Eliminate the need for code delete events in CPU profiler.

Eliminate the need for code delete events in CPU profiler.

Events are still generated for tick processor on performance testing
server to work, as soon as scripts will be updated, it will be safe
to remove code delete events emitting code.

R=erik.corry@gmail.com
BUG=v8:1466
TEST=existing tests in test-profile-generator,test-cpu-profiler and mjsunit/tools

Committed: http://code.google.com/p/v8/source/detail?r=9275

[Erik Corry, 2011-09-10 14:03:07]

LGTM!  Thanks.

[mnaganov (inactive), 2011-09-13 19:25:01]

On 2011/09/10 14:03:07, Erik Corry wrote:
> LGTM!  Thanks.

Please, take another look.

I have noticed test-log/EquivalenceOfLoggingAndTraversal failures with
difference in compiled function optimization status. And it turned out that this
was due to %SetCode was updating opimization flag, and this wasn't reflected in
log. And as now adding a new code over an old one discards the latter, code with
wrong status made its way into the code map and was accounted in the test.

[Erik Corry, 2011-09-14 10:56:52]

LGTM

http://codereview.chromium.org/7864017/diff/4001/src/log.cc
File src/log.cc (right):

http://codereview.chromium.org/7864017/diff/4001/src/log.cc#newcode1591
src/log.cc:1591: LogExistingFunction(*sfis[i], *code_objects[i]);
This is in handle code and it starts by putting its arguments into handles.  I
think it would be cleaner if it just took its argument as handles.

[mnaganov (inactive), 2011-09-14 11:16:07]

http://codereview.chromium.org/7864017/diff/4001/src/log.cc
File src/log.cc (right):

http://codereview.chromium.org/7864017/diff/4001/src/log.cc#newcode1591
src/log.cc:1591: LogExistingFunction(*sfis[i], *code_objects[i]);
On 2011/09/14 10:56:52, Erik Corry wrote:
> This is in handle code and it starts by putting its arguments into handles.  I
> think it would be cleaner if it just took its argument as handles.

Indeed. Fixed, thanks!