[molticrystal]

>hw-rpi3-5 Hello World bypassing clib altogether, and talking direcly to the hardware.

>Apparently this is a no-no, even if it is theoretically more efficient. The bare-metal and OS dev folk are pretty much limited to using this method

    mov r7, #4    /\* raw system call for write */  
    swi #0    /* print without using clib \*/

This seems to be just a call into the linux kernel, not bare metal, essentially what the clib does itself. You have to do this to do functions the kernel supports but haven't been ported to clib yet.

I haven't really looked deep into it but a google search came up with this for bare metal text printing https://github.com/bztsrc/raspi3-tutorial/tree/master/0A_pcs...

[ksaj]

Thanks. I will be updating the text shortly.

"You have to do this to do functions the kernel supports but haven't been ported to clib yet."

This is insightful. I'll reflect it in my updated text.

[molticrystal]

I meant the comment you responded to for a different thread, but yes, there is a strong mapping between the kernel of an OS and the preferred C library of that OS, because what is the point of exposing functionality to programs in userspace, like more efficient methods, if you have to crack open kernel docs to figure out how to use it, instead of a nice wrapper in your C library. And both the kernel people and the clib people want you to have access to better methods.

But as their goals and timelines are different, they are out sync, which comes into play not only in the new functionality not being included, but variations of existing functionality where the function has additional flags or parameters that the clib function hasn't added. For example, you'll often see this in areas involving threading, starting processes, and sockets. Luckily for most people you'll often find a fork of the clib or just a 3rd party library that wraps the functionality, but if you are on cutting edge or using a feature that never really caught on because its applicability is very niche, you'll have to go back to syscalls.