Issue metadata
Sign in to add a comment
|
Potential memory leak in GlobalActivityTracker |
||||||||||||||||||||||
Issue descriptionThe experiment was done by browsing the web randomly using a chrome extension for about 7 days. Sites are alexa top-10000. The chrome browser was running with native heap profiling and a memosy snapshot was taking at regular intervals. This is leading us to potential memory leaks. This leak is still showing up in recent trace: os-arch: "x86_64", os-name: "Windows NT", revision: "cdd15784955039742fe9a8235581922d41b82d78-refs/heads/master@{#492239}", product-version: "Chrome/62.0.3178.0", See attached file for full stackframe. After a quick look, I suspect |known_processes_| to keep tracks of process ids that are never removed. see https://cs.chromium.org/chromium/src/base/debug/activity_tracker.cc?type=cs&l=1417 They are supposed to be removed here: https://cs.chromium.org/chromium/src/base/debug/activity_tracker.cc?type=cs&l=1436 I didn't look at why the activity tracker doesn't get called.
,
Aug 17 2017
It's possible. Process reaping isn't done in a single place, unfortunately. Can you tell me the contents of that memory? If I know which process it represents, I might be able to find where it is reaped. Alternatively, I can do some sort of background check for processes that are no longer running.
,
Aug 17 2017
Unfortunately, I can't tell you what is in the set. My browser process is no longer running. I just launched a new experiment, I can take a look in a week. My guess will be that we are leaking renderer pids and their corresponding command-lines.
,
Aug 21 2017
,
Aug 22 2017
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by etienneb@chromium.org
, Aug 16 2017