MP Scheduler test suite: core_test_scn03.2a test case failed

[Naresh Kamboju]

Scheduler test suite core_test_scn03.2a test case failed.

LSK Android build:
https://android-build.linaro.org/builds/~linaro-android-restricted/vexpress-linaro-lsk/#build=129

Linux version:
Linux version 3.10.16-00350-gf85b6e7 (jenkins-build@ip-10-62-115-104) (gcc version 4.7.4 20130626 (prerelease) (Linaro GCC 4.7-2013.07) ) #1 SMP Wed Oct 16 01:36:30 UTC 2013

Scheduler test suite has latest patches dated Mon, 14 Oct 2013.
tasklibrary: Fix parsing priority token in case of interpolated load

root@vexpress:/data/local/fts/mpcoresuite/testcases/core_test_scn03.2a# ./run.sh
Fast CPU 3,4 Slow CPU 0,1,2
Setting NODOWN_THRESHOLD to 65
Setting NOUP_THRESHOLD to 53
This starts with a heavy task however the task priority
is modified mid way through execution below cutoff. Since the
big domain is oversubscribed, the task stays on little domain
../../common/affinity_tools Created by: Olivier Cozette, November 2012
Copyright: (C) 2012 ARM Limited
Generated load contains 4 time points.
0 : 132818 -5
400 : 132818 2
600 : -1 2
2147483647 : 0 0
#HOG on cpu 4
#HOG on cpu 3
#HOG on cpu 4
#HOG on cpu 4
../../common/affinity_tools Created by: Olivier Cozette, November 2012
Copyright: (C) 2012 ARM Limited
# Set affinity of pid 3217 to cpu 2 1 0
#HOG on cpu 3
#load_generator PID=3217 COMMAND=0,B.88.-5:2000,B.88.2:3000,end
../../common/affinity_tools Created by: Olivier Cozette, November 2012
Copyright: (C) 2012 ARM Limited
# Set affinity of pid 3217 to cpu 2 1 0 4 3
Timer expired 601 times
#Stop ftrace and store data
#Ran from 542.79 to 544.60
Trace events alr[ 212.848260] init: untracked pid 3219 exited
eady copied
[ 212.865597] init: untracked pid 3220 exited
# TRACE_EVENTS_P[ 212.879824] init: untracked pid 3221 exited
ATH=/data/local/[ 212.896225] init: untracked pid 3222 exited
fts/mpcoresuite/[ 212.912641] init: untracked pid 3223 exited
testcases/core_test_scn03.2a/../../../mpcoresuite/common/events/ EXPECTE
D_TIME_IN_END_STATE_MS=600 EXPECTED_CHANGE_TIME_MS_MIN=1800 EXPECTED_CHA
NGE_TIME_MS_MAX=2200 START_LITTLE=3217 START_LITTLE_PRIORITY=-5 START_BI
G= START_BIG_PRIORITY= END_LITTLE=3217 END_LITTLE_PRIORITY=2 END_BIG= EN
D_BIG_PRIORITY= FTRACE_START_MARKER=531.56 DISCARD_TIME_MS= CPU_FAST=3,4
 CPU_SLOW=0,1,2 ../../common/ftrace -l ../../common/libbiglittleswitch.s
o.1.0.0 -t ./trace.ftrace
#../../common/ftrace from Olivier Cozette <email address hidden>
# START STATE discard interval [206.370384, 209.723526] size 3.353142 >
 2.200000
# START STATE discard interval [210.707427, 211.852155] size 1.144728 <
1.800000
# Start state reach at inf
# from start to end -inf must be between 1.800000 and 2.200000
# End state reach at 206.370384
# End state lasted 3.353142 must be greater than 0.600000
# FAILED start state not found
# Start state big enought (> 1.800000s) found [inf,-inf] = -inf s
# Stop state big enought (> 0.600000s] found [206.370384,209.723526] = 3
.353142 s
# ../../common/libbiglittleswitch.so.1.0.0: FAILED
# Result: FAILED
#Ran from 545.10 to 545.94
FAILED

[Alex Shi (alex-shi)]

Naresh, what's the correct result of this bug? Could like commit a successful test results.

[Alex Shi (alex-shi)]

I can not access the benchmark since no permission. :(
And the bug is already known/tracked by ARM.

Tixy, I am wondering if you like to take a look on this.

[Tixy (Jon Medhurst) (tixy)]

These are ARM's tests for their big.LITTLE MP feature and they seem OK that it has always failed, so I don't think we should bother either. If you disagree, please raise a defect in the ARM Landing Team project in cards.linaro.org, that is where Landing Teams don't use Launchpad for managing bugs any more. Thanks.

[Fathi Boudra (fboudra)]

wontfix per Tixy's comment #3