[Letux-kernel] Status LX16

[H. Nikolaus Schaller]

> Am 23.12.2024 um 13:13 schrieb Paul Boddie <paul at boddie.org.uk>:
> 
> On 2024-12-23 08:09, H. Nikolaus Schaller wrote:
>> Ah, I wasn't aware that you managed to get SPI speed up. On my setup I still run SPI with DMA disabled
>> and at slow clock rate. And then get 1 fps max.
>> So this means I should simply give it a try?
> 
> I got this working in my L4Re setup, but my Linux testing only really verified that the flashing cursor in a blank console window wasn't halted by your /dev/zero stress-testing script. Obviously, a flashing cursor isn't very demanding in terms of transfers, probably involving little more than a "set window" command and then a small number of pixel values.

Ah, I see.

> A recurring problem I have had, which struck at this point, was that I couldn't get the X server and the login screen to appear so that I could properly test the transfers. Previously, I resolved that by fixing up damaged/missing files in /usr/local for the GNUstep libraries, but I couldn't figure it out in the short amount of time before the holidays arrived. I probably need to start again in terms of my distribution image just to have a reliable foundation for testing.

For me the main issue at the moment is on PVRSGX boards (like GTA04 or PandaES) where it depends on the random sequence how the HDMI and the SGX /dev/dri cards are loaded. This seems to stop the nodm service if it fails once. Then, rmmodding the sgx (of blocking completely) and running "service nodm start" makes the GUI come back again. But on LX16 we have no SGX...

> 
> But it can't hurt for you to give it a try!

Well, will take some days. At the moment the upgrade to 6.13-rc4 is blocking everything. I should think about a different strategy of (semi-)automatically running LetuxOS kernels on one (standalone) machine and developing new stuff and debugging on another one.

BR,
Nikolaus