formatted and combined logic

x02 commented

2026-10-03 09:30:35 -07:00

Contributor

fixed the unsafe where the mutable borrow was occuring on the immutable reference to self :)

x02 added 2 commits

2026-10-03 09:30:35 -07:00

feat: reformat and combined logic 43c84a1c76

removed a file i accidentally added with ':w ' bf4adb374c

husky commented

2026-10-04 00:13:27 -07:00

Owner

thank you a ton! very happy to have people helping to make my terribly hacked together code better :p

i'll test this in a bit and probably merge since everything looks good from what i can tell

thank you a ton! very happy to have people helping to make my terribly hacked together code better :p i'll test this in a bit and probably merge since everything looks good from what i can tell

x02 commented

2026-10-04 08:06:35 -07:00

Author

Contributor

you're welcome 🙏
I wasn't able to test it as i dont have the device that its built for however it did build without any warnings or errors

you're welcome 🙏 I wasn't able to test it as i dont have the device that its built for however it _did_ build without any warnings or errors

husky commented

2026-10-04 10:28:11 -07:00

Owner

just tested it, it seems a bug was introduced in the memory allocator's block merging algorithm; because debugging in this environment is a nightmare, all i can tell you is that the number of buddy blocks before the allocations does not match the number of buddy blocks after the frees (which is what the following code i've included tests; output is "mem bad ):")

    let mut memalloc: MemAlloc<1024, 128> = MemAlloc::init(mem_start, mem_end);

    let len = memalloc.blocks.count_real();

    fb_draw_string(b"mem created...", 0, 0);

    let block1 = memalloc.alloc(11);
    if let Some(block1) = block1 {
        fb_draw_string(b"alloc blck 11", 0, 10);
        let block2 = memalloc.alloc(12);
        if let Some(block2) = block2 {
            fb_draw_string(b"alloc blck 12", 0, 20);
            memalloc.free(block1);
            fb_draw_string(b"free blok 11", 0, 30);
            memalloc.free(block2);
            fb_draw_string(b"free blck 12", 0, 40);

            if memalloc.blocks.count_real() == len {
                fb_draw_string(b"mem good!", 0, 50);
            } else {
                fb_draw_string(b"mem bad ):", 0, 50);
            }
        } else {
            fb_draw_string(b"fail blck 12", 0, 20);
        }
    } else {
        fb_draw_string(b"fail blck 11", 0, 10);
    }

    fb_update();

i can look further into this later if you'd like since you don't have access to hardware for testing, however i have some changes i'd like to push to develop first

since i hadn't until now considered that other people might wanna work on this, i'll start doing that stuff on a separate branch to keep develop cleaner for merges in the future

just tested it, it seems a bug was introduced in the memory allocator's block merging algorithm; because debugging in this environment is a nightmare, all i can tell you is that the number of buddy blocks before the allocations does not match the number of buddy blocks after the frees (which is what the following code i've included tests; output is "mem bad ):") ``` let mut memalloc: MemAlloc<1024, 128> = MemAlloc::init(mem_start, mem_end); let len = memalloc.blocks.count_real(); fb_draw_string(b"mem created...", 0, 0); let block1 = memalloc.alloc(11); if let Some(block1) = block1 { fb_draw_string(b"alloc blck 11", 0, 10); let block2 = memalloc.alloc(12); if let Some(block2) = block2 { fb_draw_string(b"alloc blck 12", 0, 20); memalloc.free(block1); fb_draw_string(b"free blok 11", 0, 30); memalloc.free(block2); fb_draw_string(b"free blck 12", 0, 40); if memalloc.blocks.count_real() == len { fb_draw_string(b"mem good!", 0, 50); } else { fb_draw_string(b"mem bad ):", 0, 50); } } else { fb_draw_string(b"fail blck 12", 0, 20); } } else { fb_draw_string(b"fail blck 11", 0, 10); } fb_update(); ``` i can look further into this later if you'd like since you don't have access to hardware for testing, however i have some changes i'd like to push to develop first since i hadn't until now considered that other people might wanna work on this, i'll start doing that stuff on a separate branch to keep develop cleaner for merges in the future

World commented

2026-10-04 16:28:47 -07:00

First-time contributor

oh hell yea

x02 commented

2026-10-04 18:54:49 -07:00

Author

Contributor

alright thats cool I’ll have a look myself tonight and push any interesting changes if you can test it for me 😇

x02 added 1 commit

2026-10-04 22:12:59 -07:00

fix: updated locals to match and tried to match any other halfs that split from freed blocks 4e5c2b6a93

x02 commented

2026-10-04 22:14:49 -07:00

Author

Contributor

im not too familiar with this paging (i've only really seen XOR allocators) is there any reason its different?

im not _too_ familiar with this paging (i've only really seen XOR allocators) is there any reason its different?

x02 commented

2026-10-04 22:17:09 -07:00

Author

Contributor

that latest commit should be the fix if you could test that pls and ty

x02 changed title from ~~formatted and combined logic~~ to WIP: formatted and combined logic

2026-10-05 11:40:26 -07:00

x02 changed title from ~~WIP: formatted and combined logic~~ to formatted and combined logic

2026-10-05 11:40:35 -07:00

husky commented

2026-10-05 13:54:44 -07:00

Owner

this is pretty much just as much of a direct port of the buddy block allocator pseudocode from wikipedia as i could do, i chose this because i wanted something with speed / flexibility in between a standard free list and like a user runtime allocator

i'll test the code now!

this is pretty much just as much of a direct port of the buddy block allocator pseudocode from wikipedia as i could do, i chose this because i wanted something with speed / flexibility in between a standard free list and like a user runtime allocator i'll test the code now!

husky commented

2026-10-05 14:08:02 -07:00

Owner

seems to work fine! before i merge this, since this is the first outside contribution and i haven't yet published this, i'm going to license this repo under AGPL, do you have any issues with this?

x02 commented

2026-10-05 14:50:13 -07:00

Author

Contributor

I have no issues with that

husky merged commit dd5864063e into develop

2026-10-05 20:14:16 -07:00

husky referenced this pull request from a commit

2026-10-05 20:14:18 -07:00

Merge pull request 'formatted and combined logic' (#1) from x02/scythe:develop into develop

Rows
Columns

formatted and combined logic #1