💾 Archived View for tilde.town › ~insom › 2024-01-13.gmi captured on 2024-05-12 at 15:26:48. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

2024-01-13

I spend a lot of the day reading and thinking about things that interest me. In general this is good, but it can lead to these unproductive loops where it _feels_ like I am doing something (learning) but I am not applying anything and I'm not even certain it's sinking in. I have nothing to show for my time.

Now; not having something to show for your time is cool too. That's really what "play" is, and it's healthy to have it in your life. Most people could probably use more of it. But, for me, a life of play doesn't really lead to satisfaction.

What's more, there's no built in signal to move on from a topic (having done something with that knowledge) -- I typically move on because I've gotten bored or at least distracted by something new. This leads to me dropping a whole load of context: meaning I will have to start nearly from scratch to regain some of that knowledge.

Recently I did a pretty deep dive on an on-again-off-again thing I want to understand: how does context switching work in a kernel. Specifically, I would like to create a scheduler with a couple of tasks, pre-empt them, and switch between them.

I already knew a lot of this stuff _in theory_ -- each process needs to have its own stack and switching processes needs to push some registers, switch the stack, and pop some registers. I inuited that some timer process could be responsible for doing this: timers can be IRQ handlers and that generally means operating in kernel-space. (Whatever that means for the architecture that you're on).

That's nice, but it's not really enough information to implement anything. After an evening spent reading assembly source for an xv6 ARM port, for NetBSD and someone's helpful guide to how FreeRTOS does this on ARM -- I had much more of a clue. Probably enough to implement it. But I am not currently able to give it the time that it needs -- I don't need to (nor feel like) implementing my own scheduler right now.

So, before I let myself move on: I wrote it all up in my notes. I found gaps in my knowledge doing this. I cited a bunch of articles (keeping them out of my bookmarks and letting me close some tabs). When I come around to doing this next time, I can start much quicker. I'm parked facing down the hill.

Anyway this is a lot of words to say "notes are good" but, you know, "notes are good". There's a reason university students take them and I'm not quite sure when I stopped the habit. Perhaps it was when I trained myself to believe that knowledge (and not information) was just a search away, so why should I write things down when I could just search later.