[PATCH v6 0/8] ptp: IEEE 1588 hardware clock support

From: Richard Cochran
Date: Thu Sep 23 2010 - 13:31:03 EST

Here is the sixth version of my patch set adding PTP hardware clock
support to the Linux kernel. The main difference to v5 is that the
character device interface has been replaced with one based on the
posix clock system calls.

The first three patches add necessary background support in the posix
clock code. The last five add the new PTP hardware clock features.
Previously, I had tried to present the posix clock changes all by
themselves, but commentators asked to see the whole context.

What follows is a rather lengthy discussion of the various design
issues.

Table of Contents
=================
1 Introduction
2 Previous Discussions
3 Design Issues
3.1 Clock Operations
3.2 Character Device vs System Calls
3.2.1 Using the POSIX Clock API
3.2.2 Tuning a POSIX Clock
3.2.3 Dynamic POSIX Clock IDs
3.3 Synchronizing the Linux System Time
3.4 Ancillary PHC Operations
3.5 User timers
4 Drivers
4.1 Supported Hardware Clocks
4.2 Open Driver Issues
4.2.1 DP83640
4.2.2 IXP465


1 Introduction
~~~~~~~~~~~~~~~

The aim of this patch set is to add support for PTP hardware clocks
into the Linux kernel. In the following description, we use the
abbreviation "PHC" to mean "PTP hardware clock."

Support for obtaining timestamps from a PHC already exists via the
SO_TIMESTAMPING socket option, integrated in kernel version 2.6.30.
This patch set completes the picture by allow user space programs to
adjust the PHC and to control its ancillary features.

2 Previous Discussions
~~~~~~~~~~~~~~~~~~~~~~~

This patch set previously appeared on the netdev list. Since V5 of
the character device patch set, the discussion has moved to the
lkml.

- PTP hardware clock as a character device V5
[http://lkml.org/lkml/2010/8/16/90]

- POSIX clock tuning syscall with static clock ids
[http://lkml.org/lkml/2010/8/23/49]

- POSIX clock tuning syscall with dynamic clock ids
[http://lkml.org/lkml/2010/9/3/119]

3 Design Issues
~~~~~~~~~~~~~~~~

3.1 Clock Operations
=====================

Based on experience with several commercially available PHCs, we
identified a set of essential operations and a set of ancillary
operations.

- Basic clock operations

1. Set time
2. Get time
3. Shift the clock by a given offset atomically
4. Adjust clock frequency

- Ancillary clock features

1. Time stamp external events
2. Enable Linux PPS subsystem events
3. Periodic output signals
4. One shot or periodic alarms, with CPU interrupt

The patch set includes examples of the first two ancillary
features, and implementing the third point for a particular PHC is
fairly straightforward. The fourth point is discussed below.

3.2 Character Device vs System Calls
=====================================

This patch set started out as a class driver that exposes the PHC
as a character device with standardized ioctls. Since several clock
operations in the ioctl interface mimic the POSIX clock API, the
suggestion was made to expose the PHC as a new clockid_t.

POSIX defines the CLOCK_REALTIME, CLOCK_MONOTONIC,
CLOCK_PROCESS_CPUTIME_ID, and CLOCK_THREAD_CPUTIME_ID clock ids.
As to other possible clock ids, the standard offers the following
hint:

An implementation may also support additional clocks. The
interpretation of time values for these clocks is unspecified.

So as far as the POSIX standard is concerned, offering a clock id
to represent the PHC would be acceptable.

From discussions on the lkml, a repeated wish was to ensure that
any changes in the POSIX clock code would be general enough to
support other new hardware clocks that might appear in the future,
not just the particulars of PHCs.

3.2.1 Using the POSIX Clock API
--------------------------------

Looking at the mapping from PHC operation to the POSIX clock API,
we see that two of the basic clock operations, marked with *, have
no POSIX equivalent. The items marked NA are peculiar to PHCs and
will be discussed separately, below.

Clock Operation POSIX function
-----------------------------+-----------------------------
Set time clock_gettime
Get time clock_settime
Shift the clock *
Adjust clock frequency *
-----------------------------+-----------------------------
Time stamp external events NA
Enable PPS events NA
Periodic output signals NA
One shot or periodic alarms timer_create, timer_settime

In contrast to the standard Linux system clock, a PHC is
adjustable in hardware, for example using frequency compensation
registers or a VCO. The ability to directly tune the PHC is
essential to reap the benefit of hardware timestamping.

3.2.2 Tuning a POSIX Clock
---------------------------

The patch set introduces a new system call which allows tuning of
a POSIX clock. The function combines the 'struct timex' from the
NTP adjtimex syscall with a POSIX clock id.

clock_adjtime(clockid_t, struct timex *);

Although the timex interface is a bit baroque (John Stultz said:
"a little crufty"), using it as the basis of the new call allows
supporting the tried and true NTP semantics. By adding one
additional mode flag to the struct timex, the requirements for
PHCs are also satisfied. In the future, if new clocks appear that
require more elaborate control, then the padding at the end of the
struct reserves 44 bytes for new fields.

3.2.3 Dynamic POSIX Clock IDs
------------------------------

The reaction on the list to having a static id like CLOCK_PTP was
mostly negative. However, the idea of generating a clock id
dynamically seems to have gained acceptance. The general idea is
to advertise the available clock ids to user space via sysfs. This
patch set implements two different ways:

/sys/class/timesource/<name>/id
/sys/class/ptp/ptp_clock_X/id

Note: I am not too sure that this is exactly what people imagined,
but it is my best understanding so far. I gleaned two
different ideas about where to offer the clock id. In order
to keep just one way, I will be happy to remove the less
popular one.

3.3 Synchronizing the Linux System Time
========================================

One could offer a PHC as a combined clock source and clock event
device. The advantage of this approach would be that it obviates
the need for synchronization when the PHC is selected as the system
timer. However, some PHCs, namely the PHY based clocks, cannot be
used in this way.

Instead, the patch set provides a way to offer a Pulse Per Second
(PPS) event from the PHC to the Linux PPS subsystem. A user space
application can read the PPS events and tune the system clock, just
like when using other external time sources like radio clocks or
GPS.

3.4 Ancillary PHC Operations
=============================

Most PHCs offer hardware interfaces to the outside world, that is,
the "real world". It is important to offer support for these
operations, since leaving them out would defeat the utility of
having a PHC in the first place. These operations do not map at all
to the POSIX clock functions, but one could offer them as a
character device or via sysfs.

Fearing a lkml debate on the merits of either one, I went ahead and
implemented both cases. Both approaches adequately cover the needed
functionality, in my opinion. The code for each is in its own .c
file, so it will be easy enough to remove one of them. I don't mind
leaving both ways in, either.

3.5 User timers
================

Using the POSIX clock API gived user space the possibility to
create and use timers with timer_create and timer_settime. In the
current patch set the kernel functionality is not implemented,
since there are some issues to consider first. I see two ways to do
about this.

1. Implement the functionality anew. This approach might end up
duplicating similar code that already exists. Also, looking at
the hrtimer code, getting user timers right seems to have a
number of gotchas and thorny issues.

2. Reuse the hrtimer code. Since the hrtimer code uses a clock
event device under the hood, it might be possible (in theory) to
offer capable PHCs as clock event devices. However, the current
hrtimers are hard-coded to the event device via a per-cpu
global. Perhaps one could associate an event device with a
hrtimer via the timer itself.

At this point I am not optimistic about either approach, and I
would vote for postponing the timer issue indefinitely. The
implementation effort would be high, but the utility low.

If the Linux system time is synchronized to the PHC via the PPS
method, then using standard hrtimers would be good enough for most
purposes. Consider the time scales involved. The PHC can be
synchronized to within 100 nanoseconds of an external time source,
while timer wakeup latency (even with rt kernels) is tens of
microseconds.

4 Drivers
~~~~~~~~~~

4.1 Supported Hardware Clocks
==============================

+ Standard Linux system timer
This driver exports the standard Linux timer as a PTP clock.
Although this duplicates CLOCK_REALTIME, the code serves as a
simple example for driver development and lets people who without
special hardware try the new API.

+ Freescale eTSEC gianfar
- 2 Time stamp external triggers, programmable polarity (opt. interrupt)
- 2 Alarm registers (optional interrupt)
- 3 Periodic signals (optional interrupt)

+ National Semiconductor DP83640
- 6 GPIOs programmable as inputs or outputs
- 6 GPIOs with dedicated functions (LED/JTAG/clock) can also be
used as general inputs or outputs
- GPIO inputs can time stamp external triggers
- GPIO outputs can produce periodic signals
- 1 interrupt pin

+ Intel IXP465
- Auxiliary Slave/Master Mode Snapshot (optional interrupt)
- Target Time (optional interrupt)

4.2 Open Driver Issues
=======================

4.2.1 DP83640
--------------
In order to make this work, one line must be added into the MAC
driver. If you have the DP83640 and want to try the driver, you
need to add this one line to your MAC driver: In the
.ndo_start_xmit function, add skb_tx_timestamp(skb).

4.2.2 IXP465
-------------
I do not know how to correctly choose the timestamp "channel"
based on the port identifier:

+#define PORT2CHANNEL(p) 1
+/*
+ * PHYSICAL_ID(p->id) ?
+ * TODO - Figure out correct mapping.
+ */

Krzysztof, can you help?


Richard Cochran (8):
posix clocks: introduce a syscall for clock tuning.
posix clocks: dynamic clock ids.
posix clocks: introduce a sysfs presence.
ptp: Added a brand new class driver for ptp clocks.
ptp: Added a simulated PTP hardware clock.
ptp: Added a clock that uses the eTSEC found on the MPC85xx.
ptp: Added a clock driver for the IXP46x.
ptp: Added a clock driver for the National Semiconductor PHYTER.

Documentation/ABI/testing/sysfs-ptp | 107 +++
Documentation/ABI/testing/sysfs-timesource | 24 +
Documentation/powerpc/dts-bindings/fsl/tsec.txt | 57 ++
Documentation/ptp/ptp.txt | 94 +++
Documentation/ptp/testptp.c | 358 +++++++++
Documentation/ptp/testptp.mk | 33 +
arch/arm/include/asm/unistd.h | 1 +
arch/arm/kernel/calls.S | 1 +
arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h | 78 ++
arch/blackfin/include/asm/unistd.h | 3 +-
arch/blackfin/mach-common/entry.S | 1 +
arch/powerpc/boot/dts/mpc8313erdb.dts | 14 +
arch/powerpc/boot/dts/mpc8572ds.dts | 14 +
arch/powerpc/boot/dts/p2020ds.dts | 14 +
arch/powerpc/boot/dts/p2020rdb.dts | 14 +
arch/powerpc/include/asm/systbl.h | 1 +
arch/powerpc/include/asm/unistd.h | 3 +-
arch/x86/ia32/ia32entry.S | 1 +
arch/x86/include/asm/unistd_32.h | 3 +-
arch/x86/include/asm/unistd_64.h | 2 +
arch/x86/kernel/syscall_table_32.S | 1 +
drivers/Kconfig | 2 +
drivers/Makefile | 1 +
drivers/char/mmtimer.c | 1 +
drivers/net/Makefile | 1 +
drivers/net/arm/ixp4xx_eth.c | 191 +++++
drivers/net/gianfar_ptp.c | 447 ++++++++++++
drivers/net/gianfar_ptp_reg.h | 113 +++
drivers/net/phy/Kconfig | 29 +
drivers/net/phy/Makefile | 1 +
drivers/net/phy/dp83640.c | 887 +++++++++++++++++++++++
drivers/net/phy/dp83640_reg.h | 261 +++++++
drivers/ptp/Kconfig | 67 ++
drivers/ptp/Makefile | 8 +
drivers/ptp/ptp_chardev.c | 178 +++++
drivers/ptp/ptp_clock.c | 382 ++++++++++
drivers/ptp/ptp_ixp46x.c | 345 +++++++++
drivers/ptp/ptp_linux.c | 165 +++++
drivers/ptp/ptp_private.h | 64 ++
drivers/ptp/ptp_sysfs.c | 235 ++++++
include/linux/Kbuild | 1 +
include/linux/posix-timers.h | 14 +-
include/linux/ptp_clock.h | 79 ++
include/linux/ptp_clock_kernel.h | 139 ++++
include/linux/syscalls.h | 2 +
include/linux/time.h | 2 +
include/linux/timex.h | 3 +-
kernel/compat.c | 136 +++--
kernel/posix-cpu-timers.c | 6 +
kernel/posix-timers.c | 98 +++-
kernel/time/ntp.c | 2 +
51 files changed, 4624 insertions(+), 60 deletions(-)
create mode 100644 Documentation/ABI/testing/sysfs-ptp
create mode 100644 Documentation/ABI/testing/sysfs-timesource
create mode 100644 Documentation/ptp/ptp.txt
create mode 100644 Documentation/ptp/testptp.c
create mode 100644 Documentation/ptp/testptp.mk
create mode 100644 arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h
create mode 100644 drivers/net/gianfar_ptp.c
create mode 100644 drivers/net/gianfar_ptp_reg.h
create mode 100644 drivers/net/phy/dp83640.c
create mode 100644 drivers/net/phy/dp83640_reg.h
create mode 100644 drivers/ptp/Kconfig
create mode 100644 drivers/ptp/Makefile
create mode 100644 drivers/ptp/ptp_chardev.c
create mode 100644 drivers/ptp/ptp_clock.c
create mode 100644 drivers/ptp/ptp_ixp46x.c
create mode 100644 drivers/ptp/ptp_linux.c
create mode 100644 drivers/ptp/ptp_private.h
create mode 100644 drivers/ptp/ptp_sysfs.c
create mode 100644 include/linux/ptp_clock.h
create mode 100644 include/linux/ptp_clock_kernel.h

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/