tracing: Simplify TraceEventMemoryOverhead, use an enum insted of a map

tracing: Simplify TraceEventMemoryOverhead, use an enum insted of a map

TraceEventMemoryOverhead is a rather simple class which only purpose is
accounting various sources of memory overhead due to tracing itself.
For debugging purposes it keeps track of the actual class causing
overhead, so that they can be inspected distinctly in the tracing UI
The current implementation was using a small_map indexed by
const char* pointer, which has a bunch of problems:
- base::small_map can degenerate into a STL map when exceeding the
  inline capacity, which in turn can cause some extra impredictability,
  see "[chromium-dev] Please avoid std::unordered_map".
- The map implicitly assumes that two identical string literals will have
  identical pointers, which is not true, esepcially in component builds.
- Any sort of map is a really over-engineered solution, given that the
  sources of overheads we care about are known a priori and are a
  manageable number.

Hence, this CL is switching the implementation of
TraceEventMemoryOverhead to be just enum-base, which ditches completely
the map.

BUG=717223
Review-Url: https://codereview.chromium.org/2857543002
Cr-Commit-Position: refs/heads/master@{#468611}
Committed: https://chromium.googlesource.com/chromium/src/+/eb6696a1bfb1c32c6d2d1baf9a0dabcc632119e7

[Primiano Tucci (use gerrit), 2017-05-02 10:18:03]

The CQ bit was checked by primiano@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-05-02 10:18:15]

Dry run: CQ is trying da patch.

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

[Primiano Tucci (use gerrit), 2017-05-02 10:19:54]

primiano@chromium.org changed reviewers:
+ hjd@chromium.org

[Primiano Tucci (use gerrit), 2017-05-02 10:19:55]

hjd: can you take a look plz?
+ssid/etienneb FYI, I suspect this should indirectly resolve some of the
unpredictability of malloc we were discussing recently

[commit-bot: I haz the power, 2017-05-02 10:33:02]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-05-02 10:33:03]

Dry run: Try jobs failed on following builders:
  android_cronet on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_cron...)

[hjd, 2017-05-02 10:46:30]

On 2017/05/02 10:19:55, Primiano Tucci wrote:
> hjd: can you take a look plz?
> +ssid/etienneb FYI, I suspect this should indirectly resolve some of the
> unpredictability of malloc we were discussing recently

lgtm

[Primiano Tucci (use gerrit), 2017-05-02 11:12:39]

The CQ bit was checked by primiano@chromium.org

[Primiano Tucci (use gerrit), 2017-05-02 11:12:40]

The patchset sent to the CQ was uploaded after l-g-t-m from hjd@chromium.org
Link to the patchset: https://codereview.chromium.org/2857543002/#ps20001
(title: "Fix compiler issues + omit empty values")

[commit-bot: I haz the power, 2017-05-02 11:12:53]

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, 2017-05-02 13:01:17]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1493723559881710,
"parent_rev": "ed559df96be91ee4921375209e63b63bb5e538ab", "commit_rev":
"eb6696a1bfb1c32c6d2d1baf9a0dabcc632119e7"}

[commit-bot: I haz the power, 2017-05-02 13:01:26]

Description was changed from

==========
tracing: Simplify TraceEventMemoryOverhead, use an enum insted of a map

TraceEventMemoryOverhead is a rather simple class which only purpose is
accounting various sources of memory overhead due to tracing itself.
For debugging purposes it keeps track of the actual class causing
overhead, so that they can be inspected distinctly in the tracing UI
The current implementation was using a small_map indexed by
const char* pointer, which has a bunch of problems:
- base::small_map can degenerate into a STL map when exceeding the
  inline capacity, which in turn can cause some extra impredictability,
  see "[chromium-dev] Please avoid std::unordered_map".
- The map implicitly assumes that two identical string literals will have
  identical pointers, which is not true, esepcially in component builds.
- Any sort of map is a really over-engineered solution, given that the
  sources of overheads we care about are known a priori and are a
  manageable number.

Hence, this CL is switching the implementation of
TraceEventMemoryOverhead to be just enum-base, which ditches completely
the map.

BUG=717223
==========

to

==========
tracing: Simplify TraceEventMemoryOverhead, use an enum insted of a map

TraceEventMemoryOverhead is a rather simple class which only purpose is
accounting various sources of memory overhead due to tracing itself.
For debugging purposes it keeps track of the actual class causing
overhead, so that they can be inspected distinctly in the tracing UI
The current implementation was using a small_map indexed by
const char* pointer, which has a bunch of problems:
- base::small_map can degenerate into a STL map when exceeding the
  inline capacity, which in turn can cause some extra impredictability,
  see "[chromium-dev] Please avoid std::unordered_map".
- The map implicitly assumes that two identical string literals will have
  identical pointers, which is not true, esepcially in component builds.
- Any sort of map is a really over-engineered solution, given that the
  sources of overheads we care about are known a priori and are a
  manageable number.

Hence, this CL is switching the implementation of
TraceEventMemoryOverhead to be just enum-base, which ditches completely
the map.

BUG=717223

Review-Url: https://codereview.chromium.org/2857543002
Cr-Commit-Position: refs/heads/master@{#468611}
Committed:
https://chromium.googlesource.com/chromium/src/+/eb6696a1bfb1c32c6d2d1baf9a0d...
==========

[commit-bot: I haz the power, 2017-05-02 13:01:27]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/eb6696a1bfb1c32c6d2d1baf9a0d...