Investigate whether we need SDHCI speed patch

[cmsj]

See https://github.com/perillamint/switch-linux/commit/ec2c0a0bbc388622e35aaa5f5038f6d899fcbbee

[Jake-Jensen]

We should, as it's a known problem in many public forums. Has nothing to do with the chosen rootfs, just Nintendo being Nintendo.

[cmsj]

I'm mostly convinced of the need, but it would also need us to start taking ownership of the kernel, which we currently don't.

I've done a couple of experiments to see if I can boot a kernel that's loaded from the rootfs, but they failed and I don't have the f0f UART->Serial adapter to figure out why. I can try some more things to see if I can get it to work, but without a serial adapter I'm effectively shooting blind.

[Jake-Jensen]

A standard 1.8v serial should be enough to interface. From what I can tell, and from other sources like Fusee, it seems that the kernel has to be loaded by the exploit itself first, which then calls the SD, which naturally houses the rootfs. Apparently some people are working on it, but I just can't see how one would load the SD first, then load a kernel, as from what I can see, RCM mode doesn't have any kind of driver or interface for loading the SD mount point.

[cmsj]

the f0f boot chain loads u-boot, which in theory can load things from MMC/ext4, so it's really a question of whether or not the f0f fork of u-boot can do that.

[Jake-Jensen]

https://community.nxp.com/thread/380527 https://community.nxp.com/thread/453303

It seems uboot is fully capable of it, but f0f's? I have no idea.

I'll be picking up a serial cable here soon, so if you want to load your kernel, I can debug what the issue is.

[cmsj]

Got it working :)

https://github.com/cmsj/shofel2/commit/74fa52a6d3817f86c00ff67a37b7bd7f93aced75

[Jake-Jensen]

Oooh is it available for packaging right now? Having no joy on penguins

[cmsj]

Not yet, I'd need to extend the image builder to also build the kernel.

[CTCaer]

have you tried 20 0 000000 btw?