Merge branch 'timecounter'

Richard Cochran says:

====================
Time Counter fixes and improvements

Several PTP Hardware Clock (PHC) drivers implement the clock in
software using the timecounter/cyclecounter code. This series adds one
simple improvement and one more subtle fix to the shared timecounter
facility. Credit for this series goes to Janusz Użycki, who pointed
the issues out to me off list.

Patch #1 simply move the timecounter code into its own file. When
working on this series, it was really annoying to see half the kernel
recompile after every tweak to the timecounter stuff. There is no
reason to keep this together with the clocksource code.

Patch #2 implements an improved adjtime() method, and patches 3-10
convert all of the drivers over to the new method.

Patch #11 fixes a subtle but important issue with the timecounter WRT
frequency adjustment. As it stands now, a timecounter based PHC will
exhibit a variable frequency resolution (and variable time error)
depending on how often the clock is read.

In timecounter_read_delta(), the expression

   (delta * cc->mult) >> cc->shift;

can lose resolution from the adjusted value of 'mult'. If the value
of 'delta' is too small, then small changes in 'mult' have no effect.
However, if the delta value is large enough, then small changes in
'mult' will have an effect.

Reading the clock too often means smaller 'delta' values which in turn
will spoil the fine adjustments made to 'mult'. Up until now, this
effect did not show up in my testing. The following example explains
why.

The CPTS has an input clock of 250 MHz, and the clock source uses
mult=0x80000000 and shift=29, making the ticks to nanoseconds
conversion like this:

   ticks * 2^31
   ------------
       2^29

Imagine what happens if the clock is read every 10 milliseconds. Ten
milliseconds are about 2500000 ticks, which corresponds to about 21
bits. The product in the numerator has then 52 bits. After the shift
operation, 23 bits are preserved. This results in a frequency
adjustment resolution of about 0.1 ppm (not _too_ bad.)

A frequency resolution of 1 ppm requires 20 bits.
A frequency resolution of 1 ppb requires 30 bits.

For the 250 MHz CPTS clock, reading every 4 seconds yields a 1 ppb
resolution (which is the finest that our API allows).

However, the error can be much higher if the clock is read too often
or if time stamps occur close in time to read operations. In general
it is really not acceptable to allow the rate of clock readings to
influence the clock accuracy.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 140 insertions, 103 deletions

diff --git a/include/linux/clocksource.h b/include/linux/clocksource.h
index abcafaa20b86..9c78d15d33e4 100644
--- a/include/linux/clocksource.h
+++ b/include/linux/clocksource.h
@@ -18,8 +18,6 @@
 #include <asm/div64.h>
 #include <asm/io.h>
 
-/* clocksource cycle base type */
-typedef u64 cycle_t;
 struct clocksource;
 struct module;
 
@@ -28,106 +26,6 @@ struct module;
 #endif
 
 /**
- * struct cyclecounter - hardware abstraction for a free running counter
- *	Provides completely state-free accessors to the underlying hardware.
- *	Depending on which hardware it reads, the cycle counter may wrap
- *	around quickly. Locking rules (if necessary) have to be defined
- *	by the implementor and user of specific instances of this API.
- *
- * @read:		returns the current cycle value
- * @mask:		bitmask for two's complement
- *			subtraction of non 64 bit counters,
- *			see CLOCKSOURCE_MASK() helper macro
- * @mult:		cycle to nanosecond multiplier
- * @shift:		cycle to nanosecond divisor (power of two)
- */
-struct cyclecounter {
-	cycle_t (*read)(const struct cyclecounter *cc);
-	cycle_t mask;
-	u32 mult;
-	u32 shift;
-};
-
-/**
- * struct timecounter - layer above a %struct cyclecounter which counts nanoseconds
- *	Contains the state needed by timecounter_read() to detect
- *	cycle counter wrap around. Initialize with
- *	timecounter_init(). Also used to convert cycle counts into the
- *	corresponding nanosecond counts with timecounter_cyc2time(). Users
- *	of this code are responsible for initializing the underlying
- *	cycle counter hardware, locking issues and reading the time
- *	more often than the cycle counter wraps around. The nanosecond
- *	counter will only wrap around after ~585 years.
- *
- * @cc:			the cycle counter used by this instance
- * @cycle_last:		most recent cycle counter value seen by
- *			timecounter_read()
- * @nsec:		continuously increasing count
- */
-struct timecounter {
-	const struct cyclecounter *cc;
-	cycle_t cycle_last;
-	u64 nsec;
-};
-
-/**
- * cyclecounter_cyc2ns - converts cycle counter cycles to nanoseconds
- * @cc:		Pointer to cycle counter.
- * @cycles:	Cycles
- *
- * XXX - This could use some mult_lxl_ll() asm optimization. Same code
- * as in cyc2ns, but with unsigned result.
- */
-static inline u64 cyclecounter_cyc2ns(const struct cyclecounter *cc,
-				      cycle_t cycles)
-{
-	u64 ret = (u64)cycles;
-	ret = (ret * cc->mult) >> cc->shift;
-	return ret;
-}
-
-/**
- * timecounter_init - initialize a time counter
- * @tc:			Pointer to time counter which is to be initialized/reset
- * @cc:			A cycle counter, ready to be used.
- * @start_tstamp:	Arbitrary initial time stamp.
- *
- * After this call the current cycle register (roughly) corresponds to
- * the initial time stamp. Every call to timecounter_read() increments
- * the time stamp counter by the number of elapsed nanoseconds.
- */
-extern void timecounter_init(struct timecounter *tc,
-			     const struct cyclecounter *cc,
-			     u64 start_tstamp);
-
-/**
- * timecounter_read - return nanoseconds elapsed since timecounter_init()
- *                    plus the initial time stamp
- * @tc:          Pointer to time counter.
- *
- * In other words, keeps track of time since the same epoch as
- * the function which generated the initial time stamp.
- */
-extern u64 timecounter_read(struct timecounter *tc);
-
-/**
- * timecounter_cyc2time - convert a cycle counter to same
- *                        time base as values returned by
- *                        timecounter_read()
- * @tc:		Pointer to time counter.
- * @cycle_tstamp:	a value returned by tc->cc->read()
- *
- * Cycle counts that are converted correctly as long as they
- * fall into the interval [-1/2 max cycle count, +1/2 max cycle count],
- * with "max cycle count" == cs->mask+1.
- *
- * This allows conversion of cycle counter values which were generated
- * in the past.
- */
-extern u64 timecounter_cyc2time(struct timecounter *tc,
-				cycle_t cycle_tstamp);
-
-/**
  * struct clocksource - hardware abstraction for a free running counter
  *	Provides mostly state-free accessors to the underlying hardware.
  *	This is the structure used for system time.
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 25c791e295fd..f1e41b33462f 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -42,7 +42,7 @@
 
 #include <linux/atomic.h>
 
-#include <linux/clocksource.h>
+#include <linux/timecounter.h>
 
 #define MAX_MSIX_P_PORT		17
 #define MAX_MSIX		64
diff --git a/include/linux/timecounter.h b/include/linux/timecounter.h
new file mode 100644
index 000000000000..74f45496e6d1
--- /dev/null
+++ b/include/linux/timecounter.h
@@ -0,0 +1,136 @@
+/*
+ * linux/include/linux/timecounter.h
+ *
+ * based on code that migrated away from
+ * linux/include/linux/clocksource.h
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+#ifndef _LINUX_TIMECOUNTER_H
+#define _LINUX_TIMECOUNTER_H
+
+#include <linux/types.h>
+
+/**
+ * struct cyclecounter - hardware abstraction for a free running counter
+ *	Provides completely state-free accessors to the underlying hardware.
+ *	Depending on which hardware it reads, the cycle counter may wrap
+ *	around quickly. Locking rules (if necessary) have to be defined
+ *	by the implementor and user of specific instances of this API.
+ *
+ * @read:		returns the current cycle value
+ * @mask:		bitmask for two's complement
+ *			subtraction of non 64 bit counters,
+ *			see CLOCKSOURCE_MASK() helper macro
+ * @mult:		cycle to nanosecond multiplier
+ * @shift:		cycle to nanosecond divisor (power of two)
+ */
+struct cyclecounter {
+	cycle_t (*read)(const struct cyclecounter *cc);
+	cycle_t mask;
+	u32 mult;
+	u32 shift;
+};
+
+/**
+ * struct timecounter - layer above a %struct cyclecounter which counts nanoseconds
+ *	Contains the state needed by timecounter_read() to detect
+ *	cycle counter wrap around. Initialize with
+ *	timecounter_init(). Also used to convert cycle counts into the
+ *	corresponding nanosecond counts with timecounter_cyc2time(). Users
+ *	of this code are responsible for initializing the underlying
+ *	cycle counter hardware, locking issues and reading the time
+ *	more often than the cycle counter wraps around. The nanosecond
+ *	counter will only wrap around after ~585 years.
+ *
+ * @cc:			the cycle counter used by this instance
+ * @cycle_last:		most recent cycle counter value seen by
+ *			timecounter_read()
+ * @nsec:		continuously increasing count
+ * @mask:		bit mask for maintaining the 'frac' field
+ * @frac:		accumulated fractional nanoseconds
+ */
+struct timecounter {
+	const struct cyclecounter *cc;
+	cycle_t cycle_last;
+	u64 nsec;
+	u64 mask;
+	u64 frac;
+};
+
+/**
+ * cyclecounter_cyc2ns - converts cycle counter cycles to nanoseconds
+ * @cc:		Pointer to cycle counter.
+ * @cycles:	Cycles
+ * @mask:	bit mask for maintaining the 'frac' field
+ * @frac:	pointer to storage for the fractional nanoseconds.
+ */
+static inline u64 cyclecounter_cyc2ns(const struct cyclecounter *cc,
+				      cycle_t cycles, u64 mask, u64 *frac)
+{
+	u64 ns = (u64) cycles;
+
+	ns = (ns * cc->mult) + *frac;
+	*frac = ns & mask;
+	return ns >> cc->shift;
+}
+
+/**
+ * timecounter_adjtime - Shifts the time of the clock.
+ * @delta:	Desired change in nanoseconds.
+ */
+static inline void timecounter_adjtime(struct timecounter *tc, s64 delta)
+{
+	tc->nsec += delta;
+}
+
+/**
+ * timecounter_init - initialize a time counter
+ * @tc:			Pointer to time counter which is to be initialized/reset
+ * @cc:			A cycle counter, ready to be used.
+ * @start_tstamp:	Arbitrary initial time stamp.
+ *
+ * After this call the current cycle register (roughly) corresponds to
+ * the initial time stamp. Every call to timecounter_read() increments
+ * the time stamp counter by the number of elapsed nanoseconds.
+ */
+extern void timecounter_init(struct timecounter *tc,
+			     const struct cyclecounter *cc,
+			     u64 start_tstamp);
+
+/**
+ * timecounter_read - return nanoseconds elapsed since timecounter_init()
+ *                    plus the initial time stamp
+ * @tc:          Pointer to time counter.
+ *
+ * In other words, keeps track of time since the same epoch as
+ * the function which generated the initial time stamp.
+ */
+extern u64 timecounter_read(struct timecounter *tc);
+
+/**
+ * timecounter_cyc2time - convert a cycle counter to same
+ *                        time base as values returned by
+ *                        timecounter_read()
+ * @tc:		Pointer to time counter.
+ * @cycle_tstamp:	a value returned by tc->cc->read()
+ *
+ * Cycle counts that are converted correctly as long as they
+ * fall into the interval [-1/2 max cycle count, +1/2 max cycle count],
+ * with "max cycle count" == cs->mask+1.
+ *
+ * This allows conversion of cycle counter values which were generated
+ * in the past.
+ */
+extern u64 timecounter_cyc2time(struct timecounter *tc,
+				cycle_t cycle_tstamp);
+
+#endif
diff --git a/include/linux/types.h b/include/linux/types.h
index a0bb7048687f..62323825cff9 100644
--- a/include/linux/types.h
+++ b/include/linux/types.h
@@ -213,5 +213,8 @@ struct callback_head {
 };
 #define rcu_head callback_head
 
+/* clocksource cycle base type */
+typedef u64 cycle_t;
+
 #endif /*  __ASSEMBLY__ */
 #endif /* _LINUX_TYPES_H */