510782 - Mozilla disk activity causes hardware conflict with USB devices

Status: RESOLVED INCOMPLETE

(Firefox :: General, defect)

Description

[eric.hall]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2 (.NET CLR 3.5.30729)

This bug has been around for a very long time. Essentially, significant Firefox storage activity triggers a hardware conflict with USB devices which can bump devices off-line temporarily, or in some cases semi-permanently (until some kind of manual intervention).

This shows up most often when Firefox is used on a laptop, and most often occurs with web pages that have many large graphic objects in a scrollable list or firefox objects that store lots of "local settings". This problem does not appear when using other browsers on the same hardware to view the same web pages and is specific to Firefox storage routines.

The specific sequence of events is that you go to a web site, it loads a flash object, and then your external mouse or keyboard becomes unresponsive and jerky even though the laptop mouse and keyboard still work fine. If the flash applet continues running the external USB device bus may simply stop responding. Typically you can recover from this by closing the web page with the onboard input devices. However sometimes you have to put the laptop into standby mode, or close Firefox, or all of the above, in order to allow the computer to rescan the devices. Sometimes you have to unplug the external devices and wait for the hardware detection routines to notice that the devices have been removed and reinserted before they will work again.

It is possible to avoid the above scenario by using the Flash security settings page (http://www.macromedia.com/support/documentation/en/flashplayer/help/settings_manager03.html) to disable local storage. However many flash applets depend on local storage for things like saving audio playback levels, and some applets will not work at all if local caching is disabled, so this is not a functional workaround. But it does demonstrate that the problem is related to the way that Firefox manages local storage.

Another instance where this can happen often is when viewing multipage lists of images. Whenever the browser scrolls down a page and new images are loaded, there is a hit to disk while the browser does something (clears files from cache to make room for object? I don't know), and this causes the USB devices to go unresponsive until the activity ends. If this is unchecked, then the devices will go offline altogether, just as occurs with the flash objects that write to disk.

FWIW, I used to have this problem with a Dell Latitude D600 laptop a couple of years ago and thought it was just a disk driver issue. However I have since upgraded to an IBM/Lenovo Thinkpad T60p and the same issues appear there, and they have completely different USB hardware of course. I have also replaced the external keyboard and mouse.

Other people have reported similar problems. #498034 seems to be an instance of this bug, for example. #485725 is identical instance of the multiple-objects scrolling problem.



Reproducible: Always

Steps to Reproduce:
1. using a laptop with external USB mouse and/or keyboard, access a web page with multiple embedded objects or use a flash object with large amounts of "local" data

Actual Results:  
devices on the USB bus become jumpy and start to stutter, and will eventually stop responding altogether. on-board mouse and keyboard will continue to operate normally.


Expected Results:  
the software should not do whatever it is doing to cause the USB devices to go offline. it appears to me that soem storage operations are causing hardware conflicts with some types of USB devices.

Comment 1

[Matthias Versen [:Matti]]

A usermode application like Firefox can not cause hardware conflicts.
Firefox could make the USB bus busy if you launch Firefox from a USB device or if your user profile is located on a USB device but it can not cause the USB device to go offline that is a hardware or driver issue.

>Steps to Reproduce:
>1. using a laptop with external USB mouse and/or keyboard, access a web page
>with multiple embedded objects or use a flash object with large amounts of
>"local" data

This is of course wfm with my local system (AMD quadcore on win2003) and my Laptop (Intel dualcore with vista) but I don't store anything from Firefox on an USB disk.

Comment 2

[eric.hall]

I do not have any USB storage devices attached to the laptop.

I can record a video capture showing how Firefox causes the USB devices to die if you would like

Comment 3

[Matthias Versen [:Matti]]

No need for a video, I believe you of course but I don't think that there is anything that we could do about it because it's something outside of our control.
Gecko doesn't do anything with USB or the USB drivers directly and there must be a bug somewhere on the OS/driver level if you get such issues.

Comment 4

[eric.hall]

I understand that you are not directly doing anything with hardware that would cause conflicts with the USB device chain. However as stated this only happens with Firefox, and has happened with different laptop brands. There is even a thread about this on the Thinkpad forums where the workaround was discovered to be a separate USB controller.

You are doing something with the storage code that causes a conflict with the USB devices. Maybe you are not doing device driver I/O directly but maybe the way you are calling the storage APIs is causing a problem. 

I can replicate this at will so if there is something you want me to capture let me know.

Comment 5

[Matthias Versen [:Matti]]

Note: I'm not a developer. I'm a bug triager that filters incoming bugs for the developers.

I have 3 issues with this report:
a) I can not find the right component for this because File I/O is by many Gecko and Firefox components. Finding the right component is required because different developers are working on different components.
b) this is a hardware bug that affects some/one chipset (?). You can be sure that this is a hardware or driver bug if an external USB hub helps.
c) There is no way to find the exact things that makes it break, only the chipset driver developer or the chipset hardware developers can find the issue.

I'm sorry but I don't think that a workaround for Firefox will be added in the next years. The chipset driver should be fixed or they should tell us what's wrong with our File I/O.

Comment 6

[Tyler Downer [He/Him]]

Reporter, are you still seeing this issue with Firefox 3.6.12 or later in safe mode? If not, please close. These links can help you in your testing.
http://support.mozilla.com/kb/Safe+Mode
http://support.mozilla.com/kb/Managing+profiles
Also, please consider using the most recent Firefox 4 beta build, your bug may be resolved there.

Comment 7

[eric.hall]

The problem is still present in 3.6.12 even with -safe-mode switch

However it seems to be gone in 4.0 beta 7. I will need to do further testing to confirm but all initial testing is good.

Comment 8

[Tyler Downer [He/Him]]

No reply, INCOMPLETE. Please retest with Firefox 3.6.13 or later and a new profile (http://support.mozilla.com/kb/Managing+profiles). If you continue to see this issue with the newest firefox and a new profile, then please comment on this bug.