we are in a handler without a frame. If we are, we do not drop the topmost frame.
> 
> For example, consider the program
> 
> function inner() {
>   var s = 0;
>   for (var i = 0; i < 100000; i++) {
>     s += i * i;
>   }
>   return s;
> }
> 
> function trivial() {
>   return inner();
> }
> 
> for (var i = 0; i < 2000; i++) {
>   trivial();
> }
> 
> 
> Before this change, d8 --prof --ignition --nocrankshaft and linux-tick-processor would produce:
> 
>   [JavaScript]:
>    ticks  total  nonlib   name
>    4885   83.4%   83.5%  Function: ~trivial a.js:15:17
>     759   13.0%   13.0%  Function: ~inner a.js:7:15
> 
> After this change, we get 
> 
>  [JavaScript]:
>    ticks  total  nonlib   name
>    5486   95.9%   96.2%  Function: ~inner a.js:7:15
>       4    0.1%    0.1%  Function: ~trivial a.js:15:17
> 
> Review-Url: https://codereview.chromium.org/2667253004
> Cr-Commit-Position: refs/heads/master@{#42894}
> Committed: https://chromium.googlesource.com/v8/v8/+/d07f6540c1f9628ed2ba1fa6507c90db07ccc5f5

TBR=bmeurer@chromium.org,jarin@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2670843005
Cr-Commit-Position: refs/heads/master@{#42912}
Committed: https://chromium.googlesource.com/v8/v8/+/63dea876da2bc73015c7c462166c9eaf3401a7d7

[Michael Achenbach, 2017-02-03 07:37:24]

The CQ bit was checked by machenbach@chromium.org

[Michael Achenbach, 2017-02-03 07:37:24]

Created Revert of [profiler] Fix attribution for the top-most interpreted frame.

[commit-bot: I haz the power, 2017-02-03 07:37:26]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-03 07:37:33]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1486107444305470, "parent_rev":
"dac1327596c31e8348cd5163dae50860f763647f", "commit_rev":
"63dea876da2bc73015c7c462166c9eaf3401a7d7"}

[commit-bot: I haz the power, 2017-02-03 07:37:38]

Description was changed from

==========
Revert of [profiler] Fix attribution for the top-most interpreted frame.
(patchset #3 id:40001 of https://codereview.chromium.org/2667253004/ )

Reason for revert:
Flaky crashes on mac asan:
https://build.chromium.org/p/client.v8/builders/V8%20Mac64%20ASAN/builds/10739

Original issue's description:
> [profiler] Fix attribution for the top-most interpreted frame.
> 
> Before this change, we attributed samples for the top-most interpreter frame
to the second-topmost frame if we were in a bytecode handler with elided frame.
With this change we try to detect that we are in a handler without a frame. If
we are, we do not drop the topmost frame.
> 
> For example, consider the program
> 
> function inner() {
>   var s = 0;
>   for (var i = 0; i < 100000; i++) {
>     s += i * i;
>   }
>   return s;
> }
> 
> function trivial() {
>   return inner();
> }
> 
> for (var i = 0; i < 2000; i++) {
>   trivial();
> }
> 
> 
> Before this change, d8 --prof --ignition --nocrankshaft and
linux-tick-processor would produce:
> 
>   [JavaScript]:
>    ticks  total  nonlib   name
>    4885   83.4%   83.5%  Function: ~trivial a.js:15:17
>     759   13.0%   13.0%  Function: ~inner a.js:7:15
> 
> After this change, we get 
> 
>  [JavaScript]:
>    ticks  total  nonlib   name
>    5486   95.9%   96.2%  Function: ~inner a.js:7:15
>       4    0.1%    0.1%  Function: ~trivial a.js:15:17
> 
> Review-Url: https://codereview.chromium.org/2667253004
> Cr-Commit-Position: refs/heads/master@{#42894}
> Committed:
https://chromium.googlesource.com/v8/v8/+/d07f6540c1f9628ed2ba1fa6507c90db07c...

TBR=bmeurer@chromium.org,jarin@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
==========

to

==========
Revert of [profiler] Fix attribution for the top-most interpreted frame.
(patchset #3 id:40001 of https://codereview.chromium.org/2667253004/ )

Reason for revert:
Flaky crashes on mac asan:
https://build.chromium.org/p/client.v8/builders/V8%20Mac64%20ASAN/builds/10739

Original issue's description:
> [profiler] Fix attribution for the top-most interpreted frame.
> 
> Before this change, we attributed samples for the top-most interpreter frame
to the second-topmost frame if we were in a bytecode handler with elided frame.
With this change we try to detect that we are in a handler without a frame. If
we are, we do not drop the topmost frame.
> 
> For example, consider the program
> 
> function inner() {
>   var s = 0;
>   for (var i = 0; i < 100000; i++) {
>     s += i * i;
>   }
>   return s;
> }
> 
> function trivial() {
>   return inner();
> }
> 
> for (var i = 0; i < 2000; i++) {
>   trivial();
> }
> 
> 
> Before this change, d8 --prof --ignition --nocrankshaft and
linux-tick-processor would produce:
> 
>   [JavaScript]:
>    ticks  total  nonlib   name
>    4885   83.4%   83.5%  Function: ~trivial a.js:15:17
>     759   13.0%   13.0%  Function: ~inner a.js:7:15
> 
> After this change, we get 
> 
>  [JavaScript]:
>    ticks  total  nonlib   name
>    5486   95.9%   96.2%  Function: ~inner a.js:7:15
>       4    0.1%    0.1%  Function: ~trivial a.js:15:17
> 
> Review-Url: https://codereview.chromium.org/2667253004
> Cr-Commit-Position: refs/heads/master@{#42894}
> Committed:
https://chromium.googlesource.com/v8/v8/+/d07f6540c1f9628ed2ba1fa6507c90db07c...

TBR=bmeurer@chromium.org,jarin@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2670843005
Cr-Commit-Position: refs/heads/master@{#42912}
Committed:
https://chromium.googlesource.com/v8/v8/+/63dea876da2bc73015c7c462166c9eaf340...
==========

[commit-bot: I haz the power, 2017-02-03 07:37:39]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/v8/v8/+/63dea876da2bc73015c7c462166c9eaf340...