Kevin su

I tested the new driver in my box (the router and box are in two different rooms). As you can see, just as @WallaceWebs mentioned, the upload speed has improved significantly. Before merging this pull request, the speed was 8.25Mbps, and now it is 33.46Mbps. Idle Latency: 11.17 ms (jitter: 1.63ms, low: 10.02ms, high: 12.42ms) Download: 38.57 Mbps (data used: 22.8 MB) 97.38 ms (jitter: 31.45ms, low: 16.86ms, high: 432.50ms) Upload: 33.46 Mbps (data used: 54.8 MB) 127.94 ms (jitter: 41.37ms, low: 28.14ms, high: 628.78ms) Packet Loss: 0.0%

@Werner Great suggestion! The repo has been updated to support DKMS and Extension. https://github.com/cdhigh/armbian_sv6256p

@WallaceWebs Fabulous, this is a milestone fix.

@Delutto Thank you for your contribution! Your changes have been successfully merged to my repo https://github.com/cdhigh/armbian_sv6256p Now the driver should work with kernels from 6.0 through 7.x.

@Vinicius Guastala Thank you for your feedback. Your patches have already been merged into my repository.

@Kauã Fonseca There are SSH commands in my GitHub repository. Just SSH into your TV box and execute them one by one. sudo apt-get update sudo apt-get install -y build-essential linux-headers-$(uname -r) git git clone https://github.com/cdhigh/armbian_sv6256p.git cd armbian_sv6256p chmod +x ./parser-conf.sh make ARCH=arm64 KSRC=/lib/modules/$(uname -r)/build ls -lh ssv6x5x.ko sudo cp ./ssv6x5x-wifi.cfg /lib/firmware/ sudo cp ./ssv6x5x-sw.bin /lib/firmware/ sudo cp ./ssv6x5x.ko /lib/modules/$(uname -r)/kernel/drivers/net/wireless/ sudo depmod -a sudo modprobe ssv6x5x ip link show nmtui

I ran a speed test on my SSV6051. Here are the results: Idle Latency: 9.78 ms (jitter: 1.12ms, low: 9.33ms, high: 10.93ms) Download: 34.46 Mbps (data used: 23.6 MB) 52.01 ms (jitter: 6.52ms, low: 19.97ms, high: 109.64ms) Upload: 8.25 Mbps (data used: 12.4 MB) 99.47 ms (jitter: 26.37ms, low: 48.84ms, high: 491.13ms) Packet Loss: 0.0%

@sicxnull I ran into the same issue when flashing "Armbian-unofficial_26.02.0-trunk_Mxqpro_bookworm_current_6.12.65_minimal" (worked only in SD Card). The issue was gone when I switched to "Armbian-unofficial_25.05.0-trunk_X96q-lpddr3-v1-3_bookworm_edge_6.12.11_server" built by @Nick A

While reviewing the system logs, I noticed several ftrace warnings appearing in dmesg. After investigating the root cause, I have applied the necessary configuration updates to eliminate these logs. The driver is now running cleanly without any ftrace-related noise.

@Nick A Yes, I used the MT7601U for a few days, but I wasn’t really satisfied with it. So, with the help of AI, I decided to port the SV6256P anyway—and to my surprise, it actually worked.

@jock The driver-related knowledge is vast and quite complicated, with many pitfalls, so porting really requires strong assistance from AI. I don’t have an SSV6051, so I can’t actually complete the porting myself. Btw, the "ssv6x5x-sw.bin" I’m using is the file you posted on the forum.

I’m really happy to share that I finally managed to port the SV6256P (SSV6X5X) SDIO Wi-Fi driver from the legacy Linux 4.4 kernel to modern kernels. This repository contains my working port for Linux kernel 6.12 and newer, with updated mac80211/cfg80211 integration so the driver can build and run on recent Armbian systems. The chipset is commonly used in many low-cost TV boxes and embedded devices, but the original driver only supported legacy kernels (4.4). Now it’s functional again on modern systems, at least on my tested hardware (Allwinner H616 / X96 Q via SDIO). Repository link: https://github.com/cdhigh/armbian_sv6256p I’m quite excited to finally see this old Wi-Fi chip working on a 6.x kernel 🙂 [ 4.774151] ssv6x5x: importing configuration from /lib/firmware/ssv6x5x-wifi.cfg [ 4.778281] tu_ssv6xxx_sdio_init, probe @(____ptrval____) [ 4.779483] TU_SSV6XXX_SDIO mmc3:0001:1: Probing SDIO bus [ 4.779513] ssv6xxx_set_sdio_clk: set sdio clk 25000000Hz [ 4.799668] TU_SSV6XXX_SDIO mmc3:0001:1: vendor = 0x3030 device = 0x3030 [ 4.825421] TU_SSV6XXX_SDIO mmc3:0001:1: dataIOPort 0x10000 regIOPort 0x10020 [ 4.849181] TU_SSV6XXX_SDIO mmc3:0001:1: dataIOPort 0x10000 regIOPort 0x10020 [ 4.849530] TU_SSV6XXX_SDIO mmc3:0001:1: CHIP ID: SSV6006C0 [ 4.850373] ssv6x5x ops chk: tx=1 start=1 stop=1 config=1 add_if=1 rm_if=1 conf_filter=1 wake_txq=1 [ 4.850402] ssv6x5x chanctx chk: any=1 all=0 emulate=1 add=1 rm=1 chg=1 assign=0 unassign=0 [ 4.850409] ssv6x5x chanctx ops set but not complete [ 4.850424] Attach SSV6006 family HAL function [ 4.858285] MAC address from e-fuse [ 4.858311] EFUSE configuration [ 4.858315] Read efuse chip identity[79000000] ip link show: 6: wlan0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN mode DORMANT group default qlen 1000 link/ether xx:xx:xx:xx:xx:xx brd ff:ff:ff:ff:ff:ff

@Nick A Thank you for your excellent work! I managed to compile it myself and shared it with someone who has the same board (IK316-EMCP_V1.0). 1. Changed kernel preemption mode from full preemption to voluntary preemption. 2. Added MediaTek Wi-Fi drivers (MT7601U etc). 3. Embedded kernel headers. https://github.com/cdhigh/armbian_build/releases/tag/v20250306

@Nick A I agree with your point — for modern SBCs, the difference is probably not that significant, and it’s more of a psychological feeling. However, it would be even better if the new version could switch to CONFIG_PREEMPT_VOLUNTARY.

@Nick A Thanks for the reply. I don’t have enough Linux knowledge, so should I just wait for your new release? My box isn’t an X96Q — it’s a generic no-name one — but after trying many, many images, I found that the version you compiled happens to work perfectly and runs smoothly. So far I haven’t found any issues (the only thing missing is Wi-Fi, but that’s not a problem with your build: SV6256P). Even other images made for X96Q LPDDR3 by different people have all kinds of problems. Another small issue: your current compilation option is fully preemptive scheduling. This option is not suitable for server use. I suggest changing it to voluntary preemption or non-preemptive, thank you. The option of your images: CONFIG_PREEMPT_BUILD=y # CONFIG_PREEMPT_NONE is not set # CONFIG_PREEMPT_VOLUNTARY is not set CONFIG_PREEMPT=y # CONFIG_PREEMPT_RT is not set CONFIG_PREEMPT_COUNT=y CONFIG_PREEMPTION=y # CONFIG_PREEMPT_DYNAMIC is not set # CONFIG_SCHED_CORE is not set # CONFIG_SCHED_CLASS_EXT is not set