[pvrusb2] Compile issue

[Mike Isely]

On Thu, 8 Dec 2005, cardboil wrote:

> Mike, I can't seem to find the /dev/video0 device, despite a seemingly OK
> hotplug loading.

Unless you've created /dev/video0 yourself, you need udev installed so 
that it will magically appear when appropriate.  Using udev is not a 
strict requirement, but without it you need to manually create the entry.


>
> Also, I noticed the following problems (?) in the dmesg output.
>
> pvrusb2 i2c_attach [client=saa7115 @ 0x21 id=0x49]
>
> <snip>
>
> pvrusb2 upload of pvrusbf2 : 262144 / 262144
> pvrusb2 Probing for ivtv video decoder driver
> pvrusb2 pvr2_i2c_saa7115_cmd failed with status -22
> pvrusb2 Failed to select a viable v4l video decoder driver
>
>
> As you can see saa7115 is not getting attached (there is no "pvrusb2
> attaching saa7115 I2C client" as in the case of the others like tveeprom,
> msp3400 and tda 9887).
>
> Any pointers on what might be happening?

I think you kicked over a fairly big rock here.  Your observation is right 
on the money.  There are two things going on here.

First, any time an I2C chip driver attaches to the I2C bus exported by the 
pvrusb2 driver, the pvrusb2 driver attempts to recognize which driver it 
is and squirrel away its handle for future use.  In this case it failed to 
recognize the saa7115 module at all.

Second, there are (AFAIK) two saa7115 modules floating around and they 
have utterly incompatible APIs.  The version that is from ivtv (and also 
supplied by pvrusb2 but it is really ultimately from ivtv) is the older 
and more mature version.  Unfortunately that older version uses a 
completely non-standard API.  The V4L folks are pulling this module into 
the V4L tree, but in doing so they've reworked the API to be consistent 
with other chip drivers that are already in V4L.  Thus two modules, two 
incompatible APIs.  The pvrusb2 driver, wanting to be robust in the face 
of change, tries to detect which type of driver it is talking to.  It does 
this by sending a command to it and looking at the response; e.g. to 
detect the newer driver it will send a standard query which is 
unimplemented in the ivtv version thus the response can be used to tell 
them apart.  The pvrusb2 driver can be configured at compile time to look 
for either or both versions (examine compat.h for comments on how to 
control this).  By default it looks for everything.  Anyway, this 
detection process can result in the -22 status message you see, and in 
that case it's OK.

However in your case since the module completely failed to be recognized 
by the driver, the subsequent detection to figure out *which* version it 
had was a waste of time and failed.

The root cause is to figure out why the chip driver wasn't recognized.  A 
key clue here is the id of 0x49 that was reported.  That's a value I 
didn't expect.  The pvrusb2 driver uses that id (mostly) to recognize chip 
drivers.  We need to figure out where 0x49 came from.  I have a suspicion, 
but I'll have to chase that down later.

To dig deeper, I need to know what kernel you compiled against, whether or 
not you have ivtv installed (and which version), and also whether or not 
you have patched up or otherwise built & installed a later V4L version 
into the kernel.  All those things affect where saa7115 comes from.

You might be able to work around this by using the saa7115.ko that can be 
built with the pvrusb2 driver.  If you try that, make sure you unload the 
existing driver from memory, rename conflicting instances of the driver 
that might be floating around in your modules tree, and rerun depmod -a 
after you change things.

If we can figure out where this ID of 0x49 is coming from, I can attempt 
to adjust the pvrusb2 driver to account for it.

I know there's other e-mail here that I haven't answered yet.  I will 
fully catch up on everything later tonight.

   -Mike


-- 
                         |         Mike Isely          |     PGP fingerprint
      Spammers Die!!     |                             | 03 54 43 4D 75 E5 CC 92
                         |   isely @ pobox (dot) com   | 71 16 01 E2 B5 F5 C1 E8
                         |                             |