On Monday, August 04, 2014 5:53 AM, Ian Abbott wrote: > On 2014-08-01 22:18, H Hartley Sweeten wrote: >> The drivers that use this module with memory mapped io all have the >> ioremap'ed base address stored in the comedi_device 'mmio' member. >> >> Modify subdev_8255_io() to handle the memory mapped io. This allows >> removing the private callbacks from some of the drivers. >> >> Signed-off-by: H Hartley Sweeten <hswee...@visionengravers.com> >> Cc: Ian Abbott <abbo...@mev.co.uk> >> Cc: Greg Kroah-Hartman <gre...@linuxfoundation.org> > > What if a driver sets both dev->mmio and dev->iobase, but has the 8255 > in the dev->iobase region? Is it reasonable for the driver to "know" > that dev->mmio would be used by the 8255 subdevice driver and so it > would need to implement a callback for an 8255 in the dev->iobase region?
Hmm.. None of the existing comedi drivers have this problem but I suppose it could happen. > I think the 8255 code needs to be explicitly told whether to use > dev->iobase or dev->mmio if it handles both. That could be done either > by adding a parameter to subdev_8255_init() (and subdev_8255_init_irq(), > though that could be removed as interrupt support is incomplete and > nothing has ever used it), or by exporting a separate init function > (e.g. subdev_8255_mm_init()) to mark it as using dev->mmio. I'll redo this patch to address the possible problem. Thanks, Hartley _______________________________________________ devel mailing list de...@linuxdriverproject.org http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
