<apritzel>
IlikeTech: not exactly the table, but the actual partitions
ftg has quit [Read error: Connection reset by peer]
<apritzel>
most partitioning tools start the first partition at 1MB, when using MBR this leaves 1MB-8KB for firmware, which is enough (we need something like 700-800KB at the moment)
<apritzel>
when using GPT, the table overlaps (34 sectors), but then you can use the later offsets: 128KB or 256K. The latter gets a bit tight then
<IlikeTech>
okay interesting
<IlikeTech>
is genimage able to handle that?
<apritzel>
what is genimage? do you mean U-Boot's mkimage?
<apritzel>
IlikeTech: we always put the FIT image with U-Boot proper, TF-A, DTB and potentially crust right behind the SPL: this is u-boot-sunxi-with-spl.bin. You then "dd" this to either 8K or 128K, and be done
<IlikeTech>
ah okay, got it
<IlikeTech>
sorry, I was confusing Linux tools with buildroot tools
<apritzel>
parthiban: do you have the SD card working in Linux? I removed CLK_SET_RATE_NO_REPARENT from the clock driver, and that looked better (SD card detected, can mount and ls), but then hangs when apparently using the SD card in anger
apritzel has quit [Ping timeout: 480 seconds]
hipboi has joined #linux-sunxi
<MasterR3C0RD>
apritzel: He did say he had it working; I'm assuming you have vqmmc/vmmc supplies set up correctly?
ungeskriptet has quit [Ping timeout: 480 seconds]
ungeskriptet has joined #linux-sunxi
<parthiban>
MasterR3C0RD: Morning, regarding the patches in your v2, IMO it makes sense to keep the original set of peripherals and get merged. I will be able to do the next round from my end. This will include GPU Power Domain, GPU itself, PWM, EMAC and so. My main idea is to get the DE, TCON working. Hope that works?
<MasterR3C0RD>
parthiban: Sounds good to me; I had just been thinking if it was working already, it'd make sense to batch them together
<MasterR3C0RD>
Right now I'm trying to get the kernel running on my device; not having too much luck yet though. I'm assuming I have something slightly wrong with my DTS, but hopefully I'll figure it out
<parthiban>
MasterR3C0RD: My tree is not well organized. I will make a better version and push to git soon.
hexdump01 has joined #linux-sunxi
hexdump0815 has quit [Ping timeout: 480 seconds]
<parthiban>
apritzel: SD in u-boot works, am going to do that for the kernel now. I was figuring out SDIO for the WiFi on the other day. Not yet for SD in Kernel. Let me check now.
hipboi has quit [Quit: hipboi]
hipboi has joined #linux-sunxi
aggi has quit [Remote host closed the connection]
aggi has joined #linux-sunxi
<MasterR3C0RD>
Got a little further in getting kernel up on my device, but I think I might have some issues with my power regulators, as it shuts down pretty fast. Investigating further...
hipboi has quit []
JohnDoe_71Rus has joined #linux-sunxi
<parthiban>
I always have "pd_ignore_unused regulator_ignore_unused clk_ignore_unused" in my bootargs. Until everything is smooth, this helps.
<parthiban>
I also have "regulator-boot-on" and "regulator-always-on" in most cases.
<parthiban>
Atleast during the development part
<parthiban>
MasterR3C0RD: "sunxi-mmc 4020000.mmc: fatal err update clk timeout" I think something is wrong for sure with MMC.
<MasterR3C0RD>
Hmm, I know for sure that eMMC/mmc2 works
<parthiban>
yeah, this is the mmc0. I have SD connected in mmc0
<MasterR3C0RD>
IRQs seems to be correct
<parthiban>
I wounder how it works in u-boot
<parthiban>
Am loading the kernel, dtb from SD with ext4 F
<parthiban>
s/F/FS/
<MasterR3C0RD>
I suppose one thing to check is the clocks
<parthiban>
"clocks = <&ccu CLK_BUS_MMC0>, <&ccu CLK_MMC0>;" this one?
<MasterR3C0RD>
Yeah, more specifically the clock definitions
<MasterR3C0RD>
Do you have a vqmmc regulator set up?
<parthiban>
There is no VCC_PF pin. SD card is powered with 3.3 in my case and also the CMD line is pulled up with same 3.3v. So I used the same regulator for vmmc and vqmmc.
<MasterR3C0RD>
apritzel: Did you have any luck bringing up eMMC on your tablet? Don't quite remember if you mentioned it; SD isn't working for parthiban in Linux either
<MasterR3C0RD>
And eMMC isn't working for me on my board
apritzel has quit [Ping timeout: 480 seconds]
Schimsalabim has quit [Ping timeout: 480 seconds]
Schimsalabim has joined #linux-sunxi
apritzel has joined #linux-sunxi
Schimsalabim has quit [Ping timeout: 480 seconds]
<loki666>
apritzel: I reduced the DRAM_CLK by 48, but it's even less stable
Schimsalabim has joined #linux-sunxi
<loki666>
I also extracted the param with sunxi-fw, I got some differences from the uboot config, but applying them didn change much
<parthiban>
MasterR3C0RD: USB host works for USB1. I have the USB0 connected to a type C port. I tied to use it as ncm gadget, but success yet.
<parthiban>
s/success/no success/
<MasterR3C0RD>
parthiban: I did succeed in getting USB0 working on my device; in U-Boot as a gadget (though with some issues) and in Linux as a host
<parthiban>
"grep allwinner,sun50i-a100-usb-phy drivers/* -ri" no results in the u-boot. Which usbphy should be used?
<MasterR3C0RD>
Use a D1 compatible
<MasterR3C0RD>
For MUSB use A33 I believe
<parthiban>
"Bus usb@5100000: Port not available." I added the d1 for usbphy and a33 is already in the usb_otg
<MasterR3C0RD>
There's extra settings in U-Boot
<MasterR3C0RD>
Device Drivers -> USB devices -> enable MUSB host or gadget mode
<MasterR3C0RD>
Make sure not to enable gadget driver model
<MasterR3C0RD>
Then ARM Architecture -> usb0 vbus, id detect, .../usb1 vbus
<MasterR3C0RD>
You use the GPIO name for the pins that enable USB VBUS, detect USB vbus, and detect the ID of the USB
<parthiban>
:q
<parthiban>
I have CONFIG_USB_MUSB_HOST CONFIG_USB_MUSB_GADGET both enabled
<MasterR3C0RD>
You'll be stuck in host mode
<MasterR3C0RD>
Apparently there was an effort to make the driver compatible with DM, which would help get rid of some of these superfluous parameters
<MasterR3C0RD>
I'll also mention that I had no luck getting MUSB host mode working in U-Boot
antonyvb[m] has joined #linux-sunxi
hipboi has quit [Quit: hipboi]
apritzel has quit [Ping timeout: 480 seconds]
apritzel has joined #linux-sunxi
szemzoa has quit []
szemzoa has joined #linux-sunxi
<apritzel>
parthiban: aliases for MMC: old discussion, short: not gonna happen officially
<apritzel>
but just do like the grown ups do and use $fdtcontroladdr when booting the kernel in U-Boot, which gives you the aliases through the back door
<parthiban>
apritzel: ok, thanks
<apritzel>
and yes, those SD card clock issues are the same I see. They go away if you remove CLK_SET_RATE_NO_REPARENT from the clock driver
<apritzel>
but then it later hangs, at least on my tests
<apritzel>
eMMC seemed to work for me, though I didn't test it in anger. But I mounted a few partitions and copied a file, no issues
<apritzel>
I couldn't find any hints as to why this flag was there, we don't use this on any other SoC (for the MMC clocks)
<apritzel>
for now I have commented the AXP nodes in my DT, as I need to first figure out how the rails are assigned, without frying anything
<apritzel>
so I have a fixed regulator for vmmc, which is all we need. SD needs to work with explicit regulator programming, as the BROM needs it, and has no clue about any PMICs whatsoever
<apritzel>
s/to work with explicit/to work without explicit/
<apritzel>
regarding TF-A: I have some patches, adding to PLAT=sun50i_h6. Identification and PMIC (via I2C) work already, but no luck with SMP so far, though I fixed something up
<apritzel>
MasterR3C0RD: just a heads up: cut-off date for 6.13 is probably this Friday or so (depending on wens' availability), so if you want to send a v2: now would be the time
<apritzel>
and keep it easy, remove everything that does not work or is not ready. Or at least order the patches this way, so that the good patches come first
wingrime-ww has quit [Read error: Connection reset by peer]
wingrime-ww has joined #linux-sunxi
aperezdc has quit [Remote host closed the connection]
aperezdc has joined #linux-sunxi
aperezdc has quit [Remote host closed the connection]
aperezdc has joined #linux-sunxi
<parthiban>
I figured a way to fast forward the development using USB host mode. Attached a USB to Ethernet adapter and able to do NFS boot. This way I can only focus on display stuff.
szemzoa has quit []
szemzoa has joined #linux-sunxi
<parthiban>
MasterR3C0RD: apritzel: Otherwise, USB host works, UDC is probed fine but when I created a NCM gadget, it didn't showup in the host side. eMMC works in MMC2, but MMC0 with SD and MMC1 with SDIO WiFi card didn't work.
<parthiban>
Also my attempt to get the ethernet emac didn't go well. Failed with reset timeout.
bauen1 has joined #linux-sunxi
pg12_ has joined #linux-sunxi
pg12 has quit [Ping timeout: 480 seconds]
pg12_ is now known as pg12
szemzoa_ has joined #linux-sunxi
szemzoa has quit [Read error: Connection reset by peer]
<apritzel>
ah, MMC1 also being broken is a good hint. MMC0 and MMC1 are the same, but MMC2 is different, so this might be a clue to follow
<apritzel>
parthiban: can you try to remove CLK_SET_RATE_NO_REPARENT from the ccu-sun50i-a100.c clock driver and check again?
<apritzel>
for Ethernet: something to look out for is the power supply of the PHY
maz has joined #linux-sunxi
szemzoa has joined #linux-sunxi
szemzoa_ has quit [Read error: Connection reset by peer]
szemzoa_ has joined #linux-sunxi
szemzoa has quit [Read error: Connection reset by peer]
szemzoa has joined #linux-sunxi
szemzoa_ has quit [Read error: Connection reset by peer]
szemzoa_ has joined #linux-sunxi
szemzoa has quit [Read error: Connection reset by peer]
warpme has joined #linux-sunxi
szemzoa_ has quit [Read error: Connection reset by peer]
szemzoa has joined #linux-sunxi
<parthiban>
apritzel: phy-supply node is added with the regulator. And the regultor is always-on with "regulator_ignore_unused" in the kernel command line. IMO am missing something else as well.
<apritzel>
I mean there must be actual physical power for the PHY, regardless of any trickery you play in the DT to appease Linux
szemzoa has quit [Read error: Connection reset by peer]
szemzoa has joined #linux-sunxi
colinsane has quit [Ping timeout: 480 seconds]
hazardchem has quit [Read error: Connection reset by peer]
hazardchem has joined #linux-sunxi
colinsane has joined #linux-sunxi
szemzoa_ has joined #linux-sunxi
szemzoa has quit [Read error: Connection reset by peer]
szemzoa has joined #linux-sunxi
szemzoa_ has quit [Read error: Connection reset by peer]
<parthiban>
MasterR3C0RD: https://pastebin.com/raw/Jh6MmHtq did you test the devfreq? I tried enabling it and it seems it's crashing
<parthiban>
apritzel: I will try to tap the PHY power pins in the dev board. Good point.
szemzoa has quit []
<MasterR3C0RD>
parthiban: As I mentioned, I'm pretty sure DVFS won't work until we have proper ATF
<parthiban>
sorry, I missed
szemzoa has joined #linux-sunxi
<apritzel>
MasterR3C0RD: I am not so sure, at least on sunxi DVFS works independently from firmware
<apritzel>
if you wanna play safe, remove the PSCI node and all secondary CPUs from the DT
<MasterR3C0RD>
apritzel: Hmm, interesting. The error number's pretty weird (EBUSY)
<apritzel>
DVFS just needs a control over the CPU voltage (via the AXP), control over the frequency (to the CPU PLL via the clock driver), and some thermal sensor, to detect overheating
szemzoa has quit [Read error: Connection reset by peer]
szemzoa has joined #linux-sunxi
<apritzel>
that dump looks like a mishap in the DT
<apritzel>
I think I had this during the H616 dvfs development
<apritzel>
the call stack should give you enough breadcrumbs to follow
<MasterR3C0RD>
But if we can't confirm DVFS is working, I don't think it makes sense to submit a V2 for this cycle. That's the only actual code change; the rest are DTS changes that rely on existing code and could be done out of tree on a standard AArch64 kernel
<apritzel>
if in doubt, try to copy the approach used in sun50i-h616-cpu-opp.dtsi
<apritzel>
MasterR3C0RD: we still need the DT changes reviewed and merged in the kernel tree
<apritzel>
because this is where now U-Boot gets its DTs from
JohnDoe_71Rus has joined #linux-sunxi
<MasterR3C0RD>
In terms of the DTSI, I don't see many differences between the OPP DTSIs, aside from different speed bins/hardware settings
<apritzel>
you don't have opp-supported-hw, I dimly remember there was something nasty about this, but forgot the details
<apritzel>
I think not having this in the DT, and having the driver code as it is triggers exactly that error
colinsane has quit [Ping timeout: 480 seconds]
<MasterR3C0RD>
I won't be able to test anything until tonight, but I'll see if that's the culprit then
<apritzel>
maybe commit fa5aec9561cfc4f4 gives you some clues. We should autodetect this case (no opp-supported-hw property), but maybe that's broken?
szemzoa_ has joined #linux-sunxi
szemzoa has quit [Read error: Connection reset by peer]