Age | Commit message (Collapse) | Author |
|
Until now, we only handle isolation for One cpu. But for better test coverage,
we actually need support for more CPUs.
This patch adds support for more than one CPU in isolation script.
Change-Id: I169766adf900289f4eaf955e1d12f4d3394f642c
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I4f17592c20027cd4739e188e1c9e14c99717b027
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
So that we can isolate all CPUs.
Change-Id: I8c1baef9fe23950cc8752e7df3628af8c5d8e3bb
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Currently we were creating 'stress' on control plane CPU and then migrating it
to data plane one. Instead create it on data plane CPU only.
Change-Id: I63a38db978e211fada66eb4a646b51517519699f
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
It was done again earlier because we were hotplugging CPUs, but its not done
anymore and so this isn't required. Remove it.
Change-Id: I5844252f2c95f2c6f1335bd4e54363dc34339a48
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Just for better readability.
Change-Id: I59da04d24519c2fa50ca5dc19636256c7ed2d50d
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: Ic2dab8318e6c120ca4802a182e42399c695f4577
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: Iefcaf437dd59e8b9d314b4a168db589f877fa47e
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Until now we were only concerned about interruption from ticks and so were
checking only interrupts from arch_timer.
But there are few problems here:
- To check which row belongs to tick-timer, we need to grep for a string. It can
be "arch-timer", "twd", "NMI", etc for different platforms and so the script
was never generic.
- Also, interruption is possible from other peripherals, PPI, SPI, IPC, etc
This patch adds in another AWK script to get total number of interrupts for any
CPU (though it supports returning that for all CPUs at once, its not used
currently). We use this instead of looking at tick-timer.
Change-Id: I2aab004314459a928e5ebc505d03a4525957c40e
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Commit: 176ea92 (is-cpu-isolated: Fix name of cpuset files) missed fixing this
entry. Fix it.
Change-Id: I3a724e4fc1bf4e36a84bf7ee52a5ae37b0287cd7
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Switch to virtio block, since sd is emulated interface.
While at it, improve installing and compiling hackbench. Avoid
compiling in steps since we have scripts in place. Make sure
we have gcc and others installed for compiling.
Change-Id: I942e6bbafb09b7739156fb8d15f41c6334f16c8c
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
|
|
On Ubuntu LTP gets installed in the LAVA test execution directory.
For this reason the ltpfixup.sh wasn't working correctly so far.
It had the /opt/ltp path hardcoded. This patch fixes the issue.
Change-Id: I4bd9389d63aea954b01e7162a4a0518aa2223750
Signed-off-by: Milosz Wasilewski <milosz.wasilewski@linaro.org>
|
|
cpu-hotplug was a HACK which was added initially to get everything migrated out.
But kernel now has better support for isolation and so this can be disabled now.
Leave it commented so that it can be used for testing later.
Change-Id: I5dd7b9055f7eac45f1c7447c05be5ade4f8ef490
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
We are creating two cpusets one for data plane, name RT and other for control
plane, named GP currently.
Lets give them proper names to avoid confusion.
Change-Id: I7c4d7a1fb05ec910d7a9ac879571e27786ec2e2b
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Based on configurations we may or may not have cpuset files names prefixed with
"cpuset.". Get provision for this in our script so that it doesn't break for
those platforms which have this prefixed.
Change-Id: I8cf569d5bdd7f96ec3a86cb2aba3a767a3aa1bc1
Reported-by: Santosh Shukla <santosh.shukla@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
When running packages separately in the same LAVA job, results were
concatenated. Now result files are stored in the test execution directory
and are separate for each lava-test-shell. ltpfixup.sh script was moved
to 'common' directory. Definition was unified between Ubuntu and OE.
Change-Id: I43372b0d060d88f871407bc4b1b44c2cf94ef333
Signed-off-by: Milosz Wasilewski <milosz.wasilewski@linaro.org>
|
|
|
|
Change-Id: Ic5e831c1330ef3d9f864ee19d74208b694bc5269
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Establish basic sanity and regression tests, this is a special
case test since it actually generates traffic for other test cases
Change-Id: I711c5a4683b116ad54cc8c69e020dcf041fa7317
Signed-off-by: Weilong Chen <weilong.chen@linaro.org>
|
|
Till now this scripts was only useful for a system with two CPUs as there were
some assumptions. But there were requirements from multiple people about
supporting more platforms which might have more than two cores.
Hence this patch, it enables support for multiple cores and any core can be
isolated (leaving core 0, which can't be isolated in kernel as that's the boot
CPU).
Order of parameters is also changed based on what user may want. And so
nohz-isolation.yaml is updated in the same patch so that things don't break
inbetween.
Change-Id: I0656131a6a6d043e662d16f05dd08f172f2a3d9d
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I939381ec2240a9f0c0ba5075164f0881c4bdf252
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I56a9c51a0c61ec338ece94041b94ab643efb7285
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Just for readability.
Change-Id: Idfbd4acc853038d108244a3ca79c73096350bc8c
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I92ef53de6dbb3712b8463f6d1a7c0322b53b2730
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
- utils dir is having binaries which will be used by other tests of
the suite. No need to consider it as a subtest.
- bug : 1290766
Change-Id: I3f766d51ac423c7d4f4bc29bb892114746548855
|
|
If success of former step is precondition of later step, when the former step
fails, the later step should be skiped, e.g. if the kvm initialization fails,
or the kvm.qcow2.gz isn't downloaded correctly, all the following tests should
be skipped, otherwise more failures are generated meanlessly, thus bug reporter
and analyzer aren't clear easily about which test fails.
Change-Id: Iaaa5c6842f8c6864b309cc6e7716a966986b03ef
Signed-off-by: Hongbo Zhang <hongbo.zhang@linaro.org>
|
|
A new feature is being introduced in kernel to migrate away all timers/hrtimers
away from CPU in question. Lets use it.
Change-Id: I63dfdb69cbb281e90ea1aa5df9f8544120e2dc66
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I6ca20a368483d30488417ecae571133c7b369974
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
list
Change-Id: I3a0011210dbd09b88293a590408e381275a22bc0
|
|
Change-Id: If707fe2b58fc789411e4c2e093473c51c3f87541
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: Ib066d8be065925ac9eff9b6e106d89e0c084a52b
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: I2efc59ca64d8fb69d2ca93e300e8418c100fdd1b
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: Iba06b340a348dfe7257493cc9f1d14653f84ad74
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: I6265c3d240359f36bac16b243289f12f9dbf4f8a
|
|
Change-Id: Ie747b73c34b99b03afbf36d4d76f3382c38918cb
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
rt-tests hackbench was replaced with the version from Ingo Molnar
http://people.redhat.com/mingo/cfs-scheduler/tools/hackbench.c
hackbench is now called from both host and guest the same way:
hackbench 100 process 500
The results are parsed using min_max_avg_parse.py. Result regexp
was updated to match the output format of the script.
Change-Id: I5d437565d42f45c9411dc8b62540be7b53e0c630
Signed-off-by: Milosz Wasilewski <milosz.wasilewski@linaro.org>
|
|
We might not see infinite isolation on the first time and so need to check it
everytime. Lets do it.
Change-Id: Ic00d4e642e2a86ec7bd062af5b0964ad398d9d71
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: Ic2b83f693214cde5129284ef787a52a6ab586cb3
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
- verify using the sanity script return status, whether the subtest
is suppoerted or not
- remove the sched_mc test as its not supported
Change-Id: I1314fdf60a7938124391853472a4bb3093a5630f
Signed-off-by: Sanjay Singh Rawat <sanjay.rawat@linaro.org>
|
|
Change-Id: I57bb7b2f56d684c79e0c24706dca77254f919c35
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
pass option -k to curl to bypass SSL verification
Change-Id: I30d36b62b5d2bdf3cf8a6cfb3ec46f6ef5c89b95
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: I9f9a626ee08045250976ef0e1425d02df4f82328
|
|
With the latest updates we might end up with infinite isolation time and hence
we need to sense it and quit/PASS tests rather than waiting infinitely.
Change-Id: Id97be7ff566d7b31101b6a0ad19b450e218cc0a9
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
If we hot-unplug CPU1 and hotplug it back, then all tasks would be moved to CPU0
and that will let us have long isolation times. Actually this currently leads to
infinite isolation time.
Change-Id: I6ad0720328fa52f81491197d61ddb410ae31ee68
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Change-Id: I26964a9ca9af4e5ba98ea1efdcd3d23440d4a4be
Signed-off-by: Ciprian Barbu <ciprian.barbu@linaro.org>
|
|
Need curl on some OE images due to busybox wget don't
support https links.
Change-Id: Iea0da72b97516e1806c2680866f3ff4479131bb6
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
|
|
Change-Id: Icfeec1eebc34eeafbfa30b9deafcccb44d872afa
|
|
Change-Id: I3708395b8f56524d51e6eaa46eaabc460eb0b934
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
This adds support to print average isolation time in measurement field and
seconds in units. Similar change is done in openembedded/nohz-isolation.yaml.
Change-Id: Ibe2b545bb769cc2b4c3e6ac34fa12b8bff819e52
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Few more optimizations to reduce number of wakeup events for CPU1, which we are
trying to isolate.
Change-Id: I3751d437dad88679a402d448c3c26ce196d7b1bc
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|