alistair23-linux/arch/x86/include/asm/timer.h
Salman Qazi 4cecf6d401 sched, x86: Avoid unnecessary overflow in sched_clock
(Added the missing signed-off-by line)

In hundreds of days, the __cycles_2_ns calculation in sched_clock
has an overflow.  cyc * per_cpu(cyc2ns, cpu) exceeds 64 bits, causing
the final value to become zero.  We can solve this without losing
any precision.

We can decompose TSC into quotient and remainder of division by the
scale factor, and then use this to convert TSC into nanoseconds.

Signed-off-by: Salman Qazi <sqazi@google.com>
Acked-by: John Stultz <johnstul@us.ibm.com>
Reviewed-by: Paul Turner <pjt@google.com>
Cc: stable@kernel.org
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111115221121.7262.88871.stgit@dungbeetle.mtv.corp.google.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2011-11-16 19:51:25 +01:00

84 lines
2.3 KiB
C

#ifndef _ASM_X86_TIMER_H
#define _ASM_X86_TIMER_H
#include <linux/init.h>
#include <linux/pm.h>
#include <linux/percpu.h>
#include <linux/interrupt.h>
#define TICK_SIZE (tick_nsec / 1000)
unsigned long long native_sched_clock(void);
extern int recalibrate_cpu_khz(void);
extern int no_timer_check;
/* Accelerators for sched_clock()
* convert from cycles(64bits) => nanoseconds (64bits)
* basic equation:
* ns = cycles / (freq / ns_per_sec)
* ns = cycles * (ns_per_sec / freq)
* ns = cycles * (10^9 / (cpu_khz * 10^3))
* ns = cycles * (10^6 / cpu_khz)
*
* Then we use scaling math (suggested by george@mvista.com) to get:
* ns = cycles * (10^6 * SC / cpu_khz) / SC
* ns = cycles * cyc2ns_scale / SC
*
* And since SC is a constant power of two, we can convert the div
* into a shift.
*
* We can use khz divisor instead of mhz to keep a better precision, since
* cyc2ns_scale is limited to 10^6 * 2^10, which fits in 32 bits.
* (mathieu.desnoyers@polymtl.ca)
*
* -johnstul@us.ibm.com "math is hard, lets go shopping!"
*
* In:
*
* ns = cycles * cyc2ns_scale / SC
*
* Although we may still have enough bits to store the value of ns,
* in some cases, we may not have enough bits to store cycles * cyc2ns_scale,
* leading to an incorrect result.
*
* To avoid this, we can decompose 'cycles' into quotient and remainder
* of division by SC. Then,
*
* ns = (quot * SC + rem) * cyc2ns_scale / SC
* = quot * cyc2ns_scale + (rem * cyc2ns_scale) / SC
*
* - sqazi@google.com
*/
DECLARE_PER_CPU(unsigned long, cyc2ns);
DECLARE_PER_CPU(unsigned long long, cyc2ns_offset);
#define CYC2NS_SCALE_FACTOR 10 /* 2^10, carefully chosen */
static inline unsigned long long __cycles_2_ns(unsigned long long cyc)
{
unsigned long long quot;
unsigned long long rem;
int cpu = smp_processor_id();
unsigned long long ns = per_cpu(cyc2ns_offset, cpu);
quot = (cyc >> CYC2NS_SCALE_FACTOR);
rem = cyc & ((1ULL << CYC2NS_SCALE_FACTOR) - 1);
ns += quot * per_cpu(cyc2ns, cpu) +
((rem * per_cpu(cyc2ns, cpu)) >> CYC2NS_SCALE_FACTOR);
return ns;
}
static inline unsigned long long cycles_2_ns(unsigned long long cyc)
{
unsigned long long ns;
unsigned long flags;
local_irq_save(flags);
ns = __cycles_2_ns(cyc);
local_irq_restore(flags);
return ns;
}
#endif /* _ASM_X86_TIMER_H */