Can you do Ctrl-Alt-F2, F3 to see the console? I think there should be debug logs there, if you can't find it on USB stick.

BTW if you can't find failure logs, usually it may imply the USB stick is not visible after some firmware update, especially EC or PD - which we've seen on Glados family for PD updates (although Eve should not have explicit PD update).

thanks, i kept hitting "tab" instead of switching VTs

VT3 shows (typing by hand):
...flashrom stuff...
Firmware update completed.
Starting cr50 updater (//usr/share/cros/cr50-update.sh /)
Command: //usr/share/cros/cr50-update.sh /
[ERROR:buss.cc(427)] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[ERROR:trunks_send.cc(868)] Failed to initialize dbus proxy.
Cr50 update attempt failed (1).
Failed to update cr50 firmware.
WARNING!!! Installation of software failed. Displaying hw diagnostics
Missing hardware diagnostics.
Can't store logs on passed device '/dev/sda1': not a block device.
Finished after 0 seconds.
Failed Command: //usr/share/cros/cr50-update.sh / - Exit Code 1
ChromeosChrootPostinst complete
Syncing filesystem at end of postinst...
/usr/sbin/chromeos-install: 1: rmdir: Input/output error
Running a hw diagnostics test -- this might take a couple of minutes.
/usr/sbin/chromeos-install: 1: tee: Input/output error
/usr/sbin/chromeos-install: 1: rmdir: Input/output error
...more errors...
Can't store logs on passed device '/dev/sda1': not a block device.

i've split the cr50 failure out into  issue 782903 , but i don't think that's the underlying issue as the postinst code only warns when it fails, it doesn't actually abort.

As you can see, /dev/sda1 is lost (not a block device).

We probably need to check what happened in flashrom part, maybe it tried to update EC and caused the issue.

+Duncan - will updating Eve EC or something cause USB/PD reset so /dev/sda1 is lost?

Yes I think this is due to the EC RO update which triggers a TCPC reset to the Analogix chip when it does the sysjump and you lose USB connections.  This affects all devices with PD in some way currently, but the lack of an SD card means there is no good alternative on Eve.

In theory this should not be a hard failure in the recovery image since the recovery itself succeeds and it only affects pre-production devices where we do EC updates before WP is enabled.

A better long-term fix would be to enable EC RO software sync so the BIOS does the EC RO update instead of doing it at runtime in the recovery image.  This was prototyped but not put into a product yet. (may be too late for Eve)

Another possibility for Eve would be to use a custom override to skip the EC update in recovery mode if the version matches, which would help with the common case where the EC RO does not change.

part of the trouble is that the flash update is in the middle of the postinstall step, so losing USB access in the middle causes cascading failures

if we can add a step to read out the current EC and not update it if it matches, that sounds like a decent workaround ... the first recovery will "fail", but when people try it again, it'll succeed.

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot