Memory by CodeAnarchist · Pull Request #67 · differrari/RedactedOS

CodeAnarchist · 2026-03-05T10:18:58Z

No description provided.

differrari · 2026-03-07T19:22:40Z

bin/read/linker.ld

@@ -1,25 +1,30 @@
 ENTRY(main)
+PHDRS {


These linker files aren't used AFAIR, we can get rid of them entirely

kernel/bin/ping.c

kernel/memory/mmu_start.c

differrari · 2026-03-15T20:02:43Z

run_virt


-USE_FB=true
-USE_NET=false
+USE_FB=false


keep defaults as they are

common.mk

kernel/bin/ping.c

kernel/bin/shutdown.c

kernel/dev/driver_base.h

differrari · 2026-03-16T16:40:01Z

kernel/filesystem/virtio_9p_pci.cpp

+} __attribute__((packed)) t_clunk;
+
+bool Virtio9PDriver::clunk(virtio_device *dev, uint32_t fid, uint16_t tag) {
+    t_clunk *cmd = (t_clunk*)kalloc(dev->memory_page, sizeof(t_clunk), ALIGN_4KB, MEM_PRIV_KERNEL);


9p now uses the p9_helper to create packet, to avoid alignment issues on new versions of qemu

differrari · 2026-03-16T16:43:58Z

kernel/hw/hw.c

+        //https://www.qemu.org/docs/master/system/arm/raspi.html
+        //raspi4b has no pcie root
+        //TODO use DTB as source
+        #if QEMU


PCI(e) needs some work. the real pi has pcie, qemu does not, let's leave it as is for now

kernel/memory/page_allocator.c

kernel/bin/tracert.c

differrari · 2026-03-17T21:27:58Z

kernel/kernel.c

+extern char __bss_start[];
+extern char __bss_end[];
+void kernel_main(uint64_t board_type, uint64_t dtb_pa) {
+    for (uintptr_t p = (uintptr_t)__bss_start; p < (uintptr_t)__bss_end; p += 8) *(uint64_t*)p = 0;


Could this be a memset call? Slightly more optimal, cleaner and open to future optimization

differrari · 2026-03-21T20:06:43Z

kernel/bin/bin_mod.c

    { "monitor", monitor_procs },
 };

-process_t* load_proc(const char *full_name, const char *prog_name, int argc, const char *argv[]){


There's some new logic that needs to be included if you merge:

Any program that specifies a full path gets loaded from that path, there's some extra logic to set its bundle (.red) if it's part of the path. So if we try to exec /shared/applications/doom.red/doom.elf, it'll load that exact executable and everything before the doom.elf as the bundle

Any program that's just an executable name (ping, test, read) gets loaded as such from the bin folder

Any program not found in the bin folder gets loaded from the ones hardcoded here if it exists

differrari · 2026-03-21T20:09:08Z

kernel/input/input_dispatch.cpp

    if (buf->write_index == buf->read_index)
        buf->read_index = (buf->read_index + 1) % INPUT_BUFFER_CAPACITY;

+    if (focused_proc->sleeping) wake_process(focused_proc);


Don't wake the proces in here or register event, it'll mess with any waiting logic, and the input is registered anyway.

differrari · 2026-03-21T20:09:22Z

kernel/input/input_dispatch.cpp

        gpu_set_cursor_pressed(last_cursor_state);
        gpu_update_cursor(mouse_loc, true);
    }
+    if (focused_proc && focused_proc->sleeping) wake_process(focused_proc);


differrari · 2026-03-21T20:09:26Z

kernel/input/input_dispatch.cpp


    if (buf->write_index == buf->read_index)
        buf->read_index = (buf->read_index + 1) % INPUT_BUFFER_CAPACITY;
+    if (focused_proc->sleeping) wake_process(focused_proc);


differrari · 2026-03-21T20:11:28Z

kernel/kernel_processes/kprocess_loader.c


-    proc->pc = PHYS_TO_VIRT(((uintptr_t)func));
-    kprintf("Kernel process %s (%i) allocated with address at %llx, stack at %llx-%llx, heap at %llx. %i argument(s)", (uintptr_t)name, proc->id, proc->pc, proc->sp - proc->stack_size, proc->sp, proc->heap, argc);
+    proc->output = (kaddr_t)palloc(PROC_OUT_BUF, MEM_PRIV_KERNEL, MEM_RW, true);


Output has been moved to being generated in the scheduler when writing to it

differrari · 2026-03-21T20:17:55Z

modules/usb/common/usb_common.cpp


 extern "C" void init_usb_process(){
-    if (!input_driver->use_interrupts)
+    USBDriver *driver = input_driver;


Should be possible to put this part in a function, or even on initialization

differrari · 2026-03-21T20:31:03Z

modules/disk/virt/disk.c

+        return false;
+    }
+
+    memset(disk_cmd, 0, sizeof(virtio_blk_req));


mem is always zero'd. Here and in the header files there's no need for explicit memset or default initialize alloc'd values

differrari · 2026-03-21T20:34:29Z

modules/graph/virt/graphics.cpp

 draw_ctx* gpu_get_ctx(){
    if (!gpu_ready()) return 0;
-    return gpu_driver->get_ctx();
+    draw_ctx *ctx = gpu_driver->get_ctx();


this seems redundant

differrari · 2026-03-21T20:44:03Z

modules/audio/virt/audio.cpp

 }

 sizedptr audio_request_buffer(uint32_t device){
+    if (!audio_driver || !audio_driver->out_dev) panic("audio not ready", 0);


In this and the other funcs, should be enough to return an error, not panic

CodeAnarchist added 22 commits February 2, 2026 01:08

[sub] fixed url submodule

e8d23b8

[netpkt] palloc in netpkt, minor fixes

2aa4448

[ALLOC,WIP] refactor

fb2a1cf

[WIP ALLOC] misc, mm, fixed net memory leak

0f2afcc

[PROC]fix wx perms, enable user processes

448fe67

Merge remote-tracking branch 'upstream/main' into mem

9f6dbfc

[ALLOC] asid, finalizing the merge

235bd1a

[mmu] add global asid, vma, various fixes

de459ae

[mem] fix high-half exception vectors

77dc3d8

[mem] brk and page fault

93dc84d

[alloc] vma-demand palloc/pfree

3306fc2

[mem] make paddr_t explicit

b52e442

[mmu] no phys as pointer

8cdccc9

[mem wip] high half ker, TTBR0/TTBR1 split (not booting)

86904c3

[mem] boot, kva/pa split

1699782

[mem] rasp

bb34ddf

[mem] uaccess, VMA

426926c

[mem/drivers] minor fixes

ba0a702

[mem/proc] unify mm ownership, user memory paths

3b624fb

[mem] finalize uaccess and TTBR0 isolation

dcbd8e2

[test]test

219632f

[misc]fix stress test issues in filesystem, process and IO paths

251a20a

differrari requested changes Mar 15, 2026

View reviewed changes

differrari reviewed Mar 15, 2026

View reviewed changes

kernel/bin/ping.c Outdated Show resolved Hide resolved

kernel/bin/shutdown.c Outdated Show resolved Hide resolved

kernel/dev/driver_base.h Outdated Show resolved Hide resolved

differrari reviewed Mar 16, 2026

View reviewed changes

[misc] git requests and testing

f230b98

differrari reviewed Mar 17, 2026

View reviewed changes

kernel/bin/tracert.c Outdated Show resolved Hide resolved

CodeAnarchist added 2 commits March 20, 2026 00:33

[misc]removed brk heap in favor of full anon, cleaned up structs

d583063

[sched] fix scheduler crash, refactored to dynamic types

1493941

CodeAnarchist force-pushed the mem branch from 1493941 to d583063 Compare March 21, 2026 18:20

sub

b7deedd

differrari reviewed Mar 21, 2026

View reviewed changes

Conversation

CodeAnarchist commented Mar 5, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants