Firefly Open Source Community

   Login   |   Register   |
New_Topic

Firefly New DualBoot Firmware Release(Beta)

87

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
87
Posted at 6/27/2016 18:47:42     
Last edited by Asuuga In 6/27/2016 18:54 Editor
mac-l1 Posted at 6/24/2016 07:11
Hi guys,

Just made a prelimary kodi release for this ubuntu image. This has GPU working at 60 f ...

Hi mac-l1,

I tested your last version of kodi and it's working great !

It succesfully decoded a 1080p hevc video flawlessly. There was just sometime some jitter in the sound.
Audio through HDMI was also working.
I also tested a 4k hevc video, but it didn't work well, as the decoding was very slow.


Hi uJerry,

Ah ok, thank you for the explanation on the graphics stack architecture !
I'm sorry I can't reproduce the mouse and keyboard bug anymore... It is working fine now.

And isle, thank you for releasing the kernel source code
Reply

Use props Report

127

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
127
Posted at 6/29/2016 04:23:16     
the following is logged in the kern.log right after the keyboard/mouse freezes:


  1. Jun 29 04:21:56 localhost kernel: [  141.203907] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  2. Jun 29 04:21:56 localhost kernel: [  141.303573] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  3. Jun 29 04:21:56 localhost kernel: [  141.403575] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  4. Jun 29 04:21:56 localhost kernel: [  141.503609] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  5. Jun 29 04:21:56 localhost kernel: [  141.603654] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  6. Jun 29 04:21:56 localhost kernel: [  141.703769] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  7. Jun 29 04:21:56 localhost kernel: [  141.803587] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
  8. Jun 29 04:21:56 localhost kernel: [  141.903621] rk3x-i2c ff160000.i2c: timeout, ipd: 0x80, state: 2
Copy the code
Reply

Use props Report

58

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
58
Posted at 6/30/2016 09:57:18     
andrewsmithty Posted at 6/29/2016 04:23
the following is logged in the kern.log right after the keyboard/mouse freezes:

Hi andrewsmithty
this bug is caused by vga ddc detect. We will fix the bug in time.
temporary, if you do not use vga monitor, you can disable it in arch/arm/boot/dts/firefly-rk3288-reload.dtsor you can update this resource.img

resource.img.tar.gz

12.37 KB, Down times: 3

resource.img withour vga ddc

Reply

Use props Report

58

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
58
Posted at 6/30/2016 12:48:57     
mac-l1 Posted at 6/24/2016 18:14
Hi all,

FYI: chromium-browser can now also run with GPU acceleration, see: http://forum.odroid.co ...

Hi mac-11
I got a libmali.gbm.so from rockchip, but I am not sure if it is available.  in wayland , gdm and weston able to start with mesa, but replace to libmali.x11.so or libmali.gbm.so, weston boot fail, Even gdm also failed to start. but lightdm can start success.
gnome-session-wayland always start fail .

libmali.gbm.so.tar.gz

1.47 MB, Down times: 5

Reply

Use props Report

690

Credits

0

Prestige

0

Contribution

advanced

Rank: 4

Credits
690
Posted at 6/30/2016 15:31:09     
Last edited by mac-l1 In 6/30/2016 15:34 Editor
ujerry Posted at 6/30/2016 12:48
Hi mac-11
I got a libmali.gbm.so from rockchip, but I am not sure if it is available.  in wayland ...

Hi ujerry,

Tnx for your effort! I checked the .so binary file and it needs mali midgard r7p0. As ARM released this for ODROID XU3 I checked these user space libs and that might give a clue how to realize this for weston. I would assume that you need to make a lot of symbolic links: so for libEGL*.so, libGLES*.so and also for libgbm.so  all pointing to this libmali.gbm.so. The libgbm.so is missing for weston to get dri access the gbm way. I would also suggest lubuntu 16.04 as that has latest weston/xwayland libs.



So its a bit of effort here, getting mali r7p0, lubuntu 16.04 and then fixing weston; my guess is that it should work.

Cheers -- mac-l1
Reply

Use props Report

58

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
58
Posted at 6/30/2016 16:35:00     
mac-l1 Posted at 6/30/2016 15:31
Hi ujerry,

Tnx for your effort! I checked the .so binary file and it needs mali midgard r7p0. A ...

Hi mac-l1
yes , I did that on lubuntu 16.04 .  but chromium broswer has a bug, so we did not release lubuntu 16.04. see
https://bugs.launchpad.net/ubunt ... rowser/+bug/1563184
I will try mali r7p0 . thanks.
Reply

Use props Report

690

Credits

0

Prestige

0

Contribution

advanced

Rank: 4

Credits
690
Posted at 6/30/2016 16:48:55     
Last edited by mac-l1 In 6/30/2016 16:58 Editor
ujerry Posted at 6/30/2016 16:35
Hi mac-l1
yes , I did that on lubuntu 16.04 .  but chromium broswer has a bug, so we did not rele ...

Hi ujerry,

great that you are working on it!

and then just hope that performance will be better... memeka from the odroid community who helped me before with  libhybris has experience with mali gbm/weston r7p0 on odroid xu3: https://community.arm.com/message/35971#35971. just hope rockchip's lib has better results...

i also asked him about odroid xu3 solutions and performance, see http://forum.odroid.com/viewtopic.php?f=95&t=18905#p149077.  it seems that although our current firefly gpu is working at least, its performance is too low for what should be expected. related to glmark2-es scores he concludes:
your offscreen scores are very low. they should be 5-10 times faster. is there any DVFS on mali? can you disable it?
so wonder what the bottleneck is here. maybe you have some suggestion or can experiment a bit? so off-screen glmark2-es scores should be around 1000 instead of around 200. maybe something wrong with syncronization, memory access, frequencies/power, ... right now no clue...

cheers -- mac-l1



Reply

Use props Report

690

Credits

0

Prestige

0

Contribution

advanced

Rank: 4

Credits
690
Posted at 6/30/2016 17:04:11     
hi ujerry,

also, i have an original firefly (no reloaded) and two things are really missing with this release or at least they don't work:

1) hdmi audio: i only have audio through the jack and not though hdmi. Asuuga states above that hdmi audio works for him, so i assume he has an firefly reload? and what would be the difference? do you know how to fix hdmi audio for original firefly (not reloaded)?

2) sd card: my mmc/sd card is not detected and i cannot access this external ext4 sd card. that makes it quite hard to develop for me. do you know a way to fix access to sd card for the kernel?

hope you can help me out here. tnx!

cheers -- mac-l1
Reply

Use props Report

87

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
87
Posted at 6/30/2016 21:10:32     
Last edited by Asuuga In 6/30/2016 23:37 Editor

Hi mac-l1,

I also have an original firefly. This is what i get in the kernel logs about hdmi-audio :
  1. ubuntu@firefly:~$ dmesg | grep hdmi
  2. [    2.930992] rockchip-hdmi-audio hdmi_audio.6: ASoC: CODEC dw-hdmi-audio not registered
  3. [    2.931010] rockchip-hdmi-audio hdmi_audio.6: register card failed (-517)
  4. [    2.931017] platform hdmi_audio.6: Driver rockchip-hdmi-audio requests probe deferral
  5. [    3.974254] rockchip-hdmi-audio hdmi_audio.6: ASoC: CODEC dw-hdmi-audio not registered
  6. [    3.982201] rockchip-hdmi-audio hdmi_audio.6: register card failed (-517)
  7. [    3.989004] platform hdmi_audio.6: Driver rockchip-hdmi-audio requests probe deferral
  8. [    4.019681] rockchip-hdmi-audio hdmi_audio.6: ASoC: CODEC dw-hdmi-audio not registered
  9. [    4.027607] rockchip-hdmi-audio hdmi_audio.6: register card failed (-517)
  10. [    4.034414] platform hdmi_audio.6: Driver rockchip-hdmi-audio requests probe deferral
  11. [    4.060459] dwhdmi-rockchip ff980000.hdmi: Detected HDMI controller 0x20:0xa:0xa0:0xc1
  12. [    4.068383] dwhdmi-rockchip ff980000.hdmi: Detected HDMI config_id 0xbf:0x22:0xc2:0x0
  13. [    4.094664] dwhdmi-rockchip ff980000.hdmi: monitor does not support audio
  14. [    4.101573] dw-hdmi-audio dw-hdmi-audio: hdmi audio init success.
  15. [    4.107676] rockchip-drm display-subsystem.4: bound ff980000.hdmi (ops dw_hdmi_rockchip_ops)
  16. [    4.313451] dw-hdmi-audio dw-hdmi-audio: jack report [4]
  17. [    4.320387] rockchip-hdmi-audio hdmi_audio.6: dw-hdmi-hifi <-> ff890000.i2s mapping ok
  18. [    4.328780] input: RockchipHDMI HDMI Jack as /devices/hdmi_audio.6/sound/card1/input4
  19. [    4.345253] rockchip-hdmi-audio hdmi_audio.6: hdmi audio init success.
  20. [   32.060181] dw-hdmi-audio dw-hdmi-audio: startup.
  21. [   32.078179] dw-hdmi-audio dw-hdmi-audio: shutdown.
  22. [   32.088048] dw-hdmi-audio dw-hdmi-audio: startup.
  23. [   32.097945] dw-hdmi-audio dw-hdmi-audio: [codec_dai]: trigger enable.
  24. [   37.626900] dw-hdmi-audio dw-hdmi-audio: [codec_dai]: trigger disable.
  25. [   37.627695] dw-hdmi-audio dw-hdmi-audio: shutdown.
Copy the code

So the driver probes correctly the hardware for me.

Also, you can control audio with pavucontrol.
In the "Configuration" tab, I have 2 items : "RockchipHDMI" (obviously, for hdmi output) and "RK_ES8323" (for jack output). I selected the profile from the first one to "Digital Stereo (HDMI) Output" whereas I set the second one to "Off". If both are activated at the same time, sound is only through the jack output and jitters a lot. So I have to activate only one at a time.

When I activate the profile on RockchipHDMI : the following kernel logs appear :
  1. [  208.952332] dw-hdmi-audio dw-hdmi-audio: startup.
  2. [  208.952732] dw-hdmi-audio dw-hdmi-audio: [codec_dai]: trigger enable.
Copy the code


When I deactivate it :
  1. [  462.242270] dw-hdmi-audio dw-hdmi-audio: [codec_dai]: trigger disable.
  2. [  462.242973] dw-hdmi-audio dw-hdmi-audio: shutdown.
Copy the code


About the sdcard problem :
There is no firefly-rk3288.dts in arch/arm/boot (in the kernel sources). So I suppose the kernel has been built with the device-tree from firefly-reload (firefly-rk3288-reload.dts), which is present. Am I correct uJerry ?
And if you look in this file, you'll see that the sdmmc node is disabled. So the hardware is never being probed and is thus not available. I wonder why they deactivated it ? Maybe it was not working correctly.


Finally, I noticed another problem : the rtc is not working. But again, there is no node that describes it in the device tree.
  1. ubuntu@firefly:~$ dmesg | grep rtc
  2. [    4.483768] drivers/rtc/hctosys.c: unable to open rtc device (rtc0)
Copy the code

About the vpu support, do you plan to use the kernel driver located in arch/arm/mach-rockchip/vcodec_service.c ? Or rather use a libhybris solution ?
Reply

Use props Report

58

Credits

0

Prestige

0

Contribution

registered members

Rank: 2

Credits
58
Posted at 7/1/2016 18:44:50     
Hi mac-l1 & Asuuga:
SD card support has fixed , the code has relepse to github.  
and the dts file renamed to firefly-rk3288.dts , I am sorry for I made a mistake before. this dts is workrd for firefly-rk3288, not reload.
for HDMI  Audio  , the paulseAudio profile is set by this script
  1. usr/bin/monitormode
Copy the code

and triggered by udev.   hdmi and jack use the same i2s, So they can't work at the same time, for they request different i2s mclk. there is a way ( reconfigure hdmi audio to fit the i2smclk when es8323 worked) may can solve it. But I haven't tried it at the moment.
for RTC, I will try to fix it , Thank you for reminding me。
Reply

Use props Report

You need to log in before you can reply Login | Register

This forum Credits Rules

Quick Reply Back to top Back to list