hexdump Posted August 16, 2019 Share Posted August 16, 2019 i tried the 5.2.8 dtb, but this is not the problem - no output on hdmi neither. on my other qplus box hdmi out works fine with 5.3, so the problem seems to be specific to the eachlink box. so looks like there is no real way around some kernel bisecting to find out which commit is causing this ... Link to comment Share on other sites More sharing options...
jernej Posted August 16, 2019 Share Posted August 16, 2019 33 minutes ago, hexdump said: some kernel bisecting to find out which commit is causing this ... I would highly appreciate this, otherwise it's impossible for me to find real cause, especially without HW in question. Link to comment Share on other sites More sharing options...
hexdump Posted August 16, 2019 Share Posted August 16, 2019 @jernej - bisecting in progress - 14 rounds, so it will take a while 1 Link to comment Share on other sites More sharing options...
hexdump Posted August 17, 2019 Share Posted August 17, 2019 @jernej - bisecting done - this is the breaking commit/patch: https://patchwork.freedesktop.org/patch/311102/ - looks like they simply broke the video= cmdline parser ... as i understand https://patchwork.freedesktop.org/series/57525/ it is not accepted yet and looks like others had problems with it as well ... @PiotrO - might be interesting for you as well Link to comment Share on other sites More sharing options...
jernej Posted August 17, 2019 Share Posted August 17, 2019 @hexdump that should be non-breaking commit, so if you're not up to fixing it, please reply to that patch with bug report (you can download mbox version of that patch, import it in your e-mail client and then simply reply to it). Anyway, I would still like to fix original issue, but I really don't know where to start... Link to comment Share on other sites More sharing options...
hexdump Posted August 17, 2019 Share Posted August 17, 2019 @jernej - i did not find any way to download the message as mbox file - neither on patchwork nor on the list archive - and i'm not sure if i would be able to import it into gmail ... could you maybe please reply, reference my bisect message from above and put me on cc? (i assume you are on dri-devel - right?) a lot of thanks in advance and best wishes - hexdump Link to comment Share on other sites More sharing options...
jernej Posted August 18, 2019 Share Posted August 18, 2019 @hexdump can you try this to see if it helps with eachlink HDMI connection detection? Replace "DRM_CONNECTOR_POLL_HPD" in https://elixir.bootlin.com/linux/latest/source/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c#L2118 with "DRM_CONNECTOR_POLL_CONNECT | DRM_CONNECTOR_POLL_DISCONNECT". It's just workaround, but if it helps, it will help with understanding of the problem. Link to comment Share on other sites More sharing options...
hexdump Posted August 18, 2019 Share Posted August 18, 2019 @jernej - will give it a try tomorrow and let you know then. thanks a lot for thinking about how to get this forward. Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - i just tried your change, but sadly it does not make any difference - still no hdmi signal Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 @hexdump please provide 8-bit values from following memory locations when display is connected: 0x06000104, 0x06003000, 0x06003004 and 0x06003007. You can use devmem for that, e.g. "devmem 0x06003000 8". Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - here they are: 0x06000104 - 0x0 0x06003000 - 0x16 0x06003004 - 0xF0 0x06003007 - 0x2 i used devmem2 addr b - i hope you can obtain what you are looking for from that as well - ? best wishes - hexdump Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - sorry - the above values were when booted 5.3-rc3 with your test from above - monitor connected, but no hdmi output. the below is 5.2.8 with working hdmi out: 0x06000104 - 0x0 0x06003000 - 0x2E 0x06003004 - 0xF1 0x06003007 - 0x2 best wishes - hexdump Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 @hexdump sorry, I wasn't clear enough, I'm interested in values from non-working setup (5.2.8 without workaround), but those values will be also useful for comparison. Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - the result for 5.2.8 without your suggested change and without video=HDMI-A-1:e are the same as for the 5.3-rc3 version above (i.e. non working hdmi) - the 5.2.8 results from above are without your suggested change and with video=HDMI-A-1:e (i.e. working hdmi) Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 Thanks, now I have to compare with my working system. Currently all values seem fine, but I have to do more detailed analysis. Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 @hexdump this is a shot in the dark, but it might change something anyway. Please try this patch: http://ix.io/1SLn Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 @hexdump scratch that, this should be be more appropriate (I think): http://ix.io/1SLq Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - no change - still no signal on hdmi - tested with 5.3-rc3 (which works fine on non eachlink) with your second patch and no video=HDMI-A-1:e (just to be safe - it would be ignored anyway due to the video= parsing bug i guess) Link to comment Share on other sites More sharing options...
jernej Posted August 19, 2019 Share Posted August 19, 2019 meh, I really don't know what to do at this point without actually having HW here. There are several known issues with H6 HDMI driver and they are all strange. Maybe if I manage to fix at least one I'll be able to deduce what could be the reason for others. I have one possible solution in mind for DRAM driver though. I'll make test patch a bit later. Link to comment Share on other sites More sharing options...
hexdump Posted August 19, 2019 Share Posted August 19, 2019 @jernej - could it maybe be that the eachlink simply has a hardware design failure, so that some hdmi detection signal is not properly connected and maybe even android uses the forced hdmi connect? ... with this eachlink box nothing would surprise me: the dram seems to be strange, the serial connector is there, but one has to solder in additional stuff to actually make it work etc. looking forward to your dram ideas - do you think there is a chance to really get 3gb out of this box? or only to make it booting safe without mem=2048M? do you also have some idea what might be the problem with emmc on those h6 tv boxes? best wishes - hexdump Link to comment Share on other sites More sharing options...
jernej Posted August 20, 2019 Share Posted August 20, 2019 7 hours ago, hexdump said: could it maybe be that the eachlink simply has a hardware design failure, so that some hdmi detection signal is not properly connected and maybe even android uses the forced hdmi connect? Maybe, but I doubt. Still, you can check if pin 19 on HDMI connector is wired (HPD pin, source https://en.wikipedia.org/wiki/HDMI). From what I see, BSP also uses this for monitor detection. Anyway, there might be some HW thing, since only STBs show some type of HDMI issues (monitor detection on Eachlink, CEC issue on Tanix TX6). 7 hours ago, hexdump said: looking forward to your dram ideas - do you think there is a chance to really get 3gb out of this box? or only to make it booting safe without mem=2048M? Given that experiments confirmed my theory, I think only 2 GiB can be used. I'm trying to achieve correct detection - 2 GiB without workaround. We'll see. 7 hours ago, hexdump said: do you also have some idea what might be the problem with emmc on those h6 tv boxes? No, not really. eMMC is not so important to me, so I don't think I'll investigate this anytime soon. BTW, initial Tanix TX6 DT was merged to mainline yesterday (without ethernet for now). Link to comment Share on other sites More sharing options...
jernej Posted August 20, 2019 Share Posted August 20, 2019 @hexdump can you test this patch for DRAM issue? http://ix.io/1SRE Please remove all workarounds before test. Link to comment Share on other sites More sharing options...
hexdump Posted August 20, 2019 Share Posted August 20, 2019 @jernej - nice to have the tx6 dtb in tree now i'll have a look at your patch during the next days - you mean i should apply it to plain u-boot mainline? Link to comment Share on other sites More sharing options...
jernej Posted August 20, 2019 Share Posted August 20, 2019 3 minutes ago, hexdump said: you mean i should apply it to plain u-boot mainline? yes Link to comment Share on other sites More sharing options...
hexdump Posted August 22, 2019 Share Posted August 22, 2019 looks like the emmc problems are more generic for the h6 and not only related to tv boxes: Link to comment Share on other sites More sharing options...
jernej Posted August 22, 2019 Share Posted August 22, 2019 @hexdump Can you test this patch for eMMC? https://pastebin.com/XeUbuWGG (kindly provided by Alexis3D from #linux-sunxi). Link to comment Share on other sites More sharing options...
hexdump Posted August 22, 2019 Share Posted August 22, 2019 @jernej - yes, i'll test it - i should have more time the next days - on my list: half dq u-boot test, uboot 2gb detection on eachlink, the above emmc patch and responding to the driv-devel list regarding the video= cmdline parser regression ... it is funny, i ran across this emmc fix for the h5 today as well and had in mind to test if it might apply to the h6 too Link to comment Share on other sites More sharing options...
hexdump Posted August 23, 2019 Share Posted August 23, 2019 @jernej - good news: this patch seems to fix the emmc problem - on the eachlink i can now read and write the emmc in a stable way and on the qplus - where it was hanging on boot with emmc enabled in dtb - it does not hang anymore on boot and i can also read and write the emmc 1 Link to comment Share on other sites More sharing options...
hexdump Posted August 23, 2019 Share Posted August 23, 2019 @jernej - i just tested your eachlink dram issue patch from above, but sadly it still reports 3gb with it ... i used current mainline u-boot, plus your h6 plus half-dq patch and the eachlink h6 mini config and dtb from your u-boot tree half-dq test on tx6 mini will follow most probably tomorrow ... update: half-dq patch tested now as well successfully and reported to the u-boot list ... Link to comment Share on other sites More sharing options...
jernej Posted August 23, 2019 Share Posted August 23, 2019 @hexdump Thanks for all testing. I guess Andre will have to figure out Eachlink DRAM bug... Link to comment Share on other sites More sharing options...
Recommended Posts