I decided briefly return my focus to Turtle Kevux development following the recent 0.6.5 release of the Featureless Linux Library (FLL). Several good things were discovered. The most notable is that both musl-libc (vesion 1.2.4) and valgrind (version 3.21.0) made some relevant releases. The problems that I have been facing appear to have been resolved. Attempting to boot the kernel revealed some additional problems. These problems are primarily that the display does not display and gives the false impression that the system did not boot.
I updated the kernel to fix some of the problems with the display and at least in Qemu the command line is visible. I believe recent changes in the kernel to make booting faster have resulting in buggier displays when a good initrd with the kernel modules setup is not available. My custom initrd work has not yet been resumed and so for now I am booting into bash.
With a version of Kevux that, technically speaking, can be booted up, I have been able to finally spend some time and effort on using the Controller program as an Init program. Doing this revealed to me that I have quite a ways to go. The biggest hurdle is that I need to learn and understand how setting up the TTY safely and properly. Process management also needs to be performed for processes that are not directly managed by the Controller program.
I believe that Suckless has an sinit program that can be used as a guide to teach me how an init program must run as far as the basics. I believe I should also provide this sinit program as an option.
Going forward I plan on making much of the Suckless software available on the Turtle Kevux distribution. This Suckless software is very similar to my FLL software. The main difference being that the FLL is accepting several design costs that tend to make the software relatively larger and more complex. Consider the FLL Fat Embedded and the Suckless software as Real Embedded or Real Simple.
The Controller needs more work before the Turtle Kevux distribution is even ready to boot. Further development of the Controller program is dependent on the mass refactoring that is currently ongoing in the FLL. Using the sinit program may be the quickest way to get this distribution moving forward.
The plan right now is to make another pass of the Turtle Kevux development and to update the website with that development code for any brave fools out there to dabble in.
Kevin Day