On Monday, September 24, 2012 8:28 AM, Ian Abbott wrote: > `ni_ai_poll()` currently acquires (and later releases) the comedi > device's spin-lock iff `in_interrupt()` returns 0. However, it is only > called during processing of a `COMEDI_POLL` ioctl so `in_interrupt()` > will always return 0 in this case. Remove this test and acquire/release > the spin-lock unconditionally. This eliminates a sparse warning about > different lock contexts for basic block. Ian, How can you guarantee that the user space COMEDI_POLL ioctl wil never happen when this driver is handling its interrupt? Maybe a better solution would be for ni_ai_poll() to just return -EBUSY if it is in_interrupt(). Regards, Hartley _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
