macromorgan has quit [Read error: Connection reset by peer]
macromorgan_ has quit []
macromorgan has joined #linux-sunxi
ftg has quit [Read error: Connection reset by peer]
pmpp has left #linux-sunxi [#linux-sunxi]
pmp-p has joined #linux-sunxi
jernej_ is now known as jernej
<gediz0x539>
hello. i'm working on a A64/AXP803 board and when i try to power the board from VBUS, i get DRAM: 0 MiB. it made me think that DRAM supply regulator may not be enabled by default when powered from VBUS.
<gediz0x539>
what's the best way to handle powering an A64/AXP803 board through USB VBUS? i'm planning to manually initialize regulators on "sunxi_board_init". but it sounds dangerously straightforward.
lurchi_ has joined #linux-sunxi
lurchi__ has quit [Ping timeout: 480 seconds]
apritzel has joined #linux-sunxi
<gediz0x539>
oh... all PMIC initializations are done in BL31. i thought they were set using axp_set_dcdcX or axp_set_aldoX.
<libv>
so rasengan is purging channels that mention libera
<libv>
so i made the freenode channel +m
cmeerw has quit [Ping timeout: 480 seconds]
<veremitz>
he didn't already (with the bot) ?
<veremitz>
all the ones I was in were
<veremitz>
oh.. he didn't *get* you :D
<libv>
we mention oftc
apritzel has quit [Ping timeout: 480 seconds]
wigyori has joined #linux-sunxi
<veremitz>
Yet. :)
montjoie has quit [Quit: leaving]
montjoie has joined #linux-sunxi
evadot has joined #linux-sunxi
diego71 has quit [Quit: leaving]
diego71 has joined #linux-sunxi
diego71 has quit [Quit: leaving]
diego71 has joined #linux-sunxi
bantu has joined #linux-sunxi
apritzel has joined #linux-sunxi
apritzel has quit [Ping timeout: 480 seconds]
apritzel has joined #linux-sunxi
wigyori has quit []
<warpme_>
jernej: apritzel: inspired by jernej t95 work i got rmii eth working on h616 tanix-tx6s. infact this box is now in functional parity with orangepi zero0 which seems to be reference for h616 current state of the things (working: usb, eth, hdmi, cec; not working: GL, audio, IR, video.dec). i'm really tempted to get gl on this box. maybe you have some hints how to get panfrost on h616?
<NekoMay>
Will this project be covering the Allwinner D1 as well?
* karlp
presumes that the peripherals are all going to be ~same, so.... yes?
Xalius has joined #linux-sunxi
<apritzel>
NekoMay: there were already discussions about the D1 in here
<apritzel>
megi: I saw that you carry some sun6i CPUX PLL patch for U-Boot in some of your trees, any chance you can send this mainline?
<apritzel>
megi: I tried to follow the discussion back from 2017, but am not sure how much of this is still relevant?
<NekoMay>
apritzel: I'm only looking for a simple answer, not another discussion.
<apritzel>
NekoMay: my answer was meant to say: "Yes, there were already *technical* discussions/questions/posts around the D1 in here"
<NekoMay>
apritzel: Ah, apologies for my misunderstanding, then.
<apritzel>
no worries, the term "discussions" is indeed somewhat ambiguous
<NekoMay>
I would have gotten the board Sipeed is hocking, but they want a bit too much for what you get, sadly
<apritzel>
NekoMay: yeah, at the moment AW is apparently trying to monetize the RISC-V hype (the D1 is weaker but more expensive than it's dual-A7 sibling)
<NekoMay>
I thought it was Sipeed making the cashgrab. I guess it still could be; I've been told the D1 is only $5 so everyone might be trying to grab their piece of pie
gediz0x539 has quit [Quit: Leaving]
pentabarf has joined #linux-sunxi
<megi>
apritzel: that patch is relevant only for my kernel
<apritzel>
megi: but does it hurt? The old discussion spoke of some kind of chicken/egg problem between U-Boot and kernel, so should we have it anyway?
<apritzel>
megi: and where does this "m should be 0" come from? I don't see that in the H3 manual, that only speaks of "p should be 0 unless <288 MHz"
gediz0x539 has joined #linux-sunxi
<gediz0x539>
even though DM_MMC=y on U-Boot, "#if !CONFIG_IS_ENABLED(DM_MMC)" is somehow evaluated to true and I get a "MMC: no card present" error
<gediz0x539>
any hint?
<apritzel>
gediz0x539: SPL is not using DM
<apritzel>
gediz0x539: any chance you are looking for CONFIG_MMC0_CD_PIN?
<apritzel>
gediz0x539: that would drive the card detection in the SPL, if you define this to "", it will skip any card detection
<gediz0x539>
oh I did not know CONFIG_MMC0_CD_PIN. it's set to PF6 which seems to be the default. I'm also using this pin for CD too
<gediz0x539>
I actually try to do not skip it
<gediz0x539>
enabling CONFIG_MMC_BROKEN_CD helps too, but I still couldn't make it work
<apritzel>
gediz0x539: is the polarity different?
<apritzel>
gediz0x539: if you get into U-Boot proper, you can verify the CD functionality with "gpio input pf6", with and without a card inserted
<apritzel>
I would try that first, before you are hunting ghosts
<gediz0x539>
apritzel: it's low active like most of the A64 boards there
<gediz0x539>
okay i'm trying it now
<gediz0x539>
thanks
<apritzel>
active low sounds indeed correct. I just wasn't sure the SPL can actually deal with active high
pentabarf has quit []
pentabarf has joined #linux-sunxi
<gediz0x539>
I see
ndufresne has joined #linux-sunxi
<apritzel>
gediz0x539: yeah, the non-DM code doesn't consider any polarity, so it wouldn't work in the SPL
<apritzel>
but active low should work
Asara has quit [Quit: leaving]
Asara has joined #linux-sunxi
Asara has quit []
chalesy has joined #linux-sunxi
chalesy has quit []
Asara has joined #linux-sunxi
Asara has quit [Quit: leaving]
Asara has joined #linux-sunxi
Asara has quit []
Asara has joined #linux-sunxi
narmstrong has joined #linux-sunxi
Asara has quit [Quit: leaving]
Asara has joined #linux-sunxi
dittid[m] has joined #linux-sunxi
ndufresne is now known as Guest7
Guest7 is now known as ndufresne
cmeerw has joined #linux-sunxi
plaes has quit [Quit: leaving]
plaes has joined #linux-sunxi
plaes has quit []
plaes has joined #linux-sunxi
tnovotny has quit []
vagrantc has joined #linux-sunxi
choozy has joined #linux-sunxi
<juri_>
warpme_: wait, the x96mate is not the best supported one of these things?
<megi>
apritzel: it probably will not hurt with mainline
<megi>
m = 0 comes from experiments
<megi>
and from it not being used on H3 in BSP
<megi>
(in their ARISC firmware)
ftg has joined #linux-sunxi
Turl_ has joined #linux-sunxi
Turl has quit [Ping timeout: 480 seconds]
<apritzel>
juri_: in the (proposed) mainline patches there is no video or HDMI or AC200 PHY right now
<apritzel>
juri_: that's somewhat independent from the actual H616 TV box model, as they all seem to be very similar
pentabarf has quit []
<apritzel>
megi: so IIUC then the mainline sunxi-ng nkmp code does not make any assumptions about certain factors, right? Is that different for you?
<apritzel>
megi: I mean U-Boot couldn't care less: we just set a fixed frequency (from Kconfig), once at boot time
<apritzel>
megi: the only downside of fixing m = 0 would be a granularity of 24 MHz, which honestly doesn't matter the slightest
<apritzel>
megi: so if it helps to use m = 0, we can take this patch as well
<juri_>
apritzel: ah, thanks. i figgured support was borked. i just needed a tv box, and once i had one with right-chip probably-wrong-board, i hit the website to buy the model people were acatually hacking on.
<megi>
apritzel: yes, it would be nice, if you took the patch too
<megi>
yes, to all the other questions too
<megi>
my kernel doesn't do reparenting to 24MHz OSC when updating factors, so it's sensitive to initial set of factors from U-Boot
<megi>
mainline isn't
<megi>
reason I don't have it enabled is because I can detect timer irregularities (using Samuel's testing tool) when changing CPU frequency using cpupower (only on H3, IIRC), when reparenting is enabled
Suniel has quit [Remote host closed the connection]
<apritzel>
megi: is that timer irregularities with the Freescale workaround (back-to-back timer reads)?
<apritzel>
juri_: H616 is quite early in the mainline support stage, but so far there are no real showstoppers
<apritzel>
juri_: IIUC HDMI works already "under lab conditions", and we know what to do for Ethernet
<apritzel>
juri_: but model wise there doesn't seem to be much difference, they all follow the same path of saving money ;-)
<megi>
too bad H616 saved money on USB3.0
<megi>
one thing I like a lot about H6
<apritzel>
exactly
<apritzel>
that's a real shame
<apritzel>
but you can still get decent H6 boxes
<apritzel>
the only downside is 3GB DRAM max compared to 4GB
<megi>
looks like Opi3 is goging to be out of production, xunlong even removed H6 section from the shop
<apritzel>
that's another general problem: once we have good mainline support, the hardware phases out :-(
<megi>
H3 still seems to be produced, but this sucks, yes
<apritzel>
megi: the chip maybe, but there are no good (new) boards anymore
<apritzel>
and the H3 is getting old, in many respects (32-bit only, max 2GB DRAM, old Mali ...)
<apritzel>
I am trying to get an H3 board with eMMC and ideally GBit Ethernet, to fill a gap in my test coverage, and there is not much
<apritzel>
but there are hundreds of H616 TV boxes on eBay ;-)
<megi>
interesting, I didn't know H3 supports gigabit ethernet, but it does :)
<apritzel>
well, via an external PHY, as already the A20 did
<vagrantc>
apparently the orange pi plus fix landed ... hope it gets backported too
<vagrantc>
or at least landed in amaintainer tree
<apritzel>
megi: but you don't need an external PHY when you are happy with 100Mbit, which is a compelling argument for board vendors
<apritzel>
vagrantc: is there a Fixes: tag in there?
<apritzel>
vagrantc: then GregKH will pick it up automatically
<megi>
Orange Pi Plus2e has gigabit + emmc
<vagrantc>
apritzel: yes
<vagrantc>
at least, in the submitted patch, i haven't actually looked at exactly what got applied
<megi>
maybe someone has it unused in a closet to donate to apritzel ? :)
<apritzel>
megi: yeah, I figured, but it's a bit on the pricey side for just a test box
<vagrantc>
i have three orange pi plus boards that are basically retired ... although they don't seem very stable anymore :/
<vagrantc>
but those are H3 and i *think* had gigabit ethernet
<apritzel>
vagrantc: the OPi Plus? Looks like 100Mbit, but that would be OK, since eMMC is more important (I think the H3 MMC controller is different from the H5 and A64 one, so makes a good test candidate)
<vagrantc>
i've barely ever used the eMMC ... mostly microSD and external disk
<apritzel>
vagrantc: but then again the OPi naming is fragile, just mix up one of the {Plus,2,PC,One,e} suffixes and you have a different board ;-)
<megi>
:D
<vagrantc>
apritzel: yeah, i guess this is the orange pi plus2 , though it uses the same device tree
<apritzel>
vagrantc: you mean Orange Pi Plus 2e? ;-)
<vagrantc>
no, just the regular plus2
<vagrantc>
:)
<apritzel>
vagrantc: ah, I see, but still eMMC + GBit, it seems
<megi>
that's one beefy H3 board, it even has sata
<apritzel>
megi: yeah, but it's this cheap USB->SATA bridge that tkaiser likes to rant about so much?