|
|
#1 | |
|
Registered User
Join Date: Sep 2006
Posts: 21
|
System is a dual proc AMD running a recent linux kernel and the GeForce FX 5500 card using driver "nvidia" version 9631 and when using this proprietary driver there is trouble using v4l software such as spcaview, vlc, vgrabber on a Logitech Communicate STX model 0x046d:0x08ad plugged into an NEC based USB 2.0 pci card as the most that will be captured is the first frame, if at all, where as changing to the standard "nv" unaccelerated driver completely resolves the problem and allows full streaming including from one or even two of these identical cams simultaneously.
This bug has troubled me for many months now and its only recently that I've found changing from "nvidia" to the "nv" driver model serves as a cure all. It would be fantastic if a developer at nvidia could ask around to see if this camera is already available at your offices, and plug it into a test system to reproduce this or demonstrate that its working for them. |
|
|
|
|
|
|
#2 | |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
Does this problem persist with 1.0-9746 ?
|
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Sep 2006
Posts: 21
|
Thanks for the quick response; the problem still persists with 9746, and you can see the nvidia-bug-report for this latest release as part of this message.
One of the accelerated GL programs I run is the multiplayer game by idSoftware, known as quake4-demo. In fact although the ChangeLog doesn't state it, perhaps the game is running even better for me at this new version release, although I haven't tested that much lately. I'll stick with 9746 and better for now, when gaming. The concern though is with some of the USB camera interaction. My current guess is that there is something going on with the way video updates are done such that for example the nvidia.ko module isn't preemptable enough in certain places for solid reliability for such purposes as USB transactions. It just seems that the video driver is whats holding up the cam streaming from happenning. |
|
|
|
|
|
#4 | |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
I've got a USB webcam here which requires the gspca kernel module, and I also see the same problem when using spcaview, however when using other packages such as ekiga, it doesn't reproduce with the nvidia X driver. At this point, this looks like an application bug, not a driver bug.
|
|
|
|
|
|
|
#5 |
|
Registered User
Join Date: Sep 2006
Posts: 21
|
ekiga runs for me under the unaccelerated "nv" driver, however still no luck with accelerated driver release "nvidia" version 9746. For completeness could you please post the exact camera model name and USB id for the logitech unit that worked on your system?
I realize this thread may lead no where to resolving the issue, however here are some other things I've observed on my system, using a simple v4l capture application vgrabber which simply reads in one frame, or two frames and puts it to disk: Running it from single user mode or a virtual console generally works fine, however switching back to accelerated "nvidia" xwindows to run vgrabber from an xterm the application usually doesn't complete especially if there is mouse movement or other window activity such as scrolling a browser; the unaccelerated "nv" driver is much more forgivng in this respect and frames get read in from this particular 0x046d:0x08ad model camera just fine. |
|
|
|
|
|
#6 | |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
I'm not using a Logitech webcam, I'm just using a webcam that is supported by gspca. It seems odd that this issue would not be specific to the gspca kernel module, if it were truly a driver bug. The webcam that I'm using is:
0ac8:301b Z-Star Microelectronics Corp. ZC0301 WebCam |
|
|
|
|
![]() |
| Thread Tools | |
|
|