[pvrusb2] HVR-1950 not being found at boot (along side a pvrusb2 2900 device)

[Roger]

On Tue, 2009-06-02 at 20:08 -0500, Mike Isely wrote:
> On Tue, 2 Jun 2009, Roger wrote:
> 
> > The pvrusb2 driver is *only* picking-up the "pvrusb2 2900 device" and
> > not the "hvr-1950".  (Both are plugged into the same USB host
> > controller.)
> > 
> > At boot, only the "pvrusb2 2900" is being found and does work with
> > mythtv/mplayer.
> > 
> > The following two lines within /var/log/messages, especially the second
> > line, are ominously absent from boot:
> > 
> > Jun  2 13:09:37  usb 1-4: new high speed USB device using ehci_hcd and
> > address 10
> > Jun  2 13:09:37  usb 1-4: New USB device found, idVendor=2040,
> > idProduct=7501
> > 
> > 
> > Searching the entire past boot log for "usb 1-4" or "hcd and address 10"
> > seems absent as well until the device is replugged-in.
> > 
> > Looks like pvrusb2 is either failing to detect both devices, or the host
> > controller is only reporting one at 
> > boot.
> 
> Uh, so how did you have this working before?  What changed?

<shrugs>  I thought so too.  Maybe inhibiting i2c_scan created a unique
timing situation allowing the hvr-1950 to be initialized partially.  I'm
going from memory here.  But since removing i2c_scan module option, not
only do I nolonger get the /dev/dvb, but also obviously no /dev/video1.

So remember, prior to removing i2c_scan, I still didn't have a
functioning /dev/dvb (and likely /dev/video1?).  I still had to replug
the device to reinitialize.

> 
> > 
> > 
> > (BTW, I have removed i2c_scan as a module option and for the past
> > several boots I've noticed this activity.)
> 
> The i2c_scan option has nothing to do with this.

Granted.


> > 
> > Using in-kernel pvrusb2 driver.
> > 
> > # uname -a
> > Linux localhost2.local 2.6.29-gentoo-r4Y #9 SMP PREEMPT Tue Jun 2
> > 03:38:16 AKDT 2009 i686 Pentium III (Coppermine) GenuineIntel GNU/Linux
> > 
> 
> The mechanism in Linux which associates a newly-plugged-in device with 
> its driver is outside of the driver itself.  Said another way, the 
> pvrusb2 driver does not itself "detect" the hardware but rather the 
> kernel does the detection and then another mechanism (udev, IIRC) 
> figures out which driver should be associated with it.  The driver is 
> then called with a handle to the hardware device - at the point the 
> driver is entered the detection is already done.
> 
> If the pvrusb2 driver is not responding when you plug in an HVR-1950, 
> then something is preventing the association from working.  When the 
> pvrusb2 driver is compiled one of the elements it exports to the kernel 
> is a table of USB ids (vendor+product ids) which tell the Linux system 
> what the driver can handle.  Odds are that the table is incomplete.
> 
> A more interesting question is how that can happen.  And for that I have 
> a very simple question for you: Obviously you had this working before, 
> so WHAT DID YOU CHANGE?

As I stated above, this seems to occur only on boot with both devices
plugged in.  (NOTE: I have yet to try just booting with one device
connected -- ie. just the hvr-1950 connected at boot.)


Exactly what I'm thinking.  The in-kernel usb mechanism is failing to
detect both devices properly at the same time.  Could be it's getting
confused or something.  I should also try another kernel as this kernel
minor version is missing a specific PCI patch would could be inhibiting
this behavior.

After reading a /var/log/messages file for the first time in a year or
so, there's so much going on, it's hard to trace at times.


<shrugs> I'm not too worried about this yet as there is a work around of
reconnecting the device (after things are booted and I find the missing
device files ie. /dev/video1 & /dev/dvb.


-- 
Roger
http://rogerx.freeshell.org