tree a59baf3dd232613d1fb79692da844517c4ddf8be
parent 9eac9c203299c3ff6a516bb579c052301235d0a8
author Ruben Rodriguez Buchillon <coconutruben@chromium.org> 1545708691 +0800
committer chrome-bot <chrome-bot@chromium.org> 1545973276 -0800

power_status: fix cpu sibling, and CPUFreq/CPUIdle clustering logic

This change addresses an issue where the clustering logic for cpu
sibling groups was too strict, and relied on some assumptions that don't
always hold. More specifically
- core_siblings_list does not have to be 'd-d' but rather can be a list
of 'x,x...' where x can be 'd-d' or just d
- glob.glob does not necessarily sort the output
- even on a sorted output, on a system where some cores are turned off,
the 2nd file does not have to correspond to cpu1 but could be cpu4 (e.g.
current kukui)
- cpuidle is not guaranteed to exist just because the cpu exists.

So this change should make PLT/power tests a little more robust
especially for early bring up.

BUG=b:121465064
TEST=manual testing
test_that --autoest_dir=$LOCAL_AT $IP power_LoadTest.fast

Change-Id: Ief757c5815e168ed85a0f326675d7242212092e0
Signed-off-by: Ruben Rodriguez Buchillon <coconutruben@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/1390734
Reviewed-by: Nicolas Boichat <drinkcat@chromium.org>