0d7897408c
The lcm_data cleanup erases the first 0x400 blocks (the ext4 filesystem
metadata) of the lcm_data partition in U-Boot, so on the next boot the
block-mount layer finds a bad superblock on /lcm (mmcblk0p16, a ~3 GiB
partition on OSPv2) and runs a full mke2fs before it can be mounted:
e2fsck: Bad magic number in super-block while trying to open /dev/mmcblk0p16
mount: mounting /dev/mmcblk0p16 on /lcm failed: Invalid argument
mke2fs 1.47.0 (5-Feb-2023)
Creating filesystem with 786432 4k blocks and 196608 inodes
That reformat is the whole point (it is what keeps /lcm clean between CI
runs), but it adds enough boot time to push the login prompt just past
the ShellDriver default login_timeout of 60s on the OSPv2 (mxl25641-hdk-6)
eMMC target, which has no override, so board bringup fails with:
pexpect.exceptions.TIMEOUT: Timeout of 60 seconds exceeded during waiting for login
Bump the OSPv2 login_timeout to 80s, which is enough headroom to absorb
the reformat on top of its usual boot while still failing bringup fast on
a genuinely dead board. The Freedom (wnc-freedom) eMMC target keeps its
own 120s, which reflects its longer/different boot time expectations
rather than the lcm_data reformat.
Fixes:
|
||
|---|---|---|
| .. | ||
| cdrouter | ||
| labgrid | ||
| README.md | ||