12 november 2018 15:40
> Yes, you to restart in order to pick up the change in configuration.
> But your new option is no different in this regard.
Yes, thats true. Forgot that I have an own TLV on the side, making it possible
to change
settings by reading the configuration file again and
On Mon, Nov 12, 2018 at 07:46:11AM +, Anders Selhammer wrote:
> Yes but no. What if you change this while stack is running and ram is
> configured, then you just can set ptp_dst_mac? You need to do some
> reconfiguration, right?
Yes, you to restart in order to pick up the change in
On Mon, Nov 12, 2018 at 07:14:31AM +, Anders Selhammer wrote:
> I use more than one network namespace for the stack yes. Ports in different
> namespace.
This series really needs a cover letter that describes the use case(s)
in some detail. As it stands, we are all wondering, "why would
On Mon, Nov 12, 2018 at 06:42:37AM -0800, Richard Cochran wrote:
> On Mon, Nov 12, 2018 at 07:14:31AM +, Anders Selhammer wrote:
> > I use more than one network namespace for the stack yes. Ports in different
> > namespace.
>
> This series really needs a cover letter that describes the use
If synchronizing a PHC to the system clock, use one of the
PTP_SYS_OFFSET ioctls (if supported) to measure the offset between the
two clocks. Negate the offset and switch the timestamp before passing
them to the servo.
This makes the synchronization between PHC and system clock symmetric.
> -Original Message-
> From: Miroslav Lichvar [mailto:mlich...@redhat.com]
> Sent: Monday, November 12, 2018 8:28 AM
> To: linuxptp-devel@lists.sourceforge.net
> Subject: [Linuxptp-devel] [PATCH 2/6] sysoff: Extend API for different sysoff
> methods.
>
> The kernel supports different
This series adds support for more accurate measurement of the offset
between a PHC and the system clock.
The first patch fixes a valgrind warning.
The second, third and fourth patches add support for new ioctls.
The fifth patch improves accuracy of PHC synchronization.
The sixth patch fixes a
A clock that cannot be adjusted has no servo. Check if a servo is
present before updating the clock.
Signed-off-by: Miroslav Lichvar
---
phc2sys.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/phc2sys.c b/phc2sys.c
index b8f1ea0..a476e63 100644
--- a/phc2sys.c
+++ b/phc2sys.c
@@
> -Original Message-
> From: Miroslav Lichvar [mailto:mlich...@redhat.com]
> Sent: Monday, November 12, 2018 8:28 AM
> To: linuxptp-devel@lists.sourceforge.net
> Subject: [Linuxptp-devel] [PATCH 0/6] More accurate PHC<->system clock
> synchronization
>
> This series adds support for more
This ioctl uses cross timestamping for a more accurate measurement of
the offset. It is supported on some onboard Intel NICs using the e1000e
driver and a virtual PHC with the ptp_kvm driver.
Signed-off-by: Miroslav Lichvar
---
sysoff.c | 21 +
sysoff.h | 1 +
2 files
This fixes valgrind errors.
Signed-off-by: Miroslav Lichvar
---
sysoff.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/sysoff.c b/sysoff.c
index f7b6240..407a01c 100644
--- a/sysoff.c
+++ b/sysoff.c
@@ -18,6 +18,7 @@
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
This is a more accurate variant of the the PTP_SYS_OFFSET ioctl, which
will probably be supported in future kernel versions.
Signed-off-by: Miroslav Lichvar
---
sysoff.c | 38 --
sysoff.h | 1 +
2 files changed, 33 insertions(+), 6 deletions(-)
diff --git
The kernel supports different PTP_SYS_OFFSET* ioctls. Use the sysoff
enum to allow selecting between them in sysoff_measure().
Signed-off-by: Miroslav Lichvar
---
phc2sys.c | 12 ++--
phc_ctl.c | 8 +---
sysoff.c | 25 +
sysoff.h | 12 +++-
4 files
13 matches
Mail list logo