Other values will cause the container to fail because the shell binary
is missing. This avoids non-obvious failures when the host SHELL is
configured differently.
Reported-by: Phillipp von Rotenhan <phillipp.von_rotenhan.ext@siemens.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Allow kas-docker to use proxy configuration of docker client instead of
inheriting proxy configuration from the current environment.
Signed-off-by: Hosgor, Tolga (CT RDA DS EU TR MTS) <tolga.hosgor@siemens.com>
Signed-off-by: Daniel Wagner <daniel.wagner@siemens.com>
Analogously to bitbake's PREMIRRORS, this allows to define alternative
sources for repo URLs specified in kas files.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Daniel Wagner <daniel.wagner@siemens.com>
We need to double-quote the content of the variables in order to
preserve newlines.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Daniel Wagner <daniel.wagner@siemens.com>
Required to control oe-git-proxy exceptions.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Daniel Wagner <daniel.wagner@siemens.com>
Ensure that all file elements of the kas file parameter are properly
canonicalized and mapped onto the container volume. This is important
when kas-docker is called using relative paths while PWD is not the
repository root.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
This is complex, not just because we are passing the arguments to the
trace function, but already because they need to be protected in their
structure when unrolling them at the top level. The trick seems to be
using the argument list via set and "$@" to preserve the individual
words. Due to the trace indirection, we additionally need the single
quotes around the variable values.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
This commit enables passing the Docker container a directory containing
SSH configuration and optionally SSH identities and list of known SSH
hosts. Basically what you might expect to find in the ~/.ssh directory
of the container.
Signed-off-by: Silvano Cirujano Cuesta <silvano.cirujano-cuesta@siemens.com>
Suggested-by: Henning Schild <henning.schild@siemens.com>
I understood the message:
Setting up loop device requires root privileges
as suggestion that I should start this script as root and hit
^C before `sudo` could prompt for my password.
This patch removes that line and adds a sudo prompt to elaborate on why
root permissions are needed and what is done with them. It also adds a
check if this was unsuccessful and suggests to the user to do it
themselves. This will also be shown if sudo is not installed.
Signed-off-by: Claudius Heine <ch@denx.de>
This ensures that both UID and GID of the builder user inside the
container is aligned with the caller of kas-docker - or that of "docker
run" when "-e GROUP_ID=..." is specified.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
The usage of 'echo -e' for printing the usage message is not POSIX
conform and is not being supported in the default shell of Debian
(dash).
As the 'Rationale' of the POSIX 'printf' documentation states, it was
created "due to irreconcilable differences in the various versions of
echo extant" and it should provide better compatibility throughout
different shells.
Signed-off-by: Silvano Cirujano Cuesta <silvano.cirujano-cuesta@siemens.com>
This shall help to standardize the way how to run kas inside a docker
container on a kas file that is locally available. The pattern is
- checkout repo with kas file(s)
- go to directory where the build output should go into
- call "kas-docker build /path/to/kas.yml"
As building Isar images both require a specific docker image (that
could be changed, though) as well as additional privileges (that needs
to be changed in Isar one day), the option "--isar" selects that mode.
And because the output of an Isar build generally contains root-owned
files, the clean command is added which use docker privileges to clean
the build folder, avoiding a "sudo".
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
