@dixit I just noticed this post after answering the other one in the p44-ledchain thread Hearing that you had similiar problems with the ws2812-draiveris driver as with my p44-ledchain, I can conclude that the problem is most likely not the timing, but the hardware (missing level shifter?). Because ws2812-draiveris always meets the timing for the WS28xx - as you might have seen when porting it, it simply locks all interrupts and then bit-bangs the data. While this is good for meeting the LED chip timing (and simple to code), locking interrupts for dozens of milliseconds is a cardinal sin in a Linux driver. On the original AR9331 CPU (Omega1), where ws2812-draiveris originates from, this was the only way to get such SmartLEDs working at all, so ws2812-draiveris is as good as it can get on that chip. However, on the MT7688/Omega2 with its PWM units, the p44-ledchain way to do it, without locking interrupts, is the way to go. And that works easily, at least with WS2813 or WS2815, with all 4 PWMs running in parallel with 1000 LEDs each.

@Lazar-Demin No: for size, cost, and power reasons we need to use the built-in UART(s). I'm interested that @luz is having no trouble doing basically the same thing. It would be nice to know what he/she is doing that is different from what I am doing. I will rebuild buildroot and see if this makes a difference.

I was asking this question few months ago. I was desperately needed a good wifi driver with my own kernel build. When I was dealing with this problem, the mac80211 wasn't just slow, it was also dropping connections and crashing the OS. I've spent a LOT of time dealing with this issue, and in the end I managed to take LinkIt Smart OpenWRT image, configure it to my needs and deploy it on Omega2. It was a real hassle, because there are some hw differences and they're using an old kernel version (3.xx) with pre-compiled Ralink wifi binaries.

@Alex-Wells said in Building Staging Drivers (FBTFT) [RESOLVED]: Now my only problem is that the MT7628 SPI driver won't accept transfers larger than 16 bytes at a time, but that's an issue for another day Have a look here, @wdu has a modified version of the SPI driver eliminating some of the bugs in the original.