On Friday, September 14, 2012 9:34 AM, Ian Abbott wrote: > > None of the low-level comedi drivers using the "mite" module currently > use the comedi `attach_pci` hook for auto-configuration of supported PCI > devices (though auto-configuration is currently supported using the > `attach` hook). They also rely on a list of National Instruments PCI > devices populated when the "mite" module is loaded. > > Add the `attach_pci` hook to these drivers and disable manual attachment > of supported PCI devices via the `attach` hook used by the > `COMEDI_DEVCONFIG` ioctl. The "ni_labpc" driver still needs the > `attach` hook to support ISA boards, so just make it log an error for > PCI devices. The `attach` hook is not needed by the "ni_6527", > "ni_65xx", "ni_660x", "ni_670x", "ni_pcidio" and "ni_pcimio" drivers, so > remove it from those drivers. > > The "ni_pcidio" driver needs to load firmware for the PCI-6534 card. > That used to be possible only by using the `COMEDI_DEVCONFIG` ioctl to > call the `attach` hook, including the firmware blobs in the ioctl data. > That is no longer supported. Instead, the `attach_pci` hook uses > `request_firmware()` to get the firmware blobs. > > Change the drivers to allocate a `struct mite_struct` dynamically in the > `attach_pci` hook instead of searching the list populated when the > "mite" module was loaded. Change the "mite" module to remove this list > altogether. > > All these drivers need further work, but this is hopefully a step in the > right direction! Amen... This all looks sane to me. And it does cleanup the ni drivers a bit. For the whole series, FWIW... Reviewed-by: H Hartley Sweeten <hsweeten@xxxxxxxxxxxxxxxxxxx> _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
