Add support for CPU and heap profiles deletion.

Add support for CPU and heap profiles deletion.

R=vitalyr@chromium.org
BUG=none
TEST=none

Committed: http://code.google.com/p/v8/source/detail?r=7308

[mnaganov (inactive), 2011-03-22 08:20:40]

This is mainly needed for heap snapshots, as they consume significant amounts of
memory on engine side, but I added support for CPU profiles deletion as well.

The approach isn't ideal, as after deleting a profile, embedder will get stale
pointers for previously retrieved nodes etc., but I don't want to introduce a
full DOM-like ref counting scheme here.

This works OK for WebKit / DevTools, as DevTools UI never tries to access nodes
of deleted profiles, and in console object, there is no possibility to clear
profiles.

[Vitaly Repeshko, 2011-03-22 14:30:36]

LGTM

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h
File include/v8-profiler.h (right):

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:143: void Delete() const;
"const" looks weird here.

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:392: void Delete() const;
Ditto.

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:496: * are freed by calling the Delete class function.
Accidental edit?

http://codereview.chromium.org/6685084/diff/3001/src/api.cc
File src/api.cc (right):

http://codereview.chromium.org/6685084/diff/3001/src/api.cc#newcode5084
src/api.cc:5084: const_cast<i::CpuProfile*>(
Dropping "const" will allow to avoid this const_cast.

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc
File src/profile-generator.cc (right):

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc#new...
src/profile-generator.cc:553: if (*list_ptr) {
nit: Use explicit != NULL comparison.

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc#new...
src/profile-generator.cc:673: p;
!= NULL

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
File test/cctest/test-cpu-profiler.cc (right):

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
test/cctest/test-cpu-profiler.cc:287: CHECK_EQ(NULL,
v8::CpuProfiler::FindProfile(uid1));
It'd be nice to test this again later to make sure it won't reappear.

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
test/cctest/test-cpu-profiler.cc:307: CHECK_EQ(NULL,
v8::CpuProfiler::FindProfile(uid2));
Same here.

[mnaganov (inactive), 2011-03-22 16:03:09]

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h
File include/v8-profiler.h (right):

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:143: void Delete() const;
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> "const" looks weird here.

Hmm, maybe. I added 'const' because all methods return const pointers (to
emphasize that embedder can't do much with them). So calling a non-const Delete
will require embedder to do a const_cast. But probably it's OK, as it attracts
more attention to calls to this method.

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:392: void Delete() const;
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> Ditto.

Done.

http://codereview.chromium.org/6685084/diff/3001/include/v8-profiler.h#newcod...
include/v8-profiler.h:496: * are freed by calling the Delete class function.
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> Accidental edit?

Ah, yes. Massive renaming. Reverted.

http://codereview.chromium.org/6685084/diff/3001/src/api.cc
File src/api.cc (right):

http://codereview.chromium.org/6685084/diff/3001/src/api.cc#newcode5084
src/api.cc:5084: const_cast<i::CpuProfile*>(
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> Dropping "const" will allow to avoid this const_cast.

Done.

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc
File src/profile-generator.cc (right):

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc#new...
src/profile-generator.cc:553: if (*list_ptr) {
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> nit: Use explicit != NULL comparison.

Done.

http://codereview.chromium.org/6685084/diff/3001/src/profile-generator.cc#new...
src/profile-generator.cc:673: p;
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> != NULL

Done.

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
File test/cctest/test-cpu-profiler.cc (right):

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
test/cctest/test-cpu-profiler.cc:287: CHECK_EQ(NULL,
v8::CpuProfiler::FindProfile(uid1));
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> It'd be nice to test this again later to make sure it won't reappear.

Added a couple of times more after.

http://codereview.chromium.org/6685084/diff/3001/test/cctest/test-cpu-profile...
test/cctest/test-cpu-profiler.cc:307: CHECK_EQ(NULL,
v8::CpuProfiler::FindProfile(uid2));
On 2011/03/22 14:30:36, Vitaly Repeshko wrote:
> Same here.

Done.