flashgotService.js Unresponsive

[therube]

flashgotService.js Unresponsive

Nothing (current) in flashgot.log (now dated 05/29/2012, 09:22:38 AM) though there is also a flashgot.log.bak (exactly the same).

When this happened, I noticed CPU went to 50% (1 of 2 cores).

First I got an unresponsive on the page.
Stopped that, though CPU remained high.

Some time after, got the unresponsive on flashgotService.js.

At the same time (& still continuing & likely to & restart I would think) Error Console is being polluted with these entries, a few every second:

Code: Select all

Timestamp: 05/30/2012 04:47:57 PM
Error: can't access dead object
Source File: chrome://adblockplus-modules/content/ObjectTabs.jsm
Line: 332

Overall SeaMonkey Mem usage is 1.11GB (high & in particular for my current system & RAM, 2GB, XP x86).

FWIW:

Code: Select all

│    ├────7,936,112 B (00.71%) -- compartment([System Principal], jar:file:///C:/Mozilla/USERS/therube/extensions/%7B73a6fe31-595d-460b-a920-fcc0f8843232%7D.xpi!/components/noscriptService.js)
│    │    ├──6,238,208 B (00.56%) -- gc-heap
│    │    │  ├──5,017,008 B (00.45%) -- arena
│    │    │  │  ├──4,992,640 B (00.45%) ── unused
│    │    │  │  └─────24,368 B (00.00%) ── headers
│    │    │  ├────674,608 B (00.06%) -- objects
│    │    │  │    ├──466,336 B (00.04%) ── non-function
│    │    │  │    └──208,272 B (00.02%) ── function
│    │    │  ├────331,104 B (00.03%) -- shapes
│    │    │  │    ├──199,800 B (00.02%) ── tree
│    │    │  │    ├───74,592 B (00.01%) ── base
│    │    │  │    └───56,712 B (00.01%) ── dict
│    │    │  ├────142,368 B (00.01%) ── scripts
│    │    │  ├─────46,944 B (00.00%) ── strings
│    │    │  ├─────18,304 B (00.00%) ── type-objects
│    │    │  └──────7,872 B (00.00%) ── sundries
│    │    ├────524,288 B (00.05%) ── analysis-temporary
│    │    ├────507,232 B (00.05%) ── script-data
│    │    ├────237,936 B (00.02%) -- objects
│    │    │    ├──169,536 B (00.02%) ── slots
│    │    │    └───68,400 B (00.01%) ── elements
│    │    ├────213,504 B (00.02%) -- shapes-extra
│    │    │    ├───86,208 B (00.01%) ── tree-tables
│    │    │    ├───73,728 B (00.01%) ── compartment-tables
│    │    │    ├───31,456 B (00.00%) ── tree-shape-kids
│    │    │    └───22,112 B (00.00%) ── dict-tables
│    │    ├────134,752 B (00.01%) ── string-chars
│    │    ├─────49,152 B (00.00%) ── cross-compartment-wrappers
│    │    ├─────18,624 B (00.00%) ── mjit-data
│    │    ├─────11,968 B (00.00%) ── type-inference/script-main
│    │    └────────448 B (00.00%) ── other-sundries

[therube]

Guess I might explain a bit.

Went to download.
The FlashGot Media progress bar stuck (as shown), & then remained - on all tabs in that window.

Not too unusual for FlashGot Media progress bar to do quirky things, like "bouncing" or whatnot, so wasn't really concerned about that.

But then noticed high CPU usage.
Didn't attribute that to anything at that point.
First went looking for plugin-container.exe to kill, but there were none.
Then closed a window here, a window there, hoping to "hit" the cause, but that didn't do it.
Then got the unresponsive warning on the webpage, click Stop, thinking that would cause CPU usage to subside, but it didn't.
Then a bit later got the unresponsive script referencing flashgotService.js.

Expected to find something in FlashGot.log, but it was a day old, along with a .bak version of itself. (It should have been current. But as it wasn't suppose that what caused it not to be also was the cause of the .bak.)

Guessing it was a fluke & don't really expect anything to come of it?

I was running beta at the time.


The FWIW & the repeating adblockplus Error Console message are just, FWIW.
Happened to be looking at "compartments" earlier & also happened to notice ABP polluting Error Console.

At the same time (& still continuing & likely to & restart I would think) Error Console is being polluted with these entries, a few every second:

Code: Select all

Timestamp: 05/30/2012 04:47:57 PM
Error: can't access dead object
Source File: chrome://adblockplus-modules/content/ObjectTabs.jsm
Line: 332

<RattyAway> therube: known fallout from Bug 695480 - Remove support for Chrome -> Content leaks
<RattyAway> See:
<RattyAway> Bug 749652 - "TypeError: can't access dead object" when selecting items in the media list after closing the page
<RattyAway> Bug 749738 - Findbar triggers multiple dead object exceptions (can't close findbar)

[Giorgio Maone]

@therube:
I assume you cannot reproduce anymore, can you?
What FlashGot method were you using, exactly ("link", "selection", "all", "media")?

[therube]

> I assume you cannot reproduce anymore, can you?

Correct, I cannot.
The progress bar still remains, but I can now successfully FlashGot Link.

> What FlashGot method

Link.

Since yesterday, set the computer to sleep, then awoken this morning.
(No browser restart or build update, yet.)

flashgot.log (flashgot.log.bak) remains unchanged though suspect once I do restart it will right itself.

flashgot.fgt.done did update with a download this morning.

[Giorgio Maone]

> I assume you cannot reproduce anymore, can you?

Correct, I cannot.
The progress bar still remains, but I can now successfully FlashGot Link.

After an error or being forcibly interrupted, the progress bar remains until you close the window.
Did you try accessing the log from FlashGot Options|Advanced|Show Log, rather than aiming at the file directly? It's heavily cached in RAM to reduce startup times.

[therube]

> Did you try accessing the log from FlashGot Options|Advanced|Show Log, rather than aiming at the file directly?

No.
But I just did.

Ah, what do you know.

The last two headers:

Code: Select all

*** FlashGot 1.4.5rc2 started at Tue, 29 May 2012 13:22:38 GMT ***

*** FlashGot 1.4.5rc2 started at Thu, 31 May 2012 14:00:49 GMT ***

So yesterday still missing.

And now the .log & .log.bak only differ by today's entries, .log.bak unchanged from what it was yesterday.
File sizes are now different (1,202,359 vs 1,194,771).
But disk date/time still are unchanged from yesterday (05/29/2012, 09:22:38 AM, or the day before, if you want to look at it that way) .

> It's heavily cached in RAM

Interesting.

[therube]

So what, (reported disk) size only changes after a Show Log?
And date only changes at browser shutdown (or start up)?
(Or some other activity that would force a flush.)

[Giorgio Maone]

(Or some other activity that would force a flush.)

this.