FWIW we saw this error in the past:

https://bugs.chromium.org/p/chromium/issues/detail?id=642269

not sure whether this one is related:
graphics_Idle: FAIL: Failed: Did not see the min DVFS clock.

https://uberchromegw.corp.google.com/i/chromeos/builders/elm-paladin

Looking at these, this is not an uncommon issue:

https://stainless.corp.google.com/search?view=matrix&row=board&col=build&first_date=20171125&last_date=20171201&test=%5Egraphics%5C_Idle.*%24&exclude_cts=false&exclude_not_run=false&exclude_non_release=false&exclude_au=false&exclude_acts=false&exclude_retried=false&exclude_non_production=false
https://stainless.corp.google.com/search?view=list&first_date=20171101&last_date=20171201&test=%5Egraphics%5C_Idle.*%24&build=%5ER64.*&status=FAIL&exclude_cts=false&exclude_not_run=false&exclude_non_release=false&exclude_au=false&exclude_acts=false&exclude_retried=false&exclude_non_production=false

I'm not sure if we can tell if the frequency of this problem occurring is higher now...

hoegsberg wasn't sure on this issue, I suspect it was a CL. I'm going to mark it closed for now, if anyone feels it should still be open please re-open.

We are seeing this on the latest R65 stable build on auron_paine and lulu, it is happening repeatedly.

https://stainless.corp.google.com/search?exclude_retried=false&first_date=2018-03-27&master_builder_name=&builder_name_number=&shard=&exclude_acts=true&builder_name=&master_builder_name_number=&owner=chromeos-test&retry=&exclude_cts=false&exclude_non_production=false&hostname=&board=&test=%5Egraphics_Idle%5C.arc%24&exclude_not_run=false&build=%5ER65%5C-10323%5C.67%5C.0%24&status=FAIL&status=ERROR&status=ABORT&reason=&waterfall=chromeos(_release%7C)&suite=%5CQbvt-arc%5CE&last_date=2018-04-10&exclude_non_release=true&exclude_au=false&model=auron_paine&view=list

Strangely it passed on the prior R65 build, and that build only had a Chrome rev in it, with one CL related to mouse input (clicking back on a mouse would go back twice). It is also ok on other Auron derivatives which are very similar (auron_yuna is just a bigger auron_paine).

Is there a chance that there are some DUTs in the lab that are bad and just never pass this test?

On a more immediate concern, can anyone comment how critical this test is, and do we believe this is a real user facing failure?

I don't want to have to block the 65 release on these devices unless we really think they are broken by the build.

The tests intent is detect wasting power due to graphics being active when it should not be. Interestingly enough the test often detects that Chrome is foobar when other tests do no. So I would say worth to install the right image on the right DUT and booting/logging in!

Who's a good person to take a loot at this? Arc and graphics is way out of my knowledge range. Kristian any ideas who could take a look at this?

Also sorry for the slow reply, I must have missed it during 2 week OOO email cleanup.

Bumping this down to a P2. I have no point of reference on this bug, need someone to at point me to the right person to talk to.

I am the wrong person for this bug (I do not work on anything related to graphics). I'm going to also mark it wont-fix for lack of activity, feel free to re-open if it's still an issue.