Two observations.

1. From autoserv.DEBUG.txt, there is a cgpt pooling and expect the result to be 0 (see verify_boot_expectations in autoupdater.py).  But it never reached 0 and thus timed out.

08/09 22:50:58.243 DEBUG|          ssh_host:0180| Running (ssh) 'FILE="/mnt/stateful_partition/.labmachine" ; [ -f "$FILE" ] || ( touch "$FILE" ; start autoreboot )'
08/09 22:50:58.432 DEBUG|          ssh_host:0180| Running (ssh) 'rootdev -s'
08/09 22:50:58.651 DEBUG|        base_utils:0278| [stdout] /dev/mmcblk0p5
08/09 22:50:58.653 DEBUG|          ssh_host:0180| Running (ssh) 'cgpt show -n -i 4 -T $(rootdev -s -d)'
08/09 22:50:59.118 DEBUG|        base_utils:0278| [stdout] 4
08/09 22:51:04.147 DEBUG|          ssh_host:0180| Running (ssh) 'cgpt show -n -i 4 -T $(rootdev -s -d)'
08/09 22:51:04.361 DEBUG|        base_utils:0278| [stdout] 4
08/09 22:51:09.382 DEBUG|          ssh_host:0180| Running (ssh) 'cgpt show -n -i 4 -T $(rootdev -s -d)'
08/09 22:51:09.560 DEBUG|        base_utils:0278| [stdout] 4
[... timeout later]

On timeout, the script checks if system-services is running.  But in this case, it's not.  There is also a renderer crash (see below), but I don't know if it's related to cgpt above.

2. In crashinfo.tgz, you can see logs like...

[1974:1974:0809/224949:FATAL:login_display_host_impl.cc(903)] Renderer crash on login window

In var/log/messages, there are also activity of ui/chrome respawn.

There is no Chrome and Android change comparing to the previous successful run.  The CrOS diff is:
https://crosland.corp.google.com/log/8690.0.0..8691.0.0

The CQ looks green so this appears to be Chrome instead of Chrome OS change which introduced this repeated renderer crash which I'm assuming is why you can't get to the login screen.

Right.  I completely forgot I was chasing a Chrome PFQ issue...

Checked the diff, it contains (only) a DEPS roll.
https://chromium.googlesource.com/chromium/src/+log/54.0.2825.0..54.0.2824.0

But since it's all green now, I'm gonna assume the problem fixed.