On Fri, 2008-03-21 at 23:02 -0700, Jim Kusznir wrote: > Ok, it took a while to get more time with this, as its been a busy > week, but I finally did: > > I also have more data on symptoms. > > First, on a cold boot, starting case seems to be semi-random. > Sometimes by powering down and back up a few times, I'll get one tuner > or the other to work (it does change halfs).
Which likely points to something not on the card, but a source common to both halves of the card. It would be unlikely to have both halves of the card intermittently failing due to a failure somewhere on the card. > The failure mode also > varyies; sometimes I get perfect video with no audio, sometimes no > video, and sometimes badly distorted video with no audio. This points to various degrees of initialization/configuration of the respective CX25840. The CX25840 on each half of the card is configured by i2c bus transactions initiated by the host CPU over the PCI bus interface to the respective CX23416. (Offsets 0x7000-0x700f from the base address of the CX23416's mapped register address space/) > I discovered that by unloading ivtv and cx25840 and reloading ivtv, > the failure mode would change (sometimes better, sometimes worse). > Also, at one point this week, "the card" stopped funcitoning all > together (black screen on both inputs) without a module reload or > power cycle (something not observed up to that point). > > And now, in the process of debugging, both halfs of the card came up > OK one time (I have not been able to replicate that) > > Based on this and other facts, I suspect the card is good and have not > pulled it. However, if it will be a useful test point, i can pull it > and put it in the "old" system (old mobo, same HD image, etc. > Different PSU). Ultimately you do need to eliminate unknowns. There's no point in trying to get a bad card to work. If you can test it to verify it works properly in another setup, then you should. I don't think your card is bad, but it would be nice to know for sure. Your symptoms really point to PCI bus or power supply problems. > The only i2c module with debugging options was i2c_algo_bit, and > turning on its debugging didn't generate any additional output > (actually, I didn't really notice any additional output from any of > the debugging). Have you tried: # modprobe -r ivtv # modprobe ivtv newi2c=1 The default is to force using the new i2c code, which is normally only enabled if an IR blaster is present on the board. By using this, you'll enable new ivtv specific i2c routines to deal odd PVR card i2c behavior. (See the comments at the top of linux/drivers/media/video/ivtv/ivtv-i2c.c in the driver source.) To ensure all debugging is on as well: # modprobe -r ivtv cx25840 # modprobe cx25840 debug=1 # modprobe ivtv debug=65535 newi2c=1 > So....I'm at a loss as to what to do next....Suggestions? > > --Jim I hope this helps. -Andy _______________________________________________ ivtv-devel mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-devel
