[PATCH] improve precision of process accounting.

The unit of the cputime accouting values that are stored per process is
currently a microsecond. The CPU timer has a maximum granularity of
2**-12 microseconds. There is no benefit in storing the per process values
in the lesser precision and there is the disadvantage that the backend
has to do the rounding to microseconds. The better solution is to use
the maximum granularity of the CPU timer as cputime unit.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

1 files changed, 21 insertions, 21 deletions

diff --git a/arch/s390/include/asm/cputime.h b/arch/s390/include/asm/cputime.h
index 133ce054fc89..521726430afa 100644
--- a/arch/s390/include/asm/cputime.h
+++ b/arch/s390/include/asm/cputime.h
@@ -11,7 +11,7 @@
 
 #include <asm/div64.h>
 
-/* We want to use micro-second resolution. */
+/* We want to use full resolution of the CPU timer: 2**-12 micro-seconds. */
 
 typedef unsigned long long cputime_t;
 typedef unsigned long long cputime64_t;
@@ -53,9 +53,9 @@ __div(unsigned long long n, unsigned int base)
 #define cputime_ge(__a, __b)            ((__a) >= (__b))
 #define cputime_lt(__a, __b)            ((__a) <  (__b))
 #define cputime_le(__a, __b)            ((__a) <= (__b))
-#define cputime_to_jiffies(__ct)        (__div((__ct), 1000000 / HZ))
+#define cputime_to_jiffies(__ct)        (__div((__ct), 4096000000ULL / HZ))
 #define cputime_to_scaled(__ct)         (__ct)
-#define jiffies_to_cputime(__hz)        ((cputime_t)(__hz) * (1000000 / HZ))
+#define jiffies_to_cputime(__hz)        ((cputime_t)(__hz) * (4096000000ULL / HZ))
 
 #define cputime64_zero                  (0ULL)
 #define cputime64_add(__a, __b)         ((__a) + (__b))
@@ -64,7 +64,7 @@ __div(unsigned long long n, unsigned int base)
 static inline u64
 cputime64_to_jiffies64(cputime64_t cputime)
 {
-        do_div(cputime, 1000000 / HZ);
+        do_div(cputime, 4096000000ULL / HZ);
         return cputime;
 }
 
@@ -74,13 +74,13 @@ cputime64_to_jiffies64(cputime64_t cputime)
 static inline unsigned int
 cputime_to_msecs(const cputime_t cputime)
 {
-        return __div(cputime, 1000);
+        return __div(cputime, 4096000);
 }
 
 static inline cputime_t
 msecs_to_cputime(const unsigned int m)
 {
-        return (cputime_t) m * 1000;
+        return (cputime_t) m * 4096000;
 }
 
 /*
@@ -89,13 +89,13 @@ msecs_to_cputime(const unsigned int m)
 static inline unsigned int
 cputime_to_secs(const cputime_t cputime)
 {
-        return __div(cputime, 1000000);
+        return __div(cputime, 2048000000) >> 1;
 }
 
 static inline cputime_t
 secs_to_cputime(const unsigned int s)
 {
-        return (cputime_t) s * 1000000;
+        return (cputime_t) s * 4096000000ULL;
 }
 
 /*
@@ -104,7 +104,7 @@ secs_to_cputime(const unsigned int s)
 static inline cputime_t
 timespec_to_cputime(const struct timespec *value)
 {
-        return value->tv_nsec / 1000 + (u64) value->tv_sec * 1000000;
+        return value->tv_nsec * 4096 / 1000 + (u64) value->tv_sec * 4096000000ULL;
 }
 
 static inline void
@@ -114,12 +114,12 @@ cputime_to_timespec(const cputime_t cputime, struct timespec *value)
         register_pair rp;
 
         rp.pair = cputime >> 1;
-        asm ("dr %0,%1" : "+d" (rp) : "d" (1000000 >> 1));
-        value->tv_nsec = rp.subreg.even * 1000;
+        asm ("dr %0,%1" : "+d" (rp) : "d" (2048000000UL));
+        value->tv_nsec = rp.subreg.even * 1000 / 4096;
         value->tv_sec = rp.subreg.odd;
 #else
-        value->tv_nsec = (cputime % 1000000) * 1000;
-        value->tv_sec = cputime / 1000000;
+        value->tv_nsec = (cputime % 4096000000ULL) * 1000 / 4096;
+        value->tv_sec = cputime / 4096000000ULL;
 #endif
 }
 
@@ -131,7 +131,7 @@ cputime_to_timespec(const cputime_t cputime, struct timespec *value)
 static inline cputime_t
 timeval_to_cputime(const struct timeval *value)
 {
-        return value->tv_usec + (u64) value->tv_sec * 1000000;
+        return value->tv_usec * 4096 + (u64) value->tv_sec * 4096000000ULL;
 }
 
 static inline void
@@ -141,12 +141,12 @@ cputime_to_timeval(const cputime_t cputime, struct timeval *value)
         register_pair rp;
 
         rp.pair = cputime >> 1;
-        asm ("dr %0,%1" : "+d" (rp) : "d" (1000000 >> 1));
-        value->tv_usec = rp.subreg.even;
+        asm ("dr %0,%1" : "+d" (rp) : "d" (2048000000UL));
+        value->tv_usec = rp.subreg.even / 4096;
         value->tv_sec = rp.subreg.odd;
 #else
-        value->tv_usec = cputime % 1000000;
-        value->tv_sec = cputime / 1000000;
+        value->tv_usec = cputime % 4096000000ULL;
+        value->tv_sec = cputime / 4096000000ULL;
 #endif
 }
 
@@ -156,13 +156,13 @@ cputime_to_timeval(const cputime_t cputime, struct timeval *value)
 static inline clock_t
 cputime_to_clock_t(cputime_t cputime)
 {
-        return __div(cputime, 1000000 / USER_HZ);
+        return __div(cputime, 4096000000ULL / USER_HZ);
 }
 
 static inline cputime_t
 clock_t_to_cputime(unsigned long x)
 {
-        return (cputime_t) x * (1000000 / USER_HZ);
+        return (cputime_t) x * (4096000000ULL / USER_HZ);
 }
 
 /*
@@ -171,7 +171,7 @@ clock_t_to_cputime(unsigned long x)
 static inline clock_t
 cputime64_to_clock_t(cputime64_t cputime)
 {
-       return __div(cputime, 1000000 / USER_HZ);
+       return __div(cputime, 4096000000ULL / USER_HZ);
 }
 
 #endif /* _S390_CPUTIME_H */