This also include a ppsi commit. We must have both at the same time,
because each uses a feature of the other, to link sw-uart at build time.

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

Don't print informational message to standard error when an archive is created.

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Alessandro Rubini authored 8 years ago and Adam Wujek committed 8 years ago

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

Alessandro Rubini authored 8 years ago and Adam Wujek committed 8 years ago

... and it saves 60 bytes if CMD_CONFIG is set!

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

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

To avoid triggering .config target at clean.
Put explicit target for $(CURDIR)/.config

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

When any of exported structures is changed please update
WRPC_SHMEM_VERSION or WRS_PPSI_SHMEM_VERSION (for PPSI)

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Also:
--move removing of include/config and include/generated to distclean
--trigger distclean in ppsi
--update ppsi

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

To avoid triggering .config target at clean.
All *_defconfig and *_config from configs directory will fall into this target,
but not .config.

NOTE: target for .config is executed at include of a file (see makefile manual)

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Adam Wujek authored 8 years ago and Grzegorz Daniluk committed 8 years ago

Use make cleanall to remove files from all compilations.

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Adam Wujek authored 8 years ago and Grzegorz Daniluk committed 8 years ago

remove on clean:
--config.o
--revision.o
--$(OUTPUT).vhd
--$(OUTPUT).mif
--$(OUTPUT)_disasm.S
--include/config
--include/generated
--tools/dump-info-host.bin

arch/lm32/ram.ld is not removed

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Adam Wujek authored 8 years ago and Grzegorz Daniluk committed 8 years ago

To avoid triggering .config target at clean
Also in PPSI

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Alessandro Rubini authored 8 years ago and Grzegorz Daniluk committed 8 years ago

Now the default for configurations that do not select peer-delay
is to not even build peer-delay code. This provides for that, and
adds two configurations for peer-delay operation.

If pdelay is selected, e2e is built as well, but selection is currently
only at build time. This will be fixed later.

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

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

Alessandro Rubini authored 8 years ago and Adam Wujek committed 8 years ago

Without this, after the first build of wrpc-sw the ppsi
configuration was never changed.

So if you built e.g. "gsi_defconfig" and then reconfigure
for "gsi_pdelay_defconfig" you'll get no pdelay in there.

For me, and the next maintainers of ppsi, the new configuration option
CONFIG_PPSI_FORCE_CONFIG can be unset in order to test with personale
changes in ppsi configuration.

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

Alessandro Rubini authored 8 years ago and Adam Wujek committed 8 years ago

Also, preserve and restore current configuration, so I don't get mad
when checking with MAKEALL during my tests for new feature.

The rationale for this is that wrpc-sw will soon select different ppsi
choices according to its own configuration, through $(PPSI-CFG-y).
Actually, it already does for the host process build.

MAKEALL must obey this choice, so we trim ppsi configuration
each time we reconfigure wrpc (removing the file would spit an
error instead).

Then, our Makefile forces a ppsi reconfig if ppsi/.config is missing
*or* empty.

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

Use git config --get-all user.name as an build author, if not available use
user@hostname.

Signed-off-by: Adam Wujek <adam.wujek@cern.ch>

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

This commit is not breaking the current status, but adds the
ability to build on the host, to simulate and (mainly) to help
me hack the networking code in a simpler/faster environment.

Missing features:
	- ptp
	- build in i386 (currently x86-64 is hardwired)

Also, this commit requires a change in ppsi, that I'm not going to
commit now: we must get rid of the main function (most of that
stuff is done win wrc_ptp.c, which we must replicate  in a unix-compatible
way).

This is the ppsi change

   --- a/arch-unix/Makefile
   +++ b/arch-unix/Makefile
   @@ -4,7 +4,7 @@ A := arch-$(ARCH)

    CFLAGS += -Itools

   -OBJ-y += $A/unix-startup.o \
   +OBJ-y += \
           $A/main-loop.o \
           $A/unix-io.o \
           $A/unix-conf.o \

Local changes in this commit:
      - adding CONFIG_HOST_PROCESS and the auto-generated opposites:
      CONFIG_LM32 and CONFIG_EMBEDDED_NODE (LM32 && NODE)

      - make some config option depend on !HOST_PROCESS

      - Makefile: some dependencie...

This, again, is preliminary to host builds, and has no effect.

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

These changes have no effect at this point, but are the initial steps
introducing host builds. Some changes are minor cleanups, some depend
on CONFIG_HOST_PROCESS which is not defined at this point.

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

This allows a node built with CONFIG_VLAN to choose at run-time
whether vlans are allowed or not.  We build both pfilter rule-sets,
and according to the active vlan number we write one or the other.

The case "CONFIG_VLAN is not set" is handled by providing default
NULL pointers for beginning and end of the vlan rule-set.

As a side effect, I renamed the files and the functions in
the pfilter-builder.

This is how it works for me, with CONFIG_PFILTER_VERBOSE set,
so you see different rule-set  are used (29 rules vs. 16 rules):

    wrc# vlan set 0
    0 ("0") out of range
    Command "vlan": error -22
    wrc# vlan off
    fixing MAC adress in rule: use 22:33:44:55:66:77
    pfilter rule 00: 4.00000000
    pfilter rule 01: 1.e4466013
    [...]
    pfilter rule 28: 8.00000000
    current vlan: 0 (0x0)
    wrc# ip set 10.0.0.2
    IP-address: 10.0.0.2 (static assignment)

  (and I am now reachable by untagged frames)

    wrc# vlan set 10
    fixing MAC adress in rule: use 22:33:44:55:66:77
    fixing VLAN number in rule: use 10
    pfilter rule 00: 4.00000000
    pfilter rule 01: 1.e4466013
    [...]
    pfilter rule 15: 8.00000000
    current vlan: 10 (0xa)

  (and I am now reachable on vlan 10)

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

Conflicts:
	Makefile

WARNING: wr_switch_defconfig doesn't build, this is fixed in next commit

Unfortunately this commit makes the internal communication between
minic and lan asymmetric: on receive the tag is discarded (and the
longer header falls into the payload) while on transmit it must be
provided by lan.c.

The reason is that on receive we can trim 4 bytes from the payload,
but on send we can't add 4 bytes without a memmove.  The functions
receive pointers to two different header structures, so hopefully
user errors will be signalled by the compiler.

Still, users interact with net.c, which hides vlan completely from
them.

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

This changes the pfilter rule-set, to make it ready for vlan addition.
Moreover, it removes the choice between three rule-sets and uses one
that works for everyone (we'll add the vlan option later on).

Moreover, we prepare for the option of peer-delay PTP and UDP-based PTP.

As a side effect, one Kconfig option is removed.

We know for sure that what goes to the "streamer" protocol
is re-checked, so the streamer can get some "everything else", exactly
like the 7S wr-nic (that sends to the host everything unused).

This is the current rule set:

     - everything tagged is dropped
     - CPU: arp broadcast, not unicast (i.e. requests only).
     - CPU: PTP ethtype. Any mac address.
     - CPU: ICMP unicast, not broadcast.
     - CPU: UDP (uni/multicast), ports  bootpc, ptp-event, ptp-general.
     - Etherbone (class 7): UDP (uni/multicast) and port 0xebd0
     - Other fabric (class 6): anything not Etherbone (inverted class 7 bit)

This counts up to 28 rules. We have 4 more rules available and we may
add options, either at build time or run time to add udp ports or
other ethertypes.  Or not...

2 caveats:

  - frames for the CPU will have class bits 0x41 instead of 0x01, because
  everything not etherbone will receive bit 6. This is not a problem, as
  the CPU is not checking the class bits.

  - the wr-nic gateware, when using this new sw code base, must be
  changed to use classes 6 and 7 like everybody else, not classes 7
  and 5.

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

The makefile included trace.h for everyone, which in turns included
wrc.h.  It's better to include wrc.h straight on (we can't rely on all
files including it, for example pp_printf does not).

Meanwhile, some redundant includes are removed, and some are added
(I prefer wrc.h to be explicitly listed in the files that use it,
even if the command line set in Makefile already prepends it).

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

Otherwise, the first "make" builds tools before spitting the
error about an unconfigured system. I prefer to have the error alone.

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

This quiet=quiet_ has the effect of printing

  HOSTCC  scripts/kconfig/conf.o
  HOSTCC  scripts/kconfig/kxgettext.o
  SHIPPED scripts/kconfig/zconf.tab.c
  SHIPPED scripts/kconfig/lex.zconf.c
  SHIPPED scripts/kconfig/zconf.hash.c
  HOSTCC  scripts/kconfig/zconf.tab.o
  HOSTLD  scripts/kconfig/conf

instead of the full command lines. We could even use quite=silent_ to
have no output at all, but this would introduce a noticeable delay in
the build.

The aim of this change is making more visible the error "your wrpc is
not configured" at the first build. The full compilation of config
stuff makes the error less clear, and the initial error confuses
people (me included).

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

This reverts commit 3a3c3806.

The previous line: every object depending on .config, was better.
We had the problem that a fresh clone was not building because of a
missing autoconf.h.

The depend rule was not working anyways, because it created an empty
.depend the first time over, due to the missing autoconf.h.  I admit I
don't know how to create a depend file after configuration: make
pretends to remake all its inclusions as the first step, so a rule
like ".depend: silentoldconfig" loops forever.

Actually, Etherbone is in gateware, and this option only enables Internet
Protocol in the software.  This fixes a long-standing misunderstanding.

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

Accessing data structures in the host to get target fields is broken:
we may have different data size or alignment, or whatever. The hack
of using "-m32" to be able to access lm32 fields from x86-64 is not
clean (we imply the i386 and target have the same size/alignment).
Even worse, not everybody has an x86-64 cross-compiler for i386.

Thus, we use OFFSET_OF in the target, with the lm32 cross compiler,
to build a binary table that is then accessed by the host.

In order to link the data structure in the host we use the assembler
instead of objcopy. With objcopy we'd need to spell out the bfd name
of the host, and we'd get horrible names like
   struct dump_info _binary_dump_info_host_bin_start[];

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

This reverts commit 03062ea3.

The patch is wrong, because it duplicates

    9491e61b

 ppsi: pick new master and fix building accordingly

which does the "fixing" in a more elegant way. With both in history,
we now have

   ./ppsi/include/generated/autoconf.h:13:0: warning:
           "CONFIG_VLAN_ARRAY_SIZE" redefined
   generated/autoconf.h:14:0: note: this is the previous definition

Besides, the reverted commit was not even using new ppsi master (we
did it before this commit).

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

Use the rules from the array we are passed in the object file.
We'll use the feature when enabling vlans: at that point we'll have
two sets to choose from, according to run-time configuration.

Using the build-time configuration avoids the hairy ifdef and saves
almost 1kB of binary size.

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

Simplify access to $CONFIG_RAMSIZE. Actually, the previous way
was not working for me, as my $SHELL is not bash.

The bug is uncovered by the next commit, because the "-s" argument
to genramvhd has always been wrong for me, but I didn't notice
(so maybe the "-s" argument has no effect? I can't tell at this point)

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

"act accordingly" means pre-including ppsi/...autoconf.h within
CFLAGS, or we miss CONFIG_MAX_VLANS_PER_PORT, used in defining the
ppsi local data structure.

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

The function ptpd_netif_create_socket() still receives two arguments,
because it is being called by ppsi, and I'd better not change the API
in small steps in there.

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

y

This time I added -Wmissing-prototypes and fixed accordingly. Again, this
used to be part of -Wall and I'm disgusted by it not being so any more.

What does "all" mean, then?

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