New file test_sw_to_spec.sh is committed into new dir tools/test.
It is a script used to test the link between a switch master and a SPEC
slave.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Alessandro Rubini authored 11 years ago and Aurelio Colosimo committed 11 years ago

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

In case a port is inactive, its fd is not valid, so do not handle it
in select.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

frgn_rec_best field of pp_instance structure was not correctly initialized.
This patch also rewrites a the ports 'for loop' so that it is more readable.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

Check whether pp_instance port link is up or down. Force a new
initialization in case link brings up. Do not call state machine for
pp_instances whose link is down.

This patch is inspired to how ptp-noposix (commit 2e46712e59) handles link
status.

Still missing (and thus declared with a FIXME) the reset of slave structures.
On the single link, this lack does not cause issues. Anyway, must be fixed

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

invalidate fd value for ethernet_mode and check it again in case
unix_net_exit is called twice (though should not happen).

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Aurelio Colosimo authored 11 years ago and Alessandro Rubini committed 11 years ago

unix_net_init must be able to be called more than once,
thus calls unix_net_exit internally. Since unix_net_exit sets
NP(ppi)->ch[0].fd to -1, it must not be called once again when
its value is negative.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

This changes all "Name Surname for CERN" to "Copyright CERN, author
Name Surname".  This is trivial and makes no change, with a few
expections:

- all trivial Makefiles that had "for CERN" lines have now no
attributions, as they are trivial anyways.

- a few unrepresented files have not the copyright notice (especially
ptpdump, that I wrote in 2012).

- a few "for CERN" remain in subprojects, where I didn't touch anyrhing

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

correction field is not printed as hex value, so do not prepend
it with 0x.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

The previous implementation of wrs_time_ops was a bare call
of unix_time_ops, but this was not correct since wrs must handle
time on fpga, not the system time. wrs_calc_timeout is still using
unix_time_ops, since it just needs a monotonic timestamping, without
requiring the best precision.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

The select must be always called with the least required timeout.
This patch fixes a bug which caused an undesired delay when a state
machine, after receiving a packet, needed to be called soon (e.g.
when it has just changed its status: its desired delay is 0).

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

The coding for wr timestamping functions has been mostly copied from
ptp-noposix project: file libptpnetif/ptpd_netif.c, commit 2e46712e.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

According to previous patch, since wrc is a single interface architecture,
set nlinks=0 so that pp_links configuration will not be searched for in
ptp engine.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

The pp_links mechanism, currently based on /etc/ppsi.conf file, shall not
be mandatory, since not all archs will use it. Thus, this patch reduces
at its minimum the dependency on ppi->nlinks. If an arch is not using
the pp_links, it just has to define nlinks=0 in pp_globals struct, and
the engine will work as well, assuming one pp_instance has been
configured in the main.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

No functional modification at all.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

The same values used in ptp-noposix are copied here.
Reference file in ptp-noposix is PTPWRd/constants.h, last changed by
commit f57bc6cd.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

A shorter interval between delay req messages leads to a faster
synchronization. This parameter was increased in the past because of some
stability issues, but at present it can be changed back to the
same value as ptp-noposix.

Signed-off-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

Sometimes the macros may already be defined (e.g.: arch-bathos).

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This allows the arch-specific Makefile to set CONFIG_NO_PRINTF, it it
knows no printf is needed from ppsi (applies to wrpc and bathos).

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Some compilers complain about using "%i" with Integer32, so cast it
away. Which is bad in itself, but I see no other clean option;
fortunately, it's only diagnostic messages.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Poul-Henning Kamp authored 11 years ago and Alessandro Rubini committed 11 years ago

stay in sync.

We need to make WR-specific things while managing the time
and network operations (even if proto-ext-whiterabbit does most
of the stuff without calling the time operations).

Thus, open-code the wrs operation, even if temporary they
just call the unix ones (which, however, are static to their
own files, and only the overall data structure is exported).

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Otherwise, a non-superuser Unix run takes 100% cpu time sending
to stderr the same error forever.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

The addition of a trailing \n (suggested by me) cannot be made
if in /etc/ppsi.conf exists: configuration is a constant string

This also simplifies the code and makes better error management: we
can use fprintf here, and fopen/fread as well.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>