On Mon, Nov 27, 2017 at 4:05 PM, Andy Shevchenko andriy.shevchenko@linux.intel.com wrote:
On Mon, 2017-11-27 at 14:19 +0100, Arnd Bergmann wrote:
timespec overflows in 2038 on 32-bit architectures, and the getnstimeofday() suffers from possible time jumps, so the timestamps here are better done using ktime_get(), which has neither of those problems.
In case of ov2680, we don't seem to use the timestamp at all, so I just remove it.
ktime_t timedelay = ns_to_ktime( min((u32)abs(dev->number_of_steps) *
DELAY_PER_STEP_NS,
(u32)DELAY_MAX_PER_STEP_NS),
};
(u32)DELAY_MAX_PER_STEP_NS));
Since you are touching this, it might make sense to convert to
min_t(u32, ...)
...and locate lines something like:
ktime_t timeday = ns_to_ktime(min_t(u32, param1, param2));
From my pov will make readability better.
Yes, good idea. Re-sending the patch now. Thanks for taking a look,
Arnd