foxyc partial logs don't count. Note that it should be debug log. You can enable it by hand by creating /storage/.kodi/userdata/advancedsettings.xml with following content:
<advancedsettings>
<loglevel hide="false">1</loglevel>
</advancedsettings>
full dmesg output would be also useful.
I really don't know what might be wrong. You said RPi worked ok in such case? Debugging drivers is not easy - I'm using separate sd card with ALARM rootfs (arch linux for arm) and modified U-Boot to boot kernel over network. Then I can modify kernel, build it, boot it over tftpd and test it.
aelray dmesg shows that HDMI is always in disconnected state. Then it's no wonder why you don't get any picture. Just to be clear, is HDMI cable actually connected to the board and only TV is off? That case should work as I often boot it in such way. However, if you plug HDMI cable later, when board is already booted, then I imagine there would be problems, but I don't plan to work on that anytime soon.
This is a single-core system, I believe.
Oh, I mixed up few SoCs. That's strange then. You can always press "o" during playback to check if video is HW decoded or not.
However I did get to dmesg output through ssh1, and there is more
Nothing strange here. But it kernel actually crashed, you shouldn't be able to get ssh connection. Is there any file with "kodi_crashlog" in same folder than normal kodi log?
That is odd then, because the OSD reports CPU significantly below 100% when stuttering like that, typically 40-60%
Maybe ffmpeg decode mpeg4 only on one core.
dmesg: It seems that after Kodi starts up, the output on UART stops. Is that intended (where should I get it, then)?
That strongly suggest that Linux crashes. Can you add "ignore_loglevel" to kernel command line in extlinux.conf, along with previous modifications we talked about?
FPS are reported only for UI at the moment, so this is surely lower for menus (not much is going on) and it doesn't match video playback. mpeg4 is slow because there is no HW acceleration and it's decoded by CPU. I have no idea what's going on with H264. For that, I would need at least Kodi debug log and dmesg output.
It's always only happening when unplugged. So if started without hdmi plugged I can't just plug it in, but have to restart.
Yeah, I imagine this could be a problem, but I'm sorry to say that this is not really a priority to fix.
am seeing a strange shimmer in the Kodi GUI (starts as soon as the splash screen shows up. It's as if some lines would render the contents of the previous line, instead of its own pixels.
Can you make a photo of that or short video? Anyway, it's unlikely that I can help with that issue because I don't have any A10 board. Does that prevent to be able to navigate menu and start video playback?
lumpi Just to be clear, when TV was off, was HDMI cable connected during boot? From dmesg doesn't seems so. I think there are some issues if you connect HDMI cable after board booted, no matter if TV is on or off.
I uploaded new test images for A10:
LibreELEC-A10.arm-9.80-devel-20200216105054-0b8cbed-mele.img.gz
LibreELEC-A10.arm-9.80-devel-20200216111916-0b8cbed-inet97fv2.img.gz
It turns out that A10 has Cortex A8 with vfpv3 for floating point coprocessor as opposed to Cortex A7 and vfpv4 on A20. This means optimizations were set wrong for A10 and of course init script crashed.
uh, A10 is Cortex-A8, but A20 is Cortex-A7. So probably A10 really needs new device type with proper CPU settings. I'll rework project a bit and upload new test image.
what would I need to clean to make either of these changes apply in the resulting image?
for second one you need just to rebuild busybox package, but for first one I suggest you rebuild whole image.
lumpi Sorry, it seems dmesg overflow, that produces much more output than anticipated. Can you re-run with "drm.debug=0x16"?
Would dmesg and kodi log as you suggested before help? Then i can try it.
yes, especially dmesg after you appended kernel arguments with that debug command
You tell then the h3 defaults to a 4k signal?
Depends on TV/monitor connected. If EDID is read, then it switches to recommended resolution. If not, it switches to 1024x768 and no sound. My TV is 4K, so it will switch to that. In your case, it should switch to 1080p. EDID should be readable even if TV/monitor is turned off, but I'm not sure if that means only in standby or also when unplugged.
Is there a way to check that it (and busybox executable) is actually on the built image..?
Not sure, but currently initramfs is in cpio archive, which starts with "070701" string.
Well, my experience is opposite. I often leave TV (LG B8) turned off and only power on SBC (doesn't matter if A64, H3 or H6). When Kodi starts, TV turns on and image is what you would expected - Kodi GUI and resolution is 4K. I heard that some devices disable DDC when they're powered off. If that's what your TVs do, then your situation is (almost) expected. Although hearing sound would suggest otherwise. Provide Kodi debug log and dmesg output and I'll check if there is anything which would explain behaviour you observed. Even better, add "drm.debug=0x1f" to extlinux.conf and get that dmesg output.
Am I doing this right? I decompiled the DTB, set status="disabled" on the first enabled usb node (and the phy which the subsequent attempt to boot complained about), compiled and replaced sun4i-a10-a1000.dtb.
yes, seems good.
but the error message is the same:
So wifi just appears to be last driver loaded, but not the cause.
jernej , which /init is this referring to? This one?
Yes.
