<smaeul>
wart_: yes, more or less anything that triggers an interrupt can wake up from crust. and there is a wire from the wifi chip to the A64 specifically intended for wake on WLAN
<wart_>
smaeul: hi again. where was the patch applied? i'm on pmos and zcat /proc/config.gz | grep WOWLAN doesn't show anything.
<wart_>
Linux pine64-pinephone 5.14.6 #3-postmarketos-allwinner SMP PREEMPT Tue Oct 19 12:25:49 UTC 20 aarch64 GNU/Linux
<apritzel>
Hi, does anyone know what U-Boot's "A charger is plugged into the OTG" message is really supposed to mean? I mean isn't external VBUS power expected when using gadget mode?
<apritzel>
MoeIcenowy, wens, smaeul ^^^^
<mripard>
apritzel: it's to prevent the other situation
<mripard>
where you would act as host, but there's already a charger sending 5V on VBUS
<apritzel>
I see, thanks
<apritzel>
I keep getting this message now repeatedly, after smaeul's latest patch ...
<apritzel>
when doing ums
<mripard>
52defe8f6570 is the commit that introduced it
<apritzel>
yeah, I found this yesterday after some digging as well, but couldn't really make sense of it
<apritzel>
as this gets called when gadget mode is initialised
<apritzel>
so "charger" is somewhat misleading, it just means: there is external VBUS power on the port?
<apritzel>
and this is no good for host mode?
<mripard>
yeah
<apritzel>
mripard: many thanks, so I need to dig deeper why this is printed in gadget mode context
<apritzel>
do we use MUSB host mode on the older SoCs like A20?
<mripard>
I think we did for things like tablets
<mripard>
especially in U-Boot where you don't have touchscreen support or a virtual keyboard
<apritzel>
right, makes sense
cnxsoft has quit []
<hlauer>
gamiee: same as bananaPro: Ampak AP6181. It's not working so far with vanilla, the wlan0 interface is recognized but doesn't show any wifi extensions
<gamiee>
@hlauer: hmm, not sure how it is on M1+, but I guess you need to add firmware for it and it should work
<hlauer>
@gamiee: firmware is set up and: brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac43362-sdio for chip BCM43362/1. Same message as on BananaPro
wart_ has quit [Quit: WeeChat 3.3]
<gamiee>
so no errors?
<hlauer>
only "no clm_blob available", which I have on the banaPro, too. Last message is: brcmf_c_preinit_dcmds: Firmware: BCM43362/1 wl0: Dec 5 2019 07:26:04 version 5.90.254 FWID 01-0
<hlauer>
Same as on the bananaPro, where wifi works
sunshavi has quit [Remote host closed the connection]
<apritzel>
mripard: I think I found the reason: this code was copied, from musb-new/sunxi.c to board.c, but the context is different there
<apritzel>
in the musb code it's about MUSB host mode, but in board.c it's in g_dnl_board_usb_cable_connected(), so explicitly for gadget mode
<apritzel>
where the message doesn't make much sense, I believe
<apritzel>
(because we *want* external VUSB in this case)
machinehum has joined #linux-sunxi
<machinehum>
D1 looks interesting
<machinehum>
Any non sketchy places to buy a board?
<apritzel>
interesting as in: "broken and wimpy as you would expect from Allwinner"?
<machinehum>
uhh
<machinehum>
I mean I just read the wiki
<machinehum>
It doesn't work?
wart_ has joined #linux-sunxi
<apritzel>
for some definition of "work" it does, but that's about it
<jernej>
machinehum: it has non-standard MMU config registers, in contradiction to RISC-V standard, so debate is still ongoing, how to implement that, if at all
<jernej>
I'm not sure whose fault is that, it might be that alibaba subsidiary (its their core) implemented it that way and AW just copy design
<apritzel>
jernej: I think AW took the Alibaba core "as is"
<apritzel>
so in fairness it's more Alibaba to blame here
<jernej>
machinehum: If you don't mind using patches that will never land upstream, you can make it work
<jernej>
pretty well, I presume
<apritzel>
machinehum: ... given that you are happy with a single(!) core at 1GHz, and the usual underwhelming I/O devices from Allwinner ;-)
<jernej>
yeah, I was a bit surprised with that single core design
<jernej>
but peripherals are more or less standard
<jernej>
apriztel: why peripheral disappoints you most?
<jernej>
*which
<apritzel>
jernej: well, it's just the normal, old and slow stuff: MMC, USB 2.0, single channel 32-bit DRAM, you name it
<apritzel>
I mean Allwinner was already at PCIe and USB 3.0, at least
<apritzel>
but they seem to have lost the floppy with the RTL ...
<jernej>
ah, that
<jernej>
I suspect they didn't go with speedier interfaces due to slow, single core
<jernej>
there is also no GPU
<jakllsch>
apritzel: heh, allwinner pcie
<jernej>
I'll take it as a proof of concept SoC
<apritzel>
I get the impression that AW is focusing on very distinct use cases, and trying to build as cheap and tailored as possible
<apritzel>
like all those newer Cortex-A7 SoCs
<jernej>
btw, I saw mentions of pcie in some dts files in BSP
<apritzel>
jakllsch: well, at least they tried, and I was hoping for the next SoC to do it properly (I know ...)
<jernej>
does somebody remember where to look for that "paging" register?
<jernej>
I want to check if the issue is still there
<apritzel>
jernej: MoeIcenowy probably knows, or check her PCIe hypervisor repo
<apritzel>
jernej: for which SoC? T723? A200?
Mangy_Dog has joined #linux-sunxi
<jernej>
apritzel: sun50iw10
<jernej>
whatever that is
apritzel has quit [Ping timeout: 480 seconds]
Mangy_Dog has quit [Remote host closed the connection]
JohnDoe_71Rus has quit []
apritzel has joined #linux-sunxi
<apritzel>
jernej: ah, that one ;-)
<valdikss>
To be honest, I think they stick to older cores mainly because they don't want to switch to a newer tech process, and the newer cores are all adapted to 12- nm
<apritzel>
valdikss: ah, btw: thanks for the USB OTG heads up in U-Boot, there are indeed a bug or two in there
<apritzel>
I think I have a fix for that charger message coming up repeatedly, and also for not having CONFIG_USB_ETHER defined
<valdikss>
apritzel: I'm glad I've helped to find it :D
<apritzel>
turns out that enabling USB_ETHER triggers the USB gadget initialisation code as a side effect