Messages in this thread | | | Date | Sat, 6 Mar 2010 05:14:38 +0800 | Subject | Different definition in div_sc and clocksource_cyc2ns | From | Lin Mac <> |
| |
hi,
While working with the clocksource, I found that the relation of cycle and nanosecond of div_sc and clocksource_cyc2ns is different.
linux-2.6.33 div_sc in include/linux/clockchips.h suggests that clock_ticks = (nanoseconds * factor) >> shift. 98/* 99 * Calculate a multiplication factor for scaled math, which is used to convert 100 * nanoseconds based values to clock ticks: 101 * 102 * clock_ticks = (nanoseconds * factor) >> shift. 103 * 104 * div_sc is the rearranged equation to calculate a factor from a given clock 105 * ticks / nanoseconds ratio: 106 * 107 * factor = (clock_ticks << shift) / nanoseconds 108 */ 109static inline unsigned long div_sc(unsigned long ticks, unsigned long nsec, 110 int shift) 111{ 112 uint64_t tmp = ((uint64_t)ticks) << shift; 113 114 do_div(tmp, nsec); 115 return (unsigned long) tmp; 116}
However, clocksource_cyc2ns in include/linux/clocksource.h suggest nanoseconds=(cycles*mult)>>shift. 268static inline s64 clocksource_cyc2ns(cycle_t cycles, u32 mult, u32 shift) 269{ 270 return ((u64) cycles * mult) >> shift; 271}
But there are many others using div_sc, I don't think they got issues with it. So what am I missing?
Best Regards, Mac Lin
| |