This commit separates to its own directory the timing code specific
to bere linux implementations.  If you build with ARCH=bare-i386
or ARCH=bare-i386-64, TIME=bare is automatically selected.

Please note how bare builds don't allow externally-set TIME=, as the
architecture Makefile forces "TIME := bare" (not "TIME ?= bare").

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

This is the first implementation of the concept ot "TIME=" on the
configuration of ppsi.  Time-related code that used to live in arch-unix
is now split into its own directory. If you have special timing hardware,
you can write your own pp_network_operations and pp_time_operations
in your own directory and select them using TIME= .

We'll do Linux-specific hardware timestamping soon, and WhiteRabbit as
well.

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

The unix time and network operations are going to be used as a fall-back
solution by other timing engines.

For example, hardware-specific timestamping capabilities can be
selected (such as the Linux ioctl API) but if the Ethernet interface
(and thus the pp_instance) doesn't support the feature, related code
will be able to set back the unix operations in place.

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

The contents of this directory are pretty portable anyways, and
the internal files were actually called "posix-io.c" and so on.
So let's get rid of the too-politica "gnu-linux" naming and of
the too-little-political "posix" at the same time.  We build
on Unix roots, so let's call our architecture like that.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Aurelio Colosimo <aurelio@aureliocolosimo.it>

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

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

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

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

Since we have a call to bmc inside this function, we do not need to
unpack Announce message locally.

This patch fixes a bug with white rabbit extension, since it did not
update the foreign master record with most recent wrFlags. As a result,
at the end of wr handshaking, the slave (very often) did not handle
correctly the parentWrModeOn flag.

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

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

This partially reverts commit 9bd032a6, see wrspec.v2.06-07-2011, page 28

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

This fixes a bug in proto-ext-whiterabbit, since the White Rabbit next_state
was overwritten by bmc.

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>

This fixes the correct behaviour of ppsi in arch-wrpc, where adjust_freq
is not available.
Still some more fixes may be needed in servo.c. I'm not sure about the
parameter -adj to be passed both in adjust_freq and adjust_offset case.
Anyway, this has been tested on two SPECs and reaches synchronization.

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

if the user passes "-DDIAG_PUTS=sth~ in USER_CFLAGS at build time, al
diagnostic messages are directed to this special puts (through sprintf
to a local buffer) instead of going to the normal printf (and thus the
default puts).

This is going to be used by wrpc-sw, to leave the default output channel
clean for the user shell.

Unfortunately, frame dumps cannot go to this special puts, as
lib/dump-funcs.c just call printf and I don't know how to fix that
without a massive change in the code.

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

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

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

Commit 7373cc02 introduced a workaround for a suboptimal choice in
wrpc-sw about global variables (that were "common").  Commit 9ff25f86
of wrpc-sw fixed it, so this removes the workaround (which, btw,
prevents compilation of ppsi as a submodule of current wrpc-sw).

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

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

call t24p calibration every time PTP is set to Slave mode. This
is a port of commit 279f2c6b of wrc-sw, by Grzegorz Daniluk, since
wrc_ptp_ppsi.c was still part of the supermodule.

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

This bugfix is child of the fixed conf startup procedure: read conf, parse
cmd line, and then open pp_globals by using its current rt_opts.

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

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

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

pp_open_globals must be called after pp_parse_cmdline; ppg->rt_opts is thus
set explicitly before pp_parse_cmdline, since this parsing affect rt_opts.

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

rt_opts parameter was always passed as NULL. With this patch, users who do not
want default rt_opts provided by proto-standard need to set their rt_opts
explicitly.

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

Default Dataset slaveOnly variable is global and cannot be used in multilink
context

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

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

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

This major patch implements the core of Best Master Clock Algorithm for
multilink context (e.g. Boundary Clock).

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

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

Will be useful for global bmc_state_decision computation

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

This partially reverts commit 16707d53, since the IEEE spec states that foreign
masters list must be kept updated for each link, then Ebest search among all
links must be performed.

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

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

Two main problems where found when compiling into wrpc-sw, commit
e00d99f82:

- wrpc_sw/include/hw/memlayout.h defines some symbols into a header file
(I think they should be defined there as "extern" and actually declared
somewhere else);
- there was an include to be removed

The simple solution for the first issue is to define __REGS_H where
the compiler complains; the better solution, to me, is to follow include
chain to discover who is including memlayout.h, or define those symbols as
extern.

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>

posix_main_loop now gets a struct pp_globals ptr and simultaneously
handles all configured links. posix_net_check_pkt has been changed too,
in order to perform a global select on all fd's.

This patch is temporary since it may need debug & test.

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

POSIX_ARCH was defined buy never used, so redefine it for pp_globals, since
it will be more useful.

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

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

Command line now acts globally. If multilink context (ppsi.conf defined and
cointaining more than one link), the behaviour of -g and -e flags is common
to each link. On the other way, -b flag can not be used in this case.

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