partitions: Add variable for device on which MBL lives

[jonathanhaigh-arm]

It will be useful for our rootfs installer to know the name of the storage device on which the MBL partitions live, so add a variables for the devices on which each partition resides.

In particular, this is useful because we will no longer be able to rely on finding rootfs partitions by their labels.

The values of these variables will be written to the "part-info" directory on devices. Currently all the MBL partitions must be on the same device but this will be useful in the future if we start to support using multiple storage devices.

For IOTMBL-2656: Robust update: don't rely on root filesystem labels

Jenkins builds

• http://jenkins.mbed-linux.arm.com/job/jh-test/204/ (cancelled)
• http://jenkins.mbed-linux.arm.com/job/jh-test/205/ (in progress; after variable name change)

Lava results

• http://e120856-lin/results.php?image_version=custom&image_number=205&custom_version=jh-test&lava_query=&lava_owner=&failures=false&submitter=jonhai01&sort=by_device. Failures:
  • https://lava.mbedcloudtesting.com/scheduler/job/85660 (Wi-Fi access; common failure).
  • https://lava.mbedcloudtesting.com/scheduler/job/85663 (Wi-Fi access; common failure).
  • https://lava.mbedcloudtesting.com/scheduler/job/85649 (lxc-create-action timeout; common failure).
  • https://lava.mbedcloudtesting.com/scheduler/job/85647 (lxc-create-action timeout; common failure).
  • https://lava.mbedcloudtesting.com/scheduler/job/85661 (optee_test failure, similar to https://github.com/ARMmbed/meta-mbl/pull/795 / https://jira.arm.com/browse/IOTMBL-2659).
  • https://lava.mbedcloudtesting.com/scheduler/job/85659 (rootfs image update via pelion; SSH failure after update, https://jira.arm.com/browse/IOTMBL-2660).

Manual testing

• Ran each image from artifactory for jh-test job 205 and checked that the _DEVICE_NAME_ files in /config/factory/part-info look correct.

Related PRs

• https://github.com/ARMmbed/mbl-core/pull/252 (automate the manual testing I did here).

[jonathanhaigh-arm]

@rwalton-arm, what do you think of https://lava.mbedcloudtesting.com/scheduler/job/85659? Does the test need to check that networking is up properly before doing the /etc/build mtime test?

[edit]: spoke to Rob IRL. Apparently this is something that just happens occasionally and is an issue with our SSH server. I'll raise a JIRA defect... https://jira.arm.com/browse/IOTMBL-2660.