kernel-ark/include/linux/jiffies.h

282 lines
10 KiB
C
Raw Normal View History

#ifndef _LINUX_JIFFIES_H
#define _LINUX_JIFFIES_H
#include <linux/calc64.h>
#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/time.h>
#include <linux/timex.h>
#include <asm/param.h> /* for HZ */
/*
* The following defines establish the engineering parameters of the PLL
* model. The HZ variable establishes the timer interrupt frequency, 100 Hz
* for the SunOS kernel, 256 Hz for the Ultrix kernel and 1024 Hz for the
* OSF/1 kernel. The SHIFT_HZ define expresses the same value as the
* nearest power of two in order to avoid hardware multiply operations.
*/
#if HZ >= 12 && HZ < 24
# define SHIFT_HZ 4
#elif HZ >= 24 && HZ < 48
# define SHIFT_HZ 5
#elif HZ >= 48 && HZ < 96
# define SHIFT_HZ 6
#elif HZ >= 96 && HZ < 192
# define SHIFT_HZ 7
#elif HZ >= 192 && HZ < 384
# define SHIFT_HZ 8
#elif HZ >= 384 && HZ < 768
# define SHIFT_HZ 9
#elif HZ >= 768 && HZ < 1536
# define SHIFT_HZ 10
#else
# error You lose.
#endif
/* LATCH is used in the interval timer and ftape setup. */
#define LATCH ((CLOCK_TICK_RATE + HZ/2) / HZ) /* For divider */
/* Suppose we want to devide two numbers NOM and DEN: NOM/DEN, the we can
* improve accuracy by shifting LSH bits, hence calculating:
* (NOM << LSH) / DEN
* This however means trouble for large NOM, because (NOM << LSH) may no
* longer fit in 32 bits. The following way of calculating this gives us
* some slack, under the following conditions:
* - (NOM / DEN) fits in (32 - LSH) bits.
* - (NOM % DEN) fits in (32 - LSH) bits.
*/
#define SH_DIV(NOM,DEN,LSH) ( (((NOM) / (DEN)) << (LSH)) \
+ ((((NOM) % (DEN)) << (LSH)) + (DEN) / 2) / (DEN))
/* HZ is the requested value. ACTHZ is actual HZ ("<< 8" is for accuracy) */
#define ACTHZ (SH_DIV (CLOCK_TICK_RATE, LATCH, 8))
/* TICK_NSEC is the time between ticks in nsec assuming real ACTHZ */
#define TICK_NSEC (SH_DIV (1000000UL * 1000, ACTHZ, 8))
/* TICK_USEC is the time between ticks in usec assuming fake USER_HZ */
#define TICK_USEC ((1000000UL + USER_HZ/2) / USER_HZ)
/* TICK_USEC_TO_NSEC is the time between ticks in nsec assuming real ACTHZ and */
/* a value TUSEC for TICK_USEC (can be set bij adjtimex) */
#define TICK_USEC_TO_NSEC(TUSEC) (SH_DIV (TUSEC * USER_HZ * 1000, ACTHZ, 8))
/* some arch's have a small-data section that can be accessed register-relative
* but that can only take up to, say, 4-byte variables. jiffies being part of
* an 8-byte variable may not be correctly accessed unless we force the issue
*/
#define __jiffy_data __attribute__((section(".data")))
/*
* The 64-bit value is not atomic - you MUST NOT read it
* without sampling the sequence number in xtime_lock.
* get_jiffies_64() will do this for you as appropriate.
*/
extern u64 __jiffy_data jiffies_64;
extern unsigned long volatile __jiffy_data jiffies;
#if (BITS_PER_LONG < 64)
u64 get_jiffies_64(void);
#else
static inline u64 get_jiffies_64(void)
{
return (u64)jiffies;
}
#endif
/*
* These inlines deal with timer wrapping correctly. You are
* strongly encouraged to use them
* 1. Because people otherwise forget
* 2. Because if the timer wrap changes in future you won't have to
* alter your driver code.
*
* time_after(a,b) returns true if the time a is after time b.
*
* Do this with "<0" and ">=0" to only test the sign of the result. A
* good compiler would generate better code (and a really good compiler
* wouldn't care). Gcc is currently neither.
*/
#define time_after(a,b) \
(typecheck(unsigned long, a) && \
typecheck(unsigned long, b) && \
((long)(b) - (long)(a) < 0))
#define time_before(a,b) time_after(b,a)
#define time_after_eq(a,b) \
(typecheck(unsigned long, a) && \
typecheck(unsigned long, b) && \
((long)(a) - (long)(b) >= 0))
#define time_before_eq(a,b) time_after_eq(b,a)
Re: [NFS] [PATCH] Attribute timeout handling and wrapping u32 jiffies I would like to discuss the idea that the current checks for attribute timeout using time_after are inadequate for 32bit architectures, since time_after works correctly only when the two timestamps being compared are within 2^31 jiffies of each other. The signed overflow caused by comparing values more than 2^31 jiffies apart will flip the result, causing incorrect assumptions of validity. 2^31 jiffies is a fairly large period of time (~25 days) when compared to the lifetime of most kernel data structures, but for long lived NFS mounts that can sit idle for months (think that for some reason autofs cannot be used), it is easy to compare inode attribute timestamps with very disparate or even bogus values (as in when jiffies have wrapped many times, where the comparison doesn't even make sense). Currently the code tests for attribute timeout by simply adding the desired amount of jiffies to the stored timestamp and comparing that with the current timestamp of obtained attribute data with time_after. This is incorrect, as it returns true for the desired timeout period and another full 2^31 range of jiffies. In testing with artificial jumps (several small jumps, not one big crank) of the jiffies I was able to reproduce a problem found in a server with very long lived NFS mounts, where attributes would not be refreshed even after touching files and directories in the server: Initial uptime: 03:42:01 up 6 min, 0 users, load average: 0.01, 0.12, 0.07 NFS volume is mounted and time is advanced: 03:38:09 up 25 days, 2 min, 0 users, load average: 1.22, 1.05, 1.08 # ls -l /local/A/foo/bar /nfs/A/foo/bar -rw-r--r-- 1 root root 0 Dec 17 03:38 /local/A/foo/bar -rw-r--r-- 1 root root 0 Nov 22 00:36 /nfs/A/foo/bar # touch /local/A/foo/bar # ls -l /local/A/foo/bar /nfs/A/foo/bar -rw-r--r-- 1 root root 0 Dec 17 03:47 /local/A/foo/bar -rw-r--r-- 1 root root 0 Nov 22 00:36 /nfs/A/foo/bar We can see the local mtime is updated, but the NFS mount still shows the old value. The patch below makes it work: Initial setup... 07:11:02 up 25 days, 1 min, 0 users, load average: 0.15, 0.03, 0.04 # ls -l /local/A/foo/bar /nfs/A/foo/bar -rw-r--r-- 1 root root 0 Jan 11 07:11 /local/A/foo/bar -rw-r--r-- 1 root root 0 Jan 11 07:11 /nfs/A/foo/bar # touch /local/A/foo/bar # ls -l /local/A/foo/bar /nfs/A/foo/bar -rw-r--r-- 1 root root 0 Jan 11 07:14 /local/A/foo/bar -rw-r--r-- 1 root root 0 Jan 11 07:14 /nfs/A/foo/bar Signed-off-by: Fabio Olive Leite <fleite@redhat.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
2007-07-27 01:59:00 +00:00
#define time_in_range(a,b,c) \
(time_after_eq(a,b) && \
time_before_eq(a,c))
/* Same as above, but does so with platform independent 64bit types.
* These must be used when utilizing jiffies_64 (i.e. return value of
* get_jiffies_64() */
#define time_after64(a,b) \
(typecheck(__u64, a) && \
typecheck(__u64, b) && \
((__s64)(b) - (__s64)(a) < 0))
#define time_before64(a,b) time_after64(b,a)
#define time_after_eq64(a,b) \
(typecheck(__u64, a) && \
typecheck(__u64, b) && \
((__s64)(a) - (__s64)(b) >= 0))
#define time_before_eq64(a,b) time_after_eq64(b,a)
/*
* Have the 32 bit jiffies value wrap 5 minutes after boot
* so jiffies wrap bugs show up earlier.
*/
#define INITIAL_JIFFIES ((unsigned long)(unsigned int) (-300*HZ))
/*
* Change timeval to jiffies, trying to avoid the
* most obvious overflows..
*
* And some not so obvious.
*
* Note that we don't want to return LONG_MAX, because
* for various timeout reasons we often end up having
* to wait "jiffies+1" in order to guarantee that we wait
* at _least_ "jiffies" - so "jiffies+1" had better still
* be positive.
*/
#define MAX_JIFFY_OFFSET ((LONG_MAX >> 1)-1)
extern unsigned long preset_lpj;
/*
* We want to do realistic conversions of time so we need to use the same
* values the update wall clock code uses as the jiffies size. This value
* is: TICK_NSEC (which is defined in timex.h). This
* is a constant and is in nanoseconds. We will used scaled math
* with a set of scales defined here as SEC_JIFFIE_SC, USEC_JIFFIE_SC and
* NSEC_JIFFIE_SC. Note that these defines contain nothing but
* constants and so are computed at compile time. SHIFT_HZ (computed in
* timex.h) adjusts the scaling for different HZ values.
* Scaled math??? What is that?
*
* Scaled math is a way to do integer math on values that would,
* otherwise, either overflow, underflow, or cause undesired div
* instructions to appear in the execution path. In short, we "scale"
* up the operands so they take more bits (more precision, less
* underflow), do the desired operation and then "scale" the result back
* by the same amount. If we do the scaling by shifting we avoid the
* costly mpy and the dastardly div instructions.
* Suppose, for example, we want to convert from seconds to jiffies
* where jiffies is defined in nanoseconds as NSEC_PER_JIFFIE. The
* simple math is: jiff = (sec * NSEC_PER_SEC) / NSEC_PER_JIFFIE; We
* observe that (NSEC_PER_SEC / NSEC_PER_JIFFIE) is a constant which we
* might calculate at compile time, however, the result will only have
* about 3-4 bits of precision (less for smaller values of HZ).
*
* So, we scale as follows:
* jiff = (sec) * (NSEC_PER_SEC / NSEC_PER_JIFFIE);
* jiff = ((sec) * ((NSEC_PER_SEC * SCALE)/ NSEC_PER_JIFFIE)) / SCALE;
* Then we make SCALE a power of two so:
* jiff = ((sec) * ((NSEC_PER_SEC << SCALE)/ NSEC_PER_JIFFIE)) >> SCALE;
* Now we define:
* #define SEC_CONV = ((NSEC_PER_SEC << SCALE)/ NSEC_PER_JIFFIE))
* jiff = (sec * SEC_CONV) >> SCALE;
*
* Often the math we use will expand beyond 32-bits so we tell C how to
* do this and pass the 64-bit result of the mpy through the ">> SCALE"
* which should take the result back to 32-bits. We want this expansion
* to capture as much precision as possible. At the same time we don't
* want to overflow so we pick the SCALE to avoid this. In this file,
* that means using a different scale for each range of HZ values (as
* defined in timex.h).
*
* For those who want to know, gcc will give a 64-bit result from a "*"
* operator if the result is a long long AND at least one of the
* operands is cast to long long (usually just prior to the "*" so as
* not to confuse it into thinking it really has a 64-bit operand,
* which, buy the way, it can do, but it take more code and at least 2
* mpys).
* We also need to be aware that one second in nanoseconds is only a
* couple of bits away from overflowing a 32-bit word, so we MUST use
* 64-bits to get the full range time in nanoseconds.
*/
/*
* Here are the scales we will use. One for seconds, nanoseconds and
* microseconds.
*
* Within the limits of cpp we do a rough cut at the SEC_JIFFIE_SC and
* check if the sign bit is set. If not, we bump the shift count by 1.
* (Gets an extra bit of precision where we can use it.)
* We know it is set for HZ = 1024 and HZ = 100 not for 1000.
* Haven't tested others.
* Limits of cpp (for #if expressions) only long (no long long), but
* then we only need the most signicant bit.
*/
#define SEC_JIFFIE_SC (31 - SHIFT_HZ)
#if !((((NSEC_PER_SEC << 2) / TICK_NSEC) << (SEC_JIFFIE_SC - 2)) & 0x80000000)
#undef SEC_JIFFIE_SC
#define SEC_JIFFIE_SC (32 - SHIFT_HZ)
#endif
#define NSEC_JIFFIE_SC (SEC_JIFFIE_SC + 29)
#define USEC_JIFFIE_SC (SEC_JIFFIE_SC + 19)
#define SEC_CONVERSION ((unsigned long)((((u64)NSEC_PER_SEC << SEC_JIFFIE_SC) +\
TICK_NSEC -1) / (u64)TICK_NSEC))
#define NSEC_CONVERSION ((unsigned long)((((u64)1 << NSEC_JIFFIE_SC) +\
TICK_NSEC -1) / (u64)TICK_NSEC))
#define USEC_CONVERSION \
((unsigned long)((((u64)NSEC_PER_USEC << USEC_JIFFIE_SC) +\
TICK_NSEC -1) / (u64)TICK_NSEC))
/*
* USEC_ROUND is used in the timeval to jiffie conversion. See there
* for more details. It is the scaled resolution rounding value. Note
* that it is a 64-bit value. Since, when it is applied, we are already
* in jiffies (albit scaled), it is nothing but the bits we will shift
* off.
*/
#define USEC_ROUND (u64)(((u64)1 << USEC_JIFFIE_SC) - 1)
/*
* The maximum jiffie value is (MAX_INT >> 1). Here we translate that
* into seconds. The 64-bit case will overflow if we are not careful,
* so use the messy SH_DIV macro to do it. Still all constants.
*/
#if BITS_PER_LONG < 64
# define MAX_SEC_IN_JIFFIES \
(long)((u64)((u64)MAX_JIFFY_OFFSET * TICK_NSEC) / NSEC_PER_SEC)
#else /* take care of overflow on 64 bits machines */
# define MAX_SEC_IN_JIFFIES \
(SH_DIV((MAX_JIFFY_OFFSET >> SEC_JIFFIE_SC) * TICK_NSEC, NSEC_PER_SEC, 1) - 1)
#endif
/*
* Convert various time units to each other:
*/
extern unsigned int jiffies_to_msecs(const unsigned long j);
extern unsigned int jiffies_to_usecs(const unsigned long j);
extern unsigned long msecs_to_jiffies(const unsigned int m);
extern unsigned long usecs_to_jiffies(const unsigned int u);
extern unsigned long timespec_to_jiffies(const struct timespec *value);
extern void jiffies_to_timespec(const unsigned long jiffies,
struct timespec *value);
extern unsigned long timeval_to_jiffies(const struct timeval *value);
extern void jiffies_to_timeval(const unsigned long jiffies,
struct timeval *value);
extern clock_t jiffies_to_clock_t(long x);
extern unsigned long clock_t_to_jiffies(unsigned long x);
extern u64 jiffies_64_to_clock_t(u64 x);
extern u64 nsec_to_clock_t(u64 x);
#define TIMESTAMP_SIZE 30
#endif