I discovered my problem with the bttv not working on a Boser HS878 with HS-1600 embedded board. The error messages were misleading, it was not an interrupt from another device. By putting tracings in the bttv-driver.c source I chased down the problem. The capture card works fine transferring images to the frame buffer for display, but fails on capture operations. It goes into the bttv_irq routine in bttv-driver and gets stuck in an infinite loop. The driver reads the interrupt register and mask from the BT878 over PCI and generates thousands of PPERR interrupts. According to the BT878A data sheet these are parity errors on the PCI bus, and it says they are serious error for evertything but display operations, that explains why the framebuffer works but capture does not. There are also FBUS interrupts intersperced with the PPERR interrupts, these are pixel data fifo overruns. So the driver is fine, the capture card is fine, the main board is hosed. This is what happens when you play with early revs of embedded hardware. I experienced some other strange bus activity and I have concluded that the main board is defective. So this one is going back to the manufacturer and another one is already on the way. This one will also be an upgrade to TM5800 processor. Thanks to those who sent me direct emails. The good news is that I now know how the bttv driver works, and I also know what the bttv-cards.c entries should be for this new card, which BTW is tiny (Mini PCI), smaller than a PCMCIA card. I will also soon be testing a combon audio and video card in this format with BT878A and ESS Solo1 (ES1938/1946) onboard, plus a firewire card with dual ports. I don't expect those to be plug-n-run either, but the potential is good and probably worth the effort. The whole system is approx 4x4 inches with capture onboard. -- Doug
