Linaro big.LITTLE Reference

[Ubuntu] data-corruption-switching test fails

Reported by Paul Larson on 2012-05-03
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
big.LITTLE Reference Switcher
New
Medium
Dave Martin

Bug Description

hwpack: http://snapshots.linaro.org/oneiric/hwpacks/vexpressdt-rtsm/70/

This passed in the non-switching modes, but on the switching one I got:
root@linaro-developer:/home/linaro/tests# data-corruption-switcher.sh -f big -c 0 -c 1 -c 2 -c 3 -s

Switching to big mode if not already in.

Start ..
Number of CPUs successfully brought up during boot = 4

Running /usr/bin/boot-a15.sh -c 0

cpu0 is big

SUCCESS!!
boot-a15 finished successfully

Switching to big mode if not already in.

Start ..
Number of CPUs successfully brought up during boot = 4

Running /usr/bin/boot-a15.sh -c 1

cpu1 is big

SUCCESS!!
boot-a15 finished successfully

Switching to big mode if not already in.

Start ..
Number of CPUs successfully brought up during boot = 4

Running /usr/bin/boot-a15.sh -c 2

cpu2 is big

SUCCESS!!
boot-a15 finished successfully

Switching to big mode if not already in.

Start ..
Number of CPUs successfully brought up during boot = 4

Running /usr/bin/boot-a15.sh -c 3

cpu3 is big

SUCCESS!!
boot-a15 finished successfully
/usr/bin/data-corruption-switcher.sh: line 229: [: : integer expression expected

Starting bigLITTLE switcher in the background

Running data-corruption.sh /data/boxes/boxes.ppm 1

***bl-agitator***
CPU count: 4
CPU0: big freq 1000000 LITTLE freq 100000
CPU1: big freq 1000000 LITTLE freq 100000
CPU2: big freq 1000000 LITTLE freq 100000
CPU3: big freq 1000000 LITTLE freq 100000
Periodic switcher time 50

*** Data Corruption Test ***

All performance values in Mpixels/sec

Bitmap Bitmap JPEG JPEG Image Image Comp Comp Decomp
Format Order Subsamp Qual Width Height Perf Ratio Perf

RGB TD GRAY 75 63 63 2.056 20.25 3.604

RGB TD 4:2:0 75 63 63 1.587 11.82 1.855

RGB TD 4:2:2 75 63 63 1.347 11.17 1.774

RGB TD 4:4:4 75 63 63 1.257 10.53 2.056

Iteration 1/1:

All performance values in Mpixels/sec

Bitmap Bitmap JPEG JPEG Image Image Comp Comp Decomp
Format Order Subsamp Qual Width Height Perf Ratio Perf

RGB TD GRAY 75 63 63 2.055 20.25 3.604

RGB TD 4:2:0 75 63 63 1.587 11.82 1.854

RGB TD 4:2:2 75 63 63 1.346 11.17 1.774

RGB TD 4:4:4 75 63 63 1.257 10.23 2.033

Data corruption test: MD5SUM_GRAY: PASS
Data corruption test: MD5SUM_420: PASS
Data corruption test: MD5SUM_422: PASS
Data corruption test: MD5SUM_444: FAIL
data-corruption.sh failed. Abort!!

Kill bigLITTLE switcher
Time elapsed: 0:01:20.570
Terminated because of SIG 15

SUCCESS!!

...so clearly there's also something wrong with the test as well if it is reporting SUCCESS!! when one of the md5sum tests fails.

Paul Larson (pwlars) on 2012-05-03
Changed in linaro-big-little-reference:
assignee: nobody → Dave Martin (dave-martin-arm)
Avik Sil (aviksil) wrote :

Though I could not reproduce the md5sum check failure, but I'm getting few bl-agitator switch failures like following:

cpu3 scaling_setspeed target 100000 current 1000000... FAIL
error on iteration 26 period 767
[...]
cpu0 scaling_setspeed target 100000 current 1000000... FAIL
error on iteration 45 period 614

FYI, the bl-agitator is running on random switching mode: bl-agitator -r -l 1000 -n

Avik Sil (aviksil) wrote :

Further observation:

Even just running 'bl-agitator -r -l 1000 -n' without running any test script shows the failures:

root@linaro-developer:~/core# bl-agitator -r -l 1000 -n
***bl-agitator***
CPU count: 4
CPU0: big freq 1000000 LITTLE freq 100000
CPU1: big freq 1000000 LITTLE freq 100000
CPU2: big freq 1000000 LITTLE freq 100000
CPU3: big freq 1000000 LITTLE freq 100000
Random switcher seed 0 limit 1000
Random switcher seed 0 limit 1000
Random switcher seed 0 limit 1000
Random switcher seed 0 limit 1000
cpu3 scaling_setspeed target 100000 current 1000000... FAIL
error on iteration 27 period 929
cpu0 scaling_setspeed target 100000 current 1000000... FAIL
error on iteration 75 period 907
^CTime elapsed: 0:03:10.1000
Terminated because of SIG 2
root@linaro-developer:~/core#

Paul Larson (pwlars) wrote :

Avik, that sounds like a different bug to me. Please open a new bug on it.

Changed in linaro-big-little-reference:
importance: Undecided → Medium
Dave Martin (dave-martin-arm) wrote :

(The issue described by https://bugs.launchpad.net/linaro-big-little-reference/+bug/995857 may not be a bug -- see that bug page.)

Dave Martin (dave-martin-arm) wrote :

Paul, can you please provide links to the filesystem and any relevant test files?

Changed in linaro-big-little-reference:
status: New → Incomplete
Paul Larson (pwlars) wrote :

The rootfs used was linaro-o-developer-tar-20120426-0.tar.gz, with the extra bits for tjbench and the test suite from the private git repository installed. It's looking like it might be a bit random or hard to reproduce. I'll try to see if I can find a better way to reproduce with current images.

Changed in linaro-big-little-reference:
status: Incomplete → New
Dave Martin (dave-martin-arm) wrote :

@Paul, Thanks -- I'll wait for the additional details before looking into this in earnest.

Do you know whether similar failures have been observed with the integrated switcher?

Paul Larson (pwlars) wrote :

Not so far

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers