linux-test-project / ltp

Linux Test Project (mailing list: https://lists.linux.it/listinfo/ltp)
https://linux-test-project.readthedocs.io/
GNU General Public License v2.0
2.28k stars 999 forks source link

run "./runltp -f power_management_tests_exclusive" faild #1015

Open yup-21 opened 1 year ago

yup-21 commented 1 year ago

I executed the testcase in the Linux-4.19.90 x86_64: ./runltp -f power_management_tests_exclusive Output result: <<>> tag=runpwtests_exclusive01 stime=1677136716 cmdline="runpwtests_exclusive01.sh" contacts="" analysis=exit <<>> <<>> sh: /sys/devices/system/cpu/sched_mc_power_savings: Permission denied sh: /sys/devices/system/cpu/sched_mc_power_savings: Permission denied sh: /sys/devices/system/cpu/sched_mc_power_savings: Permission denied sh: /sys/devices/system/cpu/sched_mc_power_savings: Permission denied sh: /sys/devices/system/cpu/sched_mc_power_savings: Permission denied

......

I looked up the "sched_mc_power_savings" interface in the kernel and found that this interface has been removed since kernel version 3.5,fix patch 8e7fbcb("sched: Remove stale power aware scheduling remnants and dysfunctional knobs"), but the ltp testcase has not been updated

Do you have plans to update this testcase, or do not execute this testcase on kernel versions 3.5 and above

metan-ucw commented 1 year ago

There is still a lot of old and broken tests in LTP we are sorting them and either fixing what is worth to be fixed or deleting tests that are broken beyond repair.

I had a brief look at the power_management and there are at least some pieces that looks reasonable so it should be cleaned up eventually. Not sure when we will manage to do so since we have a still a lot of tests like this in the queue. And as usual the fastest way to get something fixed in opensource is to send a patch as patches are always welcomed.

yup-21 commented 1 year ago

I added a kernel version detection step before executing this case, like this, if tst_kvcmp -ge "3.5"; then tst_brkm TCONF "Kernel version not supported; not " \ "running testcases" fi to make the test case passable, but obviously this did not solve the problem at all. however I am not familiar with power management testing, I look forward to your improvements, thanks.