11:38 <warpme> junari: after cleaning our work - this seems to be minimal set of changes req. for proper dram sizing on x96-q h313 box: https://github.com/warpme/minimyth2/blob/011461791aa5f74f553a6f0479ad9195023642b5/script/bootloaders/u-boot-h313/files/03-fix-x96-q-memory-sizing.patch

12:29 <warpme> junari, apritzel: you were critical souls to get this https://controlc.com/5690a8f1 :-)

12:41 <warpme> apritzel: i will need a bit of your help with getting h616 emac1 rmii (int. phy) working on current uboot. It works for me on custodians tree from 11.05.2021 - but on current custodians master (or mainline) i can't get it working. On 11.05.2021 i'm using https://github.com/warpme/minimyth2/blob/master/script/bootloaders/u-boot-h616/files/add-emac2-to-h6-h616.patch to get it working on tx6s with int phy. I updated this patch for

12:41 <warpme> current custodians master - but probably i done this wrongly.....

12:47 <warpme> btw: it looks like my h313 indeed needs update of cpu cluster control in ATF. Current 2.8.1-lts ATF gives me: [ 0.102251] smp: Bringing up secondary CPUs ...

12:47 <warpme> [ 5.238195] CPU1: failed to come online

12:47 <warpme> [ 5.242366] CPU1: failed in unknown state : 0x0

12:47 <warpme> [ 10.367505] CPU2: failed to come online

12:47 <warpme> [ 10.371657] CPU2: failed in unknown state : 0x0

12:47 <warpme> [ 15.496018] CPU3: failed to come online

12:47 <warpme> [ 15.505167] smp: Brought up 1 node, 1 CPU

12:47 <warpme> [ 15.500189] CPU3: failed in unknown state : 0x0

12:47 <apritzel> yeah, saw this, and that's expected in your case

12:48 <apritzel> warpme: so is it that TV box here: https://www.ebay.co.uk/itm/353814299552?chn=ps&mkevt=1&mkcid=28&var=623183944433

12:48 <apritzel> with those tapered corners and that "X" style casing?

12:48 <warpme> indeed mine is exactly like this one

12:49 <apritzel> that seems to be quite wide spread, at least I saw many eBay offers in the last months

12:50 <warpme> oh - i saw it for 17.8Eur. Mine cost me 20.5Eur

12:56 <warpme> i suspect such low price might be because lack of linux capability on it. This may make lower demand so supply exceedings may make price going down. Maybe....

13:06 <warpme> Funny thing is that company who sold me this box sold me 2G/16G variant. When it arrived - i spent 15min walking on vendor android trying to get info about installed RAM. No way. Vendor android is modified in a way that user can't get this info. I attached uart and discovered box has 1G instead of marketed 2G. So i fill goods complaint. Company promised me to send another box - this time with marketed (and paid by me) 2G RAM. Will see

13:06 <warpme> what will happen :-))))))

13:10 <apritzel> warpme: yeah, they tend to lie heavily in Android, had a very similar experience lately: the 4G/32G H313 box turned out to be using a 1G/8G H3(!) board

13:10 <warpme> geez!

13:12 <warpme> but i must say reaction on goods complaint submission was v.quick, full of apologies and satisfactory (assuming they will realise what they promised)

13:18 <apritzel> I think you can find out the real memory in /proc/meminfo somewhere, need to look up which number I used on the complaint

14:29 <junari> warpme: try https://github.com/iuncuim/arm-trusted-firmware/commits/t507_imp

14:29 <junari> make PLAT=sun50i_t507 DEBUG=1 bl31

14:54 <warpme> junari: so now i have: 4x cpu cores, emmc, sdcard, hdmi, gpu, audio, bt, wifi, ir working (https://controlc.com/5f893d91). To have fully working media player: Eth is missing. Many thx!

14:58 <junari> But I don't implement core power off yet. Only switch on

15:39 <apritzel> warpme: Ethernet will not easily work (with mainline), you need support for the embedded PHY in the H616 chip, which is early WIP

15:43 <warpme> apritzel : i have already working int.phy on h616. Issue is that uboot update from 11.05.2021->05.03.2023 breaks h313/h616 Eth support in kernel (for me)

15:44 <warpme> i would say: uboot update from 11.05.2021->05.03.2023 leaves hw in different state and resulting with no h313/h616 int.phy Eth working in kernel (but ext.phy i.e. in Zero2 works ok)

15:45 <apritzel> is that 11.05.2021 a mainline branch? Or one of jernej's hacks to beat the AC300 into submission?

15:45 <warpme> so maybe getting int.eth working just needs of updating kernel patch related to int.eth?

15:48 <warpme> saying "11.05.2021" i'm refering to sunxi-custodians uboot. This uboot + kernel patch gives me working eth on h616 tx6s (this box has int.eth phy)

15:50 <warpme> the same kernel patch but with new uboot (current master of sunxi-custodians) has not anymore working eth

15:51 <apritzel> what "11.05.2021" branch is this, exactly? Just checked out at this date?

15:52 <apritzel> there is no mainline kernel support for the second MAC or the internal PHY

15:53 <apritzel> there are some patches floating around, but there are not in a great shape and will need to change before they can get merged

16:18 <jernej> warpme: most likely reason for your issue is difference in PWM peripheral. It seems H616/H313 need new or at least amended PWM driver.

16:19 <jernej> the reason it might worked before but not later is that Linux relied on having PHY already enabled since it is not capable doing it on it's own

16:20 <jernej> and later version may just disable unused clocks, resets, etc.

17:12 <apritzel> jernej: I loosely looked into the PWM issue. The PWM device is quite different to the previous sun4i one, it's close to the R40 (and some other SoC).

17:12 <apritzel> there was a driver on the list years ago, but this had problems, and would need some serious fixes

17:14 <apritzel> but at least we could cover all those new PWM devices with one driver. From a first glance I don't think adding support into the existing pwm-sun4i driver is worthwhile

17:15 <jernej> I think new driver is better. But unsure if it's better to resurrect old attempts or do clean one

19:33 <jernej> apritzel: I went trough minimal changes that warpme made to dram code and I noticed that they all depend on tpr2 parameter in BSP code

19:33 <jernej> so I would like to add that too when I respin U-Boot DRAM patch series, if you don't mind

21:04 <apritzel> jernej: happy to take any patches for the H616 DRAM, however you like it

21:04 <apritzel> IIRC you said the last time that you learned some new parameter names and wanted to update the series anyway?

21:08 <jernej> yeah, I think I already pushed that on github

21:17 <jernej> it looks like tpr2 has only one useful bit, bit8, and it changes read and write latency

21:18 <jernej> it's used only for (LP)DDR3

21:37 <jernej> actually, there is more to it. just found another spot where tpr2 is used