546373 - Crashed caused by OOM related to OOPP

Status: RESOLVED WORKSFORME

(Core :: IPC, defect)

Description

[Rob Arnold [:robarnold]]

I had a window open with two tabs, one of which had been viewing a PDF but I had clicked "back" on so it was not showing (also its tab was not the active one). I clicked the close button on the window and the entire browser locked, went to using a full core and started increasing in memory usage - about 100 MB/min. I tried the usual technique of killing the plugin process but that only delayed the memory increase - it resumed and eventually killed the process, never responding to my mouse click. During this time I noticed a steady and rapid rate of messages: WM_PAINT and the nsAppShell message perfectly interleaved being dispatched. I'll attach a few stack traces and hopefully someone more knowledgeable about OOPP will have some insight.

Comment 1

[Rob Arnold [:robarnold]]

Attachment: Sample Trace 1

Comment 2

[Rob Arnold [:robarnold]]

Attachment: Sample Trace 2

Comment 3

[Rob Arnold [:robarnold]]

Attachment: Sample Trace 3

Comment 4

[Rob Arnold [:robarnold]]

Attachment: Sample Trace 4

Comment 5

[Rob Arnold [:robarnold]]

Attachment: Sample Trace 5

Comment 6

[Rob Arnold [:robarnold]]

Attachment: Trace at point of OOM Exception

Comment 7

[Jim Mathies [:jimm]]

Hey Rob, what version/nightly was this? This looks like bug 543788.

Comment 8

[Jim Mathies [:jimm]]

(The final patch for bug 543788 just landed on m-c this morning.)

Comment 9

[Rob Arnold [:robarnold]]

(In reply to comment #7)
> Hey Rob, what version/nightly was this? This looks like bug 543788.

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.3a2pre) Gecko/20100211 Minefield/3.7a2pre (.NET CLR 3.5.30729)

Comment 10

[Wayne Mery (:wsmwk)]

Rob, 
can you still reproduce this with trunk?
If not, please close WFM

Comment 11

[Rob Arnold [:robarnold]]

I have not reproduced this since.