Am Mittwoch, 25. September 2019, 13:43:44 CEST schrieb Fabian Vogt:
Am Mittwoch, 25. September 2019, 13:16:37 CEST schrieb Matthias Brugger:
On 25/09/2019 11:57, Fabian Vogt wrote:
Am Dienstag, 24. September 2019, 16:49:42 CEST schrieb Matthias Brugger:
> On 24/09/2019 13:24, Linux Kamarada wrote:
>> Hi, everyone!
>> AFAIK openSUSE does not support Raspberry Pi 4 Model B yet. 
>> I’ve bought that board, I know the very basics on how to use it 
>> and I want to help to port openSUSE to it. How can I start?
> I'm working on some (still hacky) support for RPi4.
> You can find the JeOS image in my home project .
> Beware that I was able to break the build of U-Boot yesterday. I'll try to fix
> this tonight.
I just gave that image a try and after a while staring at a console with no
activity I noticed that it's the JeOS with jeos-firstboot, so I had to remove
"console=tty" to actually see anything. Using the HDMI console won't
work without USB support.
Why? HDMI output should work.
Yes, but without any ability for interaction not really useful.
chrony-wait should be disabled, it delays booting by 10min as no
network is currently working.
Yes, networking is on my list.
What doesn't work:
- USB (expected), but the OTG port seems to be available. It's forced to host
mode by the kernel though as gadget support is not enabled
- WiFi: brcmfmac: brcmf_sdio_htclk: HT Avail timeout (1000000): clkctl 0x50
I see that you force-disabled WiFi stuff in JeOS.kiwi, I had to copy over the
Does it work reliably? I heard that it has some hick-ups.
It doesn't work at all here, it fails because of the mentioned timeout.
Ethernet: Gets detected just fine, but when actually sending packets the
kernel freezes (sometimes with
NETDEV WATCHDOG: eth0 (bcmgenet): transmit queue 1 timed out)
- Sometimes the UART (clock?) glitches and there's garbled text
Really? Did you change anything in the config.txt? I haven't experienced this up
Nope. I see it occasionally after between grub and the login prompt. I can't
100% rule out a bad connection, but as it doesn't glitch out after logging in
at all I think that's unlikely.
- u-boot, grub, kernel, sd card, serial
- I/O activity LED :D
- cpufreq: Speed of "bc" corresponds roughly to CPU frequency
So very nice progress already!
I'll try to build a kernel with dwc2
peripheral support, maybe that's enough.
Probably not though, considering it's USB-C...
That's not enough. Nicolas is working on some patches to enable PCI, which is
needed by USB.
For the USB-A ports, yes. Also for USB-C which is connected to the dwc2 which
the kernel already initializes properly?
My kernel-default wasn't picked up by OBS yet for the last three hours :-/
Success! It works fine with a USB-C -> USB-A cable. Run:
+# CONFIG_USB_DWC2_HOST is not set
-# CONFIG_USB_DWC2_DUAL_ROLE is not set
modprobe libcomposite u_ether
mount none -t configfs /mnt
mkdir configs/c.1 functions/eem.usb0
ln -s functions/eem.usb0 configs/c.1/
echo fe980000.usb > UDC
Forward a network connection, then:
echo "BOOTPROTO=dhcp" > /etc/sysconfig/network/ifcfg-usb0
and you can ssh onto the pi4 and download stuff.
localhost:~ # hdparm -tT /dev/mmcblk1
Timing cached reads: 1632 MB in 2.00 seconds = 816.32 MB/sec
HDIO_DRIVE_CMD(identify) failed: Invalid argument
Timing buffered disk reads: 126 MB in 3.01 seconds = 41.90 MB/sec
I played around with some other programs and it seems like CPU performance has
roughly doubled compared to a 3B+.
Even over X11 forwarding applications run very smoothly.
On my wish list is now to have u-boot use the device tree from the firmware for
its own initialization as well, so that hardcoding of addresses isn't necessary
anymore. That way we wouldn't need a new u-boot-rpi4 flavor and new JeOS
flavors installing it instead of u-boot-rpi3.