<naoki>
for ROCK 5B, usb_host2_xhci is already used before 7ebc77fede4bb8f1c0b1239c7093d1fa69ccc5a9, and it didn't work
franoosh has joined #linux-rockchip
Stat_headcrabed has joined #linux-rockchip
Stat_headcrabed has quit [Client Quit]
<naoki>
my final solution: &usb_host2_xhci { status = "disabled"; };
franoosh has quit [Ping timeout: 268 seconds]
Stat_headcrabed has joined #linux-rockchip
franoosh has joined #linux-rockchip
franoosh has quit [Ping timeout: 264 seconds]
Stat_headcrabed has quit [Quit: Stat_headcrabed]
Stat_headcrabed has joined #linux-rockchip
Stat_headcrabed has quit [Quit: Stat_headcrabed]
warpme has joined #linux-rockchip
franoosh has joined #linux-rockchip
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
franoosh has quit [Changing host]
franoosh has joined #linux-rockchip
franoosh has quit [Remote host closed the connection]
warpme has joined #linux-rockchip
stikonas_ has joined #linux-rockchip
stikonas_ has quit [Quit: Konversation terminated!]
raster has joined #linux-rockchip
<CounterPillow>
Kwiboo: does PCIe3 in mainline u-boot on rock5b work? I get pcie_dw_rockchip pcie@fe170000: PCIe-4 Link Fail when I try "pci enum"
<CounterPillow>
"clk dump" and the lack of pmic stuff seems like a no :)
<CounterPillow>
okay, enabling rockchip spi support, iodomain support and common clk framework did change up the error a little
<CounterPillow>
and now shows more clocks
Livio has joined #linux-rockchip
<CounterPillow>
Current problem is `pcie_dw_rockchip pcie@fe190000: Can't get reset: -2`, which isn't really the PCIe controller I'm using but also that reset is just regular rockchip CRU which should be part of ARCH_ROCKCHIP
<CounterPillow>
Alright, pcie@fe170000 (which is what I'm using) is also not getting its reset with -2
Livio has quit [Ping timeout: 260 seconds]
<Kwiboo>
CounterPillow: yeah, I think two pcie is enabled on rock5b and it will repetedly report link failed when running "pci enum"
<Kwiboo>
first time all ports is enumerated, second time is will still try to enumerate failed ports and stops without printing probed pci devices
<CounterPillow>
It seems my .config changes on top of defconfig made things worse as now not even the PCIe NIC works :x
<Kwiboo>
naoki: will re-run some tests on my boards later, first have few series of patches to push out
<Kwiboo>
CounterPillow: hehe, yeah, last time I tested everything was working with defconfig as-is, but "pci enum" (or possible pcie driver) is not playing nice
<Kwiboo>
maybe it will behave better if it still report successful probe even when link fails
<Kwiboo>
that way pci enum should print out all probed pci devices as expected
<CounterPillow>
ok, with defconfig + Rockchip SPI enabled (don't know if that does anything) the PCIe ethernet is back but the x4 PCIe3 is back to 'pcie_dw_rockchip pcie@fe170000: PCIe-4 Link Fail', so I guess link fail is the better state than not getting reset
<CounterPillow>
oh hey
<CounterPillow>
I'm silly
<CounterPillow>
my NVMe is detected :D
<Kwiboo>
qschulz: events could probably be used, to my knowledge only when MMC UHS is used u-boot would try to change io voltage, not sure many rockchip board use or have use of uhs mode in u-boot
<Kwiboo>
CounterPillow: yeah, nvme probably gets detected, probably the m2-e slot that report link failed
<Kwiboo>
and continued "pci enum" will continue to probe the failed pcie device and not show any successfully probed pci devices
<CounterPillow>
Yep, I don't know what I did to think the nvme didn't work. I think I enabled `lsblk` and it didn't show anything? At that point I might've already messed too much with the .config
<Kwiboo>
I think you may need to run both "pci enum" and "nvme scan" and then it should be fully detected
<CounterPillow>
Ah, and I guess the standard bootflow does that for me
<Kwiboo>
correct, once it has tested mmc devices it will move to nvme and pcie+nvme gets probed/scanned
<diederik>
CounterPillow: In case it helps, here's my 6.10-rc2 kernel config (based on Debian's): https://dpaste.com/47UVLKXXS
<naoki>
Kwiboo: thanks. but strange, I don't think both my rock 5a/5b are broken, and all my USB3 devices (GbE, SSD) are broken too...
<naoki>
my ROCK 5A rev. is X1.2. it should be same as V1.1 from our doc, but I'll ask our engineer
<Kwiboo>
I just tested my ROCK 5B v1.46 and all 4 usb ports also seem to work in u-boot, I do get a "cannot reset port 1!?" on both 5A and 5B but my usb memory is working
<naoki>
I can see that error and no USB SSD found...
<naoki>
I want to know how it works on other rk3588(s) boards
<Kwiboo>
I know that u-boot is fragile and some usb devices is not quick enough to respond/reset and that may result in u-boot not working
<Kwiboo>
*device not working in uboot
<naoki>
hmm
<naoki>
I tried to add retry count and wait time in usb_hub.c
<Kwiboo>
the dwc3 code in uboot is pretty far behind linux version, so there is some bad/strange behavior with some attached devices
<naoki>
but it doesn't work on Linux kernel...
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<diederik>
I have a hard time believing the patch was submitted without verification that it actually worked (and subsequently accepted without verification)
<Kwiboo>
tested in linux, and my usb3 memory (corsair voyager vega) work on all ports, but my usb3 ssd (samsung t5) does not work on upper usb3 port on my 5B
<naoki>
oh
<naoki>
usb usb6-port1: Cannot enable. Maybe the USB cable is bad?
<Kwiboo>
on my it say: usb usb2-port1: Cannot enable. Maybe the USB cable is bad?
<Kwiboo>
will test on my 5B v1.42 now, and on orangepi 5 after that
<naoki>
usb2 is ohci for me (vanilla defconfig, linux-next)
<naoki>
Kwiboo: thank you so much!
<naoki>
I don't think disabling usb_host2_xhci is final solution for everyone ;)
<naoki>
btw "rockchip,dp-lane-mux = <2 3>;" will do something on u-boot? (now I'm talking about OTG port, not HOST port)
<Kwiboo>
I got similar result on my 5B v1.42, memory stick work on both usb3 port in linux and uboot, but not my ssd only on lower usb3 port
<Kwiboo>
I think the usb phy driver should be in pretty good sync with linux mainline driver, but if there is issue in linux driver there is probably same issue in u-boot
<naoki>
hmm
<naoki>
btw there is no problem on ROCK 3A
<Kwiboo>
same result on my orangepi 5 rk3588s, in linux: usb usb2-port1: Cannot enable. Maybe the USB cable is bad?
<Kwiboo>
with the usb3 ssd
<Kwiboo>
but my usb3 corsair voyager vega memorystick
<Kwiboo>
works in both u-boot and linux
<Kwiboo>
diederik: yeah, looks like some devices works as they should and others does not
<diederik>
yeah looks like it. Your possible explanation for 30mins ago seems plausible