Very Odd Issue W/ A CentOS 6 System
Now, this is one we have an issue with: it’s got a bttv card, and motion running on it. It’s a Dell PE R720. For some reason, it has never liked the card: 20 min after reboot, it says fatal bus error… but nothing’s wrong, and it runs just fine. Well, expect that we need to keep it on the
2.6.32-431.20.3 kernel, because *everything* later, it randomly starts rebooting, several times a day.
But we’ve never had anything like this, and I’ve never seen anything like this: the bttv module was throwing a ton of IRQ lockup errors. Restarting motion didn’t help. So I was wondering if the issue was with the bttv kernel module, so I rmmoded that, then modprobe bttv… and that was all she wrote. The system lost network connectivity.
Ok, I go over to the rack, and plug in the monitor-on-a-stick, and that’s where it got very strange, and I’ve never seen anything like this: I turn up the monitor… and see it announcing it had seen a new USB device (the keyboard)… and no login, it kept printing copying data [
When it hit 100%, it rebooted.
So – has anyone ever seen this behavior? Anyone have an idea what it was copying?
mark
9 thoughts on - Very Odd Issue W/ A CentOS 6 System
Isn’t that just it dealing with the kdump?
jh
Hi,
What is the make/model of the card?
Are you supplying options when loading the bttv module?
Regards
Phil
Phil Wyett wrote:
lspci says:
Brooktree Corporation Bt878 Audio Capture (rev 11)
On and off, I’ve been googling to find if there’s something I can tell it, or the kernel on boot. Currently, I have, in /etc/modprobe.d/bttv:
options i2c-algo-bit bit_test=1
options bttv gbuffers card
Hi,
The options line 1 is used widely and only worth changing if all else fails.
The second options line option of combfilter by docs indicate there is no 2 option and is enabled by default, so you can remove that option to test.
I would enable two possible compatibility options and see if that helps.
triton1=0/1 – for Triton1 (+others) compatibility.
vsfx=0/1 – yet another chipset bug compatibility bit
The triton1 insmod option sets the EN_TBFX bit in the control register. The vsfx insmod option does the same for EN_VSFX bit. If you have stability problems you can try if one of these options makes your box work solid.
Source: http://xawdecode.sourceforge.net/aideUS/htmlpage/BTTV-param.htm
Regards
Phil
Hi, Phil,
Phil Wyett wrote:
Let me add to that: lspci -s -vv gives me this
SERR-
05:00.0 Multimedia video controller: Brooktree Corporation Bt878 Video Capture (rev 11)
Subsystem: iTuner Spectra8 CardA Input0
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop-
ParErr+ Stepping- SERR+ FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
Ok, thanks.
Thanks very much. I’ll try those as I have the chance; unfortunately, it’s not something I can do frequently, as the server’s in use, and exporting project directories to users… oh, and it is actually running motion for surveillance of the room…
mark
Phil Wyett wrote:
Ok, I’ve been doing some research, partly refreshing my memory, since we bought these 2-3 years ago. As you note, above, it *says* it’s an iTuner Spectra8; *however*, when I look at bttv-cards.c, what I see is that *it*
thinks that such a card has *one* video and one audio. In fact, this is a Bt878, and it has *four* inputs for each, with four chips on the card, one for each channel. The nearest thing to it *appears* to be a Provideo PV150, card
Hi Mark,
It did specify spectra 8 in your lspci output. Below is product page.
http://www.ituner.com/spectra.htm
In my research I did find this related page.
http://www.linuxconsulting.ro/spectra8/
This does state that with latest internal driver, it is detected and works as the ProVision 150. The card=98,… is correct.
The matter of reducing all the settings to 4 rather than 8 may help.
coring is a luma setting. See:
https://www.linuxtv.org/wiki/index.php/Bttv#insmod_Options
Not sure and someone else maybe can shed better light on the setting. By default it is disabled and I would tend to think it can be removed.
Regards
Phil
Phil Wyett wrote:
Any thoughts on the number of buffers?
mark
Hi Mark,
You have the setting of 16 which maybe correct for how many feeds you are taking/processing. If less I maybe tempted to drop it to the default
8. If the device is running as close to or at default and no hickups –
all the better. :-)
Regards
Phil