sven-ola

@cal5582 Seems like lots of NVME needs to disable ASPM. For my Samsung 9b1 NVME, I solved this with a kernel patch, see https://github.com/armbian/build/blob/main/patch/kernel/archive/spacemit-6.18/013-Disable-ASPM-L0-for-Samsung-PM9B1-NVME.patch HTH // Sven-Ola

@JooJoo Bee666 Thanks for the logs. They are basically the same, but kbd and mouse swapped between Bus2 and Bus4. I don't see a reason, why one is not booting. Seems best to disable USB boot (already done in armbian/build::main) with an exception for R2S. @cal5582 If you compile Armbian, there's a Debian package for the kernel (such as inux-image-current-spacemit_26.08.0-trunk_riscv64__6.18.44-S1b8a-D0000-Pa617-C999999-H20f3-HK01ba-Vc222-Bb84f-R448a.deb) but no source*.deb. To compile on the board, you can copy the kernel source tree from cache/sources/linux-kernel-worktree/ (the 6.18__spacemit__riscv64 directory) e.g. to /usr/src. Then simply issue "make" in there. Since the compiler already is pre-installed for DKMS-bcmdhd (wifi) this should work. After ~3 hours, you should be able to "make install" and "make modules_install". Check in /boot if you need to adapt for a different linux-image file name (extlinux.conf). HTH // Sven-Ola

Hi @rm_! Upcoming images willhave USB boot disabled with an exception for the R2S. Also, currently there is some timeout, where systemd-udevd waits for two minutes at early start. Unknown why, may be triggered by the default KERNEL_BTF=yes on image compile.

Hi @JooJoo Bee666! Can you elaborate a bit on the faulty keyboard / mouse combination? Such as: plug them in after boot in the faulty configuration, issue a lsusb -v, an paste the output here. Background: the SpacemiT K1 maintainer is trying to disable USB boot altogether, and at least the Orange Pi R2S folks will complain because those board have no SD card with makes updates without UART a lot harder... Ref: https://github.com/armbian/build/pull/10462 Edit: I tried all keyboard / mouses that I have but no combination of my hardware triggered an exception like this. Also, I've added lsusb -v from my board for comparison. LG // Sven-Ola lsusb.cap

@JooJoo Bee666 Your description "immediate bootloop" is vague. How do you determine that bootloop? Does it display something on HDMI or on UART? If yes, send image/video or text log. Also, your SD card may be faulty. Same bootloop if no SD is inserted? HTH // Sven-Ola

@sivert The "edge" kernel 7.x is more or less the stock kernel from upstream. It's missing some specific K1 drivers developed by SpacemiT basically. I think, upstream devs may implement their own version, thus may need some time. I am not sure, if upstream has the necessary SoC register documentation or if they persue some reverse-engineering on this. I recommend sticking with "current". HTH // Sven-Ola

Hi @cal5582, your options are (rising difficulty) Do not install updates Set linux-kernel packages on hold (apt-mark hold...) Make a dkms config for your extra drivers (see above, "b2c2-flexcop-usb") Make your own Debian repo for updates and change /etc/apt sources in userpatches/customize_image.sh Start contributing to Armbian and convince devs to include your changes to kernel cfg HTH // Sven-Ola

As an addon to @sivert: Ubuntu has dkms "3.0.11-1ubuntu13" while Debian has the newer "3.2.2-1~deb13u1". There is a difference in /etc/kernel/prerm.d/dkms: Ubuntu starts "dkms remove..." and Debian starts "dkms kernel_prerm". Both should work normally, because if you update the kernel there are two kernels: the old one and the incoming one. If you happen to have only one kernel, and thus only one compiled dkms module version, the Ubuntu "remove" completely removes anything dkms-module related. Now Armbian is offering a kernel update to the absolute same kernel version number, thus simply overwriting linux-image. Also, a "dpkg-reconfigure linux-image-current-spacemit" (with only one Ubuntu kernel version installed) will also remove the Wifi module. I am not sure how to fix this...

Hey @sivert. The Ubuntu "dkms" package is borked somehow. It removes the dkms config from the file system during kernel update. The Debian "dkms" package works as expected, it removes the modules and rebuilds them on kernel update. The bcmdhd driver is of -ähh- questionable code quality and will never made it in the official kernel. For that reason, I have added an internal package named bcmdhd-spacemit-sdio-dkms that contains driver source in /usr/src and a dkms.conf which is build and installed during Armbian compile.sh. Internal means: that package is installed during build but is not installable via Armbian package repo. Using "dpkg-reconfigure bcmdhd-spacemit-sdio-dkms" is a shorter command than the previously posted "dkms add, dkms build, dkms install" sequence. If your build does not include the bcmdhd-spacemit-sdio-dkms package and therefore does not have /usr/src/bcmdhd-spacemit-sdio-101.10.591.84.37-4/ directory, then something else is wrong here. HTH // Sven-Ola

Hi @sivert! The downloadable Ubuntu noble image has a bcmdhd module for Wifi, so basically Wifi is working. While checking, I discovered that the bcmdhd kernel module is removed and not rebuilt during kernel update. You need to re-add it manually until I know whats wrong with that: dpkg-reconfigure bcmdhd-spacemit-sdio-dkms The bcmdhd automatic rebuild seems to work with the Debian Trixie image. Also, that "edge" Image is meant for devel. No UART, no devel so to speak 😉 HTH // Sven-Ola

Hi @NHO, copy the Armbian kernel and boot with this: yes probably. You will also need the correct *.dtb in /boot - you need to rename / replace k1*.dtb->x1*.dtb e.g. if you run the original uboot. The GPU probably does not work, since Mesa does / will not support the K1 imagination GPU. You need the Bianbu GPU binaries and userspace tools (...wayland...) that matches them. The spacemit-gpu-addon.sh script (see green [solution] button on top of this page) uses a copied / stripped down version of the Bianbu Repo to install Gnome/Wayland but I doubt that this works under Gentoo Linux.

@rm_ you have a capable platform at hand, why not build edge on your RV2? No docker b/c this is riscv64, but this should work anyhow: git clone https://github.com/armbian/build.git cd build && ./compile.sh BOARD=orangepirv2 BRANCH=edge RELEASE=trixie \ BUILD_MINIMAL=yes KERNEL_CONFIGURE=no KERNEL_BTF=no KERNEL_GIT=shallow Mine is running, but AFAICR this is under 4h. You may want to place board upright for better ventilation 😉

@c0rnelius thanks for the u-boot-only image. May help to sort things out. Also, the Armbian.com download pages now have downloadable RV2 and R2S images, so I can stop uploading my locally build ones. I grabbed a second NVME and placed it in the RV2 lower 2280 M.2 slot, while my upper 2230 M.2 slot concurrently holds another NVME. There is a minor quirks between U-Boot and Kernel: U-Boot output shows PCIE-0: Link up (Gen2-x2, Bus0) and PCIE-2: Link up (Gen2-x2, Bus2). U-Boot identifies 2280 as Device 0, and 2230 as Device 1. Linux swapped to 2230=/dev/nvme0n1 and 2280=/dev/nvme1n1. Both NVMEs are bootable, i.e the 2280 is booted if present, if not RV2 boots from 2230. A stopped U-Boot shows the following info => nvme info Device 0: Vendor: 0x144d Rev: 1B7QCXE7 Prod: S3EUNX0HB05786X [ 6.900] Type: Hard Disk [ 6.903] Capacity: 476940.0 MB = 465.7 GB (976773168 x 512) [ 6.910] Device 1: Vendor: 0x144d Rev: HPS4NHBV Prod: S774NX0XB34550 [ 6.917] Type: Hard Disk [ 6.920] Capacity: 122104.3 MB = 119.2 GB (250069680 x 512) while the Linux kernel has this: root@orangepirv2:~# df -h / Filesystem Size Used Avail Use% Mounted on /dev/nvme1n1p1 455G 1.7G 448G 1% / root@orangepirv2:~# lspci 0001:00:00.0 PCI bridge: SpacemiT X60 PCIe 2.0 x2 Root Complex (rev 01) 0001:01:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller SM961/PM961/SM963 0002:00:00.0 PCI bridge: SpacemiT X60 PCIe 2.0 x2 Root Complex (rev 01) 0002:01:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller PM9B1 (DRAM-less) (rev 02) root@orangepirv2:~# hdparm -t /dev/nvme?n1 /dev/nvme0n1: Timing buffered disk reads: 1446 MB in 3.01 seconds = 480.91 MB/sec /dev/nvme1n1: Timing buffered disk reads: 1760 MB in 3.00 seconds = 585.89 MB/sec HTH // Sven-Ola

@rm_ No, SD and NVME are working with Armbian edge 7.1, while USB does not. root@orangepirv2:~# uname -a Linux orangepirv2 7.1.0-rc3-edge-spacemit #3 SMP PREEMPT_DYNAMIC Sun May 10 21:08:09 UTC 2026 riscv64 GNU/Linux root@orangepirv2:~# blkid /dev/nvme0n1p1: LABEL="armbi_root" UUID="fdbd4ab2-2238-4a62-a079-74ff0582c8d2" BLOCK_SIZE="4096" TYPE="ext4" PARTUUID="00406d4f-01" /dev/mmcblk0p1: LABEL="armbi_root" UUID="63ee7593-e111-4547-ac2f-6bdb8519ce11" BLOCK_SIZE="4096" TYPE="ext4" PARTUUID="9ebf8299-01" /dev/zram0: UUID="3be872bc-b408-46c3-8ec9-d27500adc1dd" TYPE="swap" /dev/zram1: LABEL="log2ram" UUID="bace3aba-15ea-4833-a7ad-f7d9e39aa0c5" BLOCK_SIZE="4096" TYPE="ext4" root@orangepirv2:~# lsusb root@orangepirv2:~# lspci 0001:00:00.0 PCI bridge: SpacemiT X60 PCIe 2.0 x2 Root Complex (rev 01) 0002:00:00.0 PCI bridge: SpacemiT X60 PCIe 2.0 x2 Root Complex (rev 01) 0002:01:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller PM9B1 (DRAM-less) (rev 02) Anyhow, thanks for the Phoronix link // Sven-Ola

Hi @armfan. From what I am able to extract from the flood of braindumps you have a non-working NVME with 2TB that is not working for some reason. If it really stops sometimes in the middle of u-boot::nvme::init then you may have some power issue with this NVME. For comparison: I placed my NVME in the lower m.2 and booted via MTD the u-boot armbian-installed from the "current" image downloadable from my site. See attached capture file. HTH // Sven-Ola nvme-lower-uboot-may-26-2026.cap