<Armbian-Discord> <r​pardini> Well if that solves it -- I definitely can't come up with a better one.
<Armbian-Discord> <r​pardini> A damn game of chase due to upstream's failure to address the mess that is meson_gx_mmc
<Armbian-Discord> <r​pardini> I personally would bet on Vyacheslav's patch (endorsed by Neil... right), and patch amlogic,mmc-phase directly in the DT for the boards we know need it (AXG's, bpim5, others?)
<Armbian-Discord> <r​pardini> We then take a 2nd try to submit that patch upstream after a small manifesto 😉
<Armbian-Discord> <B​ogdan> Patch is ok, but idk how to sort board phase clk 🙂 It depends on emmc chip, that used in board. But sometimes in different versions of same board used different emmc chips, that can work only on 180, or only on 270 phase clk
<Armbian-Discord> <r​pardini> well. I'd say as long as we standardise (vs upstream, which proposes no solution at all) on amlogic,mmc-phase... we're free to patch it / overlay it as we see fit. I confess the reason some require 270 and others 180 is over my head, and why is meson64 the one to suffer from this the most
<Armbian-Discord> <r​pardini> meanwhile, in Radxa-land: https://github.com/radxa-repo/bsp/blob/main/linux/latest/0020-radxa-zero/0002-Revert-mmc-meson-gx-add-SDIO-interrupt-support.patch ("SDIO interrupt support" -- that they allow in...)
<Armbian-Discord> <B​ogdan> Using overlay will require action from user, so if it can be patched without it - it would be great.
<Armbian-Discord> <r​pardini> Will do. BpiM5, any others?
<Armbian-Discord> <B​ogdan> Idk others, that supported by armbian. I have some GT-Kings (supported by mainline) with same problem. Need to check BpiM5 with 270 phase first I think.
<Armbian-Discord> <r​pardini> hmm, gt-kings are g12B aren't they. I thought g12b were a bit less affected by this 😐
<Armbian-Discord> <B​ogdan> In CoreELEC kernel i saw some code with phases selection on fly, maybe it will be useful
<Armbian-Discord> <B​ogdan> However they use vendor kernel
<Armbian-Discord> <r​pardini> yeah, we've mainline only. (I'd have added gt-king's to Armbian, had anyone sent me a few... heh)
<Armbian-Discord> <B​ogdan> Sure, I contribute some stuff to armbian meson64, so I know how it works. I mean that maybe Neil can help with some information about 180/270 specifics
<Armbian-Discord> <r​pardini> We've talked about it, let me find
<Armbian-Discord> <B​ogdan> I found some talks in November
<Armbian-Discord> <r​pardini> Hmm. Things were done in DT instead of using a new compatible, but the gist is the same: https://discord.com/channels/854735915313659944/858684386806595584/1039157121066995783
<Armbian-Discord> <B​ogdan> I think we need to patch only supported affected devices for now, because it mainline problem, not armbian
<Armbian-Discord> <B​ogdan> Maybe later someone will found proper solution how to do this right
<Armbian-Discord> <r​pardini> LibreELEC is just patching 270 (like we did before) for everything...
<Armbian-Discord> <B​ogdan> Chewitt use this trick long time ago 🙂
<Armbian-Discord> <r​pardini> yep -- and except for the N2+'s UHS mode, worked for everything, until the Radxa Zero
<Armbian-Discord> <r​pardini> so before was: 1 patch, 270 everything, N2+ is slow, Radxa Zero broken.
<Armbian-Discord> <r​pardini> and now is: 1 patch, 1 DT hack for AXG, 1 DT hack for BpiM5
<Armbian-Discord> <B​ogdan> Looks almost the same😄
<Armbian-Discord> <r​pardini> yes. only my dignity is completely gone 😄
<Armbian-Discord> <B​ogdan> If I will have some ideas how to fix it - I will write and create PR 🙂
<Armbian-Discord> <r​pardini> Maybe if address one or two of the issues raised by https://lore.kernel.org/linux-amlogic/1jk03y37vs.fsf@starbuckisacylon.baylibre.com/t/#u
<Armbian-Discord> <r​pardini> it's cosmetic, but kind of thing will get us further
<Armbian-Discord> <B​ogdan> As Jerome says, he tried to do calculating values on fly, but it unstable
<Armbian-Discord> <r​pardini> Yes, the last of Brunet's replies is more sane.
<Armbian-Discord> <B​ogdan> When he develop driver in 2017, I remember that tries in git history
<Armbian-Discord> <B​ogdan> AML engeneers do shitty controller, use a lot of tricks in vendor driver, and advise not-working solutions to mainline linux developers 😄
<Armbian-Discord> <r​pardini> Anyway. Any rework there will take a long time to land, and we've broken boards now -- I still bet on amlogic,mmc-phase and small DT hacks as less-terrible-option
<Armbian-Discord> <B​ogdan> Yes, I bet on this too. Any rework will get a lot of time, and possibly will do regressions
<Armbian-Discord> <r​pardini> Ok I'm patching both sd & eMMC; https://github.com/rpardini/armbian-build/commit/9336adcf08dd132220ef389122010772ffbb5d93 -- if OK will PR
<Armbian-Discord> <r​pardini> - PR: https://github.com/armbian/build/pull/4674
<Armbian-Discord> <r​pardini> damn, now I realized you're @chbgdn 😄
<Armbian-Discord> <B​ogdan> Yeah, i got mail about PR. Need to tag guy, who report this problem on BPiM5 for testing image 🙂
<Armbian-Discord> <B​ogdan> @Maner_ can you test image from @rpardini ?
<Armbian-Discord> <r​pardini> I also sent to a person on JIRA. Let's wait. Igor might also have to test
<Armbian-Discord> <M​aner_> I dont have access to my bananapi until the end of january
ebach has joined #armbian-amlogic
<Armbian-Discord> <r​pardini> Meanwhile @IgorPec worked his magic
<Armbian-Discord> <I​gorPec> need approval https://github.com/armbian/build/pull/4675
ebach has quit [Quit: ebach]
<Armbian-Discord> <r​pardini> building before approving 😉 1 minute
<Armbian-Discord> <r​pardini> done
<Armbian-Discord> <r​pardini> approved
<Armbian-Discord> <r​pardini> merged
ebach has joined #armbian-amlogic
<Armbian-Discord> <s​erkanp> hi all, maybe it's a duplicate, but i need help on s905x3 soc chip based armbian linux, main thread was here https://forum.armbian.com/topic/22293-analog-audio-jack-is-not-working/#comment-157201
<Armbian-Discord> <s​erkanp> so i dont want to waste your time.. i cant get analog audio output from this device which stock image came with device (armbian debian) only supports hdmi audio
ebach has quit [Ping timeout: 246 seconds]