[Webkit-unassigned] [Bug 139812] New: REGRESSION(176609): Memory leak in Canvas/reuse.html performance test

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Dec 19 00:35:39 PST 2014


https://bugs.webkit.org/show_bug.cgi?id=139812

            Bug ID: 139812
           Summary: REGRESSION(176609): Memory leak in Canvas/reuse.html
                    performance test
    Classification: Unclassified
           Product: WebKit
           Version: 528+ (Nightly build)
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Critical
          Priority: P1
         Component: New Bugs
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: ossy at webkit.org
                CC: akling at apple.com, cdumez at apple.com,
                    cgarcia at igalia.com, clopez at igalia.com,
                    gyuyoung.kim at webkit.org, ryuan.choi at navercorp.com
            Blocks: 139140

at least on EFL and GTK bots, because there is no malloc measuring on Mac bots.

before:
--------
https://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release%20%28Perf%29/builds/1382

Running Canvas/reuse.html (33 of 141)
RESULT Canvas: reuse: Runs= 70.7723901718 runs/s
median= 63.5780801678 runs/s, stdev= 13.4939255648 runs/s, min= 62.5708614903 runs/s, max= 97.1289885232 runs/s
RESULT Canvas: reuse: JSHeap= 267628.8 bytes
median= 269721.0 bytes, stdev= 3148.69720093 bytes, min= 261820.0 bytes, max= 269721.0 bytes
RESULT Canvas: reuse: Malloc= 1009948.4 bytes
median= 985512.0 bytes, stdev= 75773.0039181 bytes, min= 979760.0 bytes, max= 1235248.0 bytes
Finished: 25.664726 s


after:
-------
https://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release%20%28Perf%29/builds/1383

Running Canvas/reuse.html (33 of 141)
RESULT Canvas: reuse: Runs= 49.7027479102 runs/s
median= 49.8827160444 runs/s, stdev= 22.7074182517 runs/s, min= 6.19658159381 runs/s, max= 124.313478825 runs/s
RESULT Canvas: reuse: JSHeap= 267201.6 bytes
median= 268569.0 bytes, stdev= 3111.07465609 bytes, min= 261788.0 bytes, max= 269689.0 bytes
RESULT Canvas: reuse: Malloc= 11552632.0 bytes
median= 9958680.0 bytes, stdev= 3720965.90803 bytes, min= 5353752.0 bytes, max= 23708440.0 bytes
Finished: 42.262450 s


Malloc= 1 009 948.4 bytes --> Malloc= 11 552 632.0 bytes 
--> 11x memory regression

And after this change the EFL performance bot is very flakey, WTR consumes more than 
8Gb memory for Canvas/reuse.html which causes swapping and causes OOM or timeout 
regularly. And of course swapping makes the whole performance measuring unreliable.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20141219/ec253da0/attachment-0002.html>


More information about the webkit-unassigned mailing list