This changes the flashing procedure, because the previous one did not
work any more now that the default filesystem size is bigger.

With this commit no jffs2 images is generated any more, and the
filesystem image is not included as a payload in the initramfs booted
using the USB tools.

Rather, the initramfs downloads wrs-image.tar.gz using TFTP, from the
IP address that replied to DHCP, and untars it to the NAND partitions.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

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

This barebox version autodetects memory, but at installation time it
detects 8MB instead of 64MB.  This patch forces 64MB, like we used to
do in v2012.05 when no autodetection was there (actually, there the
pm9g45 harwired 128MB, but this had no side effects for us).
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

When porting forward the patches from v2012.05 to v2014.04, I forgot
this last patch, that allows flash-wrs to set a mac address in the
barebox binary.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

If $WRS_FAKE_FLASHER is not empty, no flashing is done but commands
are reported to the user.

If $WRS_KEEP_TEMPORARIES is not empty, all temporary files are
preserved.

I need both features to make the new flashing procedure for the
upcoming 3.4 release.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

It is a common convention to use $TMPDIR for temporaries, instead
of an hardwired /tmp pathname.  And now I need is because my /tmp is
too small on the system where I flash.

The local variable TMPDIR is thus renamed to WRSTMPDIR.
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>

the new barebox we'll introduce supports much more stuff, and is 0x63000
in size.  Thus, load 0x84000 bytes from dataflash, a multiple of the
pages size (0x420).
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Use true/false for if, not 1/empty, because "0" would return true
and it would be mileading (no bug is there, just potential)
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>

rsyslogd forks and daemonizes by default. So we can't "respawn" from
inittab. The option to avoid forking also prints massive diagnostic
information, so let's fire it once only.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This also adds the default configuration file for ppsi.  It has no
effect, because it's the default that would apply if no configuration
file is there, but it's useful as a reference, and to ease editing.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is just the basic snmpd configuration, with no customization yet.
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 userspace/wrsw_hal/rt_ipc.h pointed to rt/ which was removed (now
RT code is build in wrpc-sw instead).  This makes the file a regular
file, with the right contents
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

We have to try until 10MHz and 1-PPS is there in GrandMaster mode.
Otherwise we won't have timing at all.

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 variable, if not empty, allows to rebuild buildroot
in the previous build place, thus avoiding most build time.
This is useful for me when I just add packages and don't need
to recreate the compiler from scratch.

Well, so it is documented in the commit message.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

During the rebase of branch "http-management" I dealt with some conflicts,
and CONFIG_RSYSLOG was lost in the buildroot config. This adds it back.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Running "ifup eth0" if eth0 is configured for dhcp would make the
client try a little while and then give up. Thus, if the dhcp server
is currently down we end up without IP address. This happens after a
power failure, if the switch reboots faster than the dhcp server.

Also, avoid ifplugd, that won't help in this case (the link is up,
so no "link up" will happen later.

This new configuration works in all cases:
     - eth0 is configured with a static address in /etc/network/interfaces
     - dhcp-driven, and link goes down and then up
     - dhcp-driven, and server not ready at boot time
     - dhcp-driven and all fine since boot time
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

The original ifplugd.action used "exec" improperly. Also,
we should not force "ifconfig eth0 down" when link down is detected,
or it will bever come up again.

Fix both items, even if we are not going to use ifplugd any more in
the next commit.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

it really looks like ifplugd is not reading this file, so overwrite
the default one with an empty file, to avoid our users the effort to
track its use. The file is not empty, actually, but include this
message.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>