Opened 11 years ago

Closed 11 years ago

#2413 closed defect (worksforme)

high cpu usage on sh4 platform (octagon) #6422

Reported by: Netview Owned by:
Priority: major Component: DVBApi
Severity: medium Keywords:
Cc: Sensitive: no

Description

This Problem is reproducable at any time if oscam is restarted.
(note: oscam is used as a pure Cccam-client)
...
20:19:40 541838 p data on socket
20:19:40 541838 p cccam(r) server1: parse_msg=1
20:19:40 541838 p cccam(r) server1: cws: 3 2A38C0222E1975BCD0B3E4674F622DDE
20:19:40 541838 p cccam(r) server1: no ecm pending!
20:19:40 541838 p start client thread action 24
20:19:40 5492F0 c data from add_job action=24 client c dvbapi
20:19:40 5492F0 c ecm answer from reader sever1 for ecm 9DDC rc=0
20:19:40 5492F0 c dvbapi: write cw odd index: 0 PTI
20:19:40 5492F0 c dvbapi (1834&000000/C481/89:9DDC): found (753 ms) by server1
20:19:40 5492F0 c cw:
20:19:40 5492F0 2A 38 C0 22 2E 19 75 BC D0 B3 E4 67 4F 62 2D DE
20:19:40 5492F0 c Got sigpipe signal -> captured
20:19:40 5492F0 c Writing to pipe failed (errno=32 Broken pipe)
20:19:40 5492F0 c ending thread

Is it caused by the broken pipe?

Change History (6)

comment:1 by pooyair, 11 years ago

Resolution: invalid
Status: newclosed

I use myself oscam for my sh4 client , from r6303 till 6421 , Cpu load was very high (it use all remaining cpu resources (70%) ) but after 6422 , it's fine now like before ... (cpu load is fewer than 0.8 % for sh4 clinet ,
what percent your cpu resources use when u restart oscam? afaik above broken pipe error , are nothing to do with cpu usages , and i am afraid the above log can't say any helpful info , since from beginning of oscam 1.2 i saw that error line , but as mentioned above with rev >6422 , There is no high cpu usages with sh4 client.
if u really see high cpu usage statistics , Suggest u , Limit your config and also change some of parameter esp pmt_mode in dvbapi , and if u find which parameter cause this high cpu for u , please feedback it in forum .

Last edited 11 years ago by pooyair (previous) (diff)

comment:2 by Netview, 11 years ago

Maybe I found the problem!

'pmt_mode = 5' should be used for octagon-receiver but this will cause a high cpu usage. With pmt_mode= 4' the cpu usage is ok but a shutdown ends with a core-dump.

comment:3 by Netview, 11 years ago

Resolution: invalid
Status: closedreopened

comment:4 by pooyair, 11 years ago

afai hear here , pmt_mode 5 is mainly suitable for stapi image ,this mode works really nice and most stable one with my ipbox, it's the best option , also users with other stapi box (like Kathrein) always are feedback pmt_mode 5 caused best result (stability and speed) for them. What firmware do u use with your octagon? (does it a stapi image , or it's sh4 firmware (e2 image)?

well i use pmt_mode 5 with my ipbox (stapi) client , and head of trunk (rev>= 6422) , and won't see any high cpu usages with pmt_mode 5!

if with pmt_mode 4 u gets core-dump , could u use gdb , provide any debug log , and post it here ?

Last edited 11 years ago by pooyair (previous) (diff)

comment:5 by Netview, 11 years ago

my octagon 1028P is using the stapi image. I will check for gdb.

comment:6 by Deas, 11 years ago

Resolution: worksforme
Status: reopenedclosed

if oscam would not run on octagon at all we would have many reports about that. please try another image.

Note: See TracTickets for help on using tickets.