Dave Smith wrote:
On Fri, 6 Feb 2004, Johannes Stezenbach wrote:
Dave Smith wrote:
I'm using a NOVA-T card (grundig_29504-401/saa7146), and a
2.4.20 kernel or 2.6.2 kernel for testing.
...
Please can someone give me some pointers. This problem is stopping me
from moving over to a 2.6 kernel.
The grundig_29504-401 driver in 1.0.0-pre3 had some gross
bugs, but they were corrected in 1.0.1. Could you copy
grundig_29504-401.c from 1.0.0-pre2 into 1.0.1 and rebuild?
Does the 1.0.1 driver work for you then? If yes, my guess is
that for some reason the simpler and slightly incorrect
version of tsa5060_set_tv_freq() in 1.0.0-pre3 works better
for you. Could you copy the code in tsa5060_set_tv_freq()
from 1.0.0-pre3 into the driver in Linux 2.6.2 and see
if that works?
Please report your findings.
Thanks for the suggestions Johannes. Both of those suggestions worked, so
using the old tsa5060_set_tv_freq() function from 1.0.0-pre2, I now have
very good Bit Error Rates whilst running the 2.6.2 kernel and drivers.
Hm, that's interesting...
If your feel up for some more experiments you could check whether
the changed frequency offset (36125000 vs. 36000000) makes the
difference, or the band/chargepump setting.
So, does this mean it's a fault to my specific card, or to all the
NOVA-T's with this chipset?
I don't know, and I don't even have a card with this frontend.