[pvrusb2] Mac Mini & pvrusb2

[Mike Isely]

On Sun, 14 Apr 2008, Bjorn Danielsson wrote:

> My Mac Mini arrived last friday. I spent this weekend installing and
> configuring Slamd64 (the x86-64 version of Slackware) and this evening
> I tried the pvrusb2 driver in the 2.6.24.4 kernel. So far I have only
> tested it with "cat /dev/video0 > file.mpg" and this worked perfectly
> without any problems, patches, or special tricks. I used the firmware
> from the ivtv page. Next weekend I will play around with some of the
> ioctls, and report back here if anything unusual happens...

Glad to hear it.


> 
> The saa7115 driver produced some curious dmesg output:
> 
> saa7115 1-0021: Input:           Composite 4
> saa7115 1-0021: Video signal:    bad
> saa7115 1-0021: Frequency:       50 Hz
> saa7115 1-0021: Detected format: BW/No color
> 
> OK, I can accept the "bad" video signal, but why does it say
> "BW/No color"? The captured MPEG did have correct colors.
> (I realize that this has nothing to the pvrusb2 driver per se)

The key to that is knowing *when* it showed you that info.

There is a V4L request that can be "broadcast" to all the attached 
drivers, which basically means "printk your status".  A driver (such as 
pvrusb2) can trigger this request any time it wants.  There are 3 cases 
where this happens in pvrusb2:

1. In response to a VIDIOC_LOG_STATUS ioctl().

2. As a side effect of cat'ing out /sys/class/pvrusb2/*/debuginfo

3. As part of hardware initialization (e.g. when the device is plugged 
in).

Case (3), as part of the setup when the device is plugged in, happens 
before any streaming takes place.  I had implemented this towards the 
end of the initialization sequence as a means to log a starting point 
state in the driver.  I had found it useful when debugging issues in the 
pvrusb2 that had impacted how these client drivers were starting 
(especially cx25840).

So this is actually normal.  Having a "bad" video signal is not an 
error; it is simply the status of the driver after initialization but 
before any streaming takes place.  If you want to see something more 
interesting, cat out that debuginfo file (then check your system log) 
while you have a streaming operation in progress.

Admittedly this feature (case 3) is less useful than it used to be.  I 
should disable it.

  -Mike


-- 

Mike Isely
isely @ pobox (dot) com
PGP: 03 54 43 4D 75 E5 CC 92 71 16 01 E2 B5 F5 C1 E8