Since about the last two weeks I've been seeing Firefox 4.0 freezing up while watching Youtube videos. So I updated to 6.0.2 and things haven't changed. The freezing up would start when a video has started to play for about 30 seconds and the strange thing is that the audio continued to play, only the video didn't. Also FF stopped responding at all. This lasts for about 10 seconds, sometimes longer. I couldn't see any pattern to when the freeze lasted longer. It made no difference if the the video were already completely downloaded or still downloading.
Then I started letting the Task Manager run parallel to FF and watched what was happening. When the freeze started FF suddenly went to between 15 and 20% CPU activity and AvastSvc.exe to about 5%. I also looking into the Avast Real-Time Protection and saw that corresponding exactly to the freezes Script Shield was running and reported 'Last script scanned' to be:
c:\Documents and Settings\{user-name-changed}\Application Data\Mozilla\Firefox\Profiles\hmj6ncgs.default\extensions\{e0204bd5-9d31-402b-a99d-a6aa8ffebdca}.xpi\components\cookie-jar-selector.js
which belongs to Torbutton 1.4.3 (
https://www.torproject.org/torbutton/).
I've had Torbutton installed for several months, but never had this behavior before a couple of weeks ago.
I also noted that while on YT in My Account - Subscriptions, if I hovered the mouse pointer over a link to a video, the same thing would happen. I generally open the pages to videos I watch in this manner. If I hover over a link to a video in the main page this doesn't happen, so it may have something to do with YT prefetching and preparing to update a cookie, although it doesn't use cookies to keep track of which videos I've already watched, because if I switch to Google Chrome, the videos I've watched markings are all up to date.
Also, just while entering this post using FF with YT open on another tab SS has scanned the above script freezing FF.
If I disable Torbutton, the freezing doesn't occur. But the Torbutton add-on is actually only active if the button is pressed, so if it is not, which is the default and the case during all of the above, it should not be doing anything, including even looking at the above script. I'm not saying that it isn't, but that it would be rather odd if it were. Also the cookie-jar-selector.js script is only 21,466 bytes in size, so I wouldn't imagine that SS should need 10+ seconds to actually scan it. But that's just my speculation.
