Here comes PTP Hardware Clock (PHC) support, version 7, and hopefully
everyone will like it.
The first four patches provide infrastructure supporting dynamic POSIX
clock devices. This new code will be useful for other kinds of new
clocks, not just PHCs. The last four patches implement the PHC code.
Table of Contents
=================
1 Introduction
2 Patch Set Overview
3 Previous Discussions
4 Design Issues
4.1 Clock Operations
4.2 System Calls for Clock Tuning
4.2.1 Using the POSIX Clock API
4.2.2 Tuning a POSIX Clock
4.2.3 Dynamic POSIX Clock IDs
4.3 Synchronizing the Linux System Time
4.4 Ancillary PHC Operations
4.5 User timers
5 Drivers
5.1 Supported Hardware Clocks
5.2 Open Driver Issues
5.2.1 DP83640
5.2.2 IXP465
6 Diff Stat
1 Introduction
~~~~~~~~~~~~~~~
The aim of this patch set is to add support for PTP Hardware Clocks
(PHCs) into the Linux kernel. 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 Patch Set Overview
~~~~~~~~~~~~~~~~~~~~~
- Patch 1 adds an ADJ_SETOFFSET mode bit to the NTP timex
structure. The new bit allows a correction of a time offset.
- Patch 2 adds a new system call, clock_adjtime(), which is like the
NTP adjtimex call for POSIX clocks.
- Patches 3 and 4 add dynamic POSIX clocks.
- The remaining patches add the core PHC code, complete with three
drivers.
3 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.
- IEEE 1588 hardware clock support [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]
- IEEE 1588 hardware clock support [V6]
[http://lkml.org/lkml/2010/9/23/310]
- Dynamic clock devices [RFC]
[http://lkml.org/lkml/2010/11/4/290]
4 Design Issues
~~~~~~~~~~~~~~~~
The following treatment of the design issues previously appeared
with the V6 version of the patch set. The text has been updated to
reflect the current patch set.
4.1 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.
4.2 System Calls for Clock Tuning
==================================
4.2.1 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.
4.2.2 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 *);
Using the timex interface 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.
4.2.3 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 represent the clocks as character devices. After opening the
character device, the file descriptor may be also used as a clock
id by applying a simple transformation.
4.3 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.
4.4 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.
4.5 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.
5 Drivers
~~~~~~~~~~
5.1 Supported Hardware Clocks
==============================
+ 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)
5.2 Open Driver Issues
=======================
5.2.1 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).
5.2.2 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?
6 Diff Stat
~~~~~~~~~~~
Richard Cochran (8):
ntp: add ADJ_SETOFFSET mode bit
posix clocks: introduce a syscall for clock tuning.
posix clocks: introduce dynamic clocks
posix clocks: hook dynamic clocks into system calls
ptp: Added a brand new class driver for ptp clocks.
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/powerpc/dts-bindings/fsl/tsec.txt | 57 ++
Documentation/ptp/ptp.txt | 94 +++
Documentation/ptp/testptp.c | 352 +++++++++
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 | 444 +++++++++++
drivers/net/gianfar_ptp_reg.h | 113 +++
drivers/net/phy/Kconfig | 29 +
drivers/net/phy/Makefile | 1 +
drivers/net/phy/dp83640.c | 890 +++++++++++++++++++++++
drivers/net/phy/dp83640_reg.h | 261 +++++++
drivers/ptp/Kconfig | 53 ++
drivers/ptp/Makefile | 7 +
drivers/ptp/ptp_chardev.c | 144 ++++
drivers/ptp/ptp_clock.c | 317 ++++++++
drivers/ptp/ptp_ixp46x.c | 342 +++++++++
drivers/ptp/ptp_private.h | 68 ++
drivers/ptp/ptp_sysfs.c | 230 ++++++
include/linux/Kbuild | 1 +
include/linux/posix-clock.h | 136 ++++
include/linux/posix-timers.h | 32 +-
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 | 151 ++++-
kernel/time/Makefile | 3 +-
kernel/time/ntp.c | 26 +
kernel/time/posix-clock.c | 376 ++++++++++
52 files changed, 4911 insertions(+), 70 deletions(-)
create mode 100644 Documentation/ABI/testing/sysfs-ptp
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_private.h
create mode 100644 drivers/ptp/ptp_sysfs.c
create mode 100644 include/linux/posix-clock.h
create mode 100644 include/linux/ptp_clock.h
create mode 100644 include/linux/ptp_clock_kernel.h
create mode 100644 kernel/time/posix-clock.c
This patch adds a new mode bit into the timex structure. When set, the bit
instructs the kernel to add the given time value to the current time.
Signed-off-by: Richard Cochran <[email protected]>
---
include/linux/timex.h | 3 ++-
kernel/time/ntp.c | 26 ++++++++++++++++++++++++++
2 files changed, 28 insertions(+), 1 deletions(-)
diff --git a/include/linux/timex.h b/include/linux/timex.h
index 32d852f..82d4b24 100644
--- a/include/linux/timex.h
+++ b/include/linux/timex.h
@@ -73,7 +73,7 @@ struct timex {
long tolerance; /* clock frequency tolerance (ppm)
* (read only)
*/
- struct timeval time; /* (read only) */
+ struct timeval time; /* (read only, except for ADJ_SETOFFSET) */
long tick; /* (modified) usecs between clock ticks */
long ppsfreq; /* pps frequency (scaled ppm) (ro) */
@@ -101,6 +101,7 @@ struct timex {
#define ADJ_ESTERROR 0x0008 /* estimated time error */
#define ADJ_STATUS 0x0010 /* clock status */
#define ADJ_TIMECONST 0x0020 /* pll time constant */
+#define ADJ_SETOFFSET 0x0040 /* add 'time' to current time */
#define ADJ_TAI 0x0080 /* set TAI offset */
#define ADJ_MICRO 0x1000 /* select microsecond resolution */
#define ADJ_NANO 0x2000 /* select nanosecond resolution */
diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c
index d232189..e9e3915 100644
--- a/kernel/time/ntp.c
+++ b/kernel/time/ntp.c
@@ -454,6 +454,7 @@ static inline void process_adjtimex_modes(struct timex *txc, struct timespec *ts
int do_adjtimex(struct timex *txc)
{
struct timespec ts;
+ ktime_t delta, kt;
int result;
/* Validate the data before disabling interrupts */
@@ -482,8 +483,33 @@ int do_adjtimex(struct timex *txc)
hrtimer_cancel(&leap_timer);
}
+ if (txc->modes & ADJ_SETOFFSET) {
+ /* Validate the delta value. */
+ if (txc->time.tv_sec && txc->time.tv_usec < 0)
+ return -EINVAL;
+
+ if (txc->modes & ADJ_NANO) {
+ struct timespec tmp;
+ tmp.tv_sec = txc->time.tv_sec;
+ tmp.tv_nsec = txc->time.tv_usec;
+ delta = timespec_to_ktime(tmp);
+ } else
+ delta = timeval_to_ktime(txc->time);
+
+ /* Adding the delta should be an "atomic" operation. */
+ local_irq_disable();
+ }
+
getnstimeofday(&ts);
+ if (txc->modes & ADJ_SETOFFSET) {
+ kt = timespec_to_ktime(ts);
+ kt = ktime_add(kt, delta);
+ ts = ktime_to_timespec(kt);
+ do_settimeofday(&ts);
+ local_irq_enable();
+ }
+
write_seqlock_irq(&xtime_lock);
if (txc->modes & ADJ_ADJTIME) {
--
1.7.0.4
A new syscall is introduced that allows tuning of a POSIX clock. The
syscall is implemented for four architectures: arm, blackfin, powerpc,
and x86.
The new syscall, clock_adjtime, takes two parameters, the clock ID,
and a pointer to a struct timex. Any ADJTIMEX(2) operation may be
requested via this system call, but various POSIX clocks may or may
not support tuning.
Signed-off-by: Richard Cochran <[email protected]>
---
arch/arm/include/asm/unistd.h | 1 +
arch/arm/kernel/calls.S | 1 +
arch/blackfin/include/asm/unistd.h | 3 +-
arch/blackfin/mach-common/entry.S | 1 +
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/char/mmtimer.c | 1 +
include/linux/posix-timers.h | 4 +
include/linux/syscalls.h | 2 +
kernel/compat.c | 136 +++++++++++++++++++++++-------------
kernel/posix-cpu-timers.c | 6 ++
kernel/posix-timers.c | 35 +++++++++
16 files changed, 150 insertions(+), 51 deletions(-)
diff --git a/arch/arm/include/asm/unistd.h b/arch/arm/include/asm/unistd.h
index c891eb7..f58d881 100644
--- a/arch/arm/include/asm/unistd.h
+++ b/arch/arm/include/asm/unistd.h
@@ -396,6 +396,7 @@
#define __NR_fanotify_init (__NR_SYSCALL_BASE+367)
#define __NR_fanotify_mark (__NR_SYSCALL_BASE+368)
#define __NR_prlimit64 (__NR_SYSCALL_BASE+369)
+#define __NR_clock_adjtime (__NR_SYSCALL_BASE+370)
/*
* The following SWIs are ARM private.
diff --git a/arch/arm/kernel/calls.S b/arch/arm/kernel/calls.S
index 5c26ecc..430de4c 100644
--- a/arch/arm/kernel/calls.S
+++ b/arch/arm/kernel/calls.S
@@ -379,6 +379,7 @@
CALL(sys_fanotify_init)
CALL(sys_fanotify_mark)
CALL(sys_prlimit64)
+/* 370 */ CALL(sys_clock_adjtime)
#ifndef syscalls_counted
.equ syscalls_padding, ((NR_syscalls + 3) & ~3) - NR_syscalls
#define syscalls_counted
diff --git a/arch/blackfin/include/asm/unistd.h b/arch/blackfin/include/asm/unistd.h
index 928ae97..e640c51 100644
--- a/arch/blackfin/include/asm/unistd.h
+++ b/arch/blackfin/include/asm/unistd.h
@@ -393,8 +393,9 @@
#define __NR_fanotify_mark 372
#define __NR_prlimit64 373
#define __NR_cacheflush 374
+#define __NR_clock_adjtime 375
-#define __NR_syscall 375
+#define __NR_syscall 376
#define NR_syscalls __NR_syscall
/* Old optional stuff no one actually uses */
diff --git a/arch/blackfin/mach-common/entry.S b/arch/blackfin/mach-common/entry.S
index 2ca915e..cb6cc1e 100644
--- a/arch/blackfin/mach-common/entry.S
+++ b/arch/blackfin/mach-common/entry.S
@@ -1738,6 +1738,7 @@ ENTRY(_sys_call_table)
.long _sys_fanotify_mark
.long _sys_prlimit64
.long _sys_cacheflush
+ .long _sys_clock_adjtime /* 375 */
.rept NR_syscalls-(.-_sys_call_table)/4
.long _sys_ni_syscall
diff --git a/arch/powerpc/include/asm/systbl.h b/arch/powerpc/include/asm/systbl.h
index aa0f1eb..6a1152c 100644
--- a/arch/powerpc/include/asm/systbl.h
+++ b/arch/powerpc/include/asm/systbl.h
@@ -348,3 +348,4 @@ COMPAT_SYS_SPU(sendmsg)
COMPAT_SYS_SPU(recvmsg)
COMPAT_SYS_SPU(recvmmsg)
SYSCALL_SPU(accept4)
+COMPAT_SYS_SPU(clock_adjtime)
diff --git a/arch/powerpc/include/asm/unistd.h b/arch/powerpc/include/asm/unistd.h
index 6151937..386de07 100644
--- a/arch/powerpc/include/asm/unistd.h
+++ b/arch/powerpc/include/asm/unistd.h
@@ -367,10 +367,11 @@
#define __NR_recvmsg 342
#define __NR_recvmmsg 343
#define __NR_accept4 344
+#define __NR_clock_adjtime 345
#ifdef __KERNEL__
-#define __NR_syscalls 345
+#define __NR_syscalls 346
#define __NR__exit __NR_exit
#define NR_syscalls __NR_syscalls
diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index 518bb99..0ed7896 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -851,4 +851,5 @@ ia32_sys_call_table:
.quad sys_fanotify_init
.quad sys32_fanotify_mark
.quad sys_prlimit64 /* 340 */
+ .quad compat_sys_clock_adjtime
ia32_syscall_end:
diff --git a/arch/x86/include/asm/unistd_32.h b/arch/x86/include/asm/unistd_32.h
index b766a5e..b6f73f1 100644
--- a/arch/x86/include/asm/unistd_32.h
+++ b/arch/x86/include/asm/unistd_32.h
@@ -346,10 +346,11 @@
#define __NR_fanotify_init 338
#define __NR_fanotify_mark 339
#define __NR_prlimit64 340
+#define __NR_clock_adjtime 341
#ifdef __KERNEL__
-#define NR_syscalls 341
+#define NR_syscalls 342
#define __ARCH_WANT_IPC_PARSE_VERSION
#define __ARCH_WANT_OLD_READDIR
diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h
index 363e9b8..5ee3085 100644
--- a/arch/x86/include/asm/unistd_64.h
+++ b/arch/x86/include/asm/unistd_64.h
@@ -669,6 +669,8 @@ __SYSCALL(__NR_fanotify_init, sys_fanotify_init)
__SYSCALL(__NR_fanotify_mark, sys_fanotify_mark)
#define __NR_prlimit64 302
__SYSCALL(__NR_prlimit64, sys_prlimit64)
+#define __NR_clock_adjtime 303
+__SYSCALL(__NR_clock_adjtime, sys_clock_adjtime)
#ifndef __NO_STUBS
#define __ARCH_WANT_OLD_READDIR
diff --git a/arch/x86/kernel/syscall_table_32.S b/arch/x86/kernel/syscall_table_32.S
index b35786d..68c7b9a 100644
--- a/arch/x86/kernel/syscall_table_32.S
+++ b/arch/x86/kernel/syscall_table_32.S
@@ -340,3 +340,4 @@ ENTRY(sys_call_table)
.long sys_fanotify_init
.long sys_fanotify_mark
.long sys_prlimit64 /* 340 */
+ .long sys_clock_adjtime
diff --git a/drivers/char/mmtimer.c b/drivers/char/mmtimer.c
index e6d7562..98f2488 100644
--- a/drivers/char/mmtimer.c
+++ b/drivers/char/mmtimer.c
@@ -767,6 +767,7 @@ static struct k_clock sgi_clock = {
.res = 0,
.clock_set = sgi_clock_set,
.clock_get = sgi_clock_get,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = sgi_timer_create,
.nsleep = do_posix_clock_nonanosleep,
.timer_set = sgi_timer_set,
diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index 3e23844..b05d9b8 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -4,6 +4,7 @@
#include <linux/spinlock.h>
#include <linux/list.h>
#include <linux/sched.h>
+#include <linux/timex.h>
union cpu_time_count {
cputime_t cpu;
@@ -71,6 +72,7 @@ struct k_clock {
int (*clock_getres) (const clockid_t which_clock, struct timespec *tp);
int (*clock_set) (const clockid_t which_clock, struct timespec * tp);
int (*clock_get) (const clockid_t which_clock, struct timespec * tp);
+ int (*clock_adj) (const clockid_t which_clock, struct timex *tx);
int (*timer_create) (struct k_itimer *timer);
int (*nsleep) (const clockid_t which_clock, int flags,
struct timespec *, struct timespec __user *);
@@ -90,6 +92,7 @@ void register_posix_clock(const clockid_t clock_id, struct k_clock *new_clock);
int do_posix_clock_nonanosleep(const clockid_t, int flags, struct timespec *,
struct timespec __user *);
int do_posix_clock_nosettime(const clockid_t, struct timespec *tp);
+int do_posix_clock_noadjtime(const clockid_t, struct timex *tx);
/* function to call to trigger timer event */
int posix_timer_event(struct k_itimer *timr, int si_private);
@@ -97,6 +100,7 @@ int posix_timer_event(struct k_itimer *timr, int si_private);
int posix_cpu_clock_getres(const clockid_t which_clock, struct timespec *ts);
int posix_cpu_clock_get(const clockid_t which_clock, struct timespec *ts);
int posix_cpu_clock_set(const clockid_t which_clock, const struct timespec *ts);
+int posix_cpu_clock_adj(const clockid_t which_clock, struct timex *tx);
int posix_cpu_timer_create(struct k_itimer *timer);
int posix_cpu_nsleep(const clockid_t which_clock, int flags,
struct timespec *rqtp, struct timespec __user *rmtp);
diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
index cacc27a..c76cefb 100644
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -313,6 +313,8 @@ asmlinkage long sys_clock_settime(clockid_t which_clock,
const struct timespec __user *tp);
asmlinkage long sys_clock_gettime(clockid_t which_clock,
struct timespec __user *tp);
+asmlinkage long sys_clock_adjtime(clockid_t which_clock,
+ struct timex __user *tx);
asmlinkage long sys_clock_getres(clockid_t which_clock,
struct timespec __user *tp);
asmlinkage long sys_clock_nanosleep(clockid_t which_clock, int flags,
diff --git a/kernel/compat.c b/kernel/compat.c
index c9e2ec0..38b1d2c 100644
--- a/kernel/compat.c
+++ b/kernel/compat.c
@@ -52,6 +52,64 @@ static int compat_put_timeval(struct compat_timeval __user *o,
put_user(i->tv_usec, &o->tv_usec)) ? -EFAULT : 0;
}
+static int compat_get_timex(struct timex *txc, struct compat_timex __user *utp)
+{
+ memset(txc, 0, sizeof(struct timex));
+
+ if (!access_ok(VERIFY_READ, utp, sizeof(struct compat_timex)) ||
+ __get_user(txc->modes, &utp->modes) ||
+ __get_user(txc->offset, &utp->offset) ||
+ __get_user(txc->freq, &utp->freq) ||
+ __get_user(txc->maxerror, &utp->maxerror) ||
+ __get_user(txc->esterror, &utp->esterror) ||
+ __get_user(txc->status, &utp->status) ||
+ __get_user(txc->constant, &utp->constant) ||
+ __get_user(txc->precision, &utp->precision) ||
+ __get_user(txc->tolerance, &utp->tolerance) ||
+ __get_user(txc->time.tv_sec, &utp->time.tv_sec) ||
+ __get_user(txc->time.tv_usec, &utp->time.tv_usec) ||
+ __get_user(txc->tick, &utp->tick) ||
+ __get_user(txc->ppsfreq, &utp->ppsfreq) ||
+ __get_user(txc->jitter, &utp->jitter) ||
+ __get_user(txc->shift, &utp->shift) ||
+ __get_user(txc->stabil, &utp->stabil) ||
+ __get_user(txc->jitcnt, &utp->jitcnt) ||
+ __get_user(txc->calcnt, &utp->calcnt) ||
+ __get_user(txc->errcnt, &utp->errcnt) ||
+ __get_user(txc->stbcnt, &utp->stbcnt))
+ return -EFAULT;
+
+ return 0;
+}
+
+static int compat_put_timex(struct compat_timex __user *utp, struct timex *txc)
+{
+ if (!access_ok(VERIFY_WRITE, utp, sizeof(struct compat_timex)) ||
+ __put_user(txc->modes, &utp->modes) ||
+ __put_user(txc->offset, &utp->offset) ||
+ __put_user(txc->freq, &utp->freq) ||
+ __put_user(txc->maxerror, &utp->maxerror) ||
+ __put_user(txc->esterror, &utp->esterror) ||
+ __put_user(txc->status, &utp->status) ||
+ __put_user(txc->constant, &utp->constant) ||
+ __put_user(txc->precision, &utp->precision) ||
+ __put_user(txc->tolerance, &utp->tolerance) ||
+ __put_user(txc->time.tv_sec, &utp->time.tv_sec) ||
+ __put_user(txc->time.tv_usec, &utp->time.tv_usec) ||
+ __put_user(txc->tick, &utp->tick) ||
+ __put_user(txc->ppsfreq, &utp->ppsfreq) ||
+ __put_user(txc->jitter, &utp->jitter) ||
+ __put_user(txc->shift, &utp->shift) ||
+ __put_user(txc->stabil, &utp->stabil) ||
+ __put_user(txc->jitcnt, &utp->jitcnt) ||
+ __put_user(txc->calcnt, &utp->calcnt) ||
+ __put_user(txc->errcnt, &utp->errcnt) ||
+ __put_user(txc->stbcnt, &utp->stbcnt) ||
+ __put_user(txc->tai, &utp->tai))
+ return -EFAULT;
+ return 0;
+}
+
asmlinkage long compat_sys_gettimeofday(struct compat_timeval __user *tv,
struct timezone __user *tz)
{
@@ -617,6 +675,29 @@ long compat_sys_clock_gettime(clockid_t which_clock,
return err;
}
+long compat_sys_clock_adjtime(clockid_t which_clock,
+ struct compat_timex __user *utp)
+{
+ struct timex txc;
+ mm_segment_t oldfs;
+ int err, ret;
+
+ err = compat_get_timex(&txc, utp);
+ if (err)
+ return err;
+
+ oldfs = get_fs();
+ set_fs(KERNEL_DS);
+ ret = sys_clock_adjtime(which_clock, (struct timex __user *) &txc);
+ set_fs(oldfs);
+
+ err = compat_put_timex(utp, &txc);
+ if (err)
+ return err;
+
+ return ret;
+}
+
long compat_sys_clock_getres(clockid_t which_clock,
struct compat_timespec __user *tp)
{
@@ -951,58 +1032,17 @@ asmlinkage long compat_sys_rt_sigsuspend(compat_sigset_t __user *unewset, compat
asmlinkage long compat_sys_adjtimex(struct compat_timex __user *utp)
{
struct timex txc;
- int ret;
-
- memset(&txc, 0, sizeof(struct timex));
+ int err, ret;
- if (!access_ok(VERIFY_READ, utp, sizeof(struct compat_timex)) ||
- __get_user(txc.modes, &utp->modes) ||
- __get_user(txc.offset, &utp->offset) ||
- __get_user(txc.freq, &utp->freq) ||
- __get_user(txc.maxerror, &utp->maxerror) ||
- __get_user(txc.esterror, &utp->esterror) ||
- __get_user(txc.status, &utp->status) ||
- __get_user(txc.constant, &utp->constant) ||
- __get_user(txc.precision, &utp->precision) ||
- __get_user(txc.tolerance, &utp->tolerance) ||
- __get_user(txc.time.tv_sec, &utp->time.tv_sec) ||
- __get_user(txc.time.tv_usec, &utp->time.tv_usec) ||
- __get_user(txc.tick, &utp->tick) ||
- __get_user(txc.ppsfreq, &utp->ppsfreq) ||
- __get_user(txc.jitter, &utp->jitter) ||
- __get_user(txc.shift, &utp->shift) ||
- __get_user(txc.stabil, &utp->stabil) ||
- __get_user(txc.jitcnt, &utp->jitcnt) ||
- __get_user(txc.calcnt, &utp->calcnt) ||
- __get_user(txc.errcnt, &utp->errcnt) ||
- __get_user(txc.stbcnt, &utp->stbcnt))
- return -EFAULT;
+ err = compat_get_timex(&txc, utp);
+ if (err)
+ return err;
ret = do_adjtimex(&txc);
- if (!access_ok(VERIFY_WRITE, utp, sizeof(struct compat_timex)) ||
- __put_user(txc.modes, &utp->modes) ||
- __put_user(txc.offset, &utp->offset) ||
- __put_user(txc.freq, &utp->freq) ||
- __put_user(txc.maxerror, &utp->maxerror) ||
- __put_user(txc.esterror, &utp->esterror) ||
- __put_user(txc.status, &utp->status) ||
- __put_user(txc.constant, &utp->constant) ||
- __put_user(txc.precision, &utp->precision) ||
- __put_user(txc.tolerance, &utp->tolerance) ||
- __put_user(txc.time.tv_sec, &utp->time.tv_sec) ||
- __put_user(txc.time.tv_usec, &utp->time.tv_usec) ||
- __put_user(txc.tick, &utp->tick) ||
- __put_user(txc.ppsfreq, &utp->ppsfreq) ||
- __put_user(txc.jitter, &utp->jitter) ||
- __put_user(txc.shift, &utp->shift) ||
- __put_user(txc.stabil, &utp->stabil) ||
- __put_user(txc.jitcnt, &utp->jitcnt) ||
- __put_user(txc.calcnt, &utp->calcnt) ||
- __put_user(txc.errcnt, &utp->errcnt) ||
- __put_user(txc.stbcnt, &utp->stbcnt) ||
- __put_user(txc.tai, &utp->tai))
- ret = -EFAULT;
+ err = compat_put_timex(utp, &txc);
+ if (err)
+ return err;
return ret;
}
diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c
index 05bb717..0206116 100644
--- a/kernel/posix-cpu-timers.c
+++ b/kernel/posix-cpu-timers.c
@@ -207,6 +207,10 @@ int posix_cpu_clock_set(const clockid_t which_clock, const struct timespec *tp)
return error;
}
+int posix_cpu_clock_adj(const clockid_t which_clock, struct timex *tx)
+{
+ return -EOPNOTSUPP;
+}
/*
* Sample a per-thread clock for the given task.
@@ -1610,6 +1614,7 @@ static __init int init_posix_cpu_timers(void)
.clock_getres = process_cpu_clock_getres,
.clock_get = process_cpu_clock_get,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = process_cpu_timer_create,
.nsleep = process_cpu_nsleep,
.nsleep_restart = process_cpu_nsleep_restart,
@@ -1618,6 +1623,7 @@ static __init int init_posix_cpu_timers(void)
.clock_getres = thread_cpu_clock_getres,
.clock_get = thread_cpu_clock_get,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = thread_cpu_timer_create,
.nsleep = thread_cpu_nsleep,
.nsleep_restart = thread_cpu_nsleep_restart,
diff --git a/kernel/posix-timers.c b/kernel/posix-timers.c
index 9ca4973..502bde4 100644
--- a/kernel/posix-timers.c
+++ b/kernel/posix-timers.c
@@ -197,6 +197,11 @@ static int common_timer_create(struct k_itimer *new_timer)
return 0;
}
+static inline int common_clock_adj(const clockid_t which_clock, struct timex *t)
+{
+ return do_adjtimex(t);
+}
+
static int no_timer_create(struct k_itimer *new_timer)
{
return -EOPNOTSUPP;
@@ -273,11 +278,13 @@ static __init int init_posix_timers(void)
.clock_getres = hrtimer_get_res,
.clock_get = posix_ktime_get_ts,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
};
struct k_clock clock_monotonic_raw = {
.clock_getres = hrtimer_get_res,
.clock_get = posix_get_monotonic_raw,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = no_timer_create,
.nsleep = no_nsleep,
};
@@ -285,6 +292,7 @@ static __init int init_posix_timers(void)
.clock_getres = posix_get_coarse_res,
.clock_get = posix_get_realtime_coarse,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = no_timer_create,
.nsleep = no_nsleep,
};
@@ -292,6 +300,7 @@ static __init int init_posix_timers(void)
.clock_getres = posix_get_coarse_res,
.clock_get = posix_get_monotonic_coarse,
.clock_set = do_posix_clock_nosettime,
+ .clock_adj = do_posix_clock_noadjtime,
.timer_create = no_timer_create,
.nsleep = no_nsleep,
};
@@ -928,6 +937,12 @@ int do_posix_clock_nosettime(const clockid_t clockid, struct timespec *tp)
}
EXPORT_SYMBOL_GPL(do_posix_clock_nosettime);
+int do_posix_clock_noadjtime(const clockid_t which_clock, struct timex *t)
+{
+ return -EOPNOTSUPP;
+}
+EXPORT_SYMBOL_GPL(do_posix_clock_noadjtime);
+
int do_posix_clock_nonanosleep(const clockid_t clock, int flags,
struct timespec *t, struct timespec __user *r)
{
@@ -969,6 +984,26 @@ SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
}
+SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock,
+ struct timex __user *, utx)
+{
+ struct timex ktx;
+ int err;
+
+ if (copy_from_user(&ktx, utx, sizeof(ktx)))
+ return -EFAULT;
+
+ if (invalid_clockid(which_clock))
+ return -EINVAL;
+
+ err = CLOCK_DISPATCH(which_clock, clock_adj, (which_clock, &ktx));
+
+ if (copy_to_user(utx, &ktx, sizeof(ktx)))
+ return -EFAULT;
+
+ return err;
+}
+
SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock,
struct timespec __user *, tp)
{
--
1.7.0.4
This patch adds support for adding and removing posix clocks. The
clock lifetime cycle is patterned after usb devices. Each clock is
represented by a standard character device. In addition, the driver
may optionally implemented custom character device operations.
The dynamic posix clocks do not yet do anything useful. This patch
merely provides some needed infrastructure.
Signed-off-by: Richard Cochran <[email protected]>
---
include/linux/posix-clock.h | 111 ++++++++++++++++++++++
kernel/time/Makefile | 3 +-
kernel/time/posix-clock.c | 217 +++++++++++++++++++++++++++++++++++++++++++
3 files changed, 330 insertions(+), 1 deletions(-)
create mode 100644 include/linux/posix-clock.h
create mode 100644 kernel/time/posix-clock.c
diff --git a/include/linux/posix-clock.h b/include/linux/posix-clock.h
new file mode 100644
index 0000000..1ce7fb7
--- /dev/null
+++ b/include/linux/posix-clock.h
@@ -0,0 +1,111 @@
+/*
+ * posix-clock.h - support for dynamic clock devices
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#ifndef _LINUX_POSIX_CLOCK_H_
+#define _LINUX_POSIX_CLOCK_H_
+
+#include <linux/fs.h>
+#include <linux/poll.h>
+#include <linux/posix-timers.h>
+
+/**
+ * struct posix_clock_fops - character device operations
+ *
+ * Every posix clock is represented by a character device. Drivers may
+ * optionally offer extended capabilities by implementing these
+ * functions. The character device file operations are first handled
+ * by the clock device layer, then passed on to the driver by calling
+ * these functions.
+ *
+ * The clock device layer already uses fp->private_data, so drivers
+ * are provided their private data via the 'priv' paramenter.
+ */
+void *posix_clock_private(struct file *fp);
+
+struct posix_clock_fops {
+ int (*fasync) (void *priv, int fd, struct file *file, int on);
+ int (*mmap) (void *priv, struct vm_area_struct *vma);
+ int (*open) (void *priv, fmode_t f_mode);
+ int (*release) (void *priv);
+ long (*ioctl) (void *priv, unsigned int cmd, unsigned long arg);
+ long (*compat_ioctl) (void *priv, unsigned int cmd, unsigned long arg);
+ ssize_t (*read) (void *priv, uint flags, char __user *buf, size_t cnt);
+ unsigned int (*poll) (void *priv, struct file *file, poll_table *wait);
+};
+
+/**
+ * struct posix_clock_operations - functional interface to the clock
+ * @owner: The clock driver should set to THIS_MODULE.
+ * @fops: Optional custom character device operations
+ * @clock_adjtime: Adjust the clock
+ * @clock_gettime: Read the current time
+ * @clock_getres: Get the clock resolution
+ * @clock_settime: Set the current time value
+ * @timer_create: Create a new timer
+ * @timer_delete: Remove a previously created timer
+ * @timer_gettime: Get remaining time and interval of a timer
+ * @timer_setttime: Set a timer's initial expiration and interval
+ */
+struct posix_clock_operations {
+ struct module *owner;
+ struct posix_clock_fops fops;
+ int (*clock_adjtime)(void *priv, struct timex *tx);
+ int (*clock_gettime)(void *priv, struct timespec *ts);
+ int (*clock_getres) (void *priv, struct timespec *ts);
+ int (*clock_settime)(void *priv, const struct timespec *ts);
+ int (*timer_create) (void *priv, struct k_itimer *kit);
+ int (*timer_delete) (void *priv, struct k_itimer *kit);
+ void (*timer_gettime)(void *priv, struct k_itimer *kit,
+ struct itimerspec *tsp);
+ int (*timer_settime)(void *priv, struct k_itimer *kit, int flags,
+ struct itimerspec *tsp, struct itimerspec *old);
+};
+
+/**
+ * struct posix_clock - an opaque type
+ */
+struct posix_clock;
+
+/**
+ * posix_clock_create() - register a new clock
+ * @cops: Pointer to the clock's interface
+ * @devid: Allocated device id
+ * @priv: Private data passed back to the driver via the interface functions
+ *
+ * A clock driver calls this function to register itself with the
+ * clock device subsystem. The 'cops' argument must point to
+ * persistent data, so the caller should pass a static global.
+ *
+ * Returns a pointer to a new clock device, or PTR_ERR on error.
+ */
+struct posix_clock *posix_clock_create(struct posix_clock_operations *cops,
+ dev_t devid, void *priv);
+
+/**
+ * posix_clock_destroy() - unregister a clock
+ * @clk: Pointer obtained via posix_clock_create()
+ *
+ * A clock driver calls this function to remove itself from the clock
+ * device subsystem. The posix_clock itself will remain (in an
+ * inactive state) until its reference count drops to zero, at which
+ * point it will be deallocated.
+ */
+void posix_clock_destroy(struct posix_clock *clk);
+
+#endif
diff --git a/kernel/time/Makefile b/kernel/time/Makefile
index ee26662..bb3c9b8 100644
--- a/kernel/time/Makefile
+++ b/kernel/time/Makefile
@@ -1,4 +1,5 @@
-obj-y += timekeeping.o ntp.o clocksource.o jiffies.o timer_list.o timecompare.o timeconv.o
+obj-y += timekeeping.o ntp.o clocksource.o jiffies.o timer_list.o \
+timecompare.o timeconv.o posix-clock.o
obj-$(CONFIG_GENERIC_CLOCKEVENTS_BUILD) += clockevents.o
obj-$(CONFIG_GENERIC_CLOCKEVENTS) += tick-common.o
diff --git a/kernel/time/posix-clock.c b/kernel/time/posix-clock.c
new file mode 100644
index 0000000..e5924c9
--- /dev/null
+++ b/kernel/time/posix-clock.c
@@ -0,0 +1,217 @@
+/*
+ * posix-clock.c - support for dynamic clock devices
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/cdev.h>
+#include <linux/device.h>
+#include <linux/mutex.h>
+#include <linux/posix-clock.h>
+#include <linux/slab.h>
+
+#define MAX_CLKDEV BITS_PER_LONG
+static DECLARE_BITMAP(clocks_map, MAX_CLKDEV);
+static DEFINE_MUTEX(clocks_mutex); /* protects 'clocks_map' */
+
+struct posix_clock {
+ struct posix_clock_operations *ops;
+ struct cdev cdev;
+ struct kref kref;
+ struct mutex mux;
+ void *priv;
+ int index;
+ bool zombie;
+};
+
+static void delete_clock(struct kref *kref);
+
+static ssize_t posix_clock_read(struct file *fp, char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.read)
+ return clk->ops->fops.read(clk->priv, fp->f_flags, buf, count);
+ else
+ return -EINVAL;
+}
+
+static unsigned int posix_clock_poll(struct file *fp, poll_table *wait)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.poll)
+ return clk->ops->fops.poll(clk->priv, fp, wait);
+ else
+ return 0;
+}
+
+static int posix_clock_fasync(int fd, struct file *fp, int on)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.fasync)
+ return clk->ops->fops.fasync(clk->priv, fd, fp, on);
+ else
+ return 0;
+}
+
+static int posix_clock_mmap(struct file *fp, struct vm_area_struct *vma)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.mmap)
+ return clk->ops->fops.mmap(clk->priv, vma);
+ else
+ return -ENODEV;
+}
+
+static long posix_clock_ioctl(struct file *fp,
+ unsigned int cmd, unsigned long arg)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.ioctl)
+ return clk->ops->fops.ioctl(clk->priv, cmd, arg);
+ else
+ return -ENOTTY;
+}
+
+#ifdef CONFIG_COMPAT
+static long posix_clock_compat_ioctl(struct file *fp,
+ unsigned int cmd, unsigned long arg)
+{
+ struct posix_clock *clk = fp->private_data;
+
+ if (clk->ops->fops.compat_ioctl)
+ return clk->ops->fops.compat_ioctl(clk->priv, cmd, arg);
+ else
+ return -ENOTTY;
+}
+#endif
+
+static int posix_clock_open(struct inode *inode, struct file *fp)
+{
+ struct posix_clock *clk =
+ container_of(inode->i_cdev, struct posix_clock, cdev);
+
+ kref_get(&clk->kref);
+ fp->private_data = clk;
+
+ if (clk->ops->fops.open)
+ return clk->ops->fops.open(clk->priv, fp->f_mode);
+ else
+ return 0;
+}
+
+static int posix_clock_release(struct inode *inode, struct file *fp)
+{
+ struct posix_clock *clk = fp->private_data;
+ int err = 0;
+
+ if (clk->ops->fops.release)
+ err = clk->ops->fops.release(clk->priv);
+
+ kref_put(&clk->kref, delete_clock);
+
+ return err;
+}
+
+static const struct file_operations posix_clock_file_operations = {
+ .owner = THIS_MODULE,
+ .llseek = no_llseek,
+ .read = posix_clock_read,
+ .poll = posix_clock_poll,
+ .unlocked_ioctl = posix_clock_ioctl,
+ .open = posix_clock_open,
+ .release = posix_clock_release,
+ .fasync = posix_clock_fasync,
+ .mmap = posix_clock_mmap,
+#ifdef CONFIG_COMPAT
+ .compat_ioctl = posix_clock_compat_ioctl,
+#endif
+};
+
+struct posix_clock *posix_clock_create(struct posix_clock_operations *cops,
+ dev_t devid, void *priv)
+{
+ struct posix_clock *clk;
+ int err;
+
+ mutex_lock(&clocks_mutex);
+
+ err = -ENOMEM;
+ clk = kzalloc(sizeof(*clk), GFP_KERNEL);
+ if (!clk)
+ goto no_memory;
+
+ err = -EBUSY;
+ clk->index = find_first_zero_bit(clocks_map, MAX_CLKDEV);
+ if (clk->index < MAX_CLKDEV)
+ set_bit(clk->index, clocks_map);
+ else
+ goto no_index;
+
+ clk->ops = cops;
+ clk->priv = priv;
+ kref_init(&clk->kref);
+ mutex_init(&clk->mux);
+
+ cdev_init(&clk->cdev, &posix_clock_file_operations);
+ clk->cdev.owner = clk->ops->owner;
+ err = cdev_add(&clk->cdev, devid, 1);
+ if (err)
+ goto no_cdev;
+
+ mutex_unlock(&clocks_mutex);
+ return clk;
+
+no_cdev:
+ mutex_destroy(&clk->mux);
+ clear_bit(clk->index, clocks_map);
+no_index:
+ kfree(clk);
+no_memory:
+ mutex_unlock(&clocks_mutex);
+ return ERR_PTR(err);
+}
+EXPORT_SYMBOL_GPL(posix_clock_create);
+
+static void delete_clock(struct kref *kref)
+{
+ struct posix_clock *clk =
+ container_of(kref, struct posix_clock, kref);
+
+ mutex_lock(&clocks_mutex);
+ clear_bit(clk->index, clocks_map);
+ mutex_unlock(&clocks_mutex);
+
+ mutex_destroy(&clk->mux);
+ kfree(clk);
+}
+
+void posix_clock_destroy(struct posix_clock *clk)
+{
+ cdev_del(&clk->cdev);
+
+ mutex_lock(&clk->mux);
+ clk->zombie = true;
+ mutex_unlock(&clk->mux);
+
+ kref_put(&clk->kref, delete_clock);
+}
+EXPORT_SYMBOL_GPL(posix_clock_destroy);
--
1.7.0.4
This patch lets the various posix clock and timer system calls use
dynamic posix clocks, as well as the traditional static clocks.
For the performance critical calls (eg clock_gettime) the code path
from the traditional static clocks is preserved.
The following system calls are affected:
- clock_adjtime (brand new syscall)
- clock_gettime
- clock_getres
- clock_settime
- timer_create
- timer_delete
- timer_gettime
- timer_settime
Signed-off-by: Richard Cochran <[email protected]>
---
include/linux/posix-clock.h | 25 +++++++
include/linux/posix-timers.h | 28 +++++++-
include/linux/time.h | 2 +
kernel/posix-timers.c | 122 +++++++++++++++++++++++++++-----
kernel/time/posix-clock.c | 159 ++++++++++++++++++++++++++++++++++++++++++
5 files changed, 316 insertions(+), 20 deletions(-)
diff --git a/include/linux/posix-clock.h b/include/linux/posix-clock.h
index 1ce7fb7..7ea94f5 100644
--- a/include/linux/posix-clock.h
+++ b/include/linux/posix-clock.h
@@ -108,4 +108,29 @@ struct posix_clock *posix_clock_create(struct posix_clock_operations *cops,
*/
void posix_clock_destroy(struct posix_clock *clk);
+/**
+ * clockid_to_posix_clock() - obtain clock device pointer from a clock id
+ * @id: user space clock id
+ *
+ * Returns a pointer to the clock device, or NULL if the id is not a
+ * dynamic clock id.
+ */
+struct posix_clock *clockid_to_posix_clock(clockid_t id);
+
+/*
+ * The following functions are only to be called from posix-timers.c
+ */
+int pc_clock_adjtime(struct posix_clock *clk, struct timex *tx);
+int pc_clock_gettime(struct posix_clock *clk, struct timespec *ts);
+int pc_clock_getres(struct posix_clock *clk, struct timespec *ts);
+int pc_clock_settime(struct posix_clock *clk, const struct timespec *ts);
+
+int pc_timer_create(struct posix_clock *clk, struct k_itimer *kit);
+int pc_timer_delete(struct posix_clock *clk, struct k_itimer *kit);
+
+void pc_timer_gettime(struct posix_clock *clk, struct k_itimer *kit,
+ struct itimerspec *ts);
+int pc_timer_settime(struct posix_clock *clk, struct k_itimer *kit,
+ int flags, struct itimerspec *ts, struct itimerspec *old);
+
#endif
diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index b05d9b8..62ae296 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -18,10 +18,22 @@ struct cpu_timer_list {
int firing;
};
+/* Bit fields within a clockid:
+ *
+ * The most significant 29 bits hold either a pid or a file descriptor.
+ *
+ * Bit 2 indicates whether a cpu clock refers to a thread or a process.
+ *
+ * Bits 1 and 0 give the type: PROF=0, VIRT=1, SCHED=2, or FD=3.
+ *
+ * A clockid is invalid if bits 2, 1, and 0 all set (see also CLOCK_INVALID
+ * in include/linux/time.h)
+ */
+
#define CPUCLOCK_PID(clock) ((pid_t) ~((clock) >> 3))
#define CPUCLOCK_PERTHREAD(clock) \
(((clock) & (clockid_t) CPUCLOCK_PERTHREAD_MASK) != 0)
-#define CPUCLOCK_PID_MASK 7
+
#define CPUCLOCK_PERTHREAD_MASK 4
#define CPUCLOCK_WHICH(clock) ((clock) & (clockid_t) CPUCLOCK_CLOCK_MASK)
#define CPUCLOCK_CLOCK_MASK 3
@@ -29,12 +41,26 @@ struct cpu_timer_list {
#define CPUCLOCK_VIRT 1
#define CPUCLOCK_SCHED 2
#define CPUCLOCK_MAX 3
+#define CLOCKFD CPUCLOCK_MAX
+#define CLOCKFD_MASK (CPUCLOCK_PERTHREAD_MASK|CPUCLOCK_CLOCK_MASK)
#define MAKE_PROCESS_CPUCLOCK(pid, clock) \
((~(clockid_t) (pid) << 3) | (clockid_t) (clock))
#define MAKE_THREAD_CPUCLOCK(tid, clock) \
MAKE_PROCESS_CPUCLOCK((tid), (clock) | CPUCLOCK_PERTHREAD_MASK)
+#define FD_TO_CLOCKID(fd) ((clockid_t) (fd << 3) | CLOCKFD)
+#define CLOCKID_TO_FD(clk) (((unsigned int) clk) >> 3)
+
+static inline bool clock_is_static(const clockid_t id)
+{
+ if (0 == (id & ~CLOCKFD_MASK))
+ return true;
+ if (CLOCKFD == (id & CLOCKFD_MASK))
+ return false;
+ return true;
+}
+
/* POSIX.1b interval timer structure. */
struct k_itimer {
struct list_head list; /* free/ allocate list */
diff --git a/include/linux/time.h b/include/linux/time.h
index 9f15ac7..914c48d 100644
--- a/include/linux/time.h
+++ b/include/linux/time.h
@@ -299,6 +299,8 @@ struct itimerval {
#define CLOCKS_MASK (CLOCK_REALTIME | CLOCK_MONOTONIC)
#define CLOCKS_MONO CLOCK_MONOTONIC
+#define CLOCK_INVALID -1
+
/*
* The various flags for setting POSIX.1b interval timers:
*/
diff --git a/kernel/posix-timers.c b/kernel/posix-timers.c
index 502bde4..cca3bd5 100644
--- a/kernel/posix-timers.c
+++ b/kernel/posix-timers.c
@@ -41,6 +41,7 @@
#include <linux/init.h>
#include <linux/compiler.h>
#include <linux/idr.h>
+#include <linux/posix-clock.h>
#include <linux/posix-timers.h>
#include <linux/syscalls.h>
#include <linux/wait.h>
@@ -532,12 +533,15 @@ SYSCALL_DEFINE3(timer_create, const clockid_t, which_clock,
struct sigevent __user *, timer_event_spec,
timer_t __user *, created_timer_id)
{
+ struct posix_clock *clk_dev;
struct k_itimer *new_timer;
int error, new_timer_id;
sigevent_t event;
int it_id_set = IT_ID_NOT_SET;
- if (invalid_clockid(which_clock))
+ clk_dev = clockid_to_posix_clock(which_clock);
+
+ if (!clk_dev && invalid_clockid(which_clock))
return -EINVAL;
new_timer = alloc_posix_timer();
@@ -600,7 +604,11 @@ SYSCALL_DEFINE3(timer_create, const clockid_t, which_clock,
goto out;
}
- error = CLOCK_DISPATCH(which_clock, timer_create, (new_timer));
+ if (unlikely(clk_dev))
+ error = pc_timer_create(clk_dev, new_timer);
+ else {
+ error = CLOCK_DISPATCH(which_clock, timer_create, (new_timer));
+ }
if (error)
goto out;
@@ -712,6 +720,7 @@ common_timer_get(struct k_itimer *timr, struct itimerspec *cur_setting)
SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
struct itimerspec __user *, setting)
{
+ struct posix_clock *clk_dev;
struct k_itimer *timr;
struct itimerspec cur_setting;
unsigned long flags;
@@ -720,7 +729,15 @@ SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
if (!timr)
return -EINVAL;
- CLOCK_DISPATCH(timr->it_clock, timer_get, (timr, &cur_setting));
+ if (likely(clock_is_static(timr->it_clock))) {
+
+ CLOCK_DISPATCH(timr->it_clock, timer_get, (timr, &cur_setting));
+
+ } else {
+ clk_dev = clockid_to_posix_clock(timr->it_clock);
+ if (clk_dev)
+ pc_timer_gettime(clk_dev, timr, &cur_setting);
+ }
unlock_timer(timr, flags);
@@ -811,6 +828,7 @@ SYSCALL_DEFINE4(timer_settime, timer_t, timer_id, int, flags,
const struct itimerspec __user *, new_setting,
struct itimerspec __user *, old_setting)
{
+ struct posix_clock *clk_dev;
struct k_itimer *timr;
struct itimerspec new_spec, old_spec;
int error = 0;
@@ -831,8 +849,19 @@ retry:
if (!timr)
return -EINVAL;
- error = CLOCK_DISPATCH(timr->it_clock, timer_set,
- (timr, flags, &new_spec, rtn));
+ if (likely(clock_is_static(timr->it_clock))) {
+
+ error = CLOCK_DISPATCH(timr->it_clock, timer_set,
+ (timr, flags, &new_spec, rtn));
+
+ } else {
+ clk_dev = clockid_to_posix_clock(timr->it_clock);
+ if (clk_dev)
+ error = pc_timer_settime(clk_dev, timr,
+ flags, &new_spec, rtn);
+ else
+ error = -EINVAL;
+ }
unlock_timer(timr, flag);
if (error == TIMER_RETRY) {
@@ -858,6 +887,13 @@ static inline int common_timer_del(struct k_itimer *timer)
static inline int timer_delete_hook(struct k_itimer *timer)
{
+ struct posix_clock *clk_dev;
+
+ clk_dev = clockid_to_posix_clock(timer->it_clock);
+
+ if (clk_dev)
+ return pc_timer_delete(clk_dev, timer);
+
return CLOCK_DISPATCH(timer->it_clock, timer_del, (timer));
}
@@ -957,26 +993,51 @@ EXPORT_SYMBOL_GPL(do_posix_clock_nonanosleep);
SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock,
const struct timespec __user *, tp)
{
+ struct posix_clock *clk_dev = NULL;
struct timespec new_tp;
- if (invalid_clockid(which_clock))
- return -EINVAL;
+ if (likely(clock_is_static(which_clock))) {
+
+ if (invalid_clockid(which_clock))
+ return -EINVAL;
+ } else {
+ clk_dev = clockid_to_posix_clock(which_clock);
+ if (!clk_dev)
+ return -EINVAL;
+ }
+
if (copy_from_user(&new_tp, tp, sizeof (*tp)))
return -EFAULT;
+ if (unlikely(clk_dev))
+ return pc_clock_settime(clk_dev, &new_tp);
+
return CLOCK_DISPATCH(which_clock, clock_set, (which_clock, &new_tp));
}
SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
- struct timespec __user *,tp)
+ struct timespec __user *, tp)
{
+ struct posix_clock *clk_dev;
struct timespec kernel_tp;
int error;
- if (invalid_clockid(which_clock))
- return -EINVAL;
- error = CLOCK_DISPATCH(which_clock, clock_get,
- (which_clock, &kernel_tp));
+ if (likely(clock_is_static(which_clock))) {
+
+ if (invalid_clockid(which_clock))
+ return -EINVAL;
+
+ error = CLOCK_DISPATCH(which_clock, clock_get,
+ (which_clock, &kernel_tp));
+ } else {
+ clk_dev = clockid_to_posix_clock(which_clock);
+
+ if (!clk_dev)
+ return -EINVAL;
+
+ error = pc_clock_gettime(clk_dev, &kernel_tp);
+ }
+
if (!error && copy_to_user(tp, &kernel_tp, sizeof (kernel_tp)))
error = -EFAULT;
@@ -987,16 +1048,28 @@ SYSCALL_DEFINE2(clock_gettime, const clockid_t, which_clock,
SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock,
struct timex __user *, utx)
{
+ struct posix_clock *clk_dev;
struct timex ktx;
int err;
if (copy_from_user(&ktx, utx, sizeof(ktx)))
return -EFAULT;
- if (invalid_clockid(which_clock))
- return -EINVAL;
+ if (likely(clock_is_static(which_clock))) {
- err = CLOCK_DISPATCH(which_clock, clock_adj, (which_clock, &ktx));
+ if (invalid_clockid(which_clock))
+ return -EINVAL;
+
+ err = CLOCK_DISPATCH(which_clock, clock_adj,
+ (which_clock, &ktx));
+ } else {
+ clk_dev = clockid_to_posix_clock(which_clock);
+
+ if (!clk_dev)
+ return -EINVAL;
+
+ err = pc_clock_adjtime(clk_dev, &ktx);
+ }
if (copy_to_user(utx, &ktx, sizeof(ktx)))
return -EFAULT;
@@ -1007,14 +1080,25 @@ SYSCALL_DEFINE2(clock_adjtime, const clockid_t, which_clock,
SYSCALL_DEFINE2(clock_getres, const clockid_t, which_clock,
struct timespec __user *, tp)
{
+ struct posix_clock *clk_dev;
struct timespec rtn_tp;
int error;
- if (invalid_clockid(which_clock))
- return -EINVAL;
+ if (likely(clock_is_static(which_clock))) {
- error = CLOCK_DISPATCH(which_clock, clock_getres,
- (which_clock, &rtn_tp));
+ if (invalid_clockid(which_clock))
+ return -EINVAL;
+
+ error = CLOCK_DISPATCH(which_clock, clock_getres,
+ (which_clock, &rtn_tp));
+ } else {
+ clk_dev = clockid_to_posix_clock(which_clock);
+
+ if (!clk_dev)
+ return -EINVAL;
+
+ error = pc_clock_getres(clk_dev, &rtn_tp);
+ }
if (!error && tp && copy_to_user(tp, &rtn_tp, sizeof (rtn_tp))) {
error = -EFAULT;
diff --git a/kernel/time/posix-clock.c b/kernel/time/posix-clock.c
index e5924c9..a707c10 100644
--- a/kernel/time/posix-clock.c
+++ b/kernel/time/posix-clock.c
@@ -19,9 +19,12 @@
*/
#include <linux/cdev.h>
#include <linux/device.h>
+#include <linux/file.h>
#include <linux/mutex.h>
#include <linux/posix-clock.h>
#include <linux/slab.h>
+#include <linux/syscalls.h>
+#include <linux/uaccess.h>
#define MAX_CLKDEV BITS_PER_LONG
static DECLARE_BITMAP(clocks_map, MAX_CLKDEV);
@@ -215,3 +218,159 @@ void posix_clock_destroy(struct posix_clock *clk)
kref_put(&clk->kref, delete_clock);
}
EXPORT_SYMBOL_GPL(posix_clock_destroy);
+
+struct posix_clock *clockid_to_posix_clock(const clockid_t id)
+{
+ struct posix_clock *clk = NULL;
+ struct file *fp;
+
+ if (clock_is_static(id))
+ return NULL;
+
+ fp = fget(CLOCKID_TO_FD(id));
+ if (!fp)
+ return NULL;
+
+ if (fp->f_op->open == posix_clock_open)
+ clk = fp->private_data;
+
+ fput(fp);
+ return clk;
+}
+
+int pc_clock_adjtime(struct posix_clock *clk, struct timex *tx)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->clock_adjtime)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->clock_adjtime(clk->priv, tx);
+
+ mutex_unlock(&clk->mux);
+ return err;
+}
+
+int pc_clock_gettime(struct posix_clock *clk, struct timespec *ts)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->clock_gettime)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->clock_gettime(clk->priv, ts);
+
+ mutex_unlock(&clk->mux);
+ return err;
+}
+
+int pc_clock_getres(struct posix_clock *clk, struct timespec *ts)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->clock_getres)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->clock_getres(clk->priv, ts);
+
+ mutex_unlock(&clk->mux);
+ return err;
+}
+
+int pc_clock_settime(struct posix_clock *clk, const struct timespec *ts)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->clock_settime)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->clock_settime(clk->priv, ts);
+
+ mutex_unlock(&clk->mux);
+ return err;
+}
+
+int pc_timer_create(struct posix_clock *clk, struct k_itimer *kit)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->timer_create)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->timer_create(clk->priv, kit);
+
+ mutex_unlock(&clk->mux);
+ return err;
+}
+
+int pc_timer_delete(struct posix_clock *clk, struct k_itimer *kit)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->timer_delete)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->timer_delete(clk->priv, kit);
+
+ mutex_unlock(&clk->mux);
+
+ return err;
+}
+
+void pc_timer_gettime(struct posix_clock *clk, struct k_itimer *kit,
+ struct itimerspec *ts)
+{
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ goto out;
+ else if (!clk->ops->timer_gettime)
+ goto out;
+ else
+ clk->ops->timer_gettime(clk->priv, kit, ts);
+out:
+ mutex_unlock(&clk->mux);
+}
+
+int pc_timer_settime(struct posix_clock *clk, struct k_itimer *kit,
+ int flags, struct itimerspec *ts, struct itimerspec *old)
+{
+ int err;
+
+ mutex_lock(&clk->mux);
+
+ if (clk->zombie)
+ err = -ENODEV;
+ else if (!clk->ops->timer_settime)
+ err = -EOPNOTSUPP;
+ else
+ err = clk->ops->timer_settime(clk->priv, kit, flags, ts, old);
+
+ mutex_unlock(&clk->mux);
+
+ return err;
+}
--
1.7.0.4
This patch adds an infrastructure for hardware clocks that implement
IEEE 1588, the Precision Time Protocol (PTP). A class driver offers a
registration method to particular hardware clock drivers. Each clock is
presented as a standard POSIX clock.
The ancillary clock features are exposed in two different ways, via
the sysfs and by a character device.
Signed-off-by: Richard Cochran <[email protected]>
---
Documentation/ABI/testing/sysfs-ptp | 107 +++++++++++
Documentation/ptp/ptp.txt | 94 ++++++++++
Documentation/ptp/testptp.c | 352 +++++++++++++++++++++++++++++++++++
Documentation/ptp/testptp.mk | 33 ++++
drivers/Kconfig | 2 +
drivers/Makefile | 1 +
drivers/ptp/Kconfig | 27 +++
drivers/ptp/Makefile | 6 +
drivers/ptp/ptp_chardev.c | 144 ++++++++++++++
drivers/ptp/ptp_clock.c | 317 +++++++++++++++++++++++++++++++
drivers/ptp/ptp_private.h | 68 +++++++
drivers/ptp/ptp_sysfs.c | 230 +++++++++++++++++++++++
include/linux/Kbuild | 1 +
include/linux/ptp_clock.h | 79 ++++++++
include/linux/ptp_clock_kernel.h | 139 ++++++++++++++
15 files changed, 1600 insertions(+), 0 deletions(-)
create mode 100644 Documentation/ABI/testing/sysfs-ptp
create mode 100644 Documentation/ptp/ptp.txt
create mode 100644 Documentation/ptp/testptp.c
create mode 100644 Documentation/ptp/testptp.mk
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_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
diff --git a/Documentation/ABI/testing/sysfs-ptp b/Documentation/ABI/testing/sysfs-ptp
new file mode 100644
index 0000000..47142ce
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-ptp
@@ -0,0 +1,107 @@
+What: /sys/class/ptp/
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This directory contains files and directories
+ providing a standardized interface to the ancillary
+ features of PTP hardware clocks.
+
+What: /sys/class/ptp/ptpN/
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This directory contains the attributes of the Nth PTP
+ hardware clock registered into the PTP class driver
+ subsystem.
+
+What: /sys/class/ptp/ptpN/clock_id
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the POSIX clock ID (a non-negative
+ integer) corresponding to the PTP hardware clock. This
+ value may be passed as the first argument to the POSIX
+ clock and timer system calls. See man CLOCK_GETRES(2)
+ and TIMER_CREATE(2).
+
+What: /sys/class/ptp/ptpN/clock_name
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the name of the PTP hardware clock
+ as a human readable string.
+
+What: /sys/class/ptp/ptpN/max_adjustment
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the PTP hardware clock's maximum
+ frequency adjustment value (a positive integer) in
+ parts per billion.
+
+What: /sys/class/ptp/ptpN/n_alarms
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the number of periodic or one shot
+ alarms offer by the PTP hardware clock.
+
+What: /sys/class/ptp/ptpN/n_external_timestamps
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the number of external timestamp
+ channels offered by the PTP hardware clock.
+
+What: /sys/class/ptp/ptpN/n_periodic_outputs
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file contains the number of programmable periodic
+ output channels offered by the PTP hardware clock.
+
+What: /sys/class/ptp/ptpN/pps_avaiable
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file indicates whether the PTP hardware clock
+ supports a Pulse Per Second to the host CPU. Reading
+ "1" means that the PPS is supported, while "0" means
+ not supported.
+
+What: /sys/class/ptp/ptpN/extts_enable
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This write-only file enables or disables external
+ timestamps. To enable external timestamps, write the
+ channel index followed by a "1" into the file.
+ To disable external timestamps, write the channel
+ index followed by a "0" into the file.
+
+What: /sys/class/ptp/ptpN/fifo
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This file provides timestamps on external events, in
+ the form of three integers: channel index, seconds,
+ and nanoseconds.
+
+What: /sys/class/ptp/ptpN/period
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This write-only file enables or disables periodic
+ outputs. To enable a periodic output, write three
+ integers into the file: channel index, seconds, and
+ nanoseconds. To disable a periodic output, set seconds
+ and nanoseconds to zero.
+
+What: /sys/class/ptp/ptpN/pps_enable
+Date: September 2010
+Contact: Richard Cochran <[email protected]>
+Description:
+ This write-only file enables or disables delivery of
+ PPS events to the Linux PPS subsystem. To enable PPS
+ events, write a "1" into the file. To disable events,
+ write a "0" into the file.
diff --git a/Documentation/ptp/ptp.txt b/Documentation/ptp/ptp.txt
new file mode 100644
index 0000000..d198772
--- /dev/null
+++ b/Documentation/ptp/ptp.txt
@@ -0,0 +1,94 @@
+
+* PTP hardware clock infrastructure for Linux
+
+ This patch set introduces support for IEEE 1588 PTP clocks in
+ Linux. Together with the SO_TIMESTAMPING socket options, this
+ presents a standardized method for developing PTP user space
+ programs, synchronizing Linux with external clocks, and using the
+ ancillary features of PTP hardware clocks.
+
+ A new class driver exports a kernel interface for specific clock
+ drivers and a user space interface. The infrastructure supports a
+ complete set of PTP hardware clock functionality.
+
+ + Basic clock operations
+ - Set time
+ - Get time
+ - Shift the clock by a given offset atomically
+ - Adjust clock frequency
+
+ + Ancillary clock features
+ - One short or periodic alarms, with signal delivery to user program
+ - Time stamp external events
+ - Period output signals configurable from user space
+ - Synchronization of the Linux system time via the PPS subsystem
+
+** PTP hardware clock kernel API
+
+ A PTP clock driver registers itself with the class driver. The
+ class driver handles all of the dealings with user space. The
+ author of a clock driver need only implement the details of
+ programming the clock hardware. The clock driver notifies the class
+ driver of asynchronous events (alarms and external time stamps) via
+ a simple message passing interface.
+
+ The class driver supports multiple PTP clock drivers. In normal use
+ cases, only one PTP clock is needed. However, for testing and
+ development, it can be useful to have more than one clock in a
+ single system, in order to allow performance comparisons.
+
+** PTP hardware clock user space API
+
+ Each clock is presented as a POSIX clock id. User space can
+ discover the clock id by reading the appropriate sysfs file (see
+ Documentation/ABI/testing/sysfs-ptp for details). Using the clock
+ id, user space may call clock_gettime, clock_settime, and
+ clock_adjtime. These calls implement the basic clock operations.
+
+ The class driver also creates a character device for each
+ registered clock. User space programs may control the clock using
+ standardized ioctls. A program may query, enable, configure, and
+ disable the ancillary clock features. User space can receive time
+ stamped events via blocking read() and poll(). One shot and
+ periodic signals may be configured via the POSIX timer_settime()
+ system call.
+
+** Writing clock drivers
+
+ Clock drivers include include/linux/ptp_clock_kernel.h and register
+ themselves by presenting a 'struct ptp_clock_info' to the
+ registration method. Clock drivers must implement all of the
+ functions in the interface. If a clock does not offer a particular
+ ancillary feature, then the driver should just return -EOPNOTSUPP
+ from those functions.
+
+ Drivers must ensure that all of the methods in interface are
+ reentrant. Since most hardware implementations treat the time value
+ as a 64 bit integer accessed as two 32 bit registers, drivers
+ should use spin_lock_irqsave/spin_unlock_irqrestore to protect
+ against concurrent access. This locking cannot be accomplished in
+ class driver, since the lock may also be needed by the clock
+ driver's interrupt service routine.
+
+** Supported hardware
+
+ + Standard Linux system timer
+ - No special PTP features
+ - For use with software time stamping
+
+ + Freescale eTSEC gianfar
+ - 2 Time stamp external triggers, programmable polarity (opt. interrupt)
+ - 2 Alarm registers (optional interrupt)
+ - 3 Periodic signals (optional interrupt)
+
+ + National 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)
diff --git a/Documentation/ptp/testptp.c b/Documentation/ptp/testptp.c
new file mode 100644
index 0000000..9e389a4
--- /dev/null
+++ b/Documentation/ptp/testptp.c
@@ -0,0 +1,352 @@
+/*
+ * PTP 1588 clock support - User space test program
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <errno.h>
+#include <fcntl.h>
+#include <math.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/timex.h>
+#include <sys/types.h>
+#include <time.h>
+#include <unistd.h>
+
+#include <linux/ptp_clock.h>
+
+#define DEVICE "/dev/ptp0"
+
+#ifndef ADJ_SETOFFSET
+#define ADJ_SETOFFSET 0x0040
+#endif
+
+#ifndef CLOCK_INVALID
+#define CLOCK_INVALID -1
+#endif
+
+/* When glibc offers the syscall, this will go away. */
+#include <sys/syscall.h>
+static int clock_adjtime(clockid_t id, struct timex *tx)
+{
+ return syscall(__NR_clock_adjtime, id, tx);
+}
+
+static clockid_t get_clockid(int fd)
+{
+#define CLOCKFD 3
+#define FD_TO_CLOCKID(fd) ((clockid_t) (fd << 3) | CLOCKFD)
+
+ return FD_TO_CLOCKID(fd);
+}
+
+static void handle_alarm(int s)
+{
+ printf("received signal %d\n", s);
+}
+
+static int install_handler(int signum, void (*handler)(int))
+{
+ struct sigaction action;
+ sigset_t mask;
+
+ /* Unblock the signal. */
+ sigemptyset(&mask);
+ sigaddset(&mask, signum);
+ sigprocmask(SIG_UNBLOCK, &mask, NULL);
+
+ /* Install the signal handler. */
+ action.sa_handler = handler;
+ action.sa_flags = 0;
+ sigemptyset(&action.sa_mask);
+ sigaction(signum, &action, NULL);
+
+ return 0;
+}
+
+static void usage(char *progname)
+{
+ fprintf(stderr,
+ "usage: %s [options]\n"
+ " -a val request a one-shot alarm after 'val' seconds\n"
+ " -A val request a periodic alarm every 'val' seconds\n"
+ " -c query the ptp clock's capabilities\n"
+ " -d name device to open\n"
+ " -e val read 'val' external time stamp events\n"
+ " -f val adjust the ptp clock frequency by 'val' ppb\n"
+ " -g get the ptp clock time\n"
+ " -h prints this message\n"
+ " -p val enable output with a period of 'val' nanoseconds\n"
+ " -P val enable or disable (val=1|0) the system clock PPS\n"
+ " -s set the ptp clock time from the system time\n"
+ " -t val shift the ptp clock time by 'val' seconds\n",
+ progname);
+}
+
+int main(int argc, char *argv[])
+{
+ struct ptp_clock_caps caps;
+ struct ptp_extts_event event;
+ struct ptp_extts_request extts_request;
+ struct ptp_perout_request perout_request;
+ struct timespec ts;
+ struct timex tx;
+
+ static timer_t timerid;
+ struct itimerspec timeout;
+ struct sigevent sigevent;
+
+ char *progname;
+ int c, cnt, fd;
+
+ char *device = DEVICE;
+ clockid_t clkid;
+ int adjfreq = 0x7fffffff;
+ int adjtime = 0;
+ int capabilities = 0;
+ int extts = 0;
+ int gettime = 0;
+ int oneshot = 0;
+ int periodic = 0;
+ int perout = -1;
+ int pps = -1;
+ int settime = 0;
+
+ progname = strrchr(argv[0], '/');
+ progname = progname ? 1+progname : argv[0];
+ while (EOF != (c = getopt(argc, argv, "a:A:cd:e:f:ghp:P:st:v"))) {
+ switch (c) {
+ case 'a':
+ oneshot = atoi(optarg);
+ break;
+ case 'A':
+ periodic = atoi(optarg);
+ break;
+ case 'c':
+ capabilities = 1;
+ break;
+ case 'd':
+ device = optarg;
+ break;
+ case 'e':
+ extts = atoi(optarg);
+ break;
+ case 'f':
+ adjfreq = atoi(optarg);
+ break;
+ case 'g':
+ gettime = 1;
+ break;
+ case 'p':
+ perout = atoi(optarg);
+ break;
+ case 'P':
+ pps = atoi(optarg);
+ break;
+ case 's':
+ settime = 1;
+ break;
+ case 't':
+ adjtime = atoi(optarg);
+ break;
+ case 'h':
+ usage(progname);
+ return 0;
+ case '?':
+ default:
+ usage(progname);
+ return -1;
+ }
+ }
+
+ fd = open(device, O_RDWR);
+ if (fd < 0) {
+ fprintf(stderr, "cannot open %s: %s", device, strerror(errno));
+ return -1;
+ }
+
+ clkid = get_clockid(fd);
+ if (CLOCK_INVALID == clkid) {
+ fprintf(stderr, "failed to read clock id\n");
+ return -1;
+ }
+
+ if (capabilities) {
+ if (ioctl(fd, PTP_CLOCK_GETCAPS, &caps)) {
+ perror("PTP_CLOCK_GETCAPS");
+ } else {
+ printf("capabilities:\n"
+ " %d maximum frequency adjustment (ppb)\n"
+ " %d programmable alarms\n"
+ " %d external time stamp channels\n"
+ " %d programmable periodic signals\n"
+ " %d pulse per second\n",
+ caps.max_adj,
+ caps.n_alarm,
+ caps.n_ext_ts,
+ caps.n_per_out,
+ caps.pps);
+ }
+ }
+
+ if (0x7fffffff != adjfreq) {
+ memset(&tx, 0, sizeof(tx));
+ tx.modes = ADJ_FREQUENCY;
+ tx.freq = (long) (adjfreq * 65.536);
+ if (clock_adjtime(clkid, &tx)) {
+ perror("clock_adjtime");
+ } else {
+ puts("frequency adjustment okay");
+ }
+ }
+
+ if (adjtime) {
+ memset(&tx, 0, sizeof(tx));
+ tx.modes = ADJ_SETOFFSET;
+ tx.time.tv_sec = adjtime;
+ tx.time.tv_usec = 0;
+ if (clock_adjtime(clkid, &tx)) {
+ perror("clock_adjtime");
+ } else {
+ puts("time shift okay");
+ }
+ }
+
+ if (gettime) {
+ if (clock_gettime(clkid, &ts)) {
+ perror("clock_gettime");
+ } else {
+ printf("clock time: %ld.%09ld or %s",
+ ts.tv_sec, ts.tv_nsec, ctime(&ts.tv_sec));
+ }
+ }
+
+ if (settime) {
+ clock_gettime(CLOCK_REALTIME, &ts);
+ if (clock_settime(clkid, &ts)) {
+ perror("clock_settime");
+ } else {
+ puts("set time okay");
+ }
+ }
+
+ if (extts) {
+ memset(&extts_request, 0, sizeof(extts_request));
+ extts_request.index = 0;
+ extts_request.flags = PTP_ENABLE_FEATURE;
+ if (ioctl(fd, PTP_EXTTS_REQUEST, &extts_request)) {
+ perror("PTP_EXTTS_REQUEST");
+ extts = 0;
+ } else {
+ puts("external time stamp request okay");
+ }
+ for (; extts; extts--) {
+ cnt = read(fd, &event, sizeof(event));
+ if (cnt != sizeof(event)) {
+ perror("read");
+ break;
+ }
+ printf("event index %u at %lld.%09u\n", event.index,
+ event.t.sec, event.t.nsec);
+ fflush(stdout);
+ }
+ /* Disable the feature again. */
+ extts_request.flags = 0;
+ if (ioctl(fd, PTP_EXTTS_REQUEST, &extts_request)) {
+ perror("PTP_EXTTS_REQUEST");
+ }
+ }
+
+ if (oneshot) {
+ install_handler(SIGALRM, handle_alarm);
+ /* Create a timer. */
+ sigevent.sigev_notify = SIGEV_SIGNAL;
+ sigevent.sigev_signo = SIGALRM;
+ if (timer_create(clkid, &sigevent, &timerid)) {
+ perror("timer_create");
+ return -1;
+ }
+ /* Start the timer. */
+ memset(&timeout, 0, sizeof(timeout));
+ timeout.it_value.tv_sec = oneshot;
+ if (timer_settime(timerid, 0, &timeout, NULL)) {
+ perror("timer_settime");
+ return -1;
+ }
+ pause();
+ timer_delete(timerid);
+ }
+
+ if (periodic) {
+ install_handler(SIGALRM, handle_alarm);
+ /* Create a timer. */
+ sigevent.sigev_notify = SIGEV_SIGNAL;
+ sigevent.sigev_signo = SIGALRM;
+ if (timer_create(clkid, &sigevent, &timerid)) {
+ perror("timer_create");
+ return -1;
+ }
+ /* Start the timer. */
+ memset(&timeout, 0, sizeof(timeout));
+ timeout.it_interval.tv_sec = periodic;
+ timeout.it_value.tv_sec = periodic;
+ if (timer_settime(timerid, 0, &timeout, NULL)) {
+ perror("timer_settime");
+ return -1;
+ }
+ while (1) {
+ pause();
+ }
+ timer_delete(timerid);
+ }
+
+ if (perout >= 0) {
+ if (clock_gettime(clkid, &ts)) {
+ perror("clock_gettime");
+ return -1;
+ }
+ memset(&perout_request, 0, sizeof(perout_request));
+ perout_request.index = 0;
+ perout_request.start.sec = ts.tv_sec + 2;
+ perout_request.start.nsec = 0;
+ perout_request.period.sec = 0;
+ perout_request.period.nsec = perout;
+ if (ioctl(fd, PTP_PEROUT_REQUEST, &perout_request)) {
+ perror("PTP_PEROUT_REQUEST");
+ } else {
+ puts("periodic output request okay");
+ }
+ }
+
+ if (pps != -1) {
+ int enable = pps ? 1 : 0;
+ if (ioctl(fd, PTP_ENABLE_PPS, enable)) {
+ perror("PTP_ENABLE_PPS");
+ } else {
+ puts("pps for system time request okay");
+ }
+ }
+
+ close(fd);
+ return 0;
+}
diff --git a/Documentation/ptp/testptp.mk b/Documentation/ptp/testptp.mk
new file mode 100644
index 0000000..4ef2d97
--- /dev/null
+++ b/Documentation/ptp/testptp.mk
@@ -0,0 +1,33 @@
+# PTP 1588 clock support - User space test program
+#
+# Copyright (C) 2010 OMICRON electronics GmbH
+#
+# 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.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+CC = $(CROSS_COMPILE)gcc
+INC = -I$(KBUILD_OUTPUT)/usr/include
+CFLAGS = -Wall $(INC)
+LDLIBS = -lrt
+PROGS = testptp
+
+all: $(PROGS)
+
+testptp: testptp.o
+
+clean:
+ rm -f testptp.o
+
+distclean: clean
+ rm -f $(PROGS)
diff --git a/drivers/Kconfig b/drivers/Kconfig
index a2b902f..774fbd7 100644
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -52,6 +52,8 @@ source "drivers/spi/Kconfig"
source "drivers/pps/Kconfig"
+source "drivers/ptp/Kconfig"
+
source "drivers/gpio/Kconfig"
source "drivers/w1/Kconfig"
diff --git a/drivers/Makefile b/drivers/Makefile
index f3ebb30..8e28b9b 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -76,6 +76,7 @@ obj-$(CONFIG_I2O) += message/
obj-$(CONFIG_RTC_LIB) += rtc/
obj-y += i2c/ media/
obj-$(CONFIG_PPS) += pps/
+obj-$(CONFIG_PTP_1588_CLOCK) += ptp/
obj-$(CONFIG_W1) += w1/
obj-$(CONFIG_POWER_SUPPLY) += power/
obj-$(CONFIG_HWMON) += hwmon/
diff --git a/drivers/ptp/Kconfig b/drivers/ptp/Kconfig
new file mode 100644
index 0000000..17be208
--- /dev/null
+++ b/drivers/ptp/Kconfig
@@ -0,0 +1,27 @@
+#
+# PTP clock support configuration
+#
+
+menu "PTP clock support"
+
+config PTP_1588_CLOCK
+ tristate "PTP clock support"
+ depends on EXPERIMENTAL
+ depends on PPS
+ help
+ The IEEE 1588 standard defines a method to precisely
+ synchronize distributed clocks over Ethernet networks. The
+ standard defines a Precision Time Protocol (PTP), which can
+ be used to achieve synchronization within a few dozen
+ microseconds. In addition, with the help of special hardware
+ time stamping units, it can be possible to achieve
+ synchronization to within a few hundred nanoseconds.
+
+ This driver adds support for PTP clocks as character
+ devices. If you want to use a PTP clock, then you should
+ also enable at least one clock driver as well.
+
+ To compile this driver as a module, choose M here: the module
+ will be called ptp.
+
+endmenu
diff --git a/drivers/ptp/Makefile b/drivers/ptp/Makefile
new file mode 100644
index 0000000..480e2af
--- /dev/null
+++ b/drivers/ptp/Makefile
@@ -0,0 +1,6 @@
+#
+# Makefile for PTP 1588 clock support.
+#
+
+ptp-y := ptp_clock.o ptp_chardev.o ptp_sysfs.o
+obj-$(CONFIG_PTP_1588_CLOCK) += ptp.o
diff --git a/drivers/ptp/ptp_chardev.c b/drivers/ptp/ptp_chardev.c
new file mode 100644
index 0000000..6e90e49
--- /dev/null
+++ b/drivers/ptp/ptp_chardev.c
@@ -0,0 +1,144 @@
+/*
+ * PTP 1588 clock support - character device implementation.
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/module.h>
+#include <linux/posix-clock.h>
+#include <linux/poll.h>
+#include <linux/sched.h>
+
+#include <ptp_private.h>
+
+int ptp_open(void *priv, fmode_t fmode)
+{
+ return 0;
+}
+
+long ptp_ioctl(void *pc_private, unsigned int cmd, unsigned long arg)
+{
+ struct ptp_clock_caps caps;
+ struct ptp_clock_request req;
+ struct ptp_clock *ptp = pc_private;
+ struct ptp_clock_info *ops = ptp->info;
+ void *priv = ops->priv;
+ int enable, err = 0;
+
+ switch (cmd) {
+
+ case PTP_CLOCK_GETCAPS:
+ memset(&caps, 0, sizeof(caps));
+ caps.max_adj = ptp->info->max_adj;
+ caps.n_alarm = ptp->info->n_alarm;
+ caps.n_ext_ts = ptp->info->n_ext_ts;
+ caps.n_per_out = ptp->info->n_per_out;
+ caps.pps = ptp->info->pps;
+ err = copy_to_user((void __user *)arg, &caps, sizeof(caps));
+ break;
+
+ case PTP_EXTTS_REQUEST:
+ if (copy_from_user(&req.extts, (void __user *)arg,
+ sizeof(req.extts))) {
+ err = -EFAULT;
+ break;
+ }
+ if (req.extts.index >= ops->n_ext_ts) {
+ err = -EINVAL;
+ break;
+ }
+ req.type = PTP_CLK_REQ_EXTTS;
+ enable = req.extts.flags & PTP_ENABLE_FEATURE ? 1 : 0;
+ err = ops->enable(priv, &req, enable);
+ break;
+
+ case PTP_PEROUT_REQUEST:
+ if (copy_from_user(&req.perout, (void __user *)arg,
+ sizeof(req.perout))) {
+ err = -EFAULT;
+ break;
+ }
+ if (req.perout.index >= ops->n_per_out) {
+ err = -EINVAL;
+ break;
+ }
+ req.type = PTP_CLK_REQ_PEROUT;
+ enable = req.perout.period.sec || req.perout.period.nsec;
+ err = ops->enable(priv, &req, enable);
+ break;
+
+ case PTP_ENABLE_PPS:
+ if (!capable(CAP_SYS_TIME))
+ return -EPERM;
+ req.type = PTP_CLK_REQ_PPS;
+ enable = arg ? 1 : 0;
+ err = ops->enable(priv, &req, enable);
+ break;
+
+ default:
+ err = -ENOTTY;
+ break;
+ }
+ return err;
+}
+
+unsigned int ptp_poll(void *priv, struct file *fp, poll_table *wait)
+{
+ struct ptp_clock *ptp = priv;
+
+ poll_wait(fp, &ptp->tsev_wq, wait);
+
+ return queue_cnt(&ptp->tsevq) ? POLLIN : 0;
+}
+
+ssize_t ptp_read(void *priv, uint flags, char __user *buf, size_t cnt)
+{
+ struct ptp_clock *ptp = priv;
+ struct timestamp_event_queue *queue = &ptp->tsevq;
+ struct ptp_extts_event *event;
+ size_t qcnt;
+
+ if (mutex_lock_interruptible(&ptp->tsevq_mux))
+ return -ERESTARTSYS;
+
+ cnt = cnt / sizeof(struct ptp_extts_event);
+
+ if (wait_event_interruptible(ptp->tsev_wq,
+ (qcnt = queue_cnt(&ptp->tsevq)))) {
+ mutex_unlock(&ptp->tsevq_mux);
+ return -ERESTARTSYS;
+ }
+
+ if (cnt > qcnt)
+ cnt = qcnt;
+
+ event = &queue->buf[queue->head];
+
+ if (copy_to_user(buf, event, cnt * sizeof(struct ptp_extts_event))) {
+ mutex_unlock(&ptp->tsevq_mux);
+ return -EFAULT;
+ }
+ queue->head = (queue->head + cnt) % PTP_MAX_TIMESTAMPS;
+
+ mutex_unlock(&ptp->tsevq_mux);
+
+ return cnt * sizeof(struct ptp_extts_event);
+}
+
+int ptp_release(void *priv)
+{
+ return 0;
+}
diff --git a/drivers/ptp/ptp_clock.c b/drivers/ptp/ptp_clock.c
new file mode 100644
index 0000000..f5c9355
--- /dev/null
+++ b/drivers/ptp/ptp_clock.c
@@ -0,0 +1,317 @@
+/*
+ * PTP 1588 clock support
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/bitops.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/posix-clock.h>
+#include <linux/pps_kernel.h>
+#include <linux/slab.h>
+#include <linux/syscalls.h>
+#include <linux/uaccess.h>
+
+#include <ptp_private.h>
+
+#define PTP_MAX_ALARMS 4
+#define PTP_MAX_CLOCKS (MAX_CLOCKS/2)
+#define PTP_PPS_DEFAULTS (PPS_CAPTUREASSERT | PPS_OFFSETASSERT)
+#define PTP_PPS_EVENT PPS_CAPTUREASSERT
+#define PTP_PPS_MODE (PTP_PPS_DEFAULTS | PPS_CANWAIT | PPS_TSFMT_TSPEC)
+
+/* private globals */
+
+extern struct device_attribute ptp_dev_attrs[]; /* see ptp_sysfs.c */
+static dev_t ptp_devt;
+static struct class *ptp_class;
+
+static DECLARE_BITMAP(ptp_clocks_map, PTP_MAX_CLOCKS);
+static DEFINE_MUTEX(ptp_clocks_mutex); /* protects 'ptp_clocks_map' */
+
+/* time stamp event queue operations */
+
+static inline int queue_free(struct timestamp_event_queue *q)
+{
+ return PTP_MAX_TIMESTAMPS - queue_cnt(q) - 1;
+}
+
+static void enqueue_external_timestamp(struct timestamp_event_queue *queue,
+ struct ptp_clock_event *src)
+{
+ struct ptp_extts_event *dst;
+ u32 remainder;
+
+ dst = &queue->buf[queue->tail];
+
+ dst->index = src->index;
+ dst->t.sec = div_u64_rem(src->timestamp, 1000000000, &remainder);
+ dst->t.nsec = remainder;
+
+ if (!queue_free(queue))
+ queue->overflow++;
+
+ queue->tail = (queue->tail + 1) % PTP_MAX_TIMESTAMPS;
+}
+
+/* posix clock implementation */
+
+static int ptp_clock_getres(void *priv, struct timespec *tp)
+{
+ return 1; /* always round timer functions to one nanosecond */
+}
+
+static int ptp_clock_settime(void *priv, const struct timespec *tp)
+{
+ struct ptp_clock *ptp = priv;
+ return ptp->info->settime(ptp->info->priv, tp);
+}
+
+static int ptp_clock_gettime(void *priv, struct timespec *tp)
+{
+ struct ptp_clock *ptp = priv;
+ return ptp->info->gettime(ptp->info->priv, tp);
+}
+
+static int ptp_clock_adjtime(void *priv, struct timex *tx)
+{
+ struct ptp_clock *ptp = priv;
+ struct ptp_clock_info *ops;
+ int err = -EOPNOTSUPP;
+
+ ops = ptp->info;
+
+ if (tx->modes & ADJ_SETOFFSET) {
+ struct timespec ts;
+ ktime_t kt;
+ s64 delta;
+
+ /* Validate the delta value. */
+ if (tx->time.tv_sec && tx->time.tv_usec < 0)
+ return -EINVAL;
+
+ if (tx->modes & ADJ_NANO) {
+ ts.tv_sec = tx->time.tv_sec;
+ ts.tv_nsec = tx->time.tv_usec;
+ kt = timespec_to_ktime(ts);
+ } else
+ kt = timeval_to_ktime(tx->time);
+
+ delta = ktime_to_ns(kt);
+
+ err = ops->adjtime(ops->priv, delta);
+
+ } else if (tx->modes & ADJ_FREQUENCY) {
+ s64 ppb = 1 + tx->freq;
+ ppb *= 125;
+ ppb >>= 13;
+ err = ops->adjfreq(ops->priv, (s32)ppb);
+ }
+
+ return err;
+}
+
+static struct posix_clock_operations ptp_cops = {
+ .owner = THIS_MODULE,
+ .clock_adjtime = ptp_clock_adjtime,
+ .clock_gettime = ptp_clock_gettime,
+ .clock_getres = ptp_clock_getres,
+ .clock_settime = ptp_clock_settime,
+ .fops = {
+ .open = ptp_open,
+ .release = ptp_release,
+ .ioctl = ptp_ioctl,
+ .read = ptp_read,
+ .poll = ptp_poll,
+ },
+};
+
+/* public interface */
+
+struct ptp_clock *ptp_clock_register(struct ptp_clock_info *info)
+{
+ struct ptp_clock *ptp;
+ int err = 0, index, major = MAJOR(ptp_devt);
+
+ if (info->n_alarm > PTP_MAX_ALARMS)
+ return ERR_PTR(-EINVAL);
+
+ /* Find a free clock slot and reserve it. */
+ err = -EBUSY;
+ mutex_lock(&ptp_clocks_mutex);
+ index = find_first_zero_bit(ptp_clocks_map, PTP_MAX_CLOCKS);
+ if (index < PTP_MAX_CLOCKS)
+ set_bit(index, ptp_clocks_map);
+ else
+ goto no_slot;
+
+ /* Initialize a clock structure. */
+ err = -ENOMEM;
+ ptp = kzalloc(sizeof(struct ptp_clock), GFP_KERNEL);
+ if (ptp == NULL)
+ goto no_memory;
+
+ ptp->info = info;
+ ptp->devid = MKDEV(major, index);
+ ptp->index = index;
+ mutex_init(&ptp->tsevq_mux);
+ init_waitqueue_head(&ptp->tsev_wq);
+
+ /* Create a new device in our class. */
+ ptp->dev = device_create(ptp_class, NULL, ptp->devid, ptp,
+ "ptp%d", ptp->index);
+ if (IS_ERR(ptp->dev))
+ goto no_device;
+
+ dev_set_drvdata(ptp->dev, ptp);
+
+ err = ptp_populate_sysfs(ptp);
+ if (err)
+ goto no_sysfs;
+
+ /* Register a new PPS source. */
+ if (info->pps) {
+ struct pps_source_info pps;
+ memset(&pps, 0, sizeof(pps));
+ snprintf(pps.name, PPS_MAX_NAME_LEN, "ptp%d", index);
+ pps.mode = PTP_PPS_MODE;
+ pps.owner = info->owner;
+ err = pps_register_source(&pps, PTP_PPS_DEFAULTS);
+ if (err < 0) {
+ pr_err("failed to register pps source\n");
+ goto no_pps;
+ } else
+ ptp->pps_source = err;
+ }
+
+ /* Create a posix clock. */
+ ptp->clk = posix_clock_create(&ptp_cops, ptp->devid, ptp);
+ if (IS_ERR(ptp->clk)) {
+ pr_err("failed to create posix clock\n");
+ err = PTR_ERR(ptp->clk);
+ goto no_clock;
+ }
+
+ mutex_unlock(&ptp_clocks_mutex);
+ return ptp;
+
+no_clock:
+ if (ptp->info->pps)
+ pps_unregister_source(ptp->pps_source);
+no_pps:
+ ptp_cleanup_sysfs(ptp);
+no_sysfs:
+ device_destroy(ptp_class, ptp->devid);
+no_device:
+ mutex_destroy(&ptp->tsevq_mux);
+ kfree(ptp);
+no_memory:
+ clear_bit(index, ptp_clocks_map);
+no_slot:
+ mutex_unlock(&ptp_clocks_mutex);
+ return ERR_PTR(err);
+}
+EXPORT_SYMBOL(ptp_clock_register);
+
+int ptp_clock_unregister(struct ptp_clock *ptp)
+{
+ /* Release the clock's resources. */
+ posix_clock_destroy(ptp->clk);
+ if (ptp->info->pps)
+ pps_unregister_source(ptp->pps_source);
+ ptp_cleanup_sysfs(ptp);
+ device_destroy(ptp_class, ptp->devid);
+ mutex_destroy(&ptp->tsevq_mux);
+
+ /* Remove the clock from the bit map. */
+ mutex_lock(&ptp_clocks_mutex);
+ clear_bit(ptp->index, ptp_clocks_map);
+ mutex_unlock(&ptp_clocks_mutex);
+
+ kfree(ptp);
+
+ return 0;
+}
+EXPORT_SYMBOL(ptp_clock_unregister);
+
+void ptp_clock_event(struct ptp_clock *ptp, struct ptp_clock_event *event)
+{
+ struct timespec ts;
+ struct pps_ktime pps_ts;
+
+ switch (event->type) {
+
+ case PTP_CLOCK_ALARM:
+ break;
+
+ case PTP_CLOCK_EXTTS:
+ enqueue_external_timestamp(&ptp->tsevq, event);
+ wake_up_interruptible(&ptp->tsev_wq);
+ break;
+
+ case PTP_CLOCK_PPS:
+ getnstimeofday(&ts);
+ pps_ts.sec = ts.tv_sec;
+ pps_ts.nsec = ts.tv_nsec;
+ pps_event(ptp->pps_source, &pps_ts, PTP_PPS_EVENT, NULL);
+ break;
+ }
+}
+EXPORT_SYMBOL(ptp_clock_event);
+
+/* module operations */
+
+static void __exit ptp_exit(void)
+{
+ class_destroy(ptp_class);
+ unregister_chrdev_region(ptp_devt, PTP_MAX_CLOCKS);
+}
+
+static int __init ptp_init(void)
+{
+ int err;
+
+ ptp_class = class_create(THIS_MODULE, "ptp");
+ if (IS_ERR(ptp_class)) {
+ pr_err("ptp: failed to allocate class\n");
+ return PTR_ERR(ptp_class);
+ }
+
+ err = alloc_chrdev_region(&ptp_devt, 0, PTP_MAX_CLOCKS, "ptp");
+ if (err < 0) {
+ pr_err("ptp: failed to allocate device region\n");
+ goto no_region;
+ }
+
+ ptp_class->dev_attrs = ptp_dev_attrs;
+ pr_info("PTP clock support registered\n");
+ return 0;
+
+no_region:
+ class_destroy(ptp_class);
+ return err;
+}
+
+subsys_initcall(ptp_init);
+module_exit(ptp_exit);
+
+MODULE_AUTHOR("Richard Cochran <[email protected]>");
+MODULE_DESCRIPTION("PTP clocks support");
+MODULE_LICENSE("GPL");
diff --git a/drivers/ptp/ptp_private.h b/drivers/ptp/ptp_private.h
new file mode 100644
index 0000000..6408655
--- /dev/null
+++ b/drivers/ptp/ptp_private.h
@@ -0,0 +1,68 @@
+/*
+ * PTP 1588 clock support - private declarations for the core module.
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#ifndef _PTP_PRIVATE_H_
+#define _PTP_PRIVATE_H_
+
+#include <linux/cdev.h>
+#include <linux/mutex.h>
+#include <linux/posix-clock.h>
+#include <linux/ptp_clock.h>
+#include <linux/ptp_clock_kernel.h>
+#include <linux/time.h>
+
+#define PTP_MAX_TIMESTAMPS 128
+
+struct timestamp_event_queue {
+ struct ptp_extts_event buf[PTP_MAX_TIMESTAMPS];
+ int head;
+ int tail;
+ int overflow;
+};
+
+struct ptp_clock {
+ struct list_head list;
+ struct posix_clock *clk;
+ struct device *dev;
+ struct ptp_clock_info *info;
+ dev_t devid;
+ int index; /* index into clocks.map */
+ int pps_source;
+ struct timestamp_event_queue tsevq; /* simple fifo for time stamps */
+ struct mutex tsevq_mux; /* one process at a time reading the fifo */
+ wait_queue_head_t tsev_wq;
+ struct cdev cdev;
+};
+
+static inline int queue_cnt(struct timestamp_event_queue *q)
+{
+ int cnt = q->tail - q->head;
+ return cnt < 0 ? PTP_MAX_TIMESTAMPS + cnt : cnt;
+}
+
+int ptp_open(void *priv, fmode_t fmode);
+int ptp_release(void *priv);
+long ptp_ioctl(void *pc_private, unsigned int cmd, unsigned long arg);
+ssize_t ptp_read(void *priv, uint flags, char __user *buf, size_t cnt);
+unsigned int ptp_poll(void *priv, struct file *fp, poll_table *wait);
+
+int ptp_cleanup_sysfs(struct ptp_clock *ptp);
+int ptp_populate_sysfs(struct ptp_clock *ptp);
+
+#endif
diff --git a/drivers/ptp/ptp_sysfs.c b/drivers/ptp/ptp_sysfs.c
new file mode 100644
index 0000000..b1d1c56
--- /dev/null
+++ b/drivers/ptp/ptp_sysfs.c
@@ -0,0 +1,230 @@
+/*
+ * PTP 1588 clock support - sysfs interface.
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/capability.h>
+#include <linux/device.h>
+
+#include <ptp_private.h>
+
+static ssize_t clock_name_show(struct device *dev,
+ struct device_attribute *attr, char *page)
+{
+ struct ptp_clock *ptp = dev_get_drvdata(dev);
+ return snprintf(page, PAGE_SIZE-1, "%s\n", ptp->info->name);
+}
+
+#define PTP_SHOW_INT(name) \
+static ssize_t name##_show(struct device *dev, \
+ struct device_attribute *attr, char *page) \
+{ \
+ struct ptp_clock *ptp = dev_get_drvdata(dev); \
+ return snprintf(page, PAGE_SIZE-1, "%d\n", ptp->info->name); \
+}
+
+PTP_SHOW_INT(max_adj);
+PTP_SHOW_INT(n_alarm);
+PTP_SHOW_INT(n_ext_ts);
+PTP_SHOW_INT(n_per_out);
+PTP_SHOW_INT(pps);
+
+#define PTP_RO_ATTR(_var, _name) { \
+ .attr = { .name = __stringify(_name), .mode = 0444 }, \
+ .show = _var##_show, \
+}
+
+struct device_attribute ptp_dev_attrs[] = {
+ PTP_RO_ATTR(clock_name, clock_name),
+ PTP_RO_ATTR(max_adj, max_adjustment),
+ PTP_RO_ATTR(n_alarm, n_alarms),
+ PTP_RO_ATTR(n_ext_ts, n_external_timestamps),
+ PTP_RO_ATTR(n_per_out, n_periodic_outputs),
+ PTP_RO_ATTR(pps, pps_available),
+ __ATTR_NULL,
+};
+
+static ssize_t extts_enable_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct ptp_clock *ptp = dev_get_drvdata(dev);
+ struct ptp_clock_info *ops = ptp->info;
+ void *priv = ops->priv;
+ struct ptp_clock_request req = { .type = PTP_CLK_REQ_EXTTS };
+ int cnt, enable;
+ int err = -EINVAL;
+
+ cnt = sscanf(buf, "%u %d", &req.extts.index, &enable);
+ if (cnt != 2)
+ goto out;
+ if (req.extts.index >= ops->n_ext_ts)
+ goto out;
+
+ err = ops->enable(priv, &req, enable ? 1 : 0);
+ if (err)
+ goto out;
+
+ return count;
+out:
+ return err;
+}
+
+static ssize_t extts_fifo_show(struct device *dev,
+ struct device_attribute *attr, char *page)
+{
+ struct ptp_clock *ptp = dev_get_drvdata(dev);
+ struct timestamp_event_queue *queue = &ptp->tsevq;
+ struct ptp_extts_event *event;
+ size_t qcnt;
+ int cnt;
+
+ if (mutex_lock_interruptible(&ptp->tsevq_mux))
+ return -ERESTARTSYS;
+
+ qcnt = queue_cnt(&ptp->tsevq);
+ if (!qcnt) {
+ mutex_unlock(&ptp->tsevq_mux);
+ return 0;
+ }
+ event = &queue->buf[queue->head];
+
+ cnt = snprintf(page, PAGE_SIZE, "%u %lld %u\n",
+ event->index, event->t.sec, event->t.nsec);
+
+ queue->head = (1 + queue->head) % PTP_MAX_TIMESTAMPS;
+
+ mutex_unlock(&ptp->tsevq_mux);
+
+ return cnt;
+}
+
+static ssize_t period_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct ptp_clock *ptp = dev_get_drvdata(dev);
+ struct ptp_clock_info *ops = ptp->info;
+ void *priv = ops->priv;
+ struct ptp_clock_request req = { .type = PTP_CLK_REQ_PEROUT };
+ int cnt, enable, err = -EINVAL;
+
+ cnt = sscanf(buf, "%u %lld %u %lld %u", &req.perout.index,
+ &req.perout.start.sec, &req.perout.start.nsec,
+ &req.perout.period.sec, &req.perout.period.nsec);
+ if (cnt != 5)
+ goto out;
+ if (req.perout.index >= ops->n_per_out)
+ goto out;
+
+ enable = req.perout.period.sec || req.perout.period.nsec;
+ err = ops->enable(priv, &req, enable);
+ if (err)
+ goto out;
+
+ return count;
+out:
+ return err;
+}
+
+static ssize_t pps_enable_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct ptp_clock *ptp = dev_get_drvdata(dev);
+ struct ptp_clock_info *ops = ptp->info;
+ void *priv = ops->priv;
+ struct ptp_clock_request req = { .type = PTP_CLK_REQ_PPS };
+ int cnt, enable;
+ int err = -EINVAL;
+
+ if (!capable(CAP_SYS_TIME))
+ return -EPERM;
+
+ cnt = sscanf(buf, "%d", &enable);
+ if (cnt != 1)
+ goto out;
+
+ err = ops->enable(priv, &req, enable ? 1 : 0);
+ if (err)
+ goto out;
+
+ return count;
+out:
+ return err;
+}
+
+static DEVICE_ATTR(extts_enable, 0220, NULL, extts_enable_store);
+static DEVICE_ATTR(fifo, 0444, extts_fifo_show, NULL);
+static DEVICE_ATTR(period, 0220, NULL, period_store);
+static DEVICE_ATTR(pps_enable, 0220, NULL, pps_enable_store);
+
+int ptp_cleanup_sysfs(struct ptp_clock *ptp)
+{
+ struct device *dev = ptp->dev;
+ struct ptp_clock_info *info = ptp->info;
+
+ if (info->n_ext_ts) {
+ device_remove_file(dev, &dev_attr_extts_enable);
+ device_remove_file(dev, &dev_attr_fifo);
+ }
+ if (info->n_per_out)
+ device_remove_file(dev, &dev_attr_period);
+
+ if (info->pps)
+ device_remove_file(dev, &dev_attr_pps_enable);
+
+ return 0;
+}
+
+int ptp_populate_sysfs(struct ptp_clock *ptp)
+{
+ struct device *dev = ptp->dev;
+ struct ptp_clock_info *info = ptp->info;
+ int err;
+
+ if (info->n_ext_ts) {
+ err = device_create_file(dev, &dev_attr_extts_enable);
+ if (err)
+ goto out1;
+ err = device_create_file(dev, &dev_attr_fifo);
+ if (err)
+ goto out2;
+ }
+ if (info->n_per_out) {
+ err = device_create_file(dev, &dev_attr_period);
+ if (err)
+ goto out3;
+ }
+ if (info->pps) {
+ err = device_create_file(dev, &dev_attr_pps_enable);
+ if (err)
+ goto out4;
+ }
+ return 0;
+out4:
+ if (info->n_per_out)
+ device_remove_file(dev, &dev_attr_period);
+out3:
+ if (info->n_ext_ts)
+ device_remove_file(dev, &dev_attr_fifo);
+out2:
+ if (info->n_ext_ts)
+ device_remove_file(dev, &dev_attr_extts_enable);
+out1:
+ return err;
+}
diff --git a/include/linux/Kbuild b/include/linux/Kbuild
index 97319a8..c9dfbc7 100644
--- a/include/linux/Kbuild
+++ b/include/linux/Kbuild
@@ -295,6 +295,7 @@ header-y += ppp-comp.h
header-y += ppp_defs.h
header-y += pps.h
header-y += prctl.h
+header-y += ptp_clock.h
header-y += ptrace.h
header-y += qnx4_fs.h
header-y += qnxtypes.h
diff --git a/include/linux/ptp_clock.h b/include/linux/ptp_clock.h
new file mode 100644
index 0000000..b4ef2cb
--- /dev/null
+++ b/include/linux/ptp_clock.h
@@ -0,0 +1,79 @@
+/*
+ * PTP 1588 clock support - user space interface
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef _PTP_CLOCK_H_
+#define _PTP_CLOCK_H_
+
+#include <linux/ioctl.h>
+#include <linux/types.h>
+
+/* PTP_xxx bits, for the flags field within the request structures. */
+#define PTP_ENABLE_FEATURE (1<<0)
+#define PTP_RISING_EDGE (1<<1)
+#define PTP_FALLING_EDGE (1<<2)
+
+/*
+ * struct ptp_clock_time - represents a time value
+ *
+ * The sign of the seconds field applies to the whole value. The
+ * nanoseconds field is always unsigned. The reserved field is
+ * included for sub-nanosecond resolution, should the demand for
+ * this ever appear.
+ *
+ */
+struct ptp_clock_time {
+ __s64 sec; /* seconds */
+ __u32 nsec; /* nanoseconds */
+ __u32 reserved;
+};
+
+struct ptp_clock_caps {
+ int max_adj; /* Maximum frequency adjustment in parts per billon. */
+ int n_alarm; /* Number of programmable alarms. */
+ int n_ext_ts; /* Number of external time stamp channels. */
+ int n_per_out; /* Number of programmable periodic signals. */
+ int pps; /* Whether the clock supports a PPS callback. */
+};
+
+struct ptp_extts_request {
+ unsigned int index; /* Which channel to configure. */
+ unsigned int flags; /* Bit field for PTP_xxx flags. */
+};
+
+struct ptp_perout_request {
+ struct ptp_clock_time start; /* Absolute start time. */
+ struct ptp_clock_time period; /* Desired period, zero means disable. */
+ unsigned int index; /* Which channel to configure. */
+ unsigned int flags; /* Reserved for future use. */
+};
+
+#define PTP_CLK_MAGIC '='
+
+#define PTP_CLOCK_GETCAPS _IOR(PTP_CLK_MAGIC, 1, struct ptp_clock_caps)
+#define PTP_EXTTS_REQUEST _IOW(PTP_CLK_MAGIC, 2, struct ptp_extts_request)
+#define PTP_PEROUT_REQUEST _IOW(PTP_CLK_MAGIC, 3, struct ptp_perout_request)
+#define PTP_ENABLE_PPS _IOW(PTP_CLK_MAGIC, 4, int)
+
+struct ptp_extts_event {
+ struct ptp_clock_time t; /* Time event occured. */
+ unsigned int index; /* Which channel produced the event. */
+};
+
+#endif
diff --git a/include/linux/ptp_clock_kernel.h b/include/linux/ptp_clock_kernel.h
new file mode 100644
index 0000000..165e391
--- /dev/null
+++ b/include/linux/ptp_clock_kernel.h
@@ -0,0 +1,139 @@
+/*
+ * PTP 1588 clock support
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef _PTP_CLOCK_KERNEL_H_
+#define _PTP_CLOCK_KERNEL_H_
+
+#include <linux/ptp_clock.h>
+
+
+struct ptp_clock_request {
+ enum {
+ PTP_CLK_REQ_EXTTS,
+ PTP_CLK_REQ_PEROUT,
+ PTP_CLK_REQ_PPS,
+ } type;
+ union {
+ struct ptp_extts_request extts;
+ struct ptp_perout_request perout;
+ };
+};
+
+/**
+ * struct ptp_clock_info - decribes a PTP hardware clock
+ *
+ * @owner: The clock driver should set to THIS_MODULE.
+ * @name: A short name to identify the clock.
+ * @max_adj: The maximum possible frequency adjustment, in parts per billon.
+ * @n_alarm: The number of programmable alarms.
+ * @n_ext_ts: The number of external time stamp channels.
+ * @n_per_out: The number of programmable periodic signals.
+ * @pps: Indicates whether the clock supports a PPS callback.
+ * @priv: Passed to the clock operations, for the driver's private use.
+ *
+ * clock operations
+ *
+ * @adjfreq: Adjusts the frequency of the hardware clock.
+ * parameter delta: Desired period change in parts per billion.
+ *
+ * @adjtime: Shifts the time of the hardware clock.
+ * parameter delta: Desired change in nanoseconds.
+ *
+ * @gettime: Reads the current time from the hardware clock.
+ * parameter ts: Holds the result.
+ *
+ * @settime: Set the current time on the hardware clock.
+ * parameter ts: Time value to set.
+ *
+ * @enable: Request driver to enable or disable an ancillary feature.
+ * parameter request: Desired resource to enable or disable.
+ * parameter on: Caller passes one to enable or zero to disable.
+ *
+ * The callbacks must all return zero on success, non-zero otherwise.
+ */
+
+struct ptp_clock_info {
+ struct module *owner;
+ char name[16];
+ s32 max_adj;
+ int n_alarm;
+ int n_ext_ts;
+ int n_per_out;
+ int pps;
+ void *priv;
+ int (*adjfreq)(void *priv, s32 delta);
+ int (*adjtime)(void *priv, s64 delta);
+ int (*gettime)(void *priv, struct timespec *ts);
+ int (*settime)(void *priv, const struct timespec *ts);
+ int (*enable)(void *priv, struct ptp_clock_request *request, int on);
+};
+
+struct ptp_clock;
+
+/**
+ * ptp_clock_register() - register a PTP hardware clock driver
+ *
+ * @info: Structure describing the new clock.
+ */
+
+extern struct ptp_clock *ptp_clock_register(struct ptp_clock_info *info);
+
+/**
+ * ptp_clock_unregister() - unregister a PTP hardware clock driver
+ *
+ * @ptp: The clock to remove from service.
+ */
+
+extern int ptp_clock_unregister(struct ptp_clock *ptp);
+
+
+enum ptp_clock_events {
+ PTP_CLOCK_ALARM,
+ PTP_CLOCK_EXTTS,
+ PTP_CLOCK_PPS,
+};
+
+/**
+ * struct ptp_clock_event - decribes a PTP hardware clock event
+ *
+ * @type: One of the ptp_clock_events enumeration values.
+ * @index: Identifies the source of the event.
+ * @timestamp: When the event occured.
+ */
+
+struct ptp_clock_event {
+ int type;
+ int index;
+ u64 timestamp;
+};
+
+/**
+ * ptp_clock_event() - notify the PTP layer about an event
+ *
+ * This function should only be called from interrupt context.
+ *
+ * @ptp: The clock obtained from ptp_clock_register().
+ * @event: Message structure describing the event.
+ */
+
+extern void ptp_clock_event(struct ptp_clock *ptp,
+ struct ptp_clock_event *event);
+
+#endif
--
1.7.0.4
The eTSEC includes a PTP clock with quite a few features. This patch adds
support for the basic clock adjustment functions, plus two external time
stamps, one alarm, and the PPS callback.
Signed-off-by: Richard Cochran <[email protected]>
---
Documentation/powerpc/dts-bindings/fsl/tsec.txt | 57 +++
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 +
drivers/net/Makefile | 1 +
drivers/net/gianfar_ptp.c | 444 +++++++++++++++++++++++
drivers/net/gianfar_ptp_reg.h | 113 ++++++
drivers/ptp/Kconfig | 13 +
9 files changed, 684 insertions(+), 0 deletions(-)
create mode 100644 drivers/net/gianfar_ptp.c
create mode 100644 drivers/net/gianfar_ptp_reg.h
diff --git a/Documentation/powerpc/dts-bindings/fsl/tsec.txt b/Documentation/powerpc/dts-bindings/fsl/tsec.txt
index edb7ae1..f6edbb8 100644
--- a/Documentation/powerpc/dts-bindings/fsl/tsec.txt
+++ b/Documentation/powerpc/dts-bindings/fsl/tsec.txt
@@ -74,3 +74,60 @@ Example:
interrupt-parent = <&mpic>;
phy-handle = <&phy0>
};
+
+* Gianfar PTP clock nodes
+
+General Properties:
+
+ - compatible Should be "fsl,etsec-ptp"
+ - reg Offset and length of the register set for the device
+ - interrupts There should be at least two interrupts. Some devices
+ have as many as four PTP related interrupts.
+
+Clock Properties:
+
+ - tclk-period Timer reference clock period in nanoseconds.
+ - tmr-prsc Prescaler, divides the output clock.
+ - tmr-add Frequency compensation value.
+ - cksel 0= external clock, 1= eTSEC system clock, 3= RTC clock input.
+ Currently the driver only supports choice "1".
+ - tmr-fiper1 Fixed interval period pulse generator.
+ - tmr-fiper2 Fixed interval period pulse generator.
+ - max-adj Maximum frequency adjustment in parts per billion.
+
+ These properties set the operational parameters for the PTP
+ clock. You must choose these carefully for the clock to work right.
+ Here is how to figure good values:
+
+ TimerOsc = system clock MHz
+ tclk_period = desired clock period nanoseconds
+ NominalFreq = 1000 / tclk_period MHz
+ FreqDivRatio = TimerOsc / NominalFreq (must be greater that 1.0)
+ tmr_add = ceil(2^32 / FreqDivRatio)
+ OutputClock = NominalFreq / tmr_prsc MHz
+ PulseWidth = 1 / OutputClock microseconds
+ FiperFreq1 = desired frequency in Hz
+ FiperDiv1 = 1000000 * OutputClock / FiperFreq1
+ tmr_fiper1 = tmr_prsc * tclk_period * FiperDiv1 - tclk_period
+ max_adj = 1000000000 * (FreqDivRatio - 1.0) - 1
+
+ The calculation for tmr_fiper2 is the same as for tmr_fiper1. The
+ driver expects that tmr_fiper1 will be correctly set to produce a 1
+ Pulse Per Second (PPS) signal, since this will be offered to the PPS
+ subsystem to synchronize the Linux clock.
+
+Example:
+
+ ptp_clock@24E00 {
+ compatible = "fsl,etsec-ptp";
+ reg = <0x24E00 0xB0>;
+ interrupts = <12 0x8 13 0x8>;
+ interrupt-parent = < &ipic >;
+ tclk-period = <10>;
+ tmr-prsc = <100>;
+ tmr-add = <0x999999A4>;
+ cksel = <0x1>;
+ tmr-fiper1 = <0x3B9AC9F6>;
+ tmr-fiper2 = <0x00018696>;
+ max-adj = <659999998>;
+ };
diff --git a/arch/powerpc/boot/dts/mpc8313erdb.dts b/arch/powerpc/boot/dts/mpc8313erdb.dts
index 183f2aa..85a7eaa 100644
--- a/arch/powerpc/boot/dts/mpc8313erdb.dts
+++ b/arch/powerpc/boot/dts/mpc8313erdb.dts
@@ -208,6 +208,20 @@
sleep = <&pmc 0x00300000>;
};
+ ptp_clock@24E00 {
+ compatible = "fsl,etsec-ptp";
+ reg = <0x24E00 0xB0>;
+ interrupts = <12 0x8 13 0x8>;
+ interrupt-parent = < &ipic >;
+ tclk-period = <10>;
+ tmr-prsc = <100>;
+ tmr-add = <0x999999A4>;
+ cksel = <0x1>;
+ tmr-fiper1 = <0x3B9AC9F6>;
+ tmr-fiper2 = <0x00018696>;
+ max-adj = <659999998>;
+ };
+
enet0: ethernet@24000 {
#address-cells = <1>;
#size-cells = <1>;
diff --git a/arch/powerpc/boot/dts/mpc8572ds.dts b/arch/powerpc/boot/dts/mpc8572ds.dts
index cafc128..74208cd 100644
--- a/arch/powerpc/boot/dts/mpc8572ds.dts
+++ b/arch/powerpc/boot/dts/mpc8572ds.dts
@@ -324,6 +324,20 @@
};
};
+ ptp_clock@24E00 {
+ compatible = "fsl,etsec-ptp";
+ reg = <0x24E00 0xB0>;
+ interrupts = <68 2 69 2 70 2 71 2>;
+ interrupt-parent = < &mpic >;
+ tclk-period = <5>;
+ tmr-prsc = <200>;
+ tmr-add = <0xAAAAAAAB>;
+ cksel = <1>;
+ tmr-fiper1 = <0x3B9AC9FB>;
+ tmr-fiper2 = <0x3B9AC9FB>;
+ max-adj = <499999999>;
+ };
+
enet0: ethernet@24000 {
#address-cells = <1>;
#size-cells = <1>;
diff --git a/arch/powerpc/boot/dts/p2020ds.dts b/arch/powerpc/boot/dts/p2020ds.dts
index 1101914..39d73bb 100644
--- a/arch/powerpc/boot/dts/p2020ds.dts
+++ b/arch/powerpc/boot/dts/p2020ds.dts
@@ -336,6 +336,20 @@
phy_type = "ulpi";
};
+ ptp_clock@24E00 {
+ compatible = "fsl,etsec-ptp";
+ reg = <0x24E00 0xB0>;
+ interrupts = <68 2 69 2 70 2>;
+ interrupt-parent = < &mpic >;
+ tclk-period = <5>;
+ tmr-prsc = <200>;
+ tmr-add = <0xCCCCCCCD>;
+ cksel = <1>;
+ tmr-fiper1 = <0x3B9AC9FB>;
+ tmr-fiper2 = <0x0001869B>;
+ max-adj = <249999999>;
+ };
+
enet0: ethernet@24000 {
#address-cells = <1>;
#size-cells = <1>;
diff --git a/arch/powerpc/boot/dts/p2020rdb.dts b/arch/powerpc/boot/dts/p2020rdb.dts
index da4cb0d..5498fb9 100644
--- a/arch/powerpc/boot/dts/p2020rdb.dts
+++ b/arch/powerpc/boot/dts/p2020rdb.dts
@@ -396,6 +396,20 @@
phy_type = "ulpi";
};
+ ptp_clock@24E00 {
+ compatible = "fsl,etsec-ptp";
+ reg = <0x24E00 0xB0>;
+ interrupts = <68 2 69 2 70 2>;
+ interrupt-parent = < &mpic >;
+ tclk-period = <5>;
+ tmr-prsc = <200>;
+ tmr-add = <0xCCCCCCCD>;
+ cksel = <1>;
+ tmr-fiper1 = <0x3B9AC9FB>;
+ tmr-fiper2 = <0x0001869B>;
+ max-adj = <249999999>;
+ };
+
enet0: ethernet@24000 {
#address-cells = <1>;
#size-cells = <1>;
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index b90738d..c303f5f 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -31,6 +31,7 @@ obj-$(CONFIG_ATL2) += atlx/
obj-$(CONFIG_ATL1E) += atl1e/
obj-$(CONFIG_ATL1C) += atl1c/
obj-$(CONFIG_GIANFAR) += gianfar_driver.o
+obj-$(CONFIG_PTP_1588_CLOCK_GIANFAR) += gianfar_ptp.o
obj-$(CONFIG_TEHUTI) += tehuti.o
obj-$(CONFIG_ENIC) += enic/
obj-$(CONFIG_JME) += jme.o
diff --git a/drivers/net/gianfar_ptp.c b/drivers/net/gianfar_ptp.c
new file mode 100644
index 0000000..955aac4
--- /dev/null
+++ b/drivers/net/gianfar_ptp.c
@@ -0,0 +1,444 @@
+/*
+ * PTP 1588 clock using the eTSEC
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/device.h>
+#include <linux/hrtimer.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/timex.h>
+#include <linux/io.h>
+
+#include <linux/ptp_clock_kernel.h>
+
+#include "gianfar_ptp_reg.h"
+#include "gianfar.h"
+
+#define DRIVER "gianfar_ptp"
+#define N_ALARM 1 /* first alarm is used internally to reset fipers */
+#define N_EXT_TS 2
+#define REG_SIZE sizeof(struct gianfar_ptp_registers)
+
+struct etsects {
+ struct gianfar_ptp_registers *regs;
+ struct ptp_clock *clock;
+ int irq;
+ u64 alarm_interval; /* for periodic alarm */
+ u64 alarm_value;
+ u32 tclk_period; /* nanoseconds */
+ u32 tmr_prsc;
+ u32 tmr_add;
+ u32 cksel;
+ u32 tmr_fiper1;
+ u32 tmr_fiper2;
+};
+
+/* Private globals */
+static struct etsects the_clock;
+DEFINE_SPINLOCK(register_lock);
+
+/*
+ * Register access functions
+ */
+
+static u64 tmr_cnt_read(struct etsects *etsects)
+{
+ u64 ns;
+ u32 lo, hi;
+
+ lo = gfar_read(&etsects->regs->tmr_cnt_l);
+ hi = gfar_read(&etsects->regs->tmr_cnt_h);
+ ns = ((u64) hi) << 32;
+ ns |= lo;
+ return ns;
+}
+
+static void tmr_cnt_write(struct etsects *etsects, u64 ns)
+{
+ u32 hi = ns >> 32;
+ u32 lo = ns & 0xffffffff;
+
+ gfar_write(&etsects->regs->tmr_cnt_l, lo);
+ gfar_write(&etsects->regs->tmr_cnt_h, hi);
+}
+
+static void set_alarm(struct etsects *etsects)
+{
+ u64 ns;
+ u32 lo, hi;
+
+ ns = tmr_cnt_read(etsects) + 1500000000ULL;
+ ns = div_u64(ns, 1000000000UL) * 1000000000ULL;
+ ns -= etsects->tclk_period;
+ hi = ns >> 32;
+ lo = ns & 0xffffffff;
+ gfar_write(&etsects->regs->tmr_alarm1_l, lo);
+ gfar_write(&etsects->regs->tmr_alarm1_h, hi);
+}
+
+static void set_fipers(struct etsects *etsects)
+{
+ u32 tmr_ctrl = gfar_read(&etsects->regs->tmr_ctrl);
+
+ gfar_write(&etsects->regs->tmr_ctrl, tmr_ctrl & (~TE));
+ gfar_write(&etsects->regs->tmr_prsc, etsects->tmr_prsc);
+ gfar_write(&etsects->regs->tmr_fiper1, etsects->tmr_fiper1);
+ gfar_write(&etsects->regs->tmr_fiper2, etsects->tmr_fiper2);
+ set_alarm(etsects);
+ gfar_write(&etsects->regs->tmr_ctrl, tmr_ctrl|TE);
+}
+
+/*
+ * Interrupt service routine
+ */
+
+static irqreturn_t isr(int irq, void *priv)
+{
+ struct etsects *etsects = priv;
+ struct ptp_clock_event event;
+ u64 ns;
+ u32 ack = 0, lo, hi, mask, val;
+
+ val = gfar_read(&etsects->regs->tmr_tevent);
+
+ if (val & ETS1) {
+ ack |= ETS1;
+ hi = gfar_read(&etsects->regs->tmr_etts1_h);
+ lo = gfar_read(&etsects->regs->tmr_etts1_l);
+ event.type = PTP_CLOCK_EXTTS;
+ event.index = 0;
+ event.timestamp = ((u64) hi) << 32;
+ event.timestamp |= lo;
+ ptp_clock_event(etsects->clock, &event);
+ }
+
+ if (val & ETS2) {
+ ack |= ETS2;
+ hi = gfar_read(&etsects->regs->tmr_etts2_h);
+ lo = gfar_read(&etsects->regs->tmr_etts2_l);
+ event.type = PTP_CLOCK_EXTTS;
+ event.index = 1;
+ event.timestamp = ((u64) hi) << 32;
+ event.timestamp |= lo;
+ ptp_clock_event(etsects->clock, &event);
+ }
+
+ if (val & ALM2) {
+ ack |= ALM2;
+ if (etsects->alarm_value) {
+ event.type = PTP_CLOCK_ALARM;
+ event.index = 0;
+ event.timestamp = etsects->alarm_value;
+ ptp_clock_event(etsects->clock, &event);
+ }
+ if (etsects->alarm_interval) {
+ ns = etsects->alarm_value + etsects->alarm_interval;
+ hi = ns >> 32;
+ lo = ns & 0xffffffff;
+ spin_lock(®ister_lock);
+ gfar_write(&etsects->regs->tmr_alarm2_l, lo);
+ gfar_write(&etsects->regs->tmr_alarm2_h, hi);
+ spin_unlock(®ister_lock);
+ etsects->alarm_value = ns;
+ } else {
+ gfar_write(&etsects->regs->tmr_tevent, ALM2);
+ spin_lock(®ister_lock);
+ mask = gfar_read(&etsects->regs->tmr_temask);
+ mask &= ~ALM2EN;
+ gfar_write(&etsects->regs->tmr_temask, mask);
+ spin_unlock(®ister_lock);
+ etsects->alarm_value = 0;
+ etsects->alarm_interval = 0;
+ }
+ }
+
+ if (val & PP1) {
+ ack |= PP1;
+ event.type = PTP_CLOCK_PPS;
+ ptp_clock_event(etsects->clock, &event);
+ }
+
+ if (ack) {
+ gfar_write(&etsects->regs->tmr_tevent, ack);
+ return IRQ_HANDLED;
+ } else
+ return IRQ_NONE;
+}
+
+/*
+ * PTP clock operations
+ */
+
+static int ptp_gianfar_adjfreq(void *priv, s32 ppb)
+{
+ u64 adj;
+ u32 diff, tmr_add;
+ int neg_adj = 0;
+ struct etsects *etsects = priv;
+
+ if (ppb < 0) {
+ neg_adj = 1;
+ ppb = -ppb;
+ }
+ tmr_add = etsects->tmr_add;
+ adj = tmr_add;
+ adj *= ppb;
+ diff = div_u64(adj, 1000000000ULL);
+
+ tmr_add = neg_adj ? tmr_add - diff : tmr_add + diff;
+
+ gfar_write(&etsects->regs->tmr_add, tmr_add);
+
+ return 0;
+}
+
+static int ptp_gianfar_adjtime(void *priv, s64 delta)
+{
+ s64 now;
+ unsigned long flags;
+ struct etsects *etsects = priv;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ now = tmr_cnt_read(etsects);
+ now += delta;
+ tmr_cnt_write(etsects, now);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ set_fipers(etsects);
+
+ return 0;
+}
+
+static int ptp_gianfar_gettime(void *priv, struct timespec *ts)
+{
+ u64 ns;
+ u32 remainder;
+ unsigned long flags;
+ struct etsects *etsects = priv;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ ns = tmr_cnt_read(etsects);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ ts->tv_sec = div_u64_rem(ns, 1000000000, &remainder);
+ ts->tv_nsec = remainder;
+ return 0;
+}
+
+static int ptp_gianfar_settime(void *priv, const struct timespec *ts)
+{
+ u64 ns;
+ unsigned long flags;
+ struct etsects *etsects = priv;
+
+ ns = ts->tv_sec * 1000000000ULL;
+ ns += ts->tv_nsec;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ tmr_cnt_write(etsects, ns);
+ set_fipers(etsects);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ return 0;
+}
+
+static int ptp_gianfar_enable(void *priv, struct ptp_clock_request *rq, int on)
+{
+ struct etsects *etsects = priv;
+ unsigned long flags;
+ u32 bit, mask;
+
+ switch (rq->type) {
+ case PTP_CLK_REQ_EXTTS:
+ switch (rq->extts.index) {
+ case 0:
+ bit = ETS1EN;
+ break;
+ case 1:
+ bit = ETS2EN;
+ break;
+ default:
+ return -EINVAL;
+ }
+ spin_lock_irqsave(®ister_lock, flags);
+ mask = gfar_read(&etsects->regs->tmr_temask);
+ if (on)
+ mask |= bit;
+ else
+ mask &= ~bit;
+ gfar_write(&etsects->regs->tmr_temask, mask);
+ spin_unlock_irqrestore(®ister_lock, flags);
+ return 0;
+
+ case PTP_CLK_REQ_PPS:
+ spin_lock_irqsave(®ister_lock, flags);
+ mask = gfar_read(&etsects->regs->tmr_temask);
+ if (on)
+ mask |= PP1EN;
+ else
+ mask &= ~PP1EN;
+ gfar_write(&etsects->regs->tmr_temask, mask);
+ spin_unlock_irqrestore(®ister_lock, flags);
+ return 0;
+
+ default:
+ break;
+ }
+
+ return -EOPNOTSUPP;
+}
+
+static struct ptp_clock_info ptp_gianfar_caps = {
+ .owner = THIS_MODULE,
+ .name = "gianfar clock",
+ .max_adj = 512000,
+ .n_alarm = N_ALARM,
+ .n_ext_ts = N_EXT_TS,
+ .n_per_out = 0,
+ .pps = 1,
+ .priv = &the_clock,
+ .adjfreq = ptp_gianfar_adjfreq,
+ .adjtime = ptp_gianfar_adjtime,
+ .gettime = ptp_gianfar_gettime,
+ .settime = ptp_gianfar_settime,
+ .enable = ptp_gianfar_enable,
+};
+
+/* OF device tree */
+
+static int get_of_u32(struct device_node *node, char *str, u32 *val)
+{
+ int plen;
+ const u32 *prop = of_get_property(node, str, &plen);
+
+ if (!prop || plen != sizeof(*prop))
+ return -1;
+ *val = *prop;
+ return 0;
+}
+
+static int gianfar_ptp_probe(struct platform_device *dev,
+ const struct of_device_id *match)
+{
+ struct device_node *node = dev->dev.of_node;
+ struct etsects *etsects = &the_clock;
+ struct timespec now;
+ u32 tmr_ctrl;
+
+ if (get_of_u32(node, "tclk-period", &etsects->tclk_period) ||
+ get_of_u32(node, "tmr-prsc", &etsects->tmr_prsc) ||
+ get_of_u32(node, "tmr-add", &etsects->tmr_add) ||
+ get_of_u32(node, "cksel", &etsects->cksel) ||
+ get_of_u32(node, "tmr-fiper1", &etsects->tmr_fiper1) ||
+ get_of_u32(node, "tmr-fiper2", &etsects->tmr_fiper2) ||
+ get_of_u32(node, "max-adj", &ptp_gianfar_caps.max_adj)) {
+ pr_err("device tree node missing required elements\n");
+ return -ENODEV;
+ }
+
+ etsects->irq = irq_of_parse_and_map(node, 0);
+
+ if (etsects->irq == NO_IRQ) {
+ pr_err("irq not in device tree\n");
+ return -ENODEV;
+ }
+ if (request_irq(etsects->irq, isr, 0, DRIVER, etsects)) {
+ pr_err("request_irq failed\n");
+ return -ENODEV;
+ }
+ etsects->regs = of_iomap(node, 0);
+ if (!etsects->regs) {
+ pr_err("of_iomap ptp registers failed\n");
+ return -EINVAL;
+ }
+ getnstimeofday(&now);
+ ptp_gianfar_settime(etsects, &now);
+
+ tmr_ctrl =
+ (etsects->tclk_period & TCLK_PERIOD_MASK) << TCLK_PERIOD_SHIFT |
+ (etsects->cksel & CKSEL_MASK) << CKSEL_SHIFT;
+
+ gfar_write(&etsects->regs->tmr_ctrl, tmr_ctrl);
+ gfar_write(&etsects->regs->tmr_add, etsects->tmr_add);
+ gfar_write(&etsects->regs->tmr_prsc, etsects->tmr_prsc);
+ gfar_write(&etsects->regs->tmr_fiper1, etsects->tmr_fiper1);
+ gfar_write(&etsects->regs->tmr_fiper2, etsects->tmr_fiper2);
+ set_alarm(etsects);
+ gfar_write(&etsects->regs->tmr_ctrl, tmr_ctrl|FS|RTPE|TE);
+
+ etsects->clock = ptp_clock_register(&ptp_gianfar_caps);
+
+ return IS_ERR(etsects->clock) ? PTR_ERR(etsects->clock) : 0;
+}
+
+static int gianfar_ptp_remove(struct platform_device *dev)
+{
+ gfar_write(&the_clock.regs->tmr_temask, 0);
+ gfar_write(&the_clock.regs->tmr_ctrl, 0);
+
+ ptp_clock_unregister(the_clock.clock);
+ free_irq(the_clock.irq, &the_clock);
+ iounmap(the_clock.regs);
+
+ return 0;
+}
+
+static struct of_device_id match_table[] = {
+ { .compatible = "fsl,etsec-ptp" },
+ {},
+};
+
+static struct of_platform_driver gianfar_ptp_driver = {
+ .driver = {
+ .name = "gianfar_ptp",
+ .of_match_table = match_table,
+ .owner = THIS_MODULE,
+ },
+ .probe = gianfar_ptp_probe,
+ .remove = gianfar_ptp_remove,
+};
+
+/* module operations */
+
+static int __init ptp_gianfar_init(void)
+{
+ return of_register_platform_driver(&gianfar_ptp_driver);
+}
+
+module_init(ptp_gianfar_init);
+
+static void __exit ptp_gianfar_exit(void)
+{
+ of_unregister_platform_driver(&gianfar_ptp_driver);
+}
+
+module_exit(ptp_gianfar_exit);
+
+MODULE_AUTHOR("Richard Cochran <[email protected]>");
+MODULE_DESCRIPTION("PTP clock using the eTSEC");
+MODULE_LICENSE("GPL");
diff --git a/drivers/net/gianfar_ptp_reg.h b/drivers/net/gianfar_ptp_reg.h
new file mode 100644
index 0000000..95e171f
--- /dev/null
+++ b/drivers/net/gianfar_ptp_reg.h
@@ -0,0 +1,113 @@
+/* gianfar_ptp_reg.h
+ * Generated by regen.tcl on Thu May 13 01:38:57 PM CEST 2010
+ *
+ * PTP 1588 clock using the gianfar eTSEC
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#ifndef _GIANFAR_PTP_REG_H_
+#define _GIANFAR_PTP_REG_H_
+
+struct gianfar_ptp_registers {
+ u32 tmr_ctrl; /* Timer control register */
+ u32 tmr_tevent; /* Timestamp event register */
+ u32 tmr_temask; /* Timer event mask register */
+ u32 tmr_pevent; /* Timestamp event register */
+ u32 tmr_pemask; /* Timer event mask register */
+ u32 tmr_stat; /* Timestamp status register */
+ u32 tmr_cnt_h; /* Timer counter high register */
+ u32 tmr_cnt_l; /* Timer counter low register */
+ u32 tmr_add; /* Timer drift compensation addend register */
+ u32 tmr_acc; /* Timer accumulator register */
+ u32 tmr_prsc; /* Timer prescale */
+ u8 res1[4];
+ u32 tmroff_h; /* Timer offset high */
+ u32 tmroff_l; /* Timer offset low */
+ u8 res2[8];
+ u32 tmr_alarm1_h; /* Timer alarm 1 high register */
+ u32 tmr_alarm1_l; /* Timer alarm 1 high register */
+ u32 tmr_alarm2_h; /* Timer alarm 2 high register */
+ u32 tmr_alarm2_l; /* Timer alarm 2 high register */
+ u8 res3[48];
+ u32 tmr_fiper1; /* Timer fixed period interval */
+ u32 tmr_fiper2; /* Timer fixed period interval */
+ u32 tmr_fiper3; /* Timer fixed period interval */
+ u8 res4[20];
+ u32 tmr_etts1_h; /* Timestamp of general purpose external trigger */
+ u32 tmr_etts1_l; /* Timestamp of general purpose external trigger */
+ u32 tmr_etts2_h; /* Timestamp of general purpose external trigger */
+ u32 tmr_etts2_l; /* Timestamp of general purpose external trigger */
+};
+
+/* Bit definitions for the TMR_CTRL register */
+#define ALM1P (1<<31) /* Alarm1 output polarity */
+#define ALM2P (1<<30) /* Alarm2 output polarity */
+#define FS (1<<28) /* FIPER start indication */
+#define PP1L (1<<27) /* Fiper1 pulse loopback mode enabled. */
+#define PP2L (1<<26) /* Fiper2 pulse loopback mode enabled. */
+#define TCLK_PERIOD_SHIFT (16) /* 1588 timer reference clock period. */
+#define TCLK_PERIOD_MASK (0x3ff)
+#define RTPE (1<<15) /* Record Tx Timestamp to PAL Enable. */
+#define FRD (1<<14) /* FIPER Realignment Disable */
+#define ESFDP (1<<11) /* External Tx/Rx SFD Polarity. */
+#define ESFDE (1<<10) /* External Tx/Rx SFD Enable. */
+#define ETEP2 (1<<9) /* External trigger 2 edge polarity */
+#define ETEP1 (1<<8) /* External trigger 1 edge polarity */
+#define COPH (1<<7) /* Generated clock (TSEC_1588_GCLK) output phase. */
+#define CIPH (1<<6) /* External oscillator input clock phase. */
+#define TMSR (1<<5) /* Timer soft reset. When enabled, it resets all the timer registers and state machines. */
+#define BYP (1<<3) /* Bypass drift compensated clock */
+#define TE (1<<2) /* 1588 timer enable. If not enabled, all the timer registers and state machines are disabled. */
+#define CKSEL_SHIFT (0) /* 1588 Timer reference clock source select. */
+#define CKSEL_MASK (0x3)
+
+/* Bit definitions for the TMR_TEVENT register */
+#define ETS2 (1<<25) /* External trigger 2 timestamp sampled */
+#define ETS1 (1<<24) /* External trigger 1 timestamp sampled */
+#define ALM2 (1<<17) /* Current time equaled alarm time register 2 */
+#define ALM1 (1<<16) /* Current time equaled alarm time register 1 */
+#define PP1 (1<<7) /* Indicates that a periodic pulse has been generated based on FIPER1 register */
+#define PP2 (1<<6) /* Indicates that a periodic pulse has been generated based on FIPER2 register */
+#define PP3 (1<<5) /* Indicates that a periodic pulse has been generated based on FIPER3 register */
+
+/* Bit definitions for the TMR_TEMASK register */
+#define ETS2EN (1<<25) /* External trigger 2 timestamp sample event enable */
+#define ETS1EN (1<<24) /* External trigger 1 timestamp sample event enable */
+#define ALM2EN (1<<17) /* Timer ALM2 event enable */
+#define ALM1EN (1<<16) /* Timer ALM1 event enable */
+#define PP1EN (1<<7) /* Periodic pulse event 1 enable */
+#define PP2EN (1<<6) /* Periodic pulse event 2 enable */
+
+/* Bit definitions for the TMR_PEVENT register */
+#define TXP2 (1<<9) /* Indicates that a PTP frame has been transmitted and its timestamp is stored in TXTS2 register */
+#define TXP1 (1<<8) /* Indicates that a PTP frame has been transmitted and its timestamp is stored in TXTS1 register */
+#define RXP (1<<0) /* Indicates that a PTP frame has been received */
+
+/* Bit definitions for the TMR_PEMASK register */
+#define TXP2EN (1<<9) /* Transmit PTP packet event 2 enable */
+#define TXP1EN (1<<8) /* Transmit PTP packet event 1 enable */
+#define RXPEN (1<<0) /* Receive PTP packet event enable */
+
+/* Bit definitions for the TMR_STAT register */
+#define STAT_VEC_SHIFT (0) /* Timer general purpose status vector */
+#define STAT_VEC_MASK (0x3f)
+
+/* Bit definitions for the TMR_PRSC register */
+#define PRSC_OCK_SHIFT (0) /* Output clock division/prescale factor. */
+#define PRSC_OCK_MASK (0xffff)
+
+#endif
diff --git a/drivers/ptp/Kconfig b/drivers/ptp/Kconfig
index 17be208..a4df298 100644
--- a/drivers/ptp/Kconfig
+++ b/drivers/ptp/Kconfig
@@ -24,4 +24,17 @@ config PTP_1588_CLOCK
To compile this driver as a module, choose M here: the module
will be called ptp.
+config PTP_1588_CLOCK_GIANFAR
+ tristate "Freescale eTSEC as PTP clock"
+ depends on PTP_1588_CLOCK
+ depends on GIANFAR
+ help
+ This driver adds support for using the eTSEC as a PTP
+ clock. This clock is only useful if your PTP programs are
+ getting hardware time stamps on the PTP Ethernet packets
+ using the SO_TIMESTAMPING API.
+
+ To compile this driver as a module, choose M here: the module
+ will be called gianfar_ptp.
+
endmenu
--
1.7.0.4
This patch adds a driver for the hardware time stamping unit found on the
IXP465. The basic clock operations and an external trigger are implemented.
Signed-off-by: Richard Cochran <[email protected]>
---
arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h | 78 ++++++
drivers/net/arm/ixp4xx_eth.c | 191 ++++++++++++++
drivers/ptp/Kconfig | 13 +
drivers/ptp/Makefile | 1 +
drivers/ptp/ptp_ixp46x.c | 342 +++++++++++++++++++++++++
5 files changed, 625 insertions(+), 0 deletions(-)
create mode 100644 arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h
create mode 100644 drivers/ptp/ptp_ixp46x.c
diff --git a/arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h b/arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h
new file mode 100644
index 0000000..729a6b2
--- /dev/null
+++ b/arch/arm/mach-ixp4xx/include/mach/ixp46x_ts.h
@@ -0,0 +1,78 @@
+/*
+ * PTP 1588 clock using the IXP46X
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef _IXP46X_TS_H_
+#define _IXP46X_TS_H_
+
+#define DEFAULT_ADDEND 0xF0000029
+#define TICKS_NS_SHIFT 4
+
+struct ixp46x_channel_ctl {
+ u32 Ch_Control; /* 0x40 Time Synchronization Channel Control */
+ u32 Ch_Event; /* 0x44 Time Synchronization Channel Event */
+ u32 TxSnapLo; /* 0x48 Transmit Snapshot Low Register */
+ u32 TxSnapHi; /* 0x4C Transmit Snapshot High Register */
+ u32 RxSnapLo; /* 0x50 Receive Snapshot Low Register */
+ u32 RxSnapHi; /* 0x54 Receive Snapshot High Register */
+ u32 SrcUUIDLo; /* 0x58 Source UUID0 Low Register */
+ u32 SrcUUIDHi; /* 0x5C Sequence Identifier/Source UUID0 High */
+};
+
+struct ixp46x_ts_regs {
+ u32 Control; /* 0x00 Time Sync Control Register */
+ u32 Event; /* 0x04 Time Sync Event Register */
+ u32 Addend; /* 0x08 Time Sync Addend Register */
+ u32 Accum; /* 0x0C Time Sync Accumulator Register */
+ u32 Test; /* 0x10 Time Sync Test Register */
+ u32 Unused; /* 0x14 */
+ u32 RSysTime_Lo; /* 0x18 RawSystemTime_Low Register */
+ u32 RSysTimeHi; /* 0x1C RawSystemTime_High Register */
+ u32 SysTimeLo; /* 0x20 SystemTime_Low Register */
+ u32 SysTimeHi; /* 0x24 SystemTime_High Register */
+ u32 TrgtLo; /* 0x28 TargetTime_Low Register */
+ u32 TrgtHi; /* 0x2C TargetTime_High Register */
+ u32 ASMSLo; /* 0x30 Auxiliary Slave Mode Snapshot Low */
+ u32 ASMSHi; /* 0x34 Auxiliary Slave Mode Snapshot High */
+ u32 AMMSLo; /* 0x38 Auxiliary Master Mode Snapshot Low */
+ u32 AMMSHi; /* 0x3C Auxiliary Master Mode Snapshot High */
+
+ struct ixp46x_channel_ctl channel[3];
+};
+
+/* 0x00 Time Sync Control Register Bits */
+#define TSCR_AMM (1<<3)
+#define TSCR_ASM (1<<2)
+#define TSCR_TTM (1<<1)
+#define TSCR_RST (1<<0)
+
+/* 0x04 Time Sync Event Register Bits */
+#define TSER_SNM (1<<3)
+#define TSER_SNS (1<<2)
+#define TTIPEND (1<<1)
+
+/* 0x40 Time Synchronization Channel Control Register Bits */
+#define MASTER_MODE (1<<0)
+#define TIMESTAMP_ALL (1<<1)
+
+/* 0x44 Time Synchronization Channel Event Register Bits */
+#define TX_SNAPSHOT_LOCKED (1<<0)
+#define RX_SNAPSHOT_LOCKED (1<<1)
+
+#endif
diff --git a/drivers/net/arm/ixp4xx_eth.c b/drivers/net/arm/ixp4xx_eth.c
index 6028226..61cf4b4 100644
--- a/drivers/net/arm/ixp4xx_eth.c
+++ b/drivers/net/arm/ixp4xx_eth.c
@@ -30,9 +30,12 @@
#include <linux/etherdevice.h>
#include <linux/io.h>
#include <linux/kernel.h>
+#include <linux/net_tstamp.h>
#include <linux/phy.h>
#include <linux/platform_device.h>
+#include <linux/ptp_classify.h>
#include <linux/slab.h>
+#include <mach/ixp46x_ts.h>
#include <mach/npe.h>
#include <mach/qmgr.h>
@@ -67,6 +70,14 @@
#define RXFREE_QUEUE(port_id) (NPE_ID(port_id) + 26)
#define TXDONE_QUEUE 31
+#define PTP_SLAVE_MODE 1
+#define PTP_MASTER_MODE 2
+#define PORT2CHANNEL(p) 1
+/*
+ * PHYSICAL_ID(p->id) ?
+ * TODO - Figure out correct mapping.
+ */
+
/* TX Control Registers */
#define TX_CNTRL0_TX_EN 0x01
#define TX_CNTRL0_HALFDUPLEX 0x02
@@ -171,6 +182,8 @@ struct port {
int id; /* logical port ID */
int speed, duplex;
u8 firmware[4];
+ int hwts_tx_en;
+ int hwts_rx_en;
};
/* NPE message structure */
@@ -246,6 +259,171 @@ static int ports_open;
static struct port *npe_port_tab[MAX_NPES];
static struct dma_pool *dma_pool;
+static struct sock_filter ptp_filter[] = {
+ PTP_FILTER
+};
+
+static int match(struct sk_buff *skb, u16 uid_hi, u32 uid_lo, u16 seq)
+{
+ unsigned int type;
+ u16 *hi, *id;
+ u8 *lo, *data = skb->data;
+
+ type = sk_run_filter(skb, ptp_filter, ARRAY_SIZE(ptp_filter));
+
+ if (PTP_CLASS_V1_IPV4 == type) {
+
+ id = (u16 *)(data + 42 + 30);
+ hi = (u16 *)(data + 42 + 22);
+ lo = data + 42 + 24;
+
+ return (uid_hi == *hi &&
+ 0 == memcmp(&uid_lo, lo, sizeof(uid_lo)) &&
+ seq == *id);
+ }
+
+ return 0;
+}
+
+static void do_rx_timestamp(struct port *port, struct sk_buff *skb)
+{
+ struct skb_shared_hwtstamps *shhwtstamps;
+ struct ixp46x_ts_regs *regs;
+ u64 ns;
+ u32 ch, hi, lo, val;
+ u16 uid, seq;
+
+ if (!port->hwts_rx_en)
+ return;
+
+ ch = PORT2CHANNEL(port);
+
+ regs = (struct ixp46x_ts_regs __iomem *) IXP4XX_TIMESYNC_BASE_VIRT;
+
+ val = __raw_readl(®s->channel[ch].Ch_Event);
+
+ if (!(val & RX_SNAPSHOT_LOCKED))
+ return;
+
+ lo = __raw_readl(®s->channel[ch].SrcUUIDLo);
+ hi = __raw_readl(®s->channel[ch].SrcUUIDHi);
+
+ uid = hi & 0xffff;
+ seq = (hi >> 16) & 0xffff;
+
+ if (!match(skb, htons(uid), htonl(lo), htons(seq)))
+ goto out;
+
+ lo = __raw_readl(®s->channel[ch].RxSnapLo);
+ hi = __raw_readl(®s->channel[ch].RxSnapHi);
+ ns = ((u64) hi) << 32;
+ ns |= lo;
+ ns <<= TICKS_NS_SHIFT;
+
+ shhwtstamps = skb_hwtstamps(skb);
+ memset(shhwtstamps, 0, sizeof(*shhwtstamps));
+ shhwtstamps->hwtstamp = ns_to_ktime(ns);
+out:
+ __raw_writel(RX_SNAPSHOT_LOCKED, ®s->channel[ch].Ch_Event);
+}
+
+static void do_tx_timestamp(struct port *port, struct sk_buff *skb)
+{
+#ifdef __ARMEB__
+ struct skb_shared_hwtstamps shhwtstamps;
+ struct ixp46x_ts_regs *regs;
+ struct skb_shared_info *shtx;
+ u64 ns;
+ u32 ch, cnt, hi, lo, val;
+
+ shtx = skb_shinfo(skb);
+ if (unlikely(shtx->tx_flags & SKBTX_HW_TSTAMP && port->hwts_tx_en))
+ shtx->tx_flags |= SKBTX_IN_PROGRESS;
+ else
+ return;
+
+ ch = PORT2CHANNEL(port);
+
+ regs = (struct ixp46x_ts_regs __iomem *) IXP4XX_TIMESYNC_BASE_VIRT;
+
+ /*
+ * This really stinks, but we have to poll for the Tx time stamp.
+ * Usually, the time stamp is ready after 4 to 6 microseconds.
+ */
+ for (cnt = 0; cnt < 100; cnt++) {
+ val = __raw_readl(®s->channel[ch].Ch_Event);
+ if (val & TX_SNAPSHOT_LOCKED)
+ break;
+ udelay(1);
+ }
+ if (!(val & TX_SNAPSHOT_LOCKED)) {
+ shtx->tx_flags &= ~SKBTX_IN_PROGRESS;
+ return;
+ }
+
+ lo = __raw_readl(®s->channel[ch].TxSnapLo);
+ hi = __raw_readl(®s->channel[ch].TxSnapHi);
+ ns = ((u64) hi) << 32;
+ ns |= lo;
+ ns <<= TICKS_NS_SHIFT;
+
+ memset(&shhwtstamps, 0, sizeof(shhwtstamps));
+ shhwtstamps.hwtstamp = ns_to_ktime(ns);
+ skb_tstamp_tx(skb, &shhwtstamps);
+
+ __raw_writel(TX_SNAPSHOT_LOCKED, ®s->channel[ch].Ch_Event);
+#endif
+}
+
+static int hwtstamp_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)
+{
+ struct hwtstamp_config cfg;
+ struct ixp46x_ts_regs *regs;
+ struct port *port = netdev_priv(netdev);
+ int ch;
+
+ if (copy_from_user(&cfg, ifr->ifr_data, sizeof(cfg)))
+ return -EFAULT;
+
+ if (cfg.flags) /* reserved for future extensions */
+ return -EINVAL;
+
+ ch = PORT2CHANNEL(port);
+ regs = (struct ixp46x_ts_regs __iomem *) IXP4XX_TIMESYNC_BASE_VIRT;
+
+ switch (cfg.tx_type) {
+ case HWTSTAMP_TX_OFF:
+ port->hwts_tx_en = 0;
+ break;
+ case HWTSTAMP_TX_ON:
+ port->hwts_tx_en = 1;
+ break;
+ default:
+ return -ERANGE;
+ }
+
+ switch (cfg.rx_filter) {
+ case HWTSTAMP_FILTER_NONE:
+ port->hwts_rx_en = 0;
+ break;
+ case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+ port->hwts_rx_en = PTP_SLAVE_MODE;
+ __raw_writel(0, ®s->channel[ch].Ch_Control);
+ break;
+ case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+ port->hwts_rx_en = PTP_MASTER_MODE;
+ __raw_writel(MASTER_MODE, ®s->channel[ch].Ch_Control);
+ break;
+ default:
+ return -ERANGE;
+ }
+
+ /* Clear out any old time stamps. */
+ __raw_writel(TX_SNAPSHOT_LOCKED | RX_SNAPSHOT_LOCKED,
+ ®s->channel[ch].Ch_Event);
+
+ return copy_to_user(ifr->ifr_data, &cfg, sizeof(cfg)) ? -EFAULT : 0;
+}
static int ixp4xx_mdio_cmd(struct mii_bus *bus, int phy_id, int location,
int write, u16 cmd)
@@ -573,6 +751,7 @@ static int eth_poll(struct napi_struct *napi, int budget)
debug_pkt(dev, "eth_poll", skb->data, skb->len);
+ do_rx_timestamp(port, skb);
skb->protocol = eth_type_trans(skb, dev);
dev->stats.rx_packets++;
dev->stats.rx_bytes += skb->len;
@@ -728,6 +907,10 @@ static int eth_xmit(struct sk_buff *skb, struct net_device *dev)
#if DEBUG_TX
printk(KERN_DEBUG "%s: eth_xmit end\n", dev->name);
#endif
+
+ do_tx_timestamp(port, skb);
+ skb_tx_timestamp(skb);
+
return NETDEV_TX_OK;
}
@@ -783,6 +966,9 @@ static int eth_ioctl(struct net_device *dev, struct ifreq *req, int cmd)
if (!netif_running(dev))
return -EINVAL;
+ if (cpu_is_ixp46x() && cmd == SIOCSHWTSTAMP)
+ return hwtstamp_ioctl(dev, req, cmd);
+
return phy_mii_ioctl(port->phydev, req, cmd);
}
@@ -1171,6 +1357,11 @@ static int __devinit eth_init_one(struct platform_device *pdev)
char phy_id[MII_BUS_ID_SIZE + 3];
int err;
+ if (ptp_filter_init(ptp_filter, ARRAY_SIZE(ptp_filter))) {
+ pr_err("ixp4xx_eth: bad ptp filter\n");
+ return -EINVAL;
+ }
+
if (!(dev = alloc_etherdev(sizeof(struct port))))
return -ENOMEM;
diff --git a/drivers/ptp/Kconfig b/drivers/ptp/Kconfig
index a4df298..6f1dcbf 100644
--- a/drivers/ptp/Kconfig
+++ b/drivers/ptp/Kconfig
@@ -37,4 +37,17 @@ config PTP_1588_CLOCK_GIANFAR
To compile this driver as a module, choose M here: the module
will be called gianfar_ptp.
+config PTP_1588_CLOCK_IXP46X
+ tristate "Intel IXP46x as PTP clock"
+ depends on PTP_1588_CLOCK
+ depends on IXP4XX_ETH
+ help
+ This driver adds support for using the IXP46X as a PTP
+ clock. This clock is only useful if your PTP programs are
+ getting hardware time stamps on the PTP Ethernet packets
+ using the SO_TIMESTAMPING API.
+
+ To compile this driver as a module, choose M here: the module
+ will be called ptp_ixp46x.
+
endmenu
diff --git a/drivers/ptp/Makefile b/drivers/ptp/Makefile
index 480e2af..f6933e8 100644
--- a/drivers/ptp/Makefile
+++ b/drivers/ptp/Makefile
@@ -4,3 +4,4 @@
ptp-y := ptp_clock.o ptp_chardev.o ptp_sysfs.o
obj-$(CONFIG_PTP_1588_CLOCK) += ptp.o
+obj-$(CONFIG_PTP_1588_CLOCK_IXP46X) += ptp_ixp46x.o
diff --git a/drivers/ptp/ptp_ixp46x.c b/drivers/ptp/ptp_ixp46x.c
new file mode 100644
index 0000000..387bd0f
--- /dev/null
+++ b/drivers/ptp/ptp_ixp46x.c
@@ -0,0 +1,342 @@
+/*
+ * PTP 1588 clock using the IXP46X
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/gpio.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+
+#include <linux/ptp_clock_kernel.h>
+#include <mach/ixp46x_ts.h>
+
+#define DRIVER "ptp_ixp46x"
+#define N_EXT_TS 2
+#define MASTER_GPIO 8
+#define MASTER_IRQ 25
+#define SLAVE_GPIO 7
+#define SLAVE_IRQ 24
+
+struct ixp_clock {
+ struct ixp46x_ts_regs *regs;
+ struct ptp_clock *ptp_clock;
+ int exts0_enabled;
+ int exts1_enabled;
+};
+
+DEFINE_SPINLOCK(register_lock);
+
+/*
+ * Register access functions
+ */
+
+static inline u32 ixp_read(volatile unsigned __iomem *addr)
+{
+ u32 val;
+ val = __raw_readl(addr);
+ return val;
+}
+
+static inline void ixp_write(volatile unsigned __iomem *addr, u32 val)
+{
+ __raw_writel(val, addr);
+}
+
+static u64 sys_time_read(struct ixp46x_ts_regs *regs)
+{
+ u64 ns;
+ u32 lo, hi;
+
+ lo = ixp_read(®s->SysTimeLo);
+ hi = ixp_read(®s->SysTimeHi);
+
+ ns = ((u64) hi) << 32;
+ ns |= lo;
+ ns <<= TICKS_NS_SHIFT;
+
+ return ns;
+}
+
+static void sys_time_write(struct ixp46x_ts_regs *regs, u64 ns)
+{
+ u32 hi, lo;
+
+ ns >>= TICKS_NS_SHIFT;
+ hi = ns >> 32;
+ lo = ns & 0xffffffff;
+
+ ixp_write(®s->SysTimeLo, lo);
+ ixp_write(®s->SysTimeHi, hi);
+}
+
+/*
+ * Interrupt service routine
+ */
+
+static irqreturn_t isr(int irq, void *priv)
+{
+ struct ixp_clock *ixp_clock = priv;
+ struct ixp46x_ts_regs *regs = ixp_clock->regs;
+ struct ptp_clock_event event;
+ u32 ack = 0, lo, hi, val;
+
+ val = ixp_read(®s->Event);
+
+ if (val & TSER_SNS) {
+ ack |= TSER_SNS;
+ if (ixp_clock->exts0_enabled) {
+ hi = ixp_read(®s->ASMSHi);
+ lo = ixp_read(®s->ASMSLo);
+ event.type = PTP_CLOCK_EXTTS;
+ event.index = 0;
+ event.timestamp = ((u64) hi) << 32;
+ event.timestamp |= lo;
+ event.timestamp <<= TICKS_NS_SHIFT;
+ ptp_clock_event(ixp_clock->ptp_clock, &event);
+ }
+ }
+
+ if (val & TSER_SNM) {
+ ack |= TSER_SNM;
+ if (ixp_clock->exts1_enabled) {
+ hi = ixp_read(®s->AMMSHi);
+ lo = ixp_read(®s->AMMSLo);
+ event.type = PTP_CLOCK_EXTTS;
+ event.index = 1;
+ event.timestamp = ((u64) hi) << 32;
+ event.timestamp |= lo;
+ event.timestamp <<= TICKS_NS_SHIFT;
+ ptp_clock_event(ixp_clock->ptp_clock, &event);
+ }
+ }
+
+ if (val & TTIPEND)
+ ack |= TTIPEND; /* this bit seems to be always set */
+
+ if (ack) {
+ ixp_write(®s->Event, ack);
+ return IRQ_HANDLED;
+ } else
+ return IRQ_NONE;
+}
+
+/*
+ * PTP clock operations
+ */
+
+static int ptp_ixp_adjfreq(void *priv, s32 ppb)
+{
+ u64 adj;
+ u32 diff, addend;
+ int neg_adj = 0;
+ struct ixp_clock *ixp_clock = priv;
+ struct ixp46x_ts_regs *regs = ixp_clock->regs;
+
+ if (ppb < 0) {
+ neg_adj = 1;
+ ppb = -ppb;
+ }
+ addend = DEFAULT_ADDEND;
+ adj = addend;
+ adj *= ppb;
+ diff = div_u64(adj, 1000000000ULL);
+
+ addend = neg_adj ? addend - diff : addend + diff;
+
+ ixp_write(®s->Addend, addend);
+
+ return 0;
+}
+
+static int ptp_ixp_adjtime(void *priv, s64 delta)
+{
+ s64 now;
+ unsigned long flags;
+ struct ixp_clock *ixp_clock = priv;
+ struct ixp46x_ts_regs *regs = ixp_clock->regs;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ now = sys_time_read(regs);
+ now += delta;
+ sys_time_write(regs, now);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ return 0;
+}
+
+static int ptp_ixp_gettime(void *priv, struct timespec *ts)
+{
+ u64 ns;
+ u32 remainder;
+ unsigned long flags;
+ struct ixp_clock *ixp_clock = priv;
+ struct ixp46x_ts_regs *regs = ixp_clock->regs;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ ns = sys_time_read(regs);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ ts->tv_sec = div_u64_rem(ns, 1000000000, &remainder);
+ ts->tv_nsec = remainder;
+ return 0;
+}
+
+static int ptp_ixp_settime(void *priv, const struct timespec *ts)
+{
+ u64 ns;
+ unsigned long flags;
+ struct ixp_clock *ixp_clock = priv;
+ struct ixp46x_ts_regs *regs = ixp_clock->regs;
+
+ ns = ts->tv_sec * 1000000000ULL;
+ ns += ts->tv_nsec;
+
+ spin_lock_irqsave(®ister_lock, flags);
+
+ sys_time_write(regs, ns);
+
+ spin_unlock_irqrestore(®ister_lock, flags);
+
+ return 0;
+}
+
+static int ptp_ixp_enable(void *priv, struct ptp_clock_request *rq, int on)
+{
+ struct ixp_clock *ixp_clock = priv;
+
+ switch (rq->type) {
+ case PTP_CLK_REQ_EXTTS:
+ switch (rq->extts.index) {
+ case 0:
+ ixp_clock->exts0_enabled = on ? 1 : 0;
+ break;
+ case 1:
+ ixp_clock->exts1_enabled = on ? 1 : 0;
+ break;
+ default:
+ return -EINVAL;
+ }
+ return 0;
+ default:
+ break;
+ }
+
+ return -EOPNOTSUPP;
+}
+
+static struct ptp_clock_info ptp_ixp_caps = {
+ .owner = THIS_MODULE,
+ .name = "IXP46X timer",
+ .max_adj = 66666655,
+ .n_ext_ts = N_EXT_TS,
+ .pps = 0,
+ .priv = NULL,
+ .adjfreq = ptp_ixp_adjfreq,
+ .adjtime = ptp_ixp_adjtime,
+ .gettime = ptp_ixp_gettime,
+ .settime = ptp_ixp_settime,
+ .enable = ptp_ixp_enable,
+};
+
+/* module operations */
+
+static struct ixp_clock ixp_clock;
+
+static int setup_interrupt(int gpio)
+{
+ int irq;
+
+ gpio_line_config(gpio, IXP4XX_GPIO_IN);
+
+ irq = gpio_to_irq(gpio);
+
+ if (NO_IRQ == irq)
+ return NO_IRQ;
+
+ if (set_irq_type(irq, IRQF_TRIGGER_FALLING)) {
+ pr_err("cannot set trigger type for irq %d\n", irq);
+ return NO_IRQ;
+ }
+
+ if (request_irq(irq, isr, 0, DRIVER, &ixp_clock)) {
+ pr_err("request_irq failed for irq %d\n", irq);
+ return NO_IRQ;
+ }
+
+ return irq;
+}
+
+static void __exit ptp_ixp_exit(void)
+{
+ free_irq(MASTER_IRQ, &ixp_clock);
+ free_irq(SLAVE_IRQ, &ixp_clock);
+ ptp_clock_unregister(ixp_clock.ptp_clock);
+}
+
+static int __init ptp_ixp_init(void)
+{
+ if (!cpu_is_ixp46x())
+ return -ENODEV;
+
+ ixp_clock.regs =
+ (struct ixp46x_ts_regs __iomem *)IXP4XX_TIMESYNC_BASE_VIRT;
+
+ ptp_ixp_caps.priv = &ixp_clock;
+
+ ixp_clock.ptp_clock = ptp_clock_register(&ptp_ixp_caps);
+
+ if (IS_ERR(ixp_clock.ptp_clock))
+ return PTR_ERR(ixp_clock.ptp_clock);
+
+ ixp_write(&ixp_clock.regs->Addend, DEFAULT_ADDEND);
+ ixp_write(&ixp_clock.regs->TrgtLo, 1);
+ ixp_write(&ixp_clock.regs->TrgtHi, 0);
+ ixp_write(&ixp_clock.regs->Event, TTIPEND);
+
+ if (MASTER_IRQ != setup_interrupt(MASTER_GPIO)) {
+ pr_err("failed to setup gpio %d as irq\n", MASTER_GPIO);
+ goto no_master;
+ }
+ if (SLAVE_IRQ != setup_interrupt(SLAVE_GPIO)) {
+ pr_err("failed to setup gpio %d as irq\n", SLAVE_GPIO);
+ goto no_slave;
+ }
+
+ return 0;
+no_slave:
+ free_irq(MASTER_IRQ, &ixp_clock);
+no_master:
+ ptp_clock_unregister(ixp_clock.ptp_clock);
+ return -ENODEV;
+}
+
+module_init(ptp_ixp_init);
+module_exit(ptp_ixp_exit);
+
+MODULE_AUTHOR("Richard Cochran <[email protected]>");
+MODULE_DESCRIPTION("PTP clock using the IXP46X timer");
+MODULE_LICENSE("GPL");
--
1.7.0.4
This patch adds support for the PTP clock found on the DP83640.
The basic clock operations and one external time stamp have
been implemented.
Signed-off-by: Richard Cochran <[email protected]>
---
drivers/net/phy/Kconfig | 29 ++
drivers/net/phy/Makefile | 1 +
drivers/net/phy/dp83640.c | 890 +++++++++++++++++++++++++++++++++++++++++
drivers/net/phy/dp83640_reg.h | 261 ++++++++++++
4 files changed, 1181 insertions(+), 0 deletions(-)
create mode 100644 drivers/net/phy/dp83640.c
create mode 100644 drivers/net/phy/dp83640_reg.h
diff --git a/drivers/net/phy/Kconfig b/drivers/net/phy/Kconfig
index 35fda5a..1f1399a 100644
--- a/drivers/net/phy/Kconfig
+++ b/drivers/net/phy/Kconfig
@@ -76,6 +76,35 @@ config NATIONAL_PHY
---help---
Currently supports the DP83865 PHY.
+config DP83640_PHY
+ tristate "Driver for the National Semiconductor DP83640 PHYTER"
+ depends on PTP_1588_CLOCK
+ depends on NETWORK_PHY_TIMESTAMPING
+ ---help---
+ Supports the DP83640 PHYTER with IEEE 1588 features.
+
+ This driver adds support for using the DP83640 as a PTP
+ clock. This clock is only useful if your PTP programs are
+ getting hardware time stamps on the PTP Ethernet packets
+ using the SO_TIMESTAMPING API.
+
+ In order for this to work, your MAC driver must also
+ implement the skb_tx_timetamp() function.
+
+config DP83640_PHY_STATUS_FRAMES
+ bool "DP83640 Status Frames"
+ default y
+ depends on DP83640_PHY
+ ---help---
+ This option allows the DP83640 PHYTER driver to obtain time
+ stamps from the PHY via special status frames, rather than
+ reading over the MDIO bus. Using status frames is therefore
+ more efficient. However, if enabled, this option will cause
+ the driver to add a mutlicast address to the MAC.
+
+ Say Y here, unless your MAC does not support multicast
+ destination addresses.
+
config STE10XP
depends on PHYLIB
tristate "Driver for STMicroelectronics STe10Xp PHYs"
diff --git a/drivers/net/phy/Makefile b/drivers/net/phy/Makefile
index 13bebab..2333215 100644
--- a/drivers/net/phy/Makefile
+++ b/drivers/net/phy/Makefile
@@ -19,6 +19,7 @@ obj-$(CONFIG_FIXED_PHY) += fixed.o
obj-$(CONFIG_MDIO_BITBANG) += mdio-bitbang.o
obj-$(CONFIG_MDIO_GPIO) += mdio-gpio.o
obj-$(CONFIG_NATIONAL_PHY) += national.o
+obj-$(CONFIG_DP83640_PHY) += dp83640.o
obj-$(CONFIG_STE10XP) += ste10Xp.o
obj-$(CONFIG_MICREL_PHY) += micrel.o
obj-$(CONFIG_MDIO_OCTEON) += mdio-octeon.o
diff --git a/drivers/net/phy/dp83640.c b/drivers/net/phy/dp83640.c
new file mode 100644
index 0000000..ffb3ce0
--- /dev/null
+++ b/drivers/net/phy/dp83640.c
@@ -0,0 +1,890 @@
+/*
+ * Driver for the National Semiconductor DP83640 PHYTER
+ *
+ * Copyright (C) 2010 OMICRON electronics GmbH
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#include <linux/ethtool.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/mii.h>
+#include <linux/module.h>
+#include <linux/net_tstamp.h>
+#include <linux/netdevice.h>
+#include <linux/phy.h>
+#include <linux/ptp_classify.h>
+#include <linux/ptp_clock_kernel.h>
+
+#include "dp83640_reg.h"
+
+#ifdef CONFIG_DP83640_PHY_STATUS_FRAMES
+#define USE_STATUS_FRAMES
+#endif
+
+#define DP83640_PHY_ID 0x20005ce1
+#define PAGESEL 0x13
+#define LAYER4 0x02
+#define LAYER2 0x01
+#define MAX_RXTS 4
+#define MAX_TXTS 4
+#define N_EXT_TS 1
+#define PSF_PTPVER 2
+#define PSF_EVNT 0x4000
+#define PSF_RX 0x2000
+#define PSF_TX 0x1000
+#define EXT_EVENT 1
+#define EXT_GPIO 1
+
+#if defined(__BIG_ENDIAN)
+#define ENDIAN_FLAG 0
+#elif defined(__LITTLE_ENDIAN)
+#define ENDIAN_FLAG PSF_ENDIAN
+#endif
+
+#define SKB_PTP_TYPE(__skb) (*(unsigned int *)((__skb)->cb))
+
+struct phy_rxts {
+ u16 ns_lo; /* ns[15:0] */
+ u16 ns_hi; /* overflow[1:0], ns[29:16] */
+ u16 sec_lo; /* sec[15:0] */
+ u16 sec_hi; /* sec[31:16] */
+ u16 seqid; /* sequenceId[15:0] */
+ u16 msgtype; /* messageType[3:0], hash[11:0] */
+};
+
+struct phy_txts {
+ u16 ns_lo; /* ns[15:0] */
+ u16 ns_hi; /* overflow[1:0], ns[29:16] */
+ u16 sec_lo; /* sec[15:0] */
+ u16 sec_hi; /* sec[31:16] */
+};
+
+struct rxts {
+ struct list_head list;
+ unsigned long tmo;
+ u64 ns;
+ u16 seqid;
+ u8 msgtype;
+ u16 hash;
+};
+
+struct dp83640_private {
+ struct phy_device *phydev;
+ struct work_struct ts_work;
+ int hwts_tx_en;
+ int hwts_rx_en;
+ int layer;
+ int version;
+ /* protects extended registers from concurrent access */
+ struct mutex extreg_mux;
+ int page;
+ /* remember the last event time stamp */
+ struct phy_txts edata;
+ /* list of rx timestamps */
+ struct list_head rxts;
+ struct list_head rxpool;
+ struct rxts rx_pool_data[MAX_RXTS];
+ /* protects above three fields from concurrent access */
+ spinlock_t rx_lock;
+
+ struct sk_buff_head rx_queue;
+ struct sk_buff_head tx_queue;
+};
+
+/* globals */
+
+static struct ptp_clock *dp83640_clock;
+DEFINE_MUTEX(clock_lock); /* protects the one and only dp83640_clock */
+
+static void do_timestamp_work(struct work_struct *work);
+
+/* extended register access functions */
+
+static int ext_read(struct phy_device *phydev, int page, u32 regnum)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+ int val;
+
+ if (dp83640->page != page) {
+ phy_write(phydev, PAGESEL, page);
+ dp83640->page = page;
+ }
+ val = phy_read(phydev, regnum);
+
+ return val;
+}
+
+static void ext_write(struct phy_device *phydev, int page, u32 regnum, u16 val)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+
+ if (dp83640->page != page) {
+ phy_write(phydev, PAGESEL, page);
+ dp83640->page = page;
+ }
+ phy_write(phydev, regnum, val);
+}
+
+static int tdr_write(struct phy_device *phydev, const struct timespec *ts, u16 cmd)
+{
+ ext_write(phydev, PAGE4, PTP_TDR, ts->tv_nsec & 0xffff);/* ns[15:0] */
+ ext_write(phydev, PAGE4, PTP_TDR, ts->tv_nsec >> 16); /* ns[31:16] */
+ ext_write(phydev, PAGE4, PTP_TDR, ts->tv_sec & 0xffff); /* sec[15:0] */
+ ext_write(phydev, PAGE4, PTP_TDR, ts->tv_sec >> 16); /* sec[31:16] */
+
+ ext_write(phydev, PAGE4, PTP_CTL, cmd);
+
+ return 0;
+}
+
+/* convert phy timestamps into driver timestamps */
+
+static void phy2rxts(struct phy_rxts *p, struct rxts *rxts)
+{
+ u32 sec;
+
+ sec = p->sec_lo;
+ sec |= p->sec_hi << 16;
+
+ rxts->ns = p->ns_lo;
+ rxts->ns |= (p->ns_hi & 0x3fff) << 16;
+ rxts->ns += ((u64)sec) * 1000000000ULL;
+ rxts->seqid = p->seqid;
+ rxts->msgtype = (p->msgtype >> 12) & 0xf;
+ rxts->hash = p->msgtype & 0x0fff;
+}
+
+static u64 phy2txts(struct phy_txts *p)
+{
+ u64 ns;
+ u32 sec;
+
+ sec = p->sec_lo;
+ sec |= p->sec_hi << 16;
+
+ ns = p->ns_lo;
+ ns |= (p->ns_hi & 0x3fff) << 16;
+ ns += ((u64)sec) * 1000000000ULL;
+
+ return ns;
+}
+
+/* ptp clock methods */
+
+static int ptp_dp83640_adjfreq(void *priv, s32 ppb)
+{
+ struct dp83640_private *dp83640 = priv;
+ struct phy_device *phydev = dp83640->phydev;
+ u64 rate;
+ int neg_adj = 0;
+ u16 hi, lo;
+
+ if (ppb < 0) {
+ neg_adj = 1;
+ ppb = -ppb;
+ }
+ rate = ppb;
+ rate <<= 26;
+ rate = div_u64(rate, 1953125);
+
+ hi = (rate >> 16) & PTP_RATE_HI_MASK;
+ if (neg_adj)
+ hi |= PTP_RATE_DIR;
+
+ lo = rate & 0xffff;
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ ext_write(phydev, PAGE4, PTP_RATEH, hi);
+ ext_write(phydev, PAGE4, PTP_RATEL, lo);
+
+ mutex_unlock(&dp83640->extreg_mux);
+
+ return 0;
+}
+
+static int ptp_dp83640_adjtime(void *priv, s64 delta)
+{
+ struct dp83640_private *dp83640 = priv;
+ struct timespec ts;
+ int err;
+
+ ts = ns_to_timespec(delta);
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ err = tdr_write(dp83640->phydev, &ts, PTP_STEP_CLK);
+
+ mutex_unlock(&dp83640->extreg_mux);
+
+ return err;
+}
+
+static int ptp_dp83640_gettime(void *priv, struct timespec *ts)
+{
+ struct dp83640_private *dp83640 = priv;
+ struct phy_device *phydev = dp83640->phydev;
+ unsigned int val[4];
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ ext_write(phydev, PAGE4, PTP_CTL, PTP_RD_CLK);
+
+ val[0] = ext_read(phydev, PAGE4, PTP_TDR); /* ns[15:0] */
+ val[1] = ext_read(phydev, PAGE4, PTP_TDR); /* ns[31:16] */
+ val[2] = ext_read(phydev, PAGE4, PTP_TDR); /* sec[15:0] */
+ val[3] = ext_read(phydev, PAGE4, PTP_TDR); /* sec[31:16] */
+
+ mutex_unlock(&dp83640->extreg_mux);
+
+ ts->tv_nsec = val[0] | (val[1] << 16);
+ ts->tv_sec = val[2] | (val[3] << 16);
+
+ return 0;
+}
+
+static int ptp_dp83640_settime(void *priv, const struct timespec *ts)
+{
+ struct dp83640_private *dp83640 = priv;
+ int err;
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ err = tdr_write(dp83640->phydev, ts, PTP_LOAD_CLK);
+
+ mutex_unlock(&dp83640->extreg_mux);
+
+ return err;
+}
+
+static int ptp_dp83640_enable(void *priv, struct ptp_clock_request *rq, int on)
+{
+ struct dp83640_private *dp83640 = priv;
+ struct phy_device *phydev = dp83640->phydev;
+ u16 evnt;
+
+ switch (rq->type) {
+ case PTP_CLK_REQ_EXTTS:
+ if (rq->extts.index != 0)
+ return -EINVAL;
+ evnt = EVNT_WR | (EXT_EVENT & EVNT_SEL_MASK) << EVNT_SEL_SHIFT;
+ if (on) {
+ evnt |= (EXT_GPIO & EVNT_GPIO_MASK) << EVNT_GPIO_SHIFT;
+ evnt |= EVNT_RISE;
+ }
+ ext_write(phydev, PAGE5, PTP_EVNT, evnt);
+ return 0;
+ default:
+ break;
+ }
+
+ return -EOPNOTSUPP;
+}
+
+static struct ptp_clock_info ptp_dp83640_caps = {
+ .owner = THIS_MODULE,
+ .name = "dp83640 timer",
+ .max_adj = 1953124,
+ .n_alarm = 0,
+ .n_ext_ts = N_EXT_TS,
+ .n_per_out = 0,
+ .pps = 0,
+ .priv = NULL,
+ .adjfreq = ptp_dp83640_adjfreq,
+ .adjtime = ptp_dp83640_adjtime,
+ .gettime = ptp_dp83640_gettime,
+ .settime = ptp_dp83640_settime,
+ .enable = ptp_dp83640_enable,
+};
+
+/* status frame conditional code */
+
+#ifdef USE_STATUS_FRAMES
+
+static u8 status_frame_dst[6] = { 0x01, 0x1B, 0x19, 0x00, 0x00, 0x00 };
+static u8 status_frame_src[6] = { 0x08, 0x00, 0x17, 0x0B, 0x6B, 0x0F };
+
+static void enable_status_frames(struct phy_device *phydev, bool on)
+{
+ u16 cfg0 = 0, ver;
+
+ if (on)
+ cfg0 = PSF_EVNT_EN | PSF_RXTS_EN | PSF_TXTS_EN | ENDIAN_FLAG;
+
+ ver = (PSF_PTPVER & VERSIONPTP_MASK) << VERSIONPTP_SHIFT;
+
+ ext_write(phydev, PAGE5, PSF_CFG0, cfg0);
+ ext_write(phydev, PAGE6, PSF_CFG1, ver);
+
+ if (!phydev->attached_dev) {
+ pr_err("expected to find an attached netdevice\n");
+ BUG();
+ }
+
+ if (on) {
+ if (dev_mc_add(phydev->attached_dev, status_frame_dst))
+ pr_warning("dp83640: failed to add mc address\n");
+ } else {
+ if (dev_mc_del(phydev->attached_dev, status_frame_dst))
+ pr_warning("dp83640: failed to delete mc address\n");
+ }
+}
+
+static bool is_status_frame(struct sk_buff *skb, int type)
+{
+ struct ethhdr *h = eth_hdr(skb);
+
+ if (PTP_CLASS_V2_L2 == type &&
+ !memcmp(h->h_source, status_frame_src, sizeof(status_frame_src)))
+ return true;
+ else
+ return false;
+}
+
+static void rx_reading_work(struct dp83640_private *dp83640)
+{
+}
+
+static void tx_timestamp_work(struct dp83640_private *dp83640)
+{
+}
+
+#else /* USE_STATUS_FRAMES */
+
+static void enable_status_frames(struct phy_device *phydev, bool on)
+{
+}
+
+static bool is_status_frame(struct sk_buff *skb, int type)
+{
+ return false;
+}
+
+static void read_rxts(struct phy_device *phydev, struct rxts *rxts)
+{
+ struct phy_rxts p;
+
+ p.ns_lo = ext_read(phydev, PAGE4, PTP_RXTS);
+ p.ns_hi = ext_read(phydev, PAGE4, PTP_RXTS);
+ p.sec_lo = ext_read(phydev, PAGE4, PTP_RXTS);
+ p.sec_hi = ext_read(phydev, PAGE4, PTP_RXTS);
+ p.seqid = ext_read(phydev, PAGE4, PTP_RXTS);
+ p.msgtype = ext_read(phydev, PAGE4, PTP_RXTS);
+
+ rxts->tmo = jiffies + HZ;
+ phy2rxts(&p, rxts);
+}
+
+static u64 read_txts(struct phy_device *phydev)
+{
+ struct phy_txts p;
+
+ p.ns_lo = ext_read(phydev, PAGE4, PTP_TXTS);
+ p.ns_hi = ext_read(phydev, PAGE4, PTP_TXTS);
+ p.sec_lo = ext_read(phydev, PAGE4, PTP_TXTS);
+ p.sec_hi = ext_read(phydev, PAGE4, PTP_TXTS);
+
+ return phy2txts(&p);
+}
+
+static void rx_reading_work(struct dp83640_private *dp83640)
+{
+ struct rxts *rxts;
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ for (;;) {
+ int val = ext_read(dp83640->phydev, PAGE4, PTP_STS);
+ if (!(val & RXTS_RDY))
+ break;
+ if (list_empty(&dp83640->rxpool)) {
+ pr_warning("dp83640: rx timestamp pool is empty\n");
+ break;
+ }
+ rxts = list_first_entry(&dp83640->rxpool, struct rxts, list);
+ list_del_init(&rxts->list);
+ read_rxts(dp83640->phydev, rxts);
+ list_add_tail(&rxts->list, &dp83640->rxts);
+ }
+
+ mutex_unlock(&dp83640->extreg_mux);
+}
+
+static void tx_timestamp_work(struct dp83640_private *dp83640)
+{
+ struct skb_shared_hwtstamps shhwtstamps;
+ struct sk_buff *skb;
+ u64 ns;
+ int val;
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ while ((skb = skb_dequeue(&dp83640->tx_queue)) != NULL) {
+
+ val = ext_read(dp83640->phydev, PAGE4, PTP_STS);
+ if (!(val & TXTS_RDY)) {
+ skb_queue_head(&dp83640->tx_queue, skb);
+ break;
+ }
+ ns = read_txts(dp83640->phydev);
+ memset(&shhwtstamps, 0, sizeof(shhwtstamps));
+ shhwtstamps.hwtstamp = ns_to_ktime(ns);
+ skb_complete_tx_timestamp(skb, &shhwtstamps);
+ }
+
+ mutex_unlock(&dp83640->extreg_mux);
+}
+
+#endif /* !USE_STATUS_FRAMES */
+
+/* time stamping methods */
+
+static void decode_evnt(struct dp83640_private *dp83640,
+ struct phy_txts *phy_txts, u16 ests)
+{
+ struct ptp_clock_event event;
+ int words = (ests >> EVNT_TS_LEN_SHIFT) & EVNT_TS_LEN_MASK;
+
+ switch (words) { /* fall through in every case */
+ case 3:
+ dp83640->edata.sec_hi = phy_txts->sec_hi;
+ case 2:
+ dp83640->edata.sec_lo = phy_txts->sec_lo;
+ case 1:
+ dp83640->edata.ns_hi = phy_txts->ns_hi;
+ case 0:
+ dp83640->edata.ns_lo = phy_txts->ns_lo;
+ }
+
+ event.type = PTP_CLOCK_EXTTS;
+ event.index = 0;
+ event.timestamp = phy2txts(&dp83640->edata);
+
+ ptp_clock_event(dp83640_clock, &event);
+}
+
+static void decode_rxts(struct dp83640_private *dp83640,
+ struct phy_rxts *phy_rxts)
+{
+ struct rxts *rxts;
+ unsigned long flags;
+
+ spin_lock_irqsave(&dp83640->rx_lock, flags);
+
+ if (list_empty(&dp83640->rxpool)) {
+ pr_warning("dp83640: rx timestamp pool is empty\n");
+ goto out;
+ }
+ rxts = list_first_entry(&dp83640->rxpool, struct rxts, list);
+ list_del_init(&rxts->list);
+ phy2rxts(phy_rxts, rxts);
+ list_add_tail(&rxts->list, &dp83640->rxts);
+out:
+ spin_unlock_irqrestore(&dp83640->rx_lock, flags);
+}
+
+static void decode_txts(struct dp83640_private *dp83640,
+ struct phy_txts *phy_txts)
+{
+ struct skb_shared_hwtstamps shhwtstamps;
+ struct sk_buff *skb;
+ u64 ns;
+
+ /* We must already have the skb that triggered this. */
+
+ skb = skb_dequeue(&dp83640->tx_queue);
+
+ if (!skb) {
+ pr_warning("dp83640: have timestamp but tx_queue empty\n");
+ return;
+ }
+ ns = phy2txts(phy_txts);
+ memset(&shhwtstamps, 0, sizeof(shhwtstamps));
+ shhwtstamps.hwtstamp = ns_to_ktime(ns);
+ skb_complete_tx_timestamp(skb, &shhwtstamps);
+}
+
+static void decode_status_frame(struct dp83640_private *dp83640,
+ struct sk_buff *skb)
+{
+ struct phy_rxts *phy_rxts;
+ struct phy_txts *phy_txts;
+ u8 *ptr;
+ int len, size;
+ u16 ests, type;
+
+ ptr = skb->data + 2;
+
+ for (len = skb_headlen(skb) - 2; len > sizeof(type); len -= size) {
+
+ type = *(u16 *)ptr;
+ ests = type & 0x0fff;
+ type = type & 0xf000;
+ len -= sizeof(type);
+ ptr += sizeof(type);
+
+ if (PSF_RX == type && len >= sizeof(*phy_rxts)) {
+
+ phy_rxts = (struct phy_rxts *) ptr;
+ decode_rxts(dp83640, phy_rxts);
+ size = sizeof(*phy_rxts);
+
+ } else if (PSF_TX == type && len >= sizeof(*phy_txts)) {
+
+ phy_txts = (struct phy_txts *) ptr;
+ decode_txts(dp83640, phy_txts);
+ size = sizeof(*phy_txts);
+
+ } else if (PSF_EVNT == type && len >= sizeof(*phy_txts)) {
+
+ phy_txts = (struct phy_txts *) ptr;
+ decode_evnt(dp83640, phy_txts, ests);
+ size = sizeof(*phy_txts);
+
+ } else {
+ size = 0;
+ break;
+ }
+ ptr += size;
+ }
+}
+
+static int expired(struct rxts *rxts)
+{
+ return time_after(jiffies, rxts->tmo);
+}
+
+static int match(struct sk_buff *skb, unsigned int type, struct rxts *rxts)
+{
+ u16 *seqid;
+ u8 *msgtype, *data = skb_mac_header(skb);
+
+ /* check sequenceID, messageType, 12 bit hash of offset 20-29 */
+ /* We assume that the IPv4 header has no options. */
+
+ switch (type) {
+ case PTP_CLASS_V1_IPV4:
+ msgtype = data + 42 + 32;
+ seqid = (u16 *)(data + 42 + 30);
+ break;
+ case PTP_CLASS_V1_IPV6:
+ msgtype = data + 62 + 32;
+ seqid = (u16 *)(data + 62 + 30);
+ break;
+ case PTP_CLASS_V2_IPV4:
+ msgtype = data + 42 + 0;
+ seqid = (u16 *)(data + 42 + 30);
+ break;
+ case PTP_CLASS_V2_IPV6:
+ msgtype = data + 62 + 0;
+ seqid = (u16 *)(data + 62 + 30);
+ break;
+ case PTP_CLASS_V2_L2:
+ msgtype = data + 14 + 0;
+ seqid = (u16 *)(data + 14 + 30);
+ break;
+ case PTP_CLASS_V2_VLAN:
+ msgtype = data + 18 + 0;
+ seqid = (u16 *)(data + 18 + 30);
+ break;
+ default:
+ return 0;
+ }
+
+ return ((*msgtype & 0xf) == rxts->msgtype && *seqid == rxts->seqid);
+}
+
+static int dp83640_probe(struct phy_device *phydev)
+{
+ struct dp83640_private *dp83640;
+ int i;
+
+ dp83640 = kzalloc(sizeof(struct dp83640_private), GFP_KERNEL);
+ if (!dp83640)
+ return -ENOMEM;
+
+ dp83640->phydev = phydev;
+ INIT_WORK(&dp83640->ts_work, do_timestamp_work);
+ mutex_init(&dp83640->extreg_mux);
+
+ INIT_LIST_HEAD(&dp83640->rxts);
+ INIT_LIST_HEAD(&dp83640->rxpool);
+ for (i = 0; i < MAX_RXTS; i++)
+ list_add(&dp83640->rx_pool_data[i].list, &dp83640->rxpool);
+
+ phydev->priv = dp83640;
+
+ spin_lock_init(&dp83640->rx_lock);
+ skb_queue_head_init(&dp83640->rx_queue);
+ skb_queue_head_init(&dp83640->tx_queue);
+
+ mutex_lock(&clock_lock);
+
+ if (!dp83640_clock) {
+ ptp_dp83640_caps.priv = dp83640;
+ dp83640_clock = ptp_clock_register(&ptp_dp83640_caps);
+ if (IS_ERR(dp83640_clock)) {
+ mutex_unlock(&clock_lock);
+ kfree(dp83640);
+ return PTR_ERR(dp83640_clock);
+ }
+ }
+ mutex_unlock(&clock_lock);
+
+ return 0;
+}
+
+static void dp83640_remove(struct phy_device *phydev)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+
+ enable_status_frames(phydev, false);
+
+ cancel_work_sync(&dp83640->ts_work);
+
+ mutex_lock(&clock_lock);
+
+ if (ptp_dp83640_caps.priv == dp83640) {
+ ptp_clock_unregister(dp83640_clock);
+ dp83640_clock = NULL;
+ ptp_dp83640_caps.priv = NULL;
+ }
+ mutex_unlock(&clock_lock);
+
+ mutex_destroy(&dp83640->extreg_lock);
+
+ kfree(dp83640);
+}
+
+static int dp83640_hwtstamp(struct phy_device *phydev, struct ifreq *ifr)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+ struct hwtstamp_config cfg;
+ u16 txcfg0, rxcfg0;
+
+ if (copy_from_user(&cfg, ifr->ifr_data, sizeof(cfg)))
+ return -EFAULT;
+
+ if (cfg.flags) /* reserved for future extensions */
+ return -EINVAL;
+
+ switch (cfg.tx_type) {
+ case HWTSTAMP_TX_OFF:
+ dp83640->hwts_tx_en = 0;
+ break;
+ case HWTSTAMP_TX_ON:
+ dp83640->hwts_tx_en = 1;
+ break;
+ default:
+ return -ERANGE;
+ }
+
+ switch (cfg.rx_filter) {
+ case HWTSTAMP_FILTER_NONE:
+ dp83640->hwts_rx_en = 0;
+ dp83640->layer = 0;
+ dp83640->version = 0;
+ break;
+ case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
+ case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+ case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+ dp83640->hwts_rx_en = 1;
+ dp83640->layer = LAYER4;
+ dp83640->version = 1;
+ break;
+ case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
+ case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
+ case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
+ dp83640->hwts_rx_en = 1;
+ dp83640->layer = LAYER4;
+ dp83640->version = 2;
+ break;
+ case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
+ case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
+ case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
+ dp83640->hwts_rx_en = 1;
+ dp83640->layer = LAYER2;
+ dp83640->version = 2;
+ break;
+ case HWTSTAMP_FILTER_PTP_V2_EVENT:
+ case HWTSTAMP_FILTER_PTP_V2_SYNC:
+ case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
+ dp83640->hwts_rx_en = 1;
+ dp83640->layer = LAYER4|LAYER2;
+ dp83640->version = 2;
+ break;
+ default:
+ return -ERANGE;
+ }
+
+ txcfg0 = (dp83640->version & TX_PTP_VER_MASK) << TX_PTP_VER_SHIFT;
+ rxcfg0 = (dp83640->version & TX_PTP_VER_MASK) << TX_PTP_VER_SHIFT;
+
+ if (dp83640->layer & LAYER2) {
+ txcfg0 |= TX_L2_EN;
+ rxcfg0 |= RX_L2_EN;
+ }
+ if (dp83640->layer & LAYER4) {
+ txcfg0 |= TX_IPV6_EN | TX_IPV4_EN;
+ rxcfg0 |= RX_IPV6_EN | RX_IPV4_EN;
+ }
+
+ if (dp83640->hwts_tx_en)
+ txcfg0 |= TX_TS_EN;
+
+ if (dp83640->hwts_rx_en)
+ rxcfg0 |= RX_TS_EN;
+
+ mutex_lock(&dp83640->extreg_mux);
+
+ if (dp83640->hwts_tx_en || dp83640->hwts_rx_en) {
+ enable_status_frames(phydev, true);
+ ext_write(phydev, PAGE4, PTP_CTL, PTP_ENABLE);
+ }
+
+ ext_write(phydev, PAGE5, PTP_TXCFG0, txcfg0);
+ ext_write(phydev, PAGE5, PTP_RXCFG0, rxcfg0);
+
+ mutex_unlock(&dp83640->extreg_mux);
+
+ return copy_to_user(ifr->ifr_data, &cfg, sizeof(cfg)) ? -EFAULT : 0;
+}
+
+static void rx_timestamp_work(struct dp83640_private *dp83640)
+{
+ struct list_head *this, *next;
+ struct rxts *rxts;
+ struct skb_shared_hwtstamps *shhwtstamps;
+ struct sk_buff *skb;
+ unsigned int type;
+ unsigned long flags;
+
+ /* Deliver each deferred packet, with or without a time stamp. */
+
+ while ((skb = skb_dequeue(&dp83640->rx_queue)) != NULL) {
+ type = SKB_PTP_TYPE(skb);
+ spin_lock_irqsave(&dp83640->rx_lock, flags);
+ list_for_each_safe(this, next, &dp83640->rxts) {
+ rxts = list_entry(this, struct rxts, list);
+ if (match(skb, type, rxts)) {
+ shhwtstamps = skb_hwtstamps(skb);
+ memset(shhwtstamps, 0, sizeof(*shhwtstamps));
+ shhwtstamps->hwtstamp = ns_to_ktime(rxts->ns);
+ list_del_init(&rxts->list);
+ list_add(&rxts->list, &dp83640->rxpool);
+ break;
+ }
+ }
+ spin_unlock_irqrestore(&dp83640->rx_lock, flags);
+ netif_rx(skb);
+ }
+
+ /* Clear out expired time stamps. */
+
+ spin_lock_irqsave(&dp83640->rx_lock, flags);
+ list_for_each_safe(this, next, &dp83640->rxts) {
+ rxts = list_entry(this, struct rxts, list);
+ if (expired(rxts)) {
+ list_del_init(&rxts->list);
+ list_add(&rxts->list, &dp83640->rxpool);
+ }
+ }
+ spin_unlock_irqrestore(&dp83640->rx_lock, flags);
+}
+
+static void do_timestamp_work(struct work_struct *work)
+{
+ struct dp83640_private *dp83640 =
+ container_of(work, struct dp83640_private, ts_work);
+
+ rx_reading_work(dp83640);
+ rx_timestamp_work(dp83640);
+ tx_timestamp_work(dp83640);
+}
+
+static bool dp83640_rxtstamp(struct phy_device *phydev,
+ struct sk_buff *skb, int type)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+
+ if (!dp83640->hwts_rx_en)
+ return false;
+
+ if (is_status_frame(skb, type)) {
+ decode_status_frame(dp83640, skb);
+ /* Let the stack drop this frame. */
+ return false;
+ }
+
+ SKB_PTP_TYPE(skb) = type;
+ skb_queue_tail(&dp83640->rx_queue, skb);
+ schedule_work(&dp83640->ts_work);
+
+ return true;
+}
+
+static void dp83640_txtstamp(struct phy_device *phydev,
+ struct sk_buff *skb, int type)
+{
+ struct dp83640_private *dp83640 = phydev->priv;
+
+ if (!dp83640->hwts_tx_en) {
+ kfree_skb(skb);
+ return;
+ }
+ skb_queue_tail(&dp83640->tx_queue, skb);
+ schedule_work(&dp83640->ts_work);
+}
+
+static struct phy_driver dp83640_driver = {
+ .phy_id = DP83640_PHY_ID,
+ .phy_id_mask = 0xfffffff0,
+ .name = "NatSemi DP83640",
+ .features = PHY_BASIC_FEATURES,
+ .flags = 0,
+ .probe = dp83640_probe,
+ .remove = dp83640_remove,
+ .config_aneg = genphy_config_aneg,
+ .read_status = genphy_read_status,
+ .hwtstamp = dp83640_hwtstamp,
+ .rxtstamp = dp83640_rxtstamp,
+ .txtstamp = dp83640_txtstamp,
+ .driver = {.owner = THIS_MODULE,}
+};
+
+static int __init dp83640_init(void)
+{
+ return phy_driver_register(&dp83640_driver);
+}
+
+static void __exit dp83640_exit(void)
+{
+ phy_driver_unregister(&dp83640_driver);
+}
+
+MODULE_DESCRIPTION("National Semiconductor DP83640 PHY driver");
+MODULE_AUTHOR("Richard Cochran <[email protected]>");
+MODULE_LICENSE("GPL");
+
+module_init(dp83640_init);
+module_exit(dp83640_exit);
+
+static struct mdio_device_id dp83640_tbl[] = {
+ { DP83640_PHY_ID, 0xfffffff0 },
+ { }
+};
+
+MODULE_DEVICE_TABLE(mdio, dp83640_tbl);
diff --git a/drivers/net/phy/dp83640_reg.h b/drivers/net/phy/dp83640_reg.h
new file mode 100644
index 0000000..9d34bb6
--- /dev/null
+++ b/drivers/net/phy/dp83640_reg.h
@@ -0,0 +1,261 @@
+/* dp83640_reg.h
+ * Generated by regen.tcl on Fri Jul 23 09:45:18 AM CEST 2010
+ */
+#ifndef HAVE_DP83640_REGISTERS
+#define HAVE_DP83640_REGISTERS
+
+#define PAGE4 0x0004
+#define PTP_CTL 0x0014 /* PTP Control Register */
+#define PTP_TDR 0x0015 /* PTP Time Data Register */
+#define PTP_STS 0x0016 /* PTP Status Register */
+#define PTP_TSTS 0x0017 /* PTP Trigger Status Register */
+#define PTP_RATEL 0x0018 /* PTP Rate Low Register */
+#define PTP_RATEH 0x0019 /* PTP Rate High Register */
+#define PTP_RDCKSUM 0x001a /* PTP Read Checksum */
+#define PTP_WRCKSUM 0x001b /* PTP Write Checksum */
+#define PTP_TXTS 0x001c /* PTP Transmit Timestamp Register, in four 16-bit reads */
+#define PTP_RXTS 0x001d /* PTP Receive Timestamp Register, in six? 16-bit reads */
+#define PTP_ESTS 0x001e /* PTP Event Status Register */
+#define PTP_EDATA 0x001f /* PTP Event Data Register */
+
+#define PAGE5 0x0005
+#define PTP_TRIG 0x0014 /* PTP Trigger Configuration Register */
+#define PTP_EVNT 0x0015 /* PTP Event Configuration Register */
+#define PTP_TXCFG0 0x0016 /* PTP Transmit Configuration Register 0 */
+#define PTP_TXCFG1 0x0017 /* PTP Transmit Configuration Register 1 */
+#define PSF_CFG0 0x0018 /* PHY Status Frame Configuration Register 0 */
+#define PTP_RXCFG0 0x0019 /* PTP Receive Configuration Register 0 */
+#define PTP_RXCFG1 0x001a /* PTP Receive Configuration Register 1 */
+#define PTP_RXCFG2 0x001b /* PTP Receive Configuration Register 2 */
+#define PTP_RXCFG3 0x001c /* PTP Receive Configuration Register 3 */
+#define PTP_RXCFG4 0x001d /* PTP Receive Configuration Register 4 */
+#define PTP_TRDL 0x001e /* PTP Temporary Rate Duration Low Register */
+#define PTP_TRDH 0x001f /* PTP Temporary Rate Duration High Register */
+
+#define PAGE6 0x0006
+#define PTP_COC 0x0014 /* PTP Clock Output Control Register */
+#define PSF_CFG1 0x0015 /* PHY Status Frame Configuration Register 1 */
+#define PSF_CFG2 0x0016 /* PHY Status Frame Configuration Register 2 */
+#define PSF_CFG3 0x0017 /* PHY Status Frame Configuration Register 3 */
+#define PSF_CFG4 0x0018 /* PHY Status Frame Configuration Register 4 */
+#define PTP_SFDCFG 0x0019 /* PTP SFD Configuration Register */
+#define PTP_INTCTL 0x001a /* PTP Interrupt Control Register */
+#define PTP_CLKSRC 0x001b /* PTP Clock Source Register */
+#define PTP_ETR 0x001c /* PTP Ethernet Type Register */
+#define PTP_OFF 0x001d /* PTP Offset Register */
+#define PTP_GPIOMON 0x001e /* PTP GPIO Monitor Register */
+#define PTP_RXHASH 0x001f /* PTP Receive Hash Register */
+
+/* Bit definitions for the PTP_CTL register */
+#define TRIG_SEL_SHIFT (10) /* PTP Trigger Select */
+#define TRIG_SEL_MASK (0x7)
+#define TRIG_DIS (1<<9) /* Disable PTP Trigger */
+#define TRIG_EN (1<<8) /* Enable PTP Trigger */
+#define TRIG_READ (1<<7) /* Read PTP Trigger */
+#define TRIG_LOAD (1<<6) /* Load PTP Trigger */
+#define PTP_RD_CLK (1<<5) /* Read PTP Clock */
+#define PTP_LOAD_CLK (1<<4) /* Load PTP Clock */
+#define PTP_STEP_CLK (1<<3) /* Step PTP Clock */
+#define PTP_ENABLE (1<<2) /* Enable PTP Clock */
+#define PTP_DISABLE (1<<1) /* Disable PTP Clock */
+#define PTP_RESET (1<<0) /* Reset PTP Clock */
+
+/* Bit definitions for the PTP_STS register */
+#define TXTS_RDY (1<<11) /* Transmit Timestamp Ready */
+#define RXTS_RDY (1<<10) /* Receive Timestamp Ready */
+#define TRIG_DONE (1<<9) /* PTP Trigger Done */
+#define EVENT_RDY (1<<8) /* PTP Event Timestamp Ready */
+#define TXTS_IE (1<<3) /* Transmit Timestamp Interrupt Enable */
+#define RXTS_IE (1<<2) /* Receive Timestamp Interrupt Enable */
+#define TRIG_IE (1<<1) /* Trigger Interrupt Enable */
+#define EVENT_IE (1<<0) /* Event Interrupt Enable */
+
+/* Bit definitions for the PTP_TSTS register */
+#define TRIG7_ERROR (1<<15) /* Trigger 7 Error */
+#define TRIG7_ACTIVE (1<<14) /* Trigger 7 Active */
+#define TRIG6_ERROR (1<<13) /* Trigger 6 Error */
+#define TRIG6_ACTIVE (1<<12) /* Trigger 6 Active */
+#define TRIG5_ERROR (1<<11) /* Trigger 5 Error */
+#define TRIG5_ACTIVE (1<<10) /* Trigger 5 Active */
+#define TRIG4_ERROR (1<<9) /* Trigger 4 Error */
+#define TRIG4_ACTIVE (1<<8) /* Trigger 4 Active */
+#define TRIG3_ERROR (1<<7) /* Trigger 3 Error */
+#define TRIG3_ACTIVE (1<<6) /* Trigger 3 Active */
+#define TRIG2_ERROR (1<<5) /* Trigger 2 Error */
+#define TRIG2_ACTIVE (1<<4) /* Trigger 2 Active */
+#define TRIG1_ERROR (1<<3) /* Trigger 1 Error */
+#define TRIG1_ACTIVE (1<<2) /* Trigger 1 Active */
+#define TRIG0_ERROR (1<<1) /* Trigger 0 Error */
+#define TRIG0_ACTIVE (1<<0) /* Trigger 0 Active */
+
+/* Bit definitions for the PTP_RATEH register */
+#define PTP_RATE_DIR (1<<15) /* PTP Rate Direction */
+#define PTP_TMP_RATE (1<<14) /* PTP Temporary Rate */
+#define PTP_RATE_HI_SHIFT (0) /* PTP Rate High 10-bits */
+#define PTP_RATE_HI_MASK (0x3ff)
+
+/* Bit definitions for the PTP_ESTS register */
+#define EVNTS_MISSED_SHIFT (8) /* Indicates number of events missed */
+#define EVNTS_MISSED_MASK (0x7)
+#define EVNT_TS_LEN_SHIFT (6) /* Indicates length of the Timestamp field in 16-bit words minus 1 */
+#define EVNT_TS_LEN_MASK (0x3)
+#define EVNT_RF (1<<5) /* Indicates whether the event is a rise or falling event */
+#define EVNT_NUM_SHIFT (2) /* Indicates Event Timestamp Unit which detected an event */
+#define EVNT_NUM_MASK (0x7)
+#define MULT_EVNT (1<<1) /* Indicates multiple events were detected at the same time */
+#define EVENT_DET (1<<0) /* PTP Event Detected */
+
+/* Bit definitions for the PTP_EDATA register */
+#define E7_RISE (1<<15) /* Indicates direction of Event 7 */
+#define E7_DET (1<<14) /* Indicates Event 7 detected */
+#define E6_RISE (1<<13) /* Indicates direction of Event 6 */
+#define E6_DET (1<<12) /* Indicates Event 6 detected */
+#define E5_RISE (1<<11) /* Indicates direction of Event 5 */
+#define E5_DET (1<<10) /* Indicates Event 5 detected */
+#define E4_RISE (1<<9) /* Indicates direction of Event 4 */
+#define E4_DET (1<<8) /* Indicates Event 4 detected */
+#define E3_RISE (1<<7) /* Indicates direction of Event 3 */
+#define E3_DET (1<<6) /* Indicates Event 3 detected */
+#define E2_RISE (1<<5) /* Indicates direction of Event 2 */
+#define E2_DET (1<<4) /* Indicates Event 2 detected */
+#define E1_RISE (1<<3) /* Indicates direction of Event 1 */
+#define E1_DET (1<<2) /* Indicates Event 1 detected */
+#define E0_RISE (1<<1) /* Indicates direction of Event 0 */
+#define E0_DET (1<<0) /* Indicates Event 0 detected */
+
+/* Bit definitions for the PTP_TRIG register */
+#define TRIG_PULSE (1<<15) /* generate a Pulse rather than a single edge */
+#define TRIG_PER (1<<14) /* generate a periodic signal */
+#define TRIG_IF_LATE (1<<13) /* trigger immediately if already past */
+#define TRIG_NOTIFY (1<<12) /* Trigger Notification Enable */
+#define TRIG_GPIO_SHIFT (8) /* Trigger GPIO Connection, value 1-12 */
+#define TRIG_GPIO_MASK (0xf)
+#define TRIG_TOGGLE (1<<7) /* Trigger Toggle Mode Enable */
+#define TRIG_CSEL_SHIFT (1) /* Trigger Configuration Select */
+#define TRIG_CSEL_MASK (0x7)
+#define TRIG_WR (1<<0) /* Trigger Configuration Write */
+
+/* Bit definitions for the PTP_EVNT register */
+#define EVNT_RISE (1<<14) /* Event Rise Detect Enable */
+#define EVNT_FALL (1<<13) /* Event Fall Detect Enable */
+#define EVNT_SINGLE (1<<12) /* enable single event capture operation */
+#define EVNT_GPIO_SHIFT (8) /* Event GPIO Connection, value 1-12 */
+#define EVNT_GPIO_MASK (0xf)
+#define EVNT_SEL_SHIFT (1) /* Event Select */
+#define EVNT_SEL_MASK (0x7)
+#define EVNT_WR (1<<0) /* Event Configuration Write */
+
+/* Bit definitions for the PTP_TXCFG0 register */
+#define SYNC_1STEP (1<<15) /* insert timestamp into transmit Sync Messages */
+#define DR_INSERT (1<<13) /* Insert Delay_Req Timestamp in Delay_Resp (dangerous) */
+#define NTP_TS_EN (1<<12) /* Enable Timestamping of NTP Packets */
+#define IGNORE_2STEP (1<<11) /* Ignore Two_Step flag for One-Step operation */
+#define CRC_1STEP (1<<10) /* Disable checking of CRC for One-Step operation */
+#define CHK_1STEP (1<<9) /* Enable UDP Checksum correction for One-Step Operation */
+#define IP1588_EN (1<<8) /* Enable IEEE 1588 defined IP address filter */
+#define TX_L2_EN (1<<7) /* Layer2 Timestamp Enable */
+#define TX_IPV6_EN (1<<6) /* IPv6 Timestamp Enable */
+#define TX_IPV4_EN (1<<5) /* IPv4 Timestamp Enable */
+#define TX_PTP_VER_SHIFT (1) /* Enable Timestamp capture for IEEE 1588 version X */
+#define TX_PTP_VER_MASK (0xf)
+#define TX_TS_EN (1<<0) /* Transmit Timestamp Enable */
+
+/* Bit definitions for the PTP_TXCFG1 register */
+#define BYTE0_MASK_SHIFT (8) /* Bit mask to be used for matching Byte0 of the PTP Message */
+#define BYTE0_MASK_MASK (0xff)
+#define BYTE0_DATA_SHIFT (0) /* Data to be used for matching Byte0 of the PTP Message */
+#define BYTE0_DATA_MASK (0xff)
+
+/* Bit definitions for the PSF_CFG0 register */
+#define MAC_SRC_ADD_SHIFT (11) /* Status Frame Mac Source Address */
+#define MAC_SRC_ADD_MASK (0x3)
+#define MIN_PRE_SHIFT (8) /* Status Frame Minimum Preamble */
+#define MIN_PRE_MASK (0x7)
+#define PSF_ENDIAN (1<<7) /* Status Frame Endian Control */
+#define PSF_IPV4 (1<<6) /* Status Frame IPv4 Enable */
+#define PSF_PCF_RD (1<<5) /* Control Frame Read PHY Status Frame Enable */
+#define PSF_ERR_EN (1<<4) /* Error PHY Status Frame Enable */
+#define PSF_TXTS_EN (1<<3) /* Transmit Timestamp PHY Status Frame Enable */
+#define PSF_RXTS_EN (1<<2) /* Receive Timestamp PHY Status Frame Enable */
+#define PSF_TRIG_EN (1<<1) /* Trigger PHY Status Frame Enable */
+#define PSF_EVNT_EN (1<<0) /* Event PHY Status Frame Enable */
+
+/* Bit definitions for the PTP_RXCFG0 register */
+#define DOMAIN_EN (1<<15) /* Domain Match Enable */
+#define ALT_MAST_DIS (1<<14) /* Alternate Master Timestamp Disable */
+#define USER_IP_SEL (1<<13) /* Selects portion of IP address accessible thru PTP_RXCFG2 */
+#define USER_IP_EN (1<<12) /* Enable User-programmed IP address filter */
+#define RX_SLAVE (1<<11) /* Receive Slave Only */
+#define IP1588_EN_SHIFT (8) /* Enable IEEE 1588 defined IP address filters */
+#define IP1588_EN_MASK (0xf)
+#define RX_L2_EN (1<<7) /* Layer2 Timestamp Enable */
+#define RX_IPV6_EN (1<<6) /* IPv6 Timestamp Enable */
+#define RX_IPV4_EN (1<<5) /* IPv4 Timestamp Enable */
+#define RX_PTP_VER_SHIFT (1) /* Enable Timestamp capture for IEEE 1588 version X */
+#define RX_PTP_VER_MASK (0xf)
+#define RX_TS_EN (1<<0) /* Receive Timestamp Enable */
+
+/* Bit definitions for the PTP_RXCFG1 register */
+#define BYTE0_MASK_SHIFT (8) /* Bit mask to be used for matching Byte0 of the PTP Message */
+#define BYTE0_MASK_MASK (0xff)
+#define BYTE0_DATA_SHIFT (0) /* Data to be used for matching Byte0 of the PTP Message */
+#define BYTE0_DATA_MASK (0xff)
+
+/* Bit definitions for the PTP_RXCFG3 register */
+#define TS_MIN_IFG_SHIFT (12) /* Minimum Inter-frame Gap */
+#define TS_MIN_IFG_MASK (0xf)
+#define ACC_UDP (1<<11) /* Record Timestamp if UDP Checksum Error */
+#define ACC_CRC (1<<10) /* Record Timestamp if CRC Error */
+#define TS_APPEND (1<<9) /* Append Timestamp for L2 */
+#define TS_INSERT (1<<8) /* Enable Timestamp Insertion */
+#define PTP_DOMAIN_SHIFT (0) /* PTP Message domainNumber field */
+#define PTP_DOMAIN_MASK (0xff)
+
+/* Bit definitions for the PTP_RXCFG4 register */
+#define IPV4_UDP_MOD (1<<15) /* Enable IPV4 UDP Modification */
+#define TS_SEC_EN (1<<14) /* Enable Timestamp Seconds */
+#define TS_SEC_LEN_SHIFT (12) /* Inserted Timestamp Seconds Length */
+#define TS_SEC_LEN_MASK (0x3)
+#define RXTS_NS_OFF_SHIFT (6) /* Receive Timestamp Nanoseconds offset */
+#define RXTS_NS_OFF_MASK (0x3f)
+#define RXTS_SEC_OFF_SHIFT (0) /* Receive Timestamp Seconds offset */
+#define RXTS_SEC_OFF_MASK (0x3f)
+
+/* Bit definitions for the PTP_COC register */
+#define PTP_CLKOUT_EN (1<<15) /* PTP Clock Output Enable */
+#define PTP_CLKOUT_SEL (1<<14) /* PTP Clock Output Source Select */
+#define PTP_CLKOUT_SPEEDSEL (1<<13) /* PTP Clock Output I/O Speed Select */
+#define PTP_CLKDIV_SHIFT (0) /* PTP Clock Divide-by Value */
+#define PTP_CLKDIV_MASK (0xff)
+
+/* Bit definitions for the PSF_CFG1 register */
+#define PTPRESERVED_SHIFT (12) /* PTP v2 reserved field */
+#define PTPRESERVED_MASK (0xf)
+#define VERSIONPTP_SHIFT (8) /* PTP v2 versionPTP field */
+#define VERSIONPTP_MASK (0xf)
+#define TRANSPORT_SPECIFIC_SHIFT (4) /* PTP v2 Header transportSpecific field */
+#define TRANSPORT_SPECIFIC_MASK (0xf)
+#define MESSAGETYPE_SHIFT (0) /* PTP v2 messageType field */
+#define MESSAGETYPE_MASK (0xf)
+
+/* Bit definitions for the PTP_SFDCFG register */
+#define TX_SFD_GPIO_SHIFT (4) /* TX SFD GPIO Select, value 1-12 */
+#define TX_SFD_GPIO_MASK (0xf)
+#define RX_SFD_GPIO_SHIFT (0) /* RX SFD GPIO Select, value 1-12 */
+#define RX_SFD_GPIO_MASK (0xf)
+
+/* Bit definitions for the PTP_INTCTL register */
+#define PTP_INT_GPIO_SHIFT (0) /* PTP Interrupt GPIO Select */
+#define PTP_INT_GPIO_MASK (0xf)
+
+/* Bit definitions for the PTP_CLKSRC register */
+#define CLK_SRC_SHIFT (14) /* PTP Clock Source Select */
+#define CLK_SRC_MASK (0x3)
+#define CLK_SRC_PER_SHIFT (0) /* PTP Clock Source Period */
+#define CLK_SRC_PER_MASK (0x7f)
+
+/* Bit definitions for the PTP_OFF register */
+#define PTP_OFFSET_SHIFT (0) /* PTP Message offset from preceding header */
+#define PTP_OFFSET_MASK (0xff)
+
+#endif
--
1.7.0.4
On Thursday 16 December 2010, Richard Cochran wrote:
> A new syscall is introduced that allows tuning of a POSIX clock. The
> syscall is implemented for four architectures: arm, blackfin, powerpc,
> and x86.
>
> The new syscall, clock_adjtime, takes two parameters, the clock ID,
> and a pointer to a struct timex. Any ADJTIMEX(2) operation may be
> requested via this system call, but various POSIX clocks may or may
> not support tuning.
>
> Signed-off-by: Richard Cochran <[email protected]>
The syscall implementation looks good to me.
Acked-by: Arnd Bergmann <[email protected]>
It would be good to add it to include/asm-generic/unistd.h
as well, so that tile and future architectures get it, too.
On Thursday 16 December 2010, Richard Cochran wrote:
> This patch adds an infrastructure for hardware clocks that implement
> IEEE 1588, the Precision Time Protocol (PTP). A class driver offers a
> registration method to particular hardware clock drivers. Each clock is
> presented as a standard POSIX clock.
>
> The ancillary clock features are exposed in two different ways, via
> the sysfs and by a character device.
>
> Signed-off-by: Richard Cochran <[email protected]>
Looks fine to me, with one tiny comment:
> +
> +/* private globals */
> +
> +extern struct device_attribute ptp_dev_attrs[]; /* see ptp_sysfs.c */
The declaration is in a C file, better move it into a header in order
to make sure the definition matches the declaration.
Arnd
On Thu, Dec 16, 2010 at 04:44:24PM +0100, Richard Cochran wrote:
> This patch adds an infrastructure for hardware clocks that implement
> IEEE 1588, the Precision Time Protocol (PTP). A class driver offers a
> registration method to particular hardware clock drivers. Each clock is
> presented as a standard POSIX clock.
>
> The ancillary clock features are exposed in two different ways, via
> the sysfs and by a character device.
>
> Signed-off-by: Richard Cochran <[email protected]>
Regarding the PPS subsystem it looks ok for me.
Acked-by: Rodolfo Giometti <[email protected]>
Ciao,
Rodolfo
--
GNU/Linux Solutions e-mail: [email protected]
Linux Device Driver [email protected]
Embedded Systems phone: +39 349 2432127
UNIX programming skype: rodolfo.giometti
Freelance ICT Italia - Consulente ICT Italia - http://www.consulenti-ict.it
On Thursday 16 December 2010, Richard Cochran wrote:
> This patch adds support for adding and removing posix clocks. The
> clock lifetime cycle is patterned after usb devices. Each clock is
> represented by a standard character device. In addition, the driver
> may optionally implemented custom character device operations.
>
> The dynamic posix clocks do not yet do anything useful. This patch
> merely provides some needed infrastructure.
>
> Signed-off-by: Richard Cochran <[email protected]>
> +struct posix_clock_fops {
> + int (*fasync) (void *priv, int fd, struct file *file, int on);
> + int (*mmap) (void *priv, struct vm_area_struct *vma);
> + int (*open) (void *priv, fmode_t f_mode);
> + int (*release) (void *priv);
> + long (*ioctl) (void *priv, unsigned int cmd, unsigned long arg);
> + long (*compat_ioctl) (void *priv, unsigned int cmd, unsigned long arg);
> + ssize_t (*read) (void *priv, uint flags, char __user *buf, size_t cnt);
> + unsigned int (*poll) (void *priv, struct file *file, poll_table *wait);
> +};
Thanks for the change to a private ops structure. Three more
suggestions for this:
* I would recommend starting without a compat_ioctl operation if you can.
Just mandate that all ioctls for posix clocks are compatible and call
fops->ioctl from the posix_clock_compat_ioctl() function.
If you ever need compat_ioctl handling, it can still be added later.
* Instead of passing a void pointer, why not pass a struct posix_clock
pointer to the lower device and use container_of? That follows what
we do in other subsystems and allows you save one allocation, by
including the posix_clock into the specific clock struct like
ptp_clock.
> +struct posix_clock_operations {
> + struct module *owner;
> + struct posix_clock_fops fops;
> + int (*clock_adjtime)(void *priv, struct timex *tx);
You can easily combine the two structures and get rid of the extra
struct posix_clock_fops by moving its operations into
posix_clock_operations.
Looks really good otherwise.
Arnd
On Thu, 16 Dec 2010, Richard Cochran wrote:
> + if (txc->modes & ADJ_SETOFFSET) {
> + /* Validate the delta value. */
> + if (txc->time.tv_sec && txc->time.tv_usec < 0)
> + return -EINVAL;
> +
> + if (txc->modes & ADJ_NANO) {
> + struct timespec tmp;
> + tmp.tv_sec = txc->time.tv_sec;
> + tmp.tv_nsec = txc->time.tv_usec;
> + delta = timespec_to_ktime(tmp);
> + } else
> + delta = timeval_to_ktime(txc->time);
> +
> + /* Adding the delta should be an "atomic" operation. */
> + local_irq_disable();
I really do not like that conditional irq_disable(), especially as we
disable them further down again and we only safe the getnstimeofday()
call in the non ADJSETOFFSET code path.
So we really better do that unconditionally before getnstimeofday()
with an appropriate comment and change the write_seqlock_irq() to
write_seqlock().
> + }
> +
> getnstimeofday(&ts);
>
> + if (txc->modes & ADJ_SETOFFSET) {
> + kt = timespec_to_ktime(ts);
> + kt = ktime_add(kt, delta);
> + ts = ktime_to_timespec(kt);
> + do_settimeofday(&ts);
> + local_irq_enable();
> + }
> +
> write_seqlock_irq(&xtime_lock);
Thanks,
tglx
>
On Thu, 16 Dec 2010, Richard Cochran wrote:
> A new syscall is introduced that allows tuning of a POSIX clock. The
> syscall is implemented for four architectures: arm, blackfin, powerpc,
> and x86.
Can you please split that patch into one providing the syscall itself
and for each architecture a separate one ?
Looks good otherwise.
Reviewed-by: Thomas Gleixner <[email protected]>
On Thu, 16 Dec 2010, Richard Cochran wrote:
> --- /dev/null
> +++ b/include/linux/posix-clock.h
> +/**
> + * struct posix_clock_fops - character device operations
> + *
> + * Every posix clock is represented by a character device. Drivers may
> + * optionally offer extended capabilities by implementing these
> + * functions. The character device file operations are first handled
> + * by the clock device layer, then passed on to the driver by calling
> + * these functions.
> + *
> + * The clock device layer already uses fp->private_data, so drivers
> + * are provided their private data via the 'priv' paramenter.
> + */
> +void *posix_clock_private(struct file *fp);
Leftover ? There is neither a caller nor an implementation
> +struct posix_clock_fops {
> + int (*fasync) (void *priv, int fd, struct file *file, int on);
> + int (*mmap) (void *priv, struct vm_area_struct *vma);
> + int (*open) (void *priv, fmode_t f_mode);
> + int (*release) (void *priv);
> + long (*ioctl) (void *priv, unsigned int cmd, unsigned long arg);
> + long (*compat_ioctl) (void *priv, unsigned int cmd, unsigned long arg);
Do we really need a compat_ioctl ?
> + ssize_t (*read) (void *priv, uint flags, char __user *buf, size_t cnt);
> + unsigned int (*poll) (void *priv, struct file *file, poll_table *wait);
> --- a/kernel/time/Makefile
> +++ b/kernel/time/Makefile
> @@ -1,4 +1,5 @@
> -obj-y += timekeeping.o ntp.o clocksource.o jiffies.o timer_list.o timecompare.o timeconv.o
> +obj-y += timekeeping.o ntp.o clocksource.o jiffies.o timer_list.o \
> +timecompare.o timeconv.o posix-clock.o
Please start a new obj-y += line instead
> --- /dev/null
> +++ b/kernel/time/posix-clock.c
> +
> +#define MAX_CLKDEV BITS_PER_LONG
Please put some more space between the MAX_CLKDEV and BITS_PER_LONG. I
had to look three times to not read it as a single word.
> +static DECLARE_BITMAP(clocks_map, MAX_CLKDEV);
> +static DEFINE_MUTEX(clocks_mutex); /* protects 'clocks_map' */
Please avoid tail comments
> +struct posix_clock {
> + struct posix_clock_operations *ops;
> + struct cdev cdev;
> + struct kref kref;
> + struct mutex mux;
> + void *priv;
You can get away with that private pointer and all the void *
arguments to the various posix_clock_operations, if you mandate that
the posix_clock_operations are embedded into a clock specific data
structure.
So void *priv would become struct posix_clock_operations *clkops and
you can get your private data in the clock implementation with
container_of().
> + int index;
> + bool zombie;
Ths field is only set, but nowhere else used. What's the purpose ?
Leftover ?
> +};
> +
> +static void delete_clock(struct kref *kref);
> +
> +
> +static int posix_clock_open(struct inode *inode, struct file *fp)
> +{
> + struct posix_clock *clk =
> + container_of(inode->i_cdev, struct posix_clock, cdev);
> +
> + kref_get(&clk->kref);
> + fp->private_data = clk;
fp->private_data should only be set on success. And this will leak a
ref count when the clock open function fails.
What's that kref protecting here ?
> +
> + if (clk->ops->fops.open)
> + return clk->ops->fops.open(clk->priv, fp->f_mode);
> + else
> + return 0;
> +}
> +
> +static int posix_clock_release(struct inode *inode, struct file *fp)
> +{
> + struct posix_clock *clk = fp->private_data;
> + int err = 0;
fp->private_data should be set to NULL in the release function.
> + if (clk->ops->fops.release)
> + err = clk->ops->fops.release(clk->priv);
> +
> + kref_put(&clk->kref, delete_clock);
> +struct posix_clock *posix_clock_create(struct posix_clock_operations *cops,
> + dev_t devid, void *priv)
> +{
> + struct posix_clock *clk;
> + int err;
> +
> + mutex_lock(&clocks_mutex);
> +
> + err = -ENOMEM;
> + clk = kzalloc(sizeof(*clk), GFP_KERNEL);
> + if (!clk)
> + goto no_memory;
> +
> + err = -EBUSY;
> + clk->index = find_first_zero_bit(clocks_map, MAX_CLKDEV);
> + if (clk->index < MAX_CLKDEV)
> + set_bit(clk->index, clocks_map);
> + else
> + goto no_index;
if (clk->index >= MAX_CLKDEV)
goto no_index;
set_bit(clk->index, clocks_map);
Makes it better readable.
> +static void delete_clock(struct kref *kref)
> +{
> + struct posix_clock *clk =
> + container_of(kref, struct posix_clock, kref);
> +
> + mutex_lock(&clocks_mutex);
> + clear_bit(clk->index, clocks_map);
> + mutex_unlock(&clocks_mutex);
> +
> + mutex_destroy(&clk->mux);
> + kfree(clk);
> +}
> +
> +void posix_clock_destroy(struct posix_clock *clk)
> +{
> + cdev_del(&clk->cdev);
> +
> + mutex_lock(&clk->mux);
> + clk->zombie = true;
> + mutex_unlock(&clk->mux);
> +
> + kref_put(&clk->kref, delete_clock);
I still have some headache to understand that kref / delete_clock
magic here.
Thanks,
tglx
On Thu, 16 Dec 2010, Richard Cochran wrote:
Can you please split this into infrastructure and conversion of
posix-timer.c ?
> diff --git a/include/linux/posix-clock.h b/include/linux/posix-clock.h
> index 1ce7fb7..7ea94f5 100644
> --- a/include/linux/posix-clock.h
> +++ b/include/linux/posix-clock.h
> @@ -108,4 +108,29 @@ struct posix_clock *posix_clock_create(struct posix_clock_operations *cops,
> */
> void posix_clock_destroy(struct posix_clock *clk);
>
> +/**
> + * clockid_to_posix_clock() - obtain clock device pointer from a clock id
> + * @id: user space clock id
> + *
> + * Returns a pointer to the clock device, or NULL if the id is not a
> + * dynamic clock id.
> + */
> +struct posix_clock *clockid_to_posix_clock(clockid_t id);
> +
> +/*
> + * The following functions are only to be called from posix-timers.c
> + */
Then they should be in a header file which is not consumable by the
general public. e.g. kernel/time/posix-clock.h
> +int pc_clock_adjtime(struct posix_clock *clk, struct timex *tx);
> +int pc_clock_gettime(struct posix_clock *clk, struct timespec *ts);
> +int pc_clock_getres(struct posix_clock *clk, struct timespec *ts);
> +int pc_clock_settime(struct posix_clock *clk, const struct timespec *ts);
> +
> +int pc_timer_create(struct posix_clock *clk, struct k_itimer *kit);
> +int pc_timer_delete(struct posix_clock *clk, struct k_itimer *kit);
> +
> +void pc_timer_gettime(struct posix_clock *clk, struct k_itimer *kit,
> + struct itimerspec *ts);
> +int pc_timer_settime(struct posix_clock *clk, struct k_itimer *kit,
> + int flags, struct itimerspec *ts, struct itimerspec *old);
> +
> #endif
> +
> +static inline bool clock_is_static(const clockid_t id)
> +{
> + if (0 == (id & ~CLOCKFD_MASK))
> + return true;
> + if (CLOCKFD == (id & CLOCKFD_MASK))
> + return false;
Please use the usual kernel notation.
> + return true;
> +}
> +
> /* POSIX.1b interval timer structure. */
> struct k_itimer {
> struct list_head list; /* free/ allocate list */
> diff --git a/include/linux/time.h b/include/linux/time.h
> index 9f15ac7..914c48d 100644
> --- a/include/linux/time.h
> +++ b/include/linux/time.h
> @@ -299,6 +299,8 @@ struct itimerval {
> #define CLOCKS_MASK (CLOCK_REALTIME | CLOCK_MONOTONIC)
> #define CLOCKS_MONO CLOCK_MONOTONIC
>
> +#define CLOCK_INVALID -1
> +
> @@ -712,6 +720,7 @@ common_timer_get(struct k_itimer *timr, struct itimerspec *cur_setting)
> SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
> struct itimerspec __user *, setting)
> {
> + struct posix_clock *clk_dev;
> struct k_itimer *timr;
> struct itimerspec cur_setting;
> unsigned long flags;
> @@ -720,7 +729,15 @@ SYSCALL_DEFINE2(timer_gettime, timer_t, timer_id,
> if (!timr)
> return -EINVAL;
>
> - CLOCK_DISPATCH(timr->it_clock, timer_get, (timr, &cur_setting));
> + if (likely(clock_is_static(timr->it_clock))) {
> +
> + CLOCK_DISPATCH(timr->it_clock, timer_get, (timr, &cur_setting));
> +
> + } else {
> + clk_dev = clockid_to_posix_clock(timr->it_clock);
> + if (clk_dev)
> + pc_timer_gettime(clk_dev, timr, &cur_setting);
Why this extra step ? Why can't you call
pc_timer_gettime(timr, &cur_setting);
You already established that the timer is fd based, so let the
pc_timer_* functions deal with it.
> + }
>
> unlock_timer(timr, flags);
>
> @@ -811,6 +828,7 @@ SYSCALL_DEFINE4(timer_settime, timer_t, timer_id, int, flags,
> const struct itimerspec __user *, new_setting,
> struct itimerspec __user *, old_setting)
> {
> + struct posix_clock *clk_dev;
> struct k_itimer *timr;
> struct itimerspec new_spec, old_spec;
> int error = 0;
> @@ -831,8 +849,19 @@ retry:
> if (!timr)
> return -EINVAL;
>
> - error = CLOCK_DISPATCH(timr->it_clock, timer_set,
> - (timr, flags, &new_spec, rtn));
> + if (likely(clock_is_static(timr->it_clock))) {
> +
> + error = CLOCK_DISPATCH(timr->it_clock, timer_set,
> + (timr, flags, &new_spec, rtn));
> +
> + } else {
> + clk_dev = clockid_to_posix_clock(timr->it_clock);
> + if (clk_dev)
> + error = pc_timer_settime(clk_dev, timr,
> + flags, &new_spec, rtn);
> + else
> + error = -EINVAL;
Ditto. pc_timer_settime() can return -EINVAL when there is no valid fd.
> @@ -957,26 +993,51 @@ EXPORT_SYMBOL_GPL(do_posix_clock_nonanosleep);
> SYSCALL_DEFINE2(clock_settime, const clockid_t, which_clock,
> const struct timespec __user *, tp)
> {
> + struct posix_clock *clk_dev = NULL;
> struct timespec new_tp;
>
> - if (invalid_clockid(which_clock))
> - return -EINVAL;
> + if (likely(clock_is_static(which_clock))) {
> +
> + if (invalid_clockid(which_clock))
> + return -EINVAL;
> + } else {
> + clk_dev = clockid_to_posix_clock(which_clock);
> + if (!clk_dev)
> + return -EINVAL;
> + }
It's not a problem to check the validity of that clock fd after the
copy_from_user. That's an error case and we don't care about whether
we return EINVAL here or later. And with your current code this can
happen anyway as you don't hold a reference to the fd. And we do the
same thing for posix_cpu_timers as well. See invalid_clockid()
> if (copy_from_user(&new_tp, tp, sizeof (*tp)))
> return -EFAULT;
>
> + if (unlikely(clk_dev))
> + return pc_clock_settime(clk_dev, &new_tp);
> +
> return CLOCK_DISPATCH(which_clock, clock_set, (which_clock, &new_tp));
I really start to wonder whether we should cleanup that whole
CLOCK_DISPATCH macro magic and provide real inline functions for
each of the clock functions instead.
static inline int dispatch_clock_settime(const clockid_t id, struct timespec *tp)
{
if (id >= 0) {
return posix_clocks[id].clock_set ?
posic_clocks[id].clock_set(id, tp) : -EINVAL;
}
if (posix_cpu_clock(id))
return -EINVAL;
return pc_timers_clock_set(id, tp);
}
That is a bit of work, but the code will be simpler and we just do the
normal thing of function pointer structures. Stuff which is not
implemented does not magically become called via some common
function. There is no point in doing that. We just have to fill in the
various k_clock structs with the correct pointers in the first place
and let the NULL case return a sensible error value. The data
structure does not become larger that way. It's a little bit more init
code, but that's fine if we make the code better in general. In that
case it's not even more init code, it's just filling the data
structures which we register.
That needs to be done in two steps:
1) cleanup CLOCK_DISPATCH
2) add the pc_timer_* extras
That will make the thing way more palatable than working around the
restrictions of CLOCK_DISPATCH and adding the hell to each syscall.
The second step would be a patch consisting of exactly nine lines:
{
if (id >= 0)
return .....
if (posix_cpu_clock_id(id))
return ....
- return -EINVAL;
+ return pc_timer_xxx(...);
}
Please don't tell me now that we could even hack this into
CLOCK_DISPATCH. *shudder*
> +
> +struct posix_clock *clockid_to_posix_clock(const clockid_t id)
> +{
> + struct posix_clock *clk = NULL;
> + struct file *fp;
> +
> + if (clock_is_static(id))
> + return NULL;
> +
> + fp = fget(CLOCKID_TO_FD(id));
> + if (!fp)
> + return NULL;
> +
> + if (fp->f_op->open == posix_clock_open)
> + clk = fp->private_data;
> +
> + fput(fp);
> + return clk;
> +}
> +
> +int pc_clock_adjtime(struct posix_clock *clk, struct timex *tx)
> +{
> + int err;
> +
> + mutex_lock(&clk->mux);
> +
> + if (clk->zombie)
Uuurgh. That explains the zombie thing. That's really horrible and we
can be more clever. When we leave the file lookup to the pc_timer_*
functions, then we can simply do:
struct posix_clock_descr {
struct file *fp;
struct posix_clock *clk;
};
static int get_posix_clock(const clockid_t id, struct posix_clock_descr *cd)
{
struct file *fp = fget(CLOCKID_TO_FD(id));
if (!fp || fp->f_op->open != posix_clock_open || !fp->private_data)
return -ENODEV;
cd->fp = fp;
cd->clk = fp->private_data;
return 0;
}
static void put_posix_clock(struct posix_clock_descr *cd)
{
fput(cd->fp);
}
int pc_timer_****(....)
{
struct posix_clock_descr cd;
ret = -EOPNOTSUPP;
if (get_posix_clock(id, &cd))
return -ENODEV;
if (cd.clk->ops->whatever)
ret = cd.clk->ops->whatever(....);
put_posix_clock(&cd);
return ret;
}
That get's rid of your life time problems, of clk->mutex, clock->zombie
and makes the code simpler and more robust. And it avoids the whole
mess in posix-timers.c as well.
> + err = -ENODEV;
> + else if (!clk->ops->clock_adjtime)
> + err = -EOPNOTSUPP;
> + else
> + err = clk->ops->clock_adjtime(clk->priv, tx);
> +
> + mutex_unlock(&clk->mux);
> + return err;
> +}
Though all this feels still a bit backwards.
The whole point of this exercise is to provide dynamic clock ids and
make them available via the standard posix timer interface and aside
of that have standard file operations for these clocks to provide
special purpose ioctls, mmap and whatever. And to make it a bit more
complex these must be removable modules.
I need to read back the previous discussions, but maybe someone can
fill me in why we can't make these dynamic things not just live in
posix_clocks[MAX_CLOCKS ... MAX_DYNAMIC_CLOCKS] (there is no
requirement to have an unlimited number of those) and just get a
mapping from the fd to the clock_id? That creates a different set of
life time problems, but no real horrible ones.
Thanks,
tglx
On Thu, Dec 16, 2010 at 09:56:38PM +0100, Thomas Gleixner wrote:
> > +void *posix_clock_private(struct file *fp);
>
> Leftover ? There is neither a caller nor an implementation
Yes, you are right. Sorry.
This patch series has been through several contortions, and this
current one won't be the last!
As to the other comments, thanks for your careful review. I will adapt
the patch set accordingly.
Richard
On Fri, Dec 17, 2010 at 12:20:44AM +0100, Thomas Gleixner wrote:
> On Thu, 16 Dec 2010, Richard Cochran wrote:
> > +
> > +static inline bool clock_is_static(const clockid_t id)
> > +{
> > + if (0 == (id & ~CLOCKFD_MASK))
> > + return true;
> > + if (CLOCKFD == (id & CLOCKFD_MASK))
> > + return false;
>
> Please use the usual kernel notation.
Sorry, what do you mean?
> > return CLOCK_DISPATCH(which_clock, clock_set, (which_clock, &new_tp));
>
> I really start to wonder whether we should cleanup that whole
> CLOCK_DISPATCH macro magic and provide real inline functions for
> each of the clock functions instead.
I'm glad you suggested that.
IMHO, the CLOCK_DISPATCH thing is calling out to be eliminated, but I
didn't dare to take that upon myself.
> Please don't tell me now that we could even hack this into
> CLOCK_DISPATCH. *shudder*
;^)
> > +int pc_clock_adjtime(struct posix_clock *clk, struct timex *tx)
> > +{
> > + int err;
> > +
> > + mutex_lock(&clk->mux);
> > +
> > + if (clk->zombie)
>
> Uuurgh. That explains the zombie thing. That's really horrible and we
> can be more clever. When we leave the file lookup to the pc_timer_*
> functions, then we can simply do:
...
> That get's rid of your life time problems, of clk->mutex, clock->zombie
> and makes the code simpler and more robust. And it avoids the whole
> mess in posix-timers.c as well.
The code in the patch set is modeled after a USB driver, namely
drivers/usb/class/usbtmc.c. Alan Cox had brought up the example of a
PTP Hardware Clock appearing as a USB device. So the device might
suddenly disappear, and the zombie field is supposed to cover the case
where the hardware no longer exists, but the file pointer is still
valid.
I'll take a closer look at your suggestion...
> The whole point of this exercise is to provide dynamic clock ids and
> make them available via the standard posix timer interface and aside
> of that have standard file operations for these clocks to provide
> special purpose ioctls, mmap and whatever. And to make it a bit more
> complex these must be removable modules.
Yes, and hot pluggable, too.
> I need to read back the previous discussions, but maybe someone can
> fill me in why we can't make these dynamic things not just live in
> posix_clocks[MAX_CLOCKS ... MAX_DYNAMIC_CLOCKS] (there is no
> requirement to have an unlimited number of those) and just get a
> mapping from the fd to the clock_id? That creates a different set of
> life time problems, but no real horrible ones.
I would summarize the discussion like this:
Alan Cox was strongly in favor of using a regular file descriptor as
the reference to the dynamic clock.
John Stultz thought it wouldn't be too bad to cycle though a number of
static ids, being careful not to every assign the same static id (in
series) to two different clocks.
I implemented Alan's idea, since it seemed like maintaining the
mapping between clocks and static ids would be extra work, but without
any real benefit.
Thanks again for your review,
Richard
On Fri, 17 Dec 2010, Richard Cochran wrote:
> On Fri, Dec 17, 2010 at 12:20:44AM +0100, Thomas Gleixner wrote:
> > > +static inline bool clock_is_static(const clockid_t id)
> > > +{
> > > + if (0 == (id & ~CLOCKFD_MASK))
> > > + return true;
> > > + if (CLOCKFD == (id & CLOCKFD_MASK))
> > > + return false;
> >
> > Please use the usual kernel notation.
>
> Sorry, what do you mean?
if (!id & ~CLOCKFD_MASK)
if ((id & CLOCKFD_MASK) == CLOCKFD)
Not a real problem. I just always stumble over that coding style.
> The code in the patch set is modeled after a USB driver, namely
> drivers/usb/class/usbtmc.c. Alan Cox had brought up the example of a
> PTP Hardware Clock appearing as a USB device. So the device might
> suddenly disappear, and the zombie field is supposed to cover the case
> where the hardware no longer exists, but the file pointer is still
> valid.
Hmm ok, so you use clk->mutex to prevent the underlying device from
going away while you call a function and clk->zombie indicates that
it's gone and clk just kept around for an open file descriptor. Now it
makes sense, but that wants a big fat comment in the code. :)
Doesn't the same problem exist for the file operations of patch 3? I
think you want the very same protection there unless you want to do
the same magic in your USB driver as well.
So you could do the following:
static int get_fd_clk(struct posix_clock_descr *cd, struct file *fp)
{
cd->fp = fp;
cd->clk = fp->private_data;
mutex_lock(&cd->clk->mutex);
if (!cd->clk->zombie)
return 0;
mutex_unlock(&cd->clk->mutex);
return -ENODEV;
}
static void put_fd_clk(struct posix_clock_descr *cd)
{
mutex_unlock(&cd->clk->mutex);
}
static int get_posix_clock(const clockid_t id, struct posix_clock_descr *cd)
{
struct file *fp = fget(CLOCKID_TO_FD(id));
int ret;
if (!fp || fp->f_op->open != posix_clock_open || !fp->private_data)
return -ENODEV;
ret = get_fd_clk(cd, fp);
if (ret)
fput(fp);
return ret;
}
static void put_posix_clock(struct posix_clock_descr *cd)
{
put_fd_clk(cd);
fput(cd->fp);
}
So your fops would call get_fd_clk() and put_fd_clk() and the pc_timer
ones get_posix_clock() and put_posix_clock() or whatever sensible
function names you come up with.
Thoughts ?
> I would summarize the discussion like this:
>
> Alan Cox was strongly in favor of using a regular file descriptor as
> the reference to the dynamic clock.
>
> John Stultz thought it wouldn't be too bad to cycle though a number of
> static ids, being careful not to every assign the same static id (in
> series) to two different clocks.
>
> I implemented Alan's idea, since it seemed like maintaining the
> mapping between clocks and static ids would be extra work, but without
> any real benefit.
Yes, he's right. Was too tired to think about the clock ids assingment
when devices are removed and plugged. The fd mapping makes that all go
away.
Thanks,
tglx
On 12/17/10, Richard Cochran <[email protected]> wrote:
> This patch adds a new mode bit into the timex structure. When set, the bit
> instructs the kernel to add the given time value to the current time.
>
The proposed new control mode, ADJ_SETOFFSET, is logically the same as
ADJ_OFFSET with timex.constant == -INFINITY. So it is possible to do
the same thing without risking forward compatibility. (I mean by "risking
forward compatibility" that the mode bit 0x0040 may be defined differently
by the upstream maintainer anytime in the future.)
On Sat, Dec 18, 2010 at 05:16:52AM +0900, Kuwahara,T. wrote:
> On 12/17/10, Richard Cochran <[email protected]> wrote:
> > This patch adds a new mode bit into the timex structure. When set, the bit
> > instructs the kernel to add the given time value to the current time.
> >
>
> The proposed new control mode, ADJ_SETOFFSET, is logically the same as
> ADJ_OFFSET with timex.constant == -INFINITY. So it is possible to do
> the same thing without risking forward compatibility. (I mean by "risking
> forward compatibility" that the mode bit 0x0040 may be defined differently
> by the upstream maintainer anytime in the future.)
Can you please elaborate?
I don't see any way to use timex.constant with ADJ_OFFSET in order to
correct a time offset. The 'time_constant' in kernel/time/ntp.c is
restricted to the interval [0..MAXTC], and MAXTC is 10 in timex.h.
Richard
On Fri, Dec 17, 2010 at 11:03:46AM +0100, Thomas Gleixner wrote:
>
> Doesn't the same problem exist for the file operations of patch 3? I
> think you want the very same protection there unless you want to do
> the same magic in your USB driver as well.
Yes, you are right. I overlooked that.
> So you could do the following:
...
> So your fops would call get_fd_clk() and put_fd_clk() and the pc_timer
> ones get_posix_clock() and put_posix_clock() or whatever sensible
> function names you come up with.
>
> Thoughts ?
It is a good, concrete suggestion. I will try it that way.
Thanks,
Richard
On Sat, 2010-12-18 at 05:16 +0900, Kuwahara,T. wrote:
> On 12/17/10, Richard Cochran <[email protected]> wrote:
> > This patch adds a new mode bit into the timex structure. When set, the bit
> > instructs the kernel to add the given time value to the current time.
> >
>
> The proposed new control mode, ADJ_SETOFFSET, is logically the same as
> ADJ_OFFSET with timex.constant == -INFINITY.
I'm not sure if this is correct. Its more like settimeofday, only giving
a relative offset to jump the clock, rather then an absolute time. It
does not slew the clock over time like ADJ_OFFSET does.
> So it is possible to do
> the same thing without risking forward compatibility. (I mean by "risking
> forward compatibility" that the mode bit 0x0040 may be defined differently
> by the upstream maintainer anytime in the future.)
adjtimex is a linux specific interface, which is compatible but not
identical to the ntp specified interfaces. The ntp client code already
has Linux specific modifications, so I don't think we have to worry
about 0x40 specifically being reserved by the NTP client.
thanks
-john
On Tue, Dec 21, 2010 at 4:56 PM, Richard Cochran
<[email protected]> wrote:
> Can you please elaborate?
The timex.constant is defined as equal to the binary logarithm of the
reciprocal of
the natural frequency minus SHIFT_PLL. In other words, the following
equation holds:
log2(natural frequency) + time_constant + SHIFT_PLL = 0,
which means that decreasing time_constant increases natural frequency
exponentially.
And since a larger natural frequency gives a smaller settling time, a
sufficiently
large negative time_constant results in immediate time step, at least in theory.
> I don't see any way to use timex.constant with ADJ_OFFSET in order to
> correct a time offset.
How about this?
if (txc->modes & ADJ_OFFSET) {
if (txc->constant == INT32_MIN) {
/* step time */
} else {
/* slew time */
}
}
> The 'time_constant' in kernel/time/ntp.c is
> restricted to the interval [0..MAXTC], and MAXTC is 10 in timex.h.
Then let's just ignore the restriction. (It's possible by setting the
timex.constant
without setting the ADJ_TIMECONST flag.)
That said, I'm somehow against the idea of using the adjtimex syscall
for that purpose.
On Wed, Dec 22, 2010 at 4:37 AM, john stultz <[email protected]> wrote:
> adjtimex is a linux specific interface, which is compatible but not
> identical to the ntp specified interfaces. The ntp client code already
> has Linux specific modifications, so I don't think we have to worry
> about 0x40 specifically being reserved by the NTP client.
But struct timex is not linux-specific...
On Wed, 2010-12-22 at 06:13 +0900, Kuwahara,T. wrote:
> On Wed, Dec 22, 2010 at 4:37 AM, john stultz <[email protected]> wrote:
> > adjtimex is a linux specific interface, which is compatible but not
> > identical to the ntp specified interfaces. The ntp client code already
> > has Linux specific modifications, so I don't think we have to worry
> > about 0x40 specifically being reserved by the NTP client.
>
> But struct timex is not linux-specific...
It is if you're compiling against linux's timex.h file.
We already have a number of differences compared with BSD's timex mode
definitions:
We have ADJ_TICK: 0x4000, which is MOD_CLKB in FreeBSD.
We also have ADJ_OFFSET_SINGLESHOT and ADJ_OFFSET_SS_READ which allow
adjtimex act like the original ntp_adjtime.
The key bit is that we map the shared MOD_* definitions that the NTP
client uses to the linux specific ADJ_* values in the linux timex.h
However, your concern does bring up a good point: 0x40 is MOD_PPSMAX in
BSD, and we should at-least check to make sure that the PPS code that is
currently floating around on the lists and is in akpm's tree hasn't
already reserved that bit.
Rodolfo, Alexander: Any comments here?
thanks
-john
On Wed, 2010-12-22 at 05:57 +0900, Kuwahara,T. wrote:
> How about this?
>
> if (txc->modes & ADJ_OFFSET) {
> if (txc->constant == INT32_MIN) {
> /* step time */
> } else {
> /* slew time */
> }
> }
This looks like magic behavior. Sort of a "knock twice and then say the
password" interface. I don't see why that would be better then adding a
clear new mode flag?
> That said, I'm somehow against the idea of using the adjtimex syscall
> for that purpose.
Could you expand on why you don't like this?
thanks
-john
On Tue, Dec 21, 2010 at 01:59:44PM -0800, john stultz wrote:
> However, your concern does bring up a good point: 0x40 is MOD_PPSMAX in
> BSD, and we should at-least check to make sure that the PPS code that is
> currently floating around on the lists and is in akpm's tree hasn't
> already reserved that bit.
If the concern is only about that bit, then we can choose another one
from the range 0x0f00 instead.
It seems to me that the ability to jump the clock is a very useful
feature, at least when implementing a clock servo for PTP in user
space, if not in other cases, too. I wonder why that option is missing
from the interface in the first place.
I am not aware of any serious attempt to get any kind of standardized
PTP clock support into any other OS. As far as compatibility with NTP
is concerned, why can't we let Linux set an example and let
NTP/BSD/*nix follow us?
Richard
On Tue, Dec 21, 2010 at 02:25:55PM -0800, john stultz wrote:
> On Wed, 2010-12-22 at 05:57 +0900, Kuwahara,T. wrote:
>
> > How about this?
> >
> > if (txc->modes & ADJ_OFFSET) {
> > if (txc->constant == INT32_MIN) {
> > /* step time */
> > } else {
> > /* slew time */
> > }
> > }
>
> This looks like magic behavior. Sort of a "knock twice and then say the
> password" interface. I don't see why that would be better then adding a
> clear new mode flag?
I have to agree with John on this one. Looks very hacky to me.
Richard
On Fri, Dec 17, 2010 at 11:03:46AM +0100, Thomas Gleixner wrote:
> So you could do the following:
...
> static int get_posix_clock(const clockid_t id, struct posix_clock_descr *cd)
> {
> struct file *fp = fget(CLOCKID_TO_FD(id));
> int ret;
>
> if (!fp || fp->f_op->open != posix_clock_open || !fp->private_data)
> return -ENODEV;
> ret = get_fd_clk(cd, fp);
> if (ret)
> fput(fp);
> return ret;
> }
In order to avoid leaking a fp reference, shouldn't this go like:
static int get_posix_clock(const clockid_t id, struct posix_clock_desc *cd)
{
struct file *fp = fget(CLOCKID_TO_FD(id));
int ret = -EINVAL;
if (!fp)
return ret;
if (fp->f_op->open != posix_clock_open || !fp->private_data)
goto out;
ret = get_fd_clk(cd, fp);
out:
if (ret)
fput(fp);
return ret;
}
Thanks again for your help,
Richard
В Tue, 21 Dec 2010 13:59:44 -0800
john stultz <[email protected]> пишет:
> On Wed, 2010-12-22 at 06:13 +0900, Kuwahara,T. wrote:
> > On Wed, Dec 22, 2010 at 4:37 AM, john stultz <[email protected]> wrote:
> > > adjtimex is a linux specific interface, which is compatible but not
> > > identical to the ntp specified interfaces. The ntp client code already
> > > has Linux specific modifications, so I don't think we have to worry
> > > about 0x40 specifically being reserved by the NTP client.
> >
> > But struct timex is not linux-specific...
>
> It is if you're compiling against linux's timex.h file.
>
> We already have a number of differences compared with BSD's timex mode
> definitions:
> We have ADJ_TICK: 0x4000, which is MOD_CLKB in FreeBSD.
> We also have ADJ_OFFSET_SINGLESHOT and ADJ_OFFSET_SS_READ which allow
> adjtimex act like the original ntp_adjtime.
>
> The key bit is that we map the shared MOD_* definitions that the NTP
> client uses to the linux specific ADJ_* values in the linux timex.h
>
> However, your concern does bring up a good point: 0x40 is MOD_PPSMAX in
> BSD, and we should at-least check to make sure that the PPS code that is
> currently floating around on the lists and is in akpm's tree hasn't
> already reserved that bit.
>
> Rodolfo, Alexander: Any comments here?
No, it is used neither in my patches nor in the original code. The only
change my patches do to timex.h is adding hardpps().
--
Alexander
On Wed, Dec 22, 2010 at 7:25 AM, john stultz <[email protected]> wrote:
> I don't see why that would be better then adding a
> clear new mode flag?
In short, time step is a special case of time slew. Those are the same,
only different in one parameter, as is shown in my previous post.
That's why I said there's no need for adding a new mode.
On Thu, 2010-12-23 at 05:27 +0900, Kuwahara,T. wrote:
> On Wed, Dec 22, 2010 at 7:25 AM, john stultz <[email protected]> wrote:
> > I don't see why that would be better then adding a
> > clear new mode flag?
>
> In short, time step is a special case of time slew. Those are the same,
> only different in one parameter, as is shown in my previous post.
> That's why I said there's no need for adding a new mode.
Sure, I get that a instantaneous offset adjustment could be represented
as a instantaneous slew of the same amount. But what is the benefit of
doing this? The ADJ_OFFSET isn't really a continuous function like you
describe. For instance, you can't slew time backwards. The amount you
can slow or speed up the clock is limited, so time will always increase.
So to have a magic value way outside the bound of normal operation that
does something special seems a bit obfuscated.
ADJ_SETOFFSET isn't slewing the clock. So I think its much clearer to
add a new mode, rather then to try to wedge the functionality into a
corner case of ADJ_OFFSET slewing just because one could mathematically
represent it the same way.
I see the cost of adding it as a special case as adding extra complexity
to an already complex subsystem. Doing things that make the code easier
to understand and read makes it more maintainable. And I don't see the
gain (other then saving a bit in the bit flag) to offset the complexity
of trying to squeeze this functionality into the ADJ_OFFSET mode.
thanks
-john
On Thu, Dec 23, 2010 at 05:27:58AM +0900, Kuwahara,T. wrote:
> On Wed, Dec 22, 2010 at 7:25 AM, john stultz <[email protected]> wrote:
> > I don't see why that would be better then adding a
> > clear new mode flag?
>
> In short, time step is a special case of time slew. Those are the same,
> only different in one parameter, as is shown in my previous post.
> That's why I said there's no need for adding a new mode.
Well, in addition to the objections raised by John, your suggested
implementation is also shortsighted. The field timex.constant is
copied into time_constant in some code paths. Obviously, this would be
a bad thing when timex.constant==-huge.
So, you need to clarify the interaction between ADJ_OFFSET,
ADJ_TIMECONST, ADJ_TAI, timex.constant, time_constant, and MAXTC.
If you would fully implement your idea, I expect it would become
obvious that it a bit of a hack, both in the kernel code and in the
user space interface. But, if you disagree, please just post a patch
with the complete implementation...
Thanks,
Richard
After all, I'd prefer your earlier patchset. Leaving aside the
compatibility issue, there's no particular reason we have to re-use
the struct timex, which requires otherwise unnecessary conditional
branches as well as unit conversions. Don't you agree?
On Sun, Dec 26, 2010 at 05:38:57AM +0900, Kuwahara,T. wrote:
> After all, I'd prefer your earlier patchset. Leaving aside the
> compatibility issue, there's no particular reason we have to re-use
> the struct timex, which requires otherwise unnecessary conditional
> branches as well as unit conversions. Don't you agree?
Well, from my point of view of wanting to allow a user space clock
servo to be able to adjust a hardware clock, it would be sufficient to
offer a way to jump the clock and to adjust the frequency via one or
perhaps two new system calls.
That is indeed what I first suggested, and I still think it would be a
clean and simple interface. The NTP call is quite gross, since it
multiplexes a whole bunch of different functions through the timex
structure.
However, several reviewers on the lkml prefered to keep with the NTP
interface. It offers the needed functionality and is already well
established, despite its ugliness.
To me, the proposed ADJ_SETOFFSET is a simple and logical extenstion
of the NTP interface. Also, the implementation is straightforward. In
contrast, using a special -INF value seems a bit obtuse to me.
Richard