tree 60cf27524c9b6325c7dcb308b342ffad1c2ffa1e
parent 77b00c8550ed9a46b425105f637f9d9874761eab
author Milton Miller <mdmii@outlook.com> 1692789273 -0500
committer Milton Miller <mdmii@outlook.com> 1693417708 -0500

initfs: Relax final watchdog timeout after update

Bump the watchdog timeout from 15s to 150s aka 2.5m after update
has completed, relaxing the reaction time before reboot.  The
normal path still calls reboot -f to reboot when the console has
been flushed.

When transfering control to our initrd shutdown, systemd leaves
a watchdog running to make sure the system doesn't hang.  The
update was determined to be too short to complete updating some
spi flashes, leaving them with erased but not written images in
the spi-nor mtd, effectively bricking the machines.  So to keep it
alive we start a watchdogd daemon to ping it every 1s timeout 5s
during flash update.  However after flash update completes we were
setting the timeout to 15s leaving it armed.  While this is enough
time to flush the serial console and call reboot, its pretty tight
for a human to recognise a the shell prompt and type the echo -n V
to the watchdog device to disarm it.

Signed-off-by: Milton Miller <mdmii@outlook.com>