kernel bug that causes problems with webcams etc?

[netnomad]

hi roberts,

at the end you will find a patch that could solve this problem...

i have some troubles with webcams in tc.
one is an internal one in my acer apire one netbook, the other is an external usb-connected logitech c270.
searching in the web, i found repeatedly some hints that there could be a kernel bug,
especially when dmesg reports something like "cannot set freq 16000 to ep 0x86"
or something like that.
i tested my 4.1microcore-fluxbox-config, but also a 3.6tinycore-basic-config,
in both case this "cannot set freq 16000 to ep 0x86" appeared sometimes.
for your further information i tested three kind of basic and familiar hardware,
equipped with intel boards and grafic: centrino-laptop, atom-netbook and atom-desktop.

my goal is to use linphone with video on different hardware or at least in one sole configuration.
i load alsa, linphone and mplayer-nodeps
and the command "mplayer -v tv://" works on all hardware, all configs and versions flawless!
so the hardware is recognized.

after loading and starting linphone i get the same messages like broberts,
http://forum.tinycorelinux.net/index.php/topic,12063.msg64302.html#msg64302
but additionally "SEGMENTATION FAULT".
some users in the net think that some modules are not proper loading,
f.e. snd_usb_audio
so i blacklisted that modul snd_usb_audio in my bootloader.

dmesg of microcore 4.1 plus Xvesa plus fluxbox
Linux box 3.0.3-tinycore #90210 SMP Thu Aug 18 11:28:07 EEST 2011 i686 GNU/Linux

here you find a "3:3:3: cannot get freq at ep 0x86"

Code: [Select]

Linux video capture interface: v2.00
uvcvideo: Found UVC 1.00 device <unnamed> (046d:0825)
intel_rng: FWH not detected
input: UVC Camera (046d:0825) as /devices/pci0000:00/0000:00:1d.7/usb1/1-2/1-2:1.0/input/input5
usbcore: registered new interface driver uvcvideo
USB Video Class driver (v1.1.0)
intel_rng: FWH not detected
intel_rng: FWH not detected
HDA Intel 0000:00:1b.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
hda_intel: msi for device 1849:0888 set to 0
hda_intel: position_fix set to 1 for device 1849:0888
HDA Intel 0000:00:1b.0: setting latency timer to 64
hda_codec: ALC888: BIOS auto-probing.
intel_rng: FWH not detected
intel_rng: FWH not detected
3:3:3: cannot get freq at ep 0x86
usbcore: registered new interface driver snd-usb-audio
hda-intel: IRQ timing workaround is activated for card #0. Suggest a bigger bdl_pos_adj.


different dmesg but resembling problem with tinycore 3.6 (pretty unmodified)
kernel Linux box 2.6.33.3-tinycore #2012 SMP Wed May 12 17:05:42 EEST 2010 i686 GNU/Linux

in that version you will find various "10:3:x: cannot set freq xxxx":

Code: [Select]

usb 1-2: new high speed USB device using ehci_hcd and address 8
8:3:2: cannot set freq 24000 to ep 0x86
usbcore: registered new interface driver snd-usb-audio
usb 1-2: USB disconnect, address 8
intel_rng: FWH not detected
intel_rng: FWH not detected
intel_rng: FWH not detected
usb 1-2: new high speed USB device using ehci_hcd and address 9
Linux video capture interface: v2.00
9:3:3: cannot set freq 32000 to ep 0x86
uvcvideo: Found UVC 1.00 device <unnamed> (046d:0825)
input: UVC Camera (046d:0825) as /devices/pci0000:00/0000:00:1d.7/usb1/1-2/1-2:1.0/input/input5
usbcore: registered new interface driver uvcvideo
USB Video Class driver (v0.1.0)
usb 1-2: USB disconnect, address 9
usb 1-2: new high speed USB device using ehci_hcd and address 10
uvcvideo: Found UVC 1.00 device <unnamed> (046d:0825)
input: UVC Camera (046d:0825) as /devices/pci0000:00/0000:00:1d.7/usb1/1-2/1-2:1.0/input/input6
10:3:1: cannot set freq 16000 to ep 0x86
10:3:2: cannot set freq 24000 to ep 0x86
10:3:3: cannot set freq 32000 to ep 0x86
10:3:4: cannot set freq 48000 to ep 0x86


searching further around at different places like debian, ubuntu, fedora and archlinux
i finally found something interesting here at archlinux:
https://bbs.archlinux.org/viewtopic.php?id=121607


in the forum it was confirmed that a patch succeeded in archlinux concerning that problem:

Code: [Select]

--- a/drivers/usb/core/quirks.c.orig    2011-08-04 02:21:36.807543038 +0300
+++ b/drivers/usb/core/quirks.c 2011-08-04 02:22:26.814211036 +0300
@@ -41,6 +41,9 @@
        /* Logitech Harmony 700-series */
        { USB_DEVICE(0x046d, 0xc122), .driver_info = USB_QUIRK_DELAY_INIT },
 
+    /* Logitech Webcam C270 */
+    { USB_DEVICE(0x046d, 0x0825), .driver_info = USB_QUIRK_RESET_RESUME },
+
        /* Philips PSC805 audio device */
        { USB_DEVICE(0x0471, 0x0155), .driver_info = USB_QUIRK_RESET_RESUME },

perhaps you see here the point what could be done to solve this problem!?
every help is welcomed.

thank you for your help.

[curaga]

While that change is not in 3.0.3, and has been added in later stable 3.0 updates, it should have nothing to do with your issue. It concerns the webcam crashing after suspend and resume.

If an app segfaults, it's just about always the app's fault. That is, a linphone bug. That mplayer works correctly shows it's not likely anything to do with the kernel.

[netnomad]

hi curaga,

thank you for your reply.
do you use webcams, do you use opensource-sip-phone?
i don't want to give up, so every little info could be a step to a solution...

[curaga]

No, I'm afraid I don't. I did quick tests on two of my laptops with their integrated webcams (mplayer), both worked. Then I put tape over them

Have you reported the bug to linphone?

[netnomad]

hi curaga,

usually my netbook-webcam has a tape on it, too .
did you try it just with mplayer or did you try also linphone?
i reported the bug or misbehavior to bmarkus, do you think i should report it directly to linphone?
i'm not sure whether it could be a common configuration problem with alsa, pulseaudio
or the interaction of several multimedia-infrastructures or "sockets"...
the webcams work flawless directly or raw,
but in the interactive structure or environment of sip or webcam-conferences
more aspects have to receive more attention.

every idea or hint is welcomed.
thank you.

[curaga]

Just mplayer, I didn't test any other app.

For the issue, even if it's caused by pulse or some other component, it's still a bug in linphone - the app should never crash like that. Please do report it to linphone.

[bmarkus]

Just mplayer, I didn't test any other app.

For the issue, even if it's caused by pulse or some other component, it's still a bug in linphone - the app should never crash like that. Please do report it to linphone.

Please consider that repo version of Linphone is not the latest, so better to see what is happening after update. I'm busy with day job, so need 4-6 weeks to touch. Anybody is free to go on with Linphone

[curaga]

Yes, that would be best.