Switcher keys are broken

Description

CRTL + 1, CRTL + 2 does not work at times. It appears that the input_server does not have an updated list of domains, as it calls out domain 3.

Logs

xec-vm showed that the slot 1's domid was 13.

Looking into repro still...
Steps to reproduce:
1. Install guest, reboot.

Environment

None

Validation Steps

Doesn't repro consistently, but these steps work more often.
1. Install stable-9 build 6582
2. Install guest VM (no OS required)
3. Boot guest, reboot guest.
4. Look for (null) domains with xl list and tap errors in logs.

Activity

Show:
Andrew Jones
April 24, 2019, 9:32 PM

I have encountered the same behaviour. So far unable to repo intentionally. In my case, I believe the domid was 31, and switch was attempting to request domid of 35 (possibly vice-versa)

Chris Rogers
June 4, 2019, 9:02 PM

Finally repro'd, it's caused by (null) domains after tapdisk fails to untap the vhds. A small excerpt from /var/log/messages that I uploaded:

Will update the repro steps.

Mahantesh Salimath
June 10, 2019, 9:53 PM

Attaching the logs for error on trying to switch to dead domains because of slot cleanup during domain start-up

Chris Rogers
June 25, 2019, 2:17 PM

Test on build 6627

Nicholas Tsirakis
June 25, 2019, 3:36 PM

Still seeing a (null) domain from time to time, but the switcher keys appear to be working in these cases.

Fixed

Assignee

Unassigned

Reporter

Joel Upham

Labels

None

QA Assignee

Nicholas Tsirakis

QA Image URL

None

Epic Link

Components

Fix versions

Affects versions

Priority

Major
Configure