Search code examples
clong-integerunsigned

UL vs casting to unsigned long

What's the difference between 160000000UL and (unsigned long) 160000000?

PROBLEM:

I have a #define with the following:

#define SYS_CLK (160000000UL / 1000UL)

SYS_CLK is different depending on the system clock. I want to know if I can change it to:

#define SYS_CLK ((unsigned long)(RCC_MAX_FREQUENCY/*get freq from the system. It returns e.g. 160000000U*/) / 1000UL)

And RCC_MAX_FREQUENCY is defined as:

#define RCC_MAX_FREQUENCY           168000000U

Are they the same?

(I checked this, but it's not my question.)

Solution

  • What's the difference between 160000000UL and (unsigned long) 160000000?

    160000000UL is unsigned at pre-processing time, (unsigned long) 160000000 is signed.

    When constants are used in a macro at pre-processing time (e.g. #if math), the width (int, long, long long) and cast are irrelevant as the width used is (u)intmax_t. The sign-ness is relevant.

    In general, do not force wider constant width with L or LL unless needed. Do use U when sign-ness is needed.

    Casting to (unsigned long) or using L can force the constant to be at least 64-bit, when a value about 160,000,000 only needs 32-bit. Using an excessive wide type incurs down-convert warnings. IMO, coders too often ignore/disable such warnings as they are deemed noisy - when in fact they are useful.

    Below is OK:

    #define SYS_CLK ((unsigned long)(RCC_MAX_FREQUENCY) / 1000UL)

    Yet I would recommend:

    #define SYS_CLK (RCC_MAX_FREQUENCY / 1000u)