Re:Next steps for Rust in the kernel
by tialaramex on Wednesday September 21, 2022 @02:43PM
(#32929492)
I essentially began with C, but my guess is that you should start lower if you can.
You don't want to get fooled into thinking C's abstract machine is what your actual machine is like, because it really, really isn't. A bunch of the stuff C is deliberately vague about because it wasn't portable in the 1970s is very concrete and important to understand about on a real computer. Rust's Wrapping<u8> and Saturating<i16> have behaviour that you could buy with a real CPU (indeed your actual CPU almost certainly can do Wrapping<u8> with no extra work) but C doesn't promise anything like this and names types things like "short int" which could be anything.
On the other hand, some things C makes seem obviously concretely real, are just being simulated convincingly and that's not necessarily how your CPU works at all. Nobody really promised "pointers" are just integers you can go around doing arithmetic with for example.
You don't want to get fooled into thinking C's abstract machine is what your actual machine is like, because it really, really isn't. A bunch of the stuff C is deliberately vague about because it wasn't portable in the 1970s is very concrete and important to understand about on a real computer. Rust's Wrapping<u8> and Saturating<i16> have behaviour that you could buy with a real CPU (indeed your actual CPU almost certainly can do Wrapping<u8> with no extra work) but C doesn't promise anything like this and names types things like "short int" which could be anything.
On the other hand, some things C makes seem obviously concretely real, are just being simulated convincingly and that's not necessarily how your CPU works at all. Nobody really promised "pointers" are just integers you can go around doing arithmetic with for example.