View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0001567T99X171.00 SKB EagleSW Issuepublic2022-11-25 08:592023-02-01 14:42
Reporter(ALTech) Younkwang Jung Assigned To(SW) Peter ChenDue Date2022-12-29 14:13
PriorityhighSeveritys4-minorReproducibilityhave not tried
Status closedResolutionfixed 
Summary0001567: [Smart3][ATV10] Requests review of settings that do not kill as much as possible at lmk
DescriptionHI Kerwin

Memory leak is currently present in the homeUI.
It's been six months, but it's still not solved.
Due to this memory issue, several other services are being killed (lmk/oom)
and when certain services (especially when upgrading EDMP) are killed, the VoC increases directly.

About com.skb.btv.upgrade / com.skb.tvpatch services
SKB is requesting to set the values of oom score, oom_adj, and oom_score_adj to not kill even when memory is low(lmk).
https://jira.skbroadband.com/browse/BFXBTF-481

Please check it
Thank you
YK.jung
TagsNo tags attached.
Attach Tags

Users monitoring this issue

User List (ALTech) JunGyu Kim , (ALTech) SY Yoon

Activities

(SW) Kerwin Chen

2022-11-25 15:13

developer   ~0011319

Hi YK,

There was a modification in aml_cma.c for CTS test.
We can add SKB services as a whitelist to avoid being killed by lmk.
Could you share how to verify the result ?

Thank you !

(ALTech) Younkwang Jung

2022-11-28 11:38

developer   ~0011326

Hi Kerwin

The current test method is not clear.
First of all, the issue will be observed if we proceed with the EDMP OTA test after 2 days of monkey test.
I will check the another test method additionally.

First of all, please review two things first.
1) Please refer to aml_cma.c and check how to reduce the killing frequency of "com.skb.btv.upgrade" and "com.skb.tvpatch" processes
2) How to never kill "com.skb.btv.upgrade" and "com.skb.tvpatch" processes

Thank you
YK.Jung

(ALTech) Younkwang Jung

2022-12-05 08:49

developer   ~0011381

Hi Kewin

please update me on the progress
and if it's hard to test , make a theoretical correction and let me know the correction point, and I'll proceed with the additional test.

Thank you
YK.Jung

(SW) Kerwin Chen

2022-12-05 09:10

developer   ~0011385

Hi YK,

Please check commit in "common" folder. (9ce8f0ff6ed0ffc1b5f406d82dc121d3aec29e46)
It is what I said the modification for CTS test. (https://172.18.223.170/vaas/view.php?id=1567#c11319)
Thank you !

(ALTech) Younkwang Jung

2022-12-19 14:19

developer   ~0011519

Hi Kerwin

If I look at the code you mentioned, it seems to be the way that the processor does not kill at all. is it correct?
=========================================================================================
+++ b/drivers/amlogic/memory_ext/aml_cma.c
@@ -969,6 +969,10 @@ static unsigned long cma_shrinker_scan(struct shrinker *s,
                            tasksize <= selected_tasksize)
                                continue;
                }
+ if (!strcmp(p->comm, "s.device.statsd"))
+ continue;
+ if (p->parent && !strcmp(p->parent->comm, "s.device.statsd"))
+ continue;
                selected = p;
                selected_tasksize = tasksize;
                selected_oom_score_adj = oom_score_adj;
========================================================================================

INNOPIA has set oom score(? , I am not sure) in the framework to match com.skb.btv.upgrade/com.skb.tvpatch processor to the same value as com.skb.tv.
For your information, com.skb.tv is set to a processor score that does not kill as much as possible.
Please check this method also.

Thank you!!
YK.Jung

(SW) Kerwin Chen

2022-12-20 09:52

developer   ~0011529

Hi YK,

I can't find where to set oom score of "com.skb.tv" in Framework.
Do you have idea about this modification ?

Thank you!

(ALTech) Younkwang Jung

2023-01-09 12:16

developer   ~0011717

Hi Kerwin

https://jira.skbroadband.com/browse/BFXBTF-481

as previously mentioned ( https://mantis.cnsbg.foxconn.com/vaas/view.php?id=1567#c11385 )
please send me the FW and patch file that have been handled as exceptions for the two processors below.
com.skb.btv.upgrade / com.skb.tvpatch services

SKB wants to apply it to ATV10/ATV12 after verification.

Thank you
YK.Jung

(SW) Peter Chen

2023-01-13 09:07

developer   ~0011758

Hi YK,

Please check the FW and patch file in /release_by_fxn/smart3_test/oom

Thank you
Peter

(ALTech) Younkwang Jung

2023-01-13 12:40

developer   ~0011770

Hi Peter

The file you sent me is for ATV12.

This patch will also be applied to ATV10 and the EDMP update test need to be done on ATV10.
Please build it at ATV10(UI534) and send me the Test FW .

and please also share the patch file how you modified it.

Thank you
YK.Jung

(SW) Peter Chen

2023-01-13 17:13

developer   ~0011775

Hi YK,

I upload FW and patch file for ATV10 again to same folder. Please check it

Thank You
Peter

(ALTech) JunGyu Kim

2023-01-19 12:27

developer   ~0011838

Hi Peter,

I am checking the FW you delivered.
There is an alphabet in the FW you delivered, so OTA cannot be performed.
Please remove the alphabet and rebuild.(15.534.839)

Thank you.

(SW) Peter Chen

2023-01-19 15:00

developer   ~0011839

Hi YK,

I rebuild again and upload to same folder. Please check again.

Thanks
Peter

(ALTech) JunGyu Kim

2023-01-27 11:53

developer   ~0011857

Hi Peter,

Result of testing with the FW you delivered, It seems working well.
But SKB wants to test in OS12. As you know that, OTA cannot be performed if FW's version have alphabet.
Please remove the alphabet in FW version and rebuild OS12 include this patch.

Thank you.

(SW) Peter Chen

2023-01-30 10:50

developer   ~0011872

Hi JunGyu,

I upload the FW for ATV12 and version is 217. The patch is also included.
You can get it in the folder /release_by_fxn/smart3_test/oom/atv12/

Thanks,
Peter

(ALTech) JunGyu Kim

2023-01-31 20:30

developer   ~0011924

Hi Peter,

As a result of testing with the FW you provided, there seems to be no problem.
Please commit this modification to both ATV10 (536 branch) / ATV12 (537 branch).

Thank you
Kim

(SW) Peter Chen

2023-02-01 11:02

developer   ~0011936

Hi JunGyu,

I commited this modification to both ATV10 (536) and ATV12 (537)

Thanks,
Peter

(ALTech) JunGyu Kim

2023-02-01 14:42

developer   ~0011952

Hi Peter,

I confirmed that this modification is committed 536 / 537 branch.

Thank you for your support.
Kim

Issue History

Date Modified Username Field Change
2022-11-25 08:59 (ALTech) Younkwang Jung New Issue
2022-11-25 08:59 (ALTech) Younkwang Jung Status new => assigned
2022-11-25 08:59 (ALTech) Younkwang Jung Assigned To => (SW) Kerwin Chen
2022-11-25 08:59 (ALTech) Younkwang Jung Issue Monitored: (ALTech) SY Yoon
2022-11-25 08:59 (ALTech) Younkwang Jung Issue Monitored: (ALTech) JunGyu Kim
2022-11-25 15:13 (SW) Kerwin Chen Note Added: 0011319
2022-11-28 11:38 (ALTech) Younkwang Jung Note Added: 0011326
2022-12-05 08:49 (ALTech) Younkwang Jung Note Added: 0011381
2022-12-05 09:10 (SW) Kerwin Chen Note Added: 0011385
2022-12-19 13:13 (ALTech) Younkwang Jung Due Date => 2022-12-29 14:13
2022-12-19 14:19 (ALTech) Younkwang Jung Note Added: 0011519
2022-12-20 09:52 (SW) Kerwin Chen Note Added: 0011529
2023-01-09 12:16 (ALTech) Younkwang Jung Note Added: 0011717
2023-01-09 12:28 (SW) Kerwin Chen Assigned To (SW) Kerwin Chen => (SW) Peter Chen
2023-01-13 09:07 (SW) Peter Chen Note Added: 0011758
2023-01-13 12:40 (ALTech) Younkwang Jung Note Added: 0011770
2023-01-13 17:13 (SW) Peter Chen Note Added: 0011775
2023-01-19 12:27 (ALTech) JunGyu Kim Note Added: 0011838
2023-01-19 15:00 (SW) Peter Chen Note Added: 0011839
2023-01-27 11:53 (ALTech) JunGyu Kim Note Added: 0011857
2023-01-30 10:50 (SW) Peter Chen Note Added: 0011872
2023-01-31 20:30 (ALTech) JunGyu Kim Note Added: 0011924
2023-02-01 11:02 (SW) Peter Chen Note Added: 0011936
2023-02-01 14:42 (ALTech) JunGyu Kim Status assigned => closed
2023-02-01 14:42 (ALTech) JunGyu Kim Resolution open => fixed
2023-02-01 14:42 (ALTech) JunGyu Kim Note Added: 0011952