kat.net.nz

the website of alxce

polymorphOS status update - 2019-10-29

posted: 2019-10-29 00:00:00 +0000

quick update! didn do anything to the kernel itself today, i was mainly focusing on the allocator, but -

polymorph-allocator now has the ability to deallocate blocks properly, and it seems to work okay! i tagged v0.1.0 and published it to crates.io so it can be used in other things. i also went through and documented everything, which you can see on docs.rs.

polymorphOS status update - 2019-10-28

posted: 2019-10-28 00:00:00 +0000

today has been the day of the ~memory manager~ !!

previously, polymorphOS has been using the linked_list_allocator crate. there’s a few downsides to that crate, one of which is you can’t add non-contiguous memory blocks.

i’ve made memory managers before (see awooos/dmm) but those were in C so i couldn’t just grab one and run with it. so, i wrote a new allocator named polymorph-allocator, that takes heavy inspiration from awooos/dmm.

it’s very early in development - it can allocate memory but it doesn’t do proper deallocation (yet) - but polymorphOS boots and works using the new allocator! i’m very happy with this so far.

currently the polymorphOS kernel only tells the memory manager about one memory block (which is the heap that was created for linked_list_allocator), but soon it’ll know about all memory in the system!

polymorphOS status update - 2019-10-27

posted: 2019-10-27 00:00:00 +0000

i figured i’d start doing updates on how polymorphOS is going! so here’s the first one.

i’ve implemented a driver system that has two load phases - “early” and “late”. the “early” phase runs as early in the boot process as possible (just after control has been passed to kmain), and the “late” phase runs after the PCI bus so it can set up drivers for installed PCI devices.

i’ve also ripped out the uart_16550 crate and replaced it with my own serial port driver, loaded in the “early” phase, that accepts data to write to the serial port through the kernel event system. in order to get klog and friends to write to the serial port, i’ve added a stub impl Write struct that just sends events to the serial port driver (with the immediate flag set).

in this screenshot, the VGA output is on the left, and the serial port output is in the terminal on the right. you can see a few messages on the left (before the “End of drivers::setup_early()” message) that aren’t outputted to the serial port - these are logged before the serial port driver is initialized.

next up on the to do list:

  • timers with callbacks (think setTimeout() in JS)
  • implementing read support in my serial port driver
  • implementing a very tiny readline-esque thing for …
  • … a debugging shell over the serial port!
  • writing a new memory manager so i can use all the available memory