Fixes ifconfig down issue for LPDC ports.

Having just a small message indicating that a build error has been
detected at the end of the script is not sufficient. The tarball is
created but is invalid and might be used.
Now, when an error is detected the build procedure is immediately
aborted. The tarball is not then created.

This will keep the suid bit for:
--sudo
--screen
--ssh-keysign

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

dropbear was replaced, because it does not support correctly LDAP+kerberos.

It was almost working. With LDAP+Kerberos over PAM dropbear was prompting once
more for a password when the correct was given.

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

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

Fixed by uplift of a kernel from 3.16.37 to 3.16.38.
Even it is a local privilege escalation bug, which does not impact us,
it is worth to have it fixed.

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

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

Synthesized from wr-switch-hdl commit 1d50045

Since the buildroot anyway downloads .xz archive of the kernel for linux-headers
we can avoid downloading the gz archive for our kernel compilation.

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

Federico Vaga authored 8 years ago and Adam Wujek committed 8 years ago

Signed-off-by: Federico Vaga <federico.vaga@cern.ch>

Federico Vaga authored 8 years ago and Adam Wujek committed 8 years ago

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Federico Vaga authored 8 years ago and Adam Wujek committed 8 years ago

The version 3.16.37 is going to be supporter until 2020

Signed-off-by: Federico Vaga <federico.vaga@cern.ch>

Explicitly define which version of buildroot should be used.

It solves problem when there is more than one directory matching build/buildroot-2*

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

Federico Vaga authored 8 years ago and Adam Wujek committed 8 years ago

This way, next time we want to update the kernel we have just to change
the version without touching too much the code

Signed-off-by: Federico Vaga <federico.vaga@cern.ch>

Check md5sums of files in the wrs-firmware.tar with stored in file
checksums.md5. When checksum does not match abort the update.

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

Instead of keeping the list of archives in build/download-info which should be
downloaded for buildroot, we use "make source" to do the same job.
However, the task of building buildroot had to split into 3 tasks:
--init (unpack and configure buildroot)
--download (download the source of used buildroot packages)
--build (compile buildroot packages, etc.)

The reason for this change is to avoid updating the download-info file when
buildroot package is updated. Additionally ltrace is not downloaded as an
archive, but it is cloned git repo. After the clone it is packed into archive.
In the current state our framework cannot do the same.

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

--Remove file THIS_IS_NOT_YOUR_ROOT_FILESYSTEM. We wrap rootfs ourself, we know
  what we're doing.
--remove symlink /etc/dropbear, before creating such directory
  Creating the symlink was added in the buildroot's commit:
  e7d04dd2 package/dropbear: fix generating keys on RO file systems
--overwrite symlink for var/run->../tmp

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

make 3.82 splits the work differently. When using buildroot with
the make > 3.81 buildroot uses BR2_JLEVEL= (preferably 0 for automatic number
of jobs setting).
For make =< 3.81 BR2_JLEVEL= is ignored and build is done using only one job.
Passing -j allows to build with many jobs, but many warning messages are
printed:
make[2]: warning: -jN forced in submake: disabling jobserver mode.

Use BR_JLEVEL=0, this allows buildroot to pick automatically number of jobs.

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

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

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

The upstream address for lsof is not working. While the download script
tries buildroot.net if upstream fails, getting lsof got stuck for
no connection to the server.

With this commit it times out after 15 seconds to then proceed using
the buildroot address. I expect the problem to affect other upstream
files over time, so the commit doesn't fix the link but the tool.

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

This factorized configuration stuff out of wrap_rootfs into a build
target.  It also adds building and installing "conf" so to allow the
following commands in the target (in /wr/etc):

	  make config

	  make oldconfig     (likely after editing dot-config)

	  make defconfig

I'm unable to build menuconfig at this time, because of missing
symbols in the target's libncurses. I hope to fix this when we
bump buildroot up (thanks Adam for that!)

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

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

If you re-build after stuff has been added to Kconfig, the .config you
already have in place is wrong. This means the dot-config at runtime
will have missing items.

We thus force a silent "make oldconfig" at each build step, to ensure
user preverences are saved, but new options get the default value.

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

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

Add watchdog to FPGA and include bugfixes

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

Print "-dirty" for:
--at91boot
--startup guide
--switch's tools

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

Update required because otherwise setting of auxclk hangs the switch.

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

Alessandro Rubini authored 10 years ago and Adam Wujek committed 10 years ago

-DDEBUG in libwr forces LOG_DEBUG level in libwr/wrs-msg.c. This is
the desired behaviour, but we want it off by default now.

All other DEBUG symbols in Makefiles have no effect, so they are
removed to avoid tricking our developers in setting/unsetting them
and see no effect at all.

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

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

Benoit Rat authored 10 years ago and Alessandro Rubini committed 10 years ago

Most of the calls in HAL use TRACE to print to stdout/stderr however
the full TRACE is depending on DEBUG flag at compilation.

This is a quick hack, and the whole TRACE system should be re written
with something like:

TRACE_P_FATAL(...)
TRACE_P_ERROR(...)
TRACE_P_WARN(...)
TRACE_P_INFO(...)
TRACE_P_VERBOSE(...)
#ifdef DEBUG
TRACE_P_DEBUG(...)
#endif

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