Hi, I have made some tests trying to identify problems with the saa7134 chip on FV3K. My card mostly works and the sound is ok. But sometimes the sound is not present and I am now almost sure that the gpio setting is not the problem - the card does not react to any changes. The problem is in the sound decoding, I get the messages: Oct 19 07:45:48 localhost kernel: scanning 5.500 MHz => dc is 0 [0/0] Oct 19 07:45:48 localhost kernel: scanning 6.000 MHz => dc is 0 [0/0] Oct 19 07:45:49 localhost kernel: scanning 6.500 MHz => dc is 0 [0/0] Oct 19 07:45:49 localhost kernel: saa7134[0]/audio: carrier scan failed, using default (5.500 MHz) The default is not helpfull here, though. If it worked, it would have worked while scanning, too. Then I tried to alter the saa7134 registers with function saa_writeb(), I added an ioctl to access the driver to do this. The thing is made so that it is executed in the function 'tvaudio_init'. I tried to change audio registers more or less randomly but I could not make any change with it. However, what works is poking the whole address range 0x000-0x1ff with zeros (I hope doing such stuff will not destroy something...). After this the sound appears - auto scanning finds the 5.5MHz carrier, but the video disappears. When I unload and reload saa7134 driver, the tv works just fine. So, there is some problem with initialization which can be solved, I think. I am still trying to locate exactly where the problem is, but the problem is I can only do this when the computer stars with the 'no sound' problem. This is rare, in 10-20% of switch-ons. I am making the register 0x000-0x3ff dump so I always have the exact state of registers. I also tried to poke the values which I saved in working situation to the card when started non-working. This does not start the sound, sigh! Further on, I also have occasional problem with video. I noticed that it happens if I switch to 'grabdisplay' option in xawtv. The problem gets worse also if I run alevt together with xawtv. When the video crashes, I get the next message from xawtv: ioctl: VIDIOC_DQBUF(0,type=1,off=0,len=0,used=0,flags=0x0,ts=0,seq=0): Input/output error After this no driver reloading helps the card working again - I have to reboot. At this error, I get something in the /var/log/messages: Oct 19 20:52:10 localhost kernel: ERROR: NO_DEVICE Oct 19 20:52:10 localhost kernel: tuner: chip found @ 0xc2 Oct 19 20:52:10 localhost kernel: i2c-core.o: client [Philips PAL] registered to adapter [saa7134[0]](pos. 0). Oct 19 20:52:10 localhost kernel: i2c-core.o: adapter saa7134[0] registered as adapter 0. Oct 19 20:52:10 localhost kernel: saa7134[0]: Huh, no eeprom present (err=-5)? Oct 19 20:52:10 localhost kernel: saa7134[0]/video: DCSDT: no signal Oct 19 20:52:10 localhost kernel: ERROR: NO_DEVICE Oct 19 20:52:10 localhost kernel: saa7134[0]: i2c scan: found device @ c2 Oct 19 20:52:10 localhost kernel: saa7134[0]/video: DCSDT: found PAL I haven't yet discovered what this ERRORS mean. The third thing I also discovered is very strange and worries me a bit. I get this message in /var/log/messages: hdf: cdrom_pc_intr: The drive appears confused (ireason = 0x 1) I do have 3 devices on interrupt 9: 9: 34471 XT-PIC ide2, usb-uhci, saa7134[0] So, my cdrom is on this shared interrupt with the card and it seems that the irq handling is somehow problematic. I did not notice any strange behavior with my cdrom, it works normaly despite of these messages. I am happy that this message does not appear where I have the hard disk.... So much till now. I hope Gerd and other driver writers can say something about this and that this will help discover the bugs. I can also send the register dumps, if wanted. Stane.
