[linux-dvb] Re: Firmware won't upload.

["raid517@fairadsl.co.uk" <raid517@fairadsl.co.uk>]

Lol you can send me that mail again if you want, it is much welcomed! Well the good news is I followed your advice and things appear to have gone swimingly well. This is how my new bootup sequence now looks:

code:

Linux video capture interface: v1.00
saa7146: register extension 'budget dvb'.
saa7146: register extension 'budget dvb /w video in'.
saa7146: register extension 'budget_ci dvb'.
ACPI: PCI interrupt 0000:01:09.0[A] -> GSI 17 (level, high) -> IRQ 17
saa7146: found saa7146 @ mem e087d000 (revision 1, irq 17) (0x13c2,0x1011).
DVB: registering new adapter (TT-Budget/WinTV-NOVA-T  PCI).
TT-Budget/WinTV-NOVA-T  PCI adapter 0 has MAC addr = 00:d0:5c:22:0b:bf
saa7146: register extension 'budget_patch dvb'.
saa7146: register extension 'dvb'.
usbcore: registered new driver ttusb-dec
usbcore: registered new driver Technotrend/Hauppauge USB-Nova 
                                                             :code

Well it doesn't specifically say that the file uploaded - but at least this time it doesn't say that it didn't either. So I presume we can assume sucess with this particular issue? (I won't be testing the DVB apps until tomorrow).

In the end there was a lot of logic to what you said, it all apears to be about the sequence of events. I think that the firmware uploader simply expects that the driver will be there as a module - and when it can't see where to insert the required code (or whatever it does) it just gives up trying.

So the answer to this dilema is, don't compile these drivers into the kernel. Leave them as modules. I went ahead though and kept DVD core and my specific WinTV Nova-T driver in the kernel, but the front end section MUST be a module.

Hopefully you guys will remember this and will be able to advise others who encounter this issue in the future. Indeed a short note in the kernel. would not be such a bad thing at all. 
And to the guy who figured this out, thank you very much. I would never have worked it out on my own. You are one top dude!

Thanks for everything guys! :)

GJ

Hello,

I have a Nova-T card and experienced the same problem on a debian box. I
managed to solve it by simply compiling the code as modules and then
loading of the firmware worked.

As far as I understand your post, you compiled everything into the
kernel and tried to load the firmware during boot. If that is the case,
using modules should fix your problem.


Here are some thoughts on the problem as I experienced it, maybe someone
who has a better understanding on the bootup process can comment on it
to clarify it for me or even better lend a hand in fixing it.

The problem as I experienced it is not in the kernel not loading the
firmware, but simply that the filesystem on which the firmware is
residing is not mounted during initialization of the dvb-kernel part.
This prevents the kernel from accessing the file
/usr/lib/hotplug/firmware/tda1004x.bin. Having realized that problem, I
tried circumventing it by using an initrd image, which seemes to fail as
well, since the initrd is executed after kernel initialization, so also
after the kernel tries to load the firmware. The bootup process appears
the same when reading the error messages.

Digging in the 2.6 kernel features, I came accross an initial ramfs
filesystem which is loaded at the same time as the kernel is. This
initramfs is a cpio archieve. From what I understand about the
bootprocess, this is the place to put the firmware, since then the
kernel can load it when the kernel is initialized. However, I have not
managed to produce a working initramfs, so I haven't yet managed to
either put the firmware file in it nor be able to load it.


Although it is not ideal, I hope modularizing the code works for you as
well Q.


Hth,

Christoph Burger-Scheidlin