Add allocation tracker test for bumb pointer allocations

Add allocation tracker test for bumb pointer allocations

Test that allocations which regularly happen inline in the generated code and create objects in bump pointer space will be recorded by the allocation tracker.

BUG=chromium:277984
LOG=N
R=loislo@chromium.org, mstarzinger@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=17794

[Michael Starzinger, 2013-11-15 13:38:39]

LGTM. One optional suggestion.

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
File test/cctest/test-heap-profiler.cc (right):

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
test/cctest/test-heap-profiler.cc:2242: "  try {\n"
I guess the try-catch is used to prevent optimization. We could also use
%NeverOptimizeFunction() instead. I don't mind either way, I'll leave the
decision up to you.

[loislo, 2013-11-15 13:44:14]

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
File test/cctest/test-heap-profiler.cc (right):

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
test/cctest/test-heap-profiler.cc:2235: static const char*
inline_heap_allocation_source =
I see no reason to think that it was an inline allocation except the knowledge
about v8 internals.
So lets rename the test or add the check that it was the inline allocation.

[yurys, 2013-11-15 13:48:06]

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
File test/cctest/test-heap-profiler.cc (right):

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
test/cctest/test-heap-profiler.cc:2242: "  try {\n"
On 2013/11/15 13:38:39, Michael Starzinger wrote:
> I guess the try-catch is used to prevent optimization. We could also use
> %NeverOptimizeFunction() instead. I don't mind either way, I'll leave the
> decision up to you.
You are right, try-catch is here just to prevent optimizations. I'll change it
to %NeverOptimizeFunction as its name conveys the intent clearly.

[yurys, 2013-11-15 14:32:26]

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
File test/cctest/test-heap-profiler.cc (right):

https://codereview.chromium.org/73893005/diff/1/test/cctest/test-heap-profile...
test/cctest/test-heap-profiler.cc:2235: static const char*
inline_heap_allocation_source =
On 2013/11/15 13:44:15, loislo wrote:
> I see no reason to think that it was an inline allocation except the knowledge
> about v8 internals.
> So lets rename the test or add the check that it was the inline allocation.

Done. Added additional step which checks that the number of allocations is
incorrect if inline allocations are manually enabled. PTAL.

[loislo, 2013-11-15 14:40:58]

lgtm

[yurys, 2013-11-15 14:47:23]

Committed patchset #3 manually as r17794 (presubmit successful).