struct DSPort now has a void pointer called ext_dsport that is used by
each extension to store its own local data.  White Rabbit, thus,
is not a special case any more in this respect.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

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

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

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

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

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

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

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

This adds the name of structures for all typedefs in the header.
It has no technical effect, but allows developers (i.e. me) to
use "struct DSPort" instead of "DSPort" in new code.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

While the standard protocol doesn't expect master-only opeation, the
flag is in the data strcture and this change, needed for white-rabbit,
has no effect here.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

Since proto-standard is used to build a library that can provide
defaults for the extension, include its Makefile regardless.  This has
no technical effect at all until we remove duplicated code (next
commit).

This commit also makes some trivial and small changes in variable
assignements (the ones the user may want to change are now at the top
of the file).
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

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

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

You can pass architectures (for example "./MAKEALL spec"),
specific settings (for example "./MAKEALL CONFIG_NO_PRINTF=y")
or "-u" to ask the printout of undefined symbols in each ppsi.o
being built.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

If you "export CONFIG_NO_PRINTF=y", no printf will be built in the
binary. This is needed when ppsi is built within wrpc-sw (which has
its own pp_printf implementation).

This will become a Kconfig item, when Kconfig is added.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

Now that we have vprintf, we can save a buffer in the stack and
simplify pp_diag_printf and pp_timed_printf.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

This patch-set removes the old diag/printf.c and related files, to use
the "official" pp_printf, which is now published separately and
included here in its entirety with not changes from its upstream repo.

Please note that HAS_DIAG and HAS_FULL_DIAG is not supported any more.

While we have no Kconfig support (coming soon), please see ./MAKEALL
to find how to pass verbosity (and thus size) options to pp_printf.

All commits build for all architectures.

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

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

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

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

This is a copy of pp_printf, that originated from ppsi's diagnostic
code and is now a separate package, with a few more features.
The upstream repository is git://gitorious.org/rubi/pp-printf.git .
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This patch-set removes most of the awful pp_ libc replacements (like
pp_memcpy) that I used in the initial ptp-proposal project.  I now
think using standard names is better and easier. The build for all
architectures has no conflicts with this approah (all commits build).

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

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

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

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

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

Even the freestanding lm32 compiler has <string.h> (because gcc builds
newlib by default as a minimal fallback).  Since I want to get rid of
pp_strlen and such stuff, we'd better include <string.h>, win size_t
and use the standard definitions for standard functions.

When we port to systems with no <string.h> we can provide it inside
its own arch- directory. So no harm is done and the code is simplified.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

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

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

This file is meant to share libc functions among non-libc-using
architectures.  In this commit it still exports the pp_* version,
which are removed later in this patch-set.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

If the user has WRPCSW_ROOT in the environment, build for arch-spec
as well, because we can refer to wrpc-sw for drivers.  When building
for spec, build the whiterabbit extension as well.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>