On Sun, Nov 29, 2020 at 12:18 AM Andy Shevchenko <andy.shevchenko@xxxxxxxxx> wrote: > On Sun, Nov 29, 2020 at 12:22 AM Linus Walleij <linus.walleij@xxxxxxxxxx> wrote: > > On Sat, Nov 28, 2020 at 7:54 PM Jonathan Cameron <jic23@xxxxxxxxxx> wrote: > > > > > From: Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx> > > > > > > Inspired by Andy Shevchenko's proposal to use get_unaligned_leXX(). > > > > > > The whole one time programable memory is treated as a single 64bit > > > little endian value. Thus we can avoid a lot of messy handling > > > of fields overlapping byte boundaries by just loading and manipulating > > > it as an __le64 converted to a u64. That lets us just use FIELD_GET() > > > and GENMASK() to extract the values desired. > > > > > > Note only build tested. > > > > > > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx> > > > Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> > > > Cc: Linus Walleij <linus.walleij@xxxxxxxxxx> > > > > Are there any specific prerequisites? linux-next? > > > > When I apply this and try to compile for an ARMv7 target I get > > a lot of noise and an error: > > > > In file included from <command-line>:0:0: > > ../drivers/iio/gyro/mpu3050-core.c: In function ‘mpu3050_hw_init’: > > ../include/linux/bits.h:36:11: warning: right shift count is negative > > [-Wshift-count-negative] > > (~UL(0) >> (BITS_PER_LONG - 1 - (h)))) > > I think GENMASK_ULL() has to be used. Oh indeed. It works as long as one just try to test-compile it on a 64bit machine of course :D Yours, Linus Walleij
