[BUG] imxrt:lpi2c kconfig and source code discrepancy about nuttx

[danielappiagyei-bc]

@davids5 hello good sir, thank you again sir for creating another excellent driver so I didn't have to! Do you concur with my findings above?

[davids5]

@danielappiagyei-bc

You are welcome.

Good job unwinding it! I think I only tested IMXRT_LPI2C_DYNTIMEO (I only use the setting). Also this code is in other archs as well - so the fix may be far reaching.

I do concur. Kconfig should supply the values and the enables. All the default guard in code (setting defaults) should go away.

I also think the the value can be just timeout in us and a 32 bit value. That will give use us resolution and 4.29 S max.

ticks is also not a good unit. It needs to insure the value is 1 for the min us - but then it can be 10ms for 1 tick. (we use 1 ms tick)

So that should be addressed also.

What do you think?

BTW, I do see CONFIG_IMXRT_LPI2C_DYNTIMEO_STARTSTOP used
https://github.com/apache/nuttx/blob/master/arch/arm/src/imxrt/imxrt_lpi2c.c#L838