Manjaro arm on Amlogic Tv box?

Thanks TheMojoMan I will try it tonight when I’m home.

Sorry, did not catch this part earlier. Is linux-vim kernel 5.9.0-2 stable with Panfrost? You did indicated earler kernel 5.9.0 wasn’t and linux-vim kernel 5.9.8-1 is also not stable with Panfrost, Any changes to linux-vim kernel 5.9.8-1 being stable with Panfrost?

Just enquire, if you install cpupower
pacman -S cpupower
set it to performance “sudo cpupower frequency-set -g performance”
systemctl enable cpupower --now

  1. Does it heat up the Unit since it is always running at highest frequency

The only difference it would make would be when the box is idling (or low load), but I haven’t really seen any difference (perhaps a few degrees higher idling temp with governor set to performance).

In order to make the selection of governor be persistent across reboots you must also edit /etc/default/cpupower (only one line needs edit).

Edit: You can also fiddle with the settings of the governor (see /etc/default/cpupower) so that the CPU(s) run at a constant frequency, but not necessarily the maximum that the chip(s) support. (My Ugoos is also currently frequency capped since the dtb doesn’t exactly match my particular version of the Ugoos, which in fact supports higher clocks than specified in the dtb.)

dmesg output on Panfrost,
[ 4.573519] panfrost ffe40000.gpu: clock rate = 24000000
[ 4.575344] panfrost ffe40000.gpu: supply mali not found, using dummy reg
ulator
[ 4.582784] panfrost ffe40000.gpu: mali-g52 id 0x7212 major 0x0 minor 0x0
status 0x0
[ 4.590230] panfrost ffe40000.gpu: features: 00000000,13de77ff, issues: 0
0000000,00000400
[ 4.598348] panfrost ffe40000.gpu: Features: L2:0x07110206 Shader:0x00000
000 Tiler:0x00000809 Mem:0x1 MMU:0x00002830 AS:0xff JS:0x7
[ 4.610061] panfrost ffe40000.gpu: shader_present=0x3 l2_present=0x1
[ 4.617461] panfrost ffe40000.gpu: error powering up gpu L2
[ 4.622928] panfrost ffe40000.gpu: error powering up gpu shader
[ 4.628866] [drm] Initialized panfrost 1.1.0 20180908 for ffe40000.gpu on

Notice Panfrost error powering up gpu L2 and power up gpu shader when using the “meson-g12b-gtking-pro.dtb” that came with the this image. I have notice the “meson-g12b-gtking-pro.dtb” from Manjaro_Xfce_ Mainline linux-kernel 5.9.6-1 or 5.9.9-2 size is slightly different that KDE_VIM3 with linux-vim kernel 5.9.0-2. If I am not mistaken the "meson-g12b-gtking-pro.dtb from linux kernel 5.9.6-1/5.9.9-2 does not have the panfrost gpu error: powering up gpu L2 or error powering up gpu shader.

Will confirm shortly.

Edit: The Manjaro_VIM3_Xfce_ linux-kernel 5.9.6-1, dmesg output on Panfrost does not have Panfrost gpu powering up error.

rc rc0: meson-ir as /devices/platform/soc/ff800000.bus/ff808000.ir/rc/rc0
[ 7.353950] mmc2: queuing unknown CIS tuple 0x81 (9 bytes)
[ 7.373049] panfrost ffe40000.gpu: clock rate = 24000000
[ 7.373555] panfrost ffe40000.gpu: supply mali not found, using dummy regulator
[ 7.373808] panfrost ffe40000.gpu: mali-g52 id 0x7212 major 0x0 minor 0x0 status 0x0
[ 7.373811] panfrost ffe40000.gpu: features: 00000000,13de77ff, issues: 00000000,00000400
[ 7.373815] panfrost ffe40000.gpu: Features: L2:0x07110206 Shader:0x00000000 Tiler:0x00000809 Mem:0x1 MMU:0x00002830 AS:0xff JS:0x7
[ 7.373818] meson-vrtc ff8000a8.rtc: registered as rtc0
[ 7.373820] panfrost ffe40000.gpu: shader_present=0x3 l2_present=0x1
[ 7.373853] meson-vrtc ff8000a8.rtc: setting system clock to 1970-01-01T00:00:07 UTC (7)
[ 7.381987] meson-drm ff900000.vpu: Queued 3 outputs on vpu
[ 7.382168] meson_vdec: module is from the staging directory, the quality is unknown, you have been warned.
[ 7.383008] rc rc0: lirc_dev: driver meson-ir registered at minor = 0, raw IR receiver, no transmitter
[ 7.383097] input: meson-ir as /devices/platform/soc/ff800000.bus/ff808000.ir/rc/rc0/input6
[ 7.387246] audit: type=1130 audit(7.510:6): pid=1 uid=0 auid=4294967295 ses=4294967295 subj==unconfined msg=‘unit=systemd-networkd comm=“systemd” exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success’
[ 7.390619] meson8b-dwmac ff3f0000.ethernet: IRQ eth_wake_irq not found
[ 7.390628] meson8b-dwmac ff3f0000.ethernet: IRQ eth_lpi not found
[ 7.390799] meson8b-dwmac ff3f0000.ethernet: PTP uses main clock
[ 7.393755] [drm] Initialized panfrost 1.1.0 20180908 for ffe40000.gpu on minor 0
[ 7.399753] meson-ir ff808000.ir: receiver initialized
[ 7.422174] meson8b-dwmac ff3f0000.ethernet: User ID: 0x11, Synopsys ID: 0x37
[ 7.426008] meson8b-dwmac ff3f0000.ethernet: DWMAC1000
[ 7.426019] meson8b-dwmac ff3f0000.ethernet: DMA HW capability register supported
[ 7.426021] meson8b-dwmac ff3f0000.ethernet: RX Checksum Offload Engine supported
[ 7.426026] meson8b-dwmac ff3f0000.ethernet: COE Type 2
[ 7.426028] meson8b-dwmac ff3f0000.ethernet: TX Checksum insertion supported
[ 7.426030] meson8b-dwmac ff3f0000.ethernet: Wake-Up On Lan supported
[ 7.426129] meson8b-dwmac ff3f0000.ethernet: Normal descriptors
[ 7.426132] meson8b-dwmac ff3f0000.ethernet: Ring mode enabled
[ 7.426135] meson8b-dwmac ff3f0000.ethernet: Enable RX Mitigation via HW Watchdog Timer
[ 7.426852] libphy: stmmac: probed
[ 7.430215] mmc2: new high speed SDIO card at address 0001
[ 7.432487] libphy: mdio_mux: probed
[ 7.459732] Bluetooth: Core ver 2.22
[ 7.459765] NET: Registered protocol family 31
[ 7.459766] Bluetooth: HCI device and connection manager initialized
[ 7.459777] Bluetooth: HCI socket layer initialized
[ 7.459781] Bluetooth: L2CAP socket layer initialized
[ 7.459788] Bluetooth: SCO socket layer initialized
[ 7.474199] meson-dw-hdmi ff600000.hdmi-tx: Detected HDMI TX controller v2.01a with HDCP (meson_dw_hdmi_phy)
[ 7.474546] meson-dw-hdmi ff600000.hdmi-tx: registered DesignWare HDMI I2C bus driver
[ 7.475522] meson-drm ff900000.vpu: bound ff600000.hdmi-tx (ops meson_dw_hdmi_ops [meson_dw_hdmi])
[ 7.476031] [drm] Initialized meson 1.0.0 20161109 for ff900000.vpu on minor 1

Edit2: Interesting, if transfer the Manjaro_VIM3_Xfce linux kernel 5.9.9-2 "meson-g12b-gtking-pro.dtb’ over to Manjaro-ARM-kde=plasma-vim3-22-11-2020.img.xz the same Panfrost error powering up gpu L2 and error powering up gpu shader where as it does give any Panfrost error in Xfce. Wonder why?

If I use the Armbian “meson-g12b-gtking-pro.dtb” on Manjaro-ARM-kde=plasma-vim3-22-11-2020.img.xz, NO Panfrost error.

Can you share the command you used? Is it cpu frequency or gpu ?
Currently we use OnDemand as default cpu governor, If you think performance will fix the graphical glitch then I can test it and then with new kernel we can user performance mode as anyways amlogic soc doesn’t really get hot and doesn’t consume much power.

I would suggest not to use it until we have stable panfrost as wayland will result into black screen for now.

Everything is already present except for panfrost support :stuck_out_tongue:

No Not to get your hopes high but this is a a kernel from custom branch which have something extra which is causing panfrost unstability.

@spikerguy Yes, on my Ugoos (AM6 plus) it definitely fixes both problems (glitches, freezes) and makes the box completely stable (yay!). I did like this. (It is cpu frequency.)

If I use this command " echo performance >/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor" and reboot if will boot up in “performance” mode?

No. That does not survive a reboot. You must do like I outlined here (i.e. edit a conf. file) (and enable the cpupuwer systemd service, if you have not do so already.)

Edit: I could of course have formatted this post better, but all the info is there.

P.S. Fiddling with the /sys/devices/… variables must be done as root (or using sudo).

echo performance >/sys/devices/system/cpu/cpu0/cpufreq/scaling_go
vernor
bash: /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor: Permission denied

-> FIXED: I was using the wrong .dtb file.

-> Only a problem when using Wayland, no problem with Plasma.

Just installed “cpupower” using Package Manager.
Preparing…
Synchronizing package databases…
Warning: boot-vim3: local (2020.09-1) is newer than core (2020.05-1)
Resolving dependencies…
Checking inter-conflicts…
Downloading cpupower (5.9-1)…
Checking keyring…
Checking integrity…
Loading packages files…
Checking file conflicts…
Checking available disk space…
Installing cpupower (5.9-1)…
Running post-transaction hooks…
Reloading system manager configuration…
Arming ConditionNeedsUpdate…
Transaction successfully finished.

[jfl@GTKPro ~]$ sudo cpupower frequency-set -g performance
Setting cpu: 0
Setting cpu: 1
Setting cpu: 2
Setting cpu: 3
Setting cpu: 4
Setting cpu: 5
[jfl@GTKPro ~]$

[jfl@GTKPro ~]$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
performance

[jfl@GTKPro ~]$ systemctl enable cpupower --now
Created symlink /etc/systemd/system/multi-user.target.wants/cpupower.service → /u
sr/lib/systemd/system/cpupower.service.

Is there anything else I need to edit to ensure it boot up as “performance” mode? Somehow I am not clear. Sorry to trouble you and your patience.

There seems to be NO more screen flashes or glitches for the past 10 minutes! TQ
But just as open up two more tabs on Firefox, encountered the first Kernel Panic on my GT King Pro. So the “cpupower performance” setting seems to solve the Manjaro-ARM screen glitches or flashes but it does not the Kernel Panic on GT King Pro.

It seems linux kernel 5.9.9-2 is more stable than linux-vim kernel 5.9.0-2 at least on my GT King Pro so far (just a perception).

Also this is good news. I think it shows that the problem with graphics glitches can be fixed by setting a fixed (high) cpu frequency, and that this works (well, might work) for several boxes/platforms. At the same time, I think we know by now that (at least some of) the kernel panics require a special fix for some boxes (e.g. GTKing Pro). Still, progress, I think.

P.S. In order for this work across reboots you must edit the conf file /etc/defaults/cpupower (and edit the line with governor=…stuff…). It must be uncommented, too.

Tripole,

What to edit in “/etc/default/cpupower (only one line needs edit)” to make it boot up in governor = performance?

Ok, change

#governor='ondemand'

to

governor='performance'

(There must be no hash mark in front of the second line.)

TQ. Done.

Edit: Reboot after editing /etc/default/cpupowe with governor=‘performance’.
[jfl@GTKPro ~]$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
performance
[jfl@GTKPro ~]$

1 Like

Currently testing this and so far it seems to have fixed the graphical glitch but let me test it further and then rebuild the kernel with performance as default gpu.

Yeay no glitched since more than 30 mins now. So we will have performance mode as default :wink:

1 Like

Arrived home and tested your dtb with the vim3 image with kernel 5.9.6-1 , I m having the same issues that you described before, network is working soundcard is recognized but no sound via HDMI.
The dtb that you shared is it the wrong one? If so i think we have de same hardware with different name.
Can you share the correct one? :grin:
Today I don’t have time to test spikerguy image, I will try it tomorrow, thanks

Tripole,

Out of curiosity I check the three other linux images (that does not have screen flashes or glitches) on the governor boot up settings.

  1. Armbian-focal-desktop-kernel-5.9.0 governor=ondemand
  2. Khadas-VIM3-focal-v0.9 20200530 kernel 224 governor=performance
  3. Ubuntu-focal-kernel 206 by Darkstar governor=performance

Interesting Armbian-focal with governor ondemand (cpu freq move 499Mhz, 666Mhz to 1.99GHz according to Armbian-HTOP) there is no screen glitches on GT King Pro (but do encounter kernel panic/freezes).

I’ve been able to reproduce your errors and have glitches with my box also. This only appears when Panfrost gives the powering error.

I guess if there is a powering fault when the processor changes frequency.