On Wed, 11 Nov 2009, Stephan Diestelhorst wrote: > Hi Linux USB developers, > Laurent has suggested to contact you with the matter at hand, please see below! > > Subject: Re: [Linux-uvc-devel] BisonCam 5986:0203 kills USB > From: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> > To: linux-uvc-devel@xxxxxxxxxxxxxxxx > > On Tuesday 10 November 2009 16:59:04 Stephan Diestelhorst wrote: > > > This issue is with the built-in Bison web cam in an MSI Wind > > > U100. Once the camera is enabled (via hotkey), the entire USB > > > subsystem is not responding anymore, causing non-detection of > > > freshly attached USB devices and prevents suspending the box. > > > > > > The only fix so far is to blacklist the "uvcvideo" module or to > > > keep the cam disabled (with hotkey). > > > > > > This is with a stable 2.6.31-14-generic kernel on Ubuntu Karmic. > > > > > > A number of related resources: > > > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/435352 contains > > > many user reports. > > > > > > http://marc.info/?l=linux-kernel&m=125541673726384&w=2 a stalled > > > discussion on the LKML with more technical insight > > > > > > https://lists.berlios.de/pipermail/linux-uvc-devel/2009-September/005143.html > > > a stalled discussion on the linux-uvc-devel ML > > > > > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=04a37e0f32f9882430bc1899899d2ed91b8aaf5bis > > > (according to launchpad) the breaking changeset. > > > > > > This is a very unpleasant bug, as it affects the usability of the > > > system a lot and is a regression: Kernel 2.6.28 worked flawless > > > with the webcam. > > > > Even though the git commit that caused the regression modifies the > > uvcvideo driver, this is an issue that need to be solved at the USB > > core level. The commit's purpose is to enable autosuspend, but a > > nasty side effect kills USB on some (buggy ?) controllers because an > > interrupt URB is canceled too fast after being submitted. > > > > > Many thanks and again, sorry for cross-posting, > > > > Once again no apologize needed. You should keep cross-posting and > > report the problem to the linux-usb mailing list. I'm not familiar > > enough with the EHCI architecture to work on a fix, so the solution > > will likely come from linux-usb developers. > > Any help? There's a temporary solution (in one of the threads mentioned above) described here (the udelay(2000) call): http://marc.info/?l=linux-kernel&m=125622034713518&w=2 See also http://marc.info/?l=linux-usb&m=125628206801242&w=2 and the following messages in that thread. You should experiment and see if the 2000 can be replaced with a smaller number. I don't have time to work on a proper fix at the moment. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
