Jump to content


Photo

branch zeus


  • Please log in to reply
148 replies to this topic

Re: branch zeus #141 A.A.

  • Senior Member
  • 110 posts

+1
Neutral

Posted 15 September 2020 - 09:44

I am now building an hybrid using kernel and kodi + tools recipes of OE-A in the OpenPLi image but getting the same sigbus error.

 

I have now to verify other buses:

- audio (there is only alsa but kodi config looks for sndio and pulseaudio as well)

- video (maybe I must really install libvugles ?? DESCRIPTION = "shared library for E2 animation")

 

The launcher itself differs a bit (/usr/bin/kodi) and the one in OpenPLi doesn't look at the hdd.

 

I'll spend a bit more work on it.

A.A.



Re: branch zeus #142 Erik Slagter

  • PLi® Core member
  • 46,177 posts

+513
Excellent

Posted 18 September 2020 - 13:53

Did you already use strace to check the signal always happens at the same location in the code?


* Wavefrontier T90 with 28E/23E/19E/13E/9E/4.8E/0.8W/5W via SCR switches 2 x 2 x 6 user bands
* Ziggo digital cable TV (FTA)
I don't read PM -> if you have something to ask or to report, do it in the forum so others can benefit. I don't take freelance jobs.
Ik lees geen PM -> als je iets te vragen of te melden hebt, doe het op het forum, zodat anderen er ook wat aan hebben.

Re: branch zeus #143 A.A.

  • Senior Member
  • 110 posts

+1
Neutral

Posted 18 September 2020 - 22:14

Erik,

yes, strace, valgrind, gdb.

It always goes crazy in libc_init/mips_start -> SIGBUS at unknow addresses.

 

Anyway I did some progress because adding dependencies on

glib-2.0 \
gstreamer1.0 \
gstreamer1.0-plugins-base \

 

i get now Segmentation Fault SIGSEGV :)

I'll work on it this weekend in the free-time, I'll check the platforms again

Cheers

A.A.

 

P.S. it's 1GB unstripped...


Edited by A.A., 18 September 2020 - 22:22.


Re: branch zeus #144 A.A.

  • Senior Member
  • 110 posts

+1
Neutral

Posted 22 September 2020 - 12:52

Erik,

 

here the strace of kodi in  OpenPLi and OpenSPA (working). The problem is in OpenPLi Gles_Init does not start.

I am even using the very same OE-A kernel but the rest is abit different so are the locale and timezone accesses.

What base-differences do other distros have? Maybe these dvb-proxy drivers need some kind of servers/configuration in userland?

 

I did not add yet bluetooth and pulseaudio because that would be a distro change and I have read in the sources these are kodi's optional configuration options.

Besides, apparently the same kodi recipe runs on other models so the runtime deps should be there.

 

Long read...

 

 

OpenPLi  https://pastebin.com/L9TtpDBM

 

OpenSPA https://pastebin.com/Fw3pichZ

 

A.A.



Re: branch zeus #145 Erik Slagter

  • PLi® Core member
  • 46,177 posts

+513
Excellent

Posted 23 September 2020 - 12:16

DVB proxy indeed is the glue between the drivers and userland.


* Wavefrontier T90 with 28E/23E/19E/13E/9E/4.8E/0.8W/5W via SCR switches 2 x 2 x 6 user bands
* Ziggo digital cable TV (FTA)
I don't read PM -> if you have something to ask or to report, do it in the forum so others can benefit. I don't take freelance jobs.
Ik lees geen PM -> als je iets te vragen of te melden hebt, doe het op het forum, zodat anderen er ook wat aan hebben.

Re: branch zeus #146 A.A.

  • Senior Member
  • 110 posts

+1
Neutral

Posted 23 September 2020 - 13:29

For the dvb-proxy drivers there is indeed another layer, the vuplus-platform.

 

I thought that with these BSP changes all necessary pieces are now in place to add dvbproxy to the machine features:

  https://tinyurl.com/yxmfh4ex

 

But as in my tedious posts above there is still something missing.

 

I've ruled out that the issue is not in:

- kernel (same sources)

- kodi recipe (98% the same)

- vuplus platform (fixed)

 

So it could be some module not loaded or some other OE-A initscript (creating devices, symlinks, daemons,...)

Or maybe there is still something about the private libs and these are not correctly linked, but these appear just fine in strace.



Re: branch zeus #147 Erik Slagter

  • PLi® Core member
  • 46,177 posts

+513
Excellent

Posted 23 September 2020 - 16:00

Can you make sure the bus error is not generated inside a system call (which is very unlikely but not impossible)? If it is, all bets are off, then it most likely is a driver bug. No mentionings in the dmesg I guess?

 

There exists a tool called ltrace, it's in OE but I don't think we build it currently. Also I am afraid you do need debug symbol version of the relevant libs, which we probably don't have.

 

Just brainstorming here...


* Wavefrontier T90 with 28E/23E/19E/13E/9E/4.8E/0.8W/5W via SCR switches 2 x 2 x 6 user bands
* Ziggo digital cable TV (FTA)
I don't read PM -> if you have something to ask or to report, do it in the forum so others can benefit. I don't take freelance jobs.
Ik lees geen PM -> als je iets te vragen of te melden hebt, doe het op het forum, zodat anderen er ook wat aan hebben.

Re: branch zeus #148 A.A.

  • Senior Member
  • 110 posts

+1
Neutral

Posted 24 September 2020 - 10:04

Heh, adding more deps transformed the sigbus in sigsegv...

 

Anyway I have maybe another hint:

 

OpenPLi:  getrlimit(RLIMIT_NOFILE, {rlim_cur=1024, rlim_max=4*1024}) = 0

OpenSPA: getrlimit(RLIMIT_NOFILE, {rlim_cur=16*1024, rlim_max=16*1024}) = 0



Re: branch zeus #149 Erik Slagter

  • PLi® Core member
  • 46,177 posts

+513
Excellent

Posted 25 September 2020 - 09:24

If the limit isn't handled gracefully, yeah, possible. So I guess you could try settings the limits considerately higher, all of them, to start with.


* Wavefrontier T90 with 28E/23E/19E/13E/9E/4.8E/0.8W/5W via SCR switches 2 x 2 x 6 user bands
* Ziggo digital cable TV (FTA)
I don't read PM -> if you have something to ask or to report, do it in the forum so others can benefit. I don't take freelance jobs.
Ik lees geen PM -> als je iets te vragen of te melden hebt, doe het op het forum, zodat anderen er ook wat aan hebben.




1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users