scheduler test: extd_test_scn01.3a, extd_test_scn03.2a & extd_test_scn03.3a failed on linux-linaro

Bug #1183769 reported by Naresh Kamboju
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Linaro Stable Kernel (LSK)
Fix Released
Undecided
Unassigned
Linaro big.LITTLE
New
Medium
Robin Randhawa

Bug Description

- A task named tasklibrary start on A7. After some time, 3 tasks per cpu are pinned on A7 (little). so global load balancer should move tasklibrary to A15, it doesn't.

- A task start tasklibrary start on A15. After some time, 3 task per cpu are pinned on A15 (BIG) so global load balancer should move tasklibrary to A7, it doesn't.

-> extd_test_scn01.3a
-> extd_test_scn03.2a
-> extd_test_scn03.3a.
The failures are due to global load balance defect where it fails to migrate busy tasks if the cores already have other tasks pinned.

tags: added: bl-android bl-mp
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :

extd_test_scn01.3a :
This scenario assumes the light task is already in little domain however
its computed load is increasing due to the load pattern (due to run queue
residency). The task moves to the big domain since the little domain is
oversubscribed. The runnable task get immediatly the CPU.

extd_test_scn03.2a:
This scenario assumes the heavy task is already in big
domain however its computed load is decreasing due to
the load pattern (due to idling). The task moves to the little
domain since the big domain is oversubscribed.

extd_test_scn03.3a:
This scenario assumes the light task is already in little
domain however its computed load is increasing due to
the load pattern (due to run queue residency). The task moves to big
domain because little domain is oversubscribed.

Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
description: updated
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :
Revision history for this message
Amit Kucheria (amitk) wrote :

Assigning to ARM, since this bug occurs with the HMP patchset and tests supplied by ARM.

Changed in linaro-big-little-system:
assignee: nobody → Robin Randhawa (robin-randhawa-arm)
importance: Undecided → Medium
Revision history for this message
Naresh Kamboju (naresh-kamboju) wrote :

w.r.t. the latest scheduler test sources from ARM the following changes been made
-> extd_test_scn01.3a --> DEPRECATED
-> extd_test_scn03.2a --> PASSED
-> extd_test_scn03.3a.--> DEPRECATED

This bug can be closed.

Revision history for this message
Mark Brown (broonie) wrote :

Closed. Naresh, if a bug can be closed please go ahead and close it (at least for the LSK).

Changed in linaro-stable-kernel:
status: New → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.