Holger Kiehl wrote:
I'd suggest to use cvsps in order to get an idea what happened to the source and then check out the relevant file version using 'cvs up -dPD<DATE> <filename>' or referencing the exact file version.Can you please try to find out why it's failing in the probe function in stv0299.c or try to check out older versions of this file from CVS and report when exactly the bug was introduced?
cvsps is a nice tool to investigate the edit history of files in a CVS repository, maybe this helps you...
I did try this but as of patchset 1029 (a big one) the patching did no longer work (it might be an error on my part since it is the first time I work with CVS). But I can say it did not work with patchset 1352, 1334, 1333, 1236, 1233 and 1224.
I suppose the Metzler driver works with this card because they got paid for this driver. But anyway, if we want to have reliable KNC1-Support in the official linux kernel source tree we'll have to fix this in the LinuxTV CVS source tree anyway -- multiple diverging source trees don't really help anybody.Meanwhile someone else suggested to me to try the driver from the Metzler brothers. When loading them I get more information on the output:
After a soft or a hard reboot? I remember that the stv0299 sometimes completely hangs when the i2c timing registers accidentally get misconfigured by the probing code. Then you have to switch off your computer comletely and remove the power cable for some minutes...I now remember that I did try this driver earlier and later switched to linuxtv-dvb-1.0.0-pre3, since trying to use dvbtune did not work. I do remember dvbtune did talk once with the card but then too it would always return the error 'No such device' when trying to open /dev/dvb/adapter0/frontend0. Looking back in /var/log/messages I see that the frontend did load once but after a reboot never again: