Create ata.c

Add base ATA PIO mode driver so that filesystems like fat16 fat32 work.

.clang-format

@@ -0,0 +1,24 @@
+BasedOnStyle: Google
+IndentWidth: 4
+TabWidth: 4
+UseTab: Never
+ColumnLimit: 80
+
+DerivePointerAlignment: false
+PointerAlignment: Right
+ReferenceAlignment: Right
+
+AlignConsecutiveMacros: Consecutive
+AlignTrailingComments:
+  Kind: Always
+  OverEmptyLines: 0
+
+IncludeBlocks: Regroup
+IncludeCategories:
+  # Std headers
+  - Regex: '<[[:alnum:]_.]+>'
+    Priority: 2
+
+  # Other headers
+  - Regex: '.*'
+    Priority: 1

.clangd

@@ -0,0 +1,6 @@
+CompileFlags:
+  CompilationDatabase: build
+
+Diagnostics:
+  UnusedIncludes: Strict
+  MissingIncludes: Strict

.editorconfig

@@ -0,0 +1,12 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+insert_final_newline = true
+trim_trailing_whitespace = true
+
+[Makefile]
+indent_style = tab
+indent_size = 8
+tab_width = 8

.gitignore

@@ -1 +1,5 @@
+.build.env
 build
+cross
+.cache/
+compile_commands.json

Makefile

@@ -1,64 +1,77 @@
 AS = nasm
-CC = gcc
+ASFLAGS = -f elf32 -g -F dwarf
-CFLAGS = -std=c11 -m32 -ffreestanding -c -fno-stack-protector -fno-pie
+CC = i386-elf-gcc
-LD = ld
+LD = i386-elf-ld
-QEMU = qemu-system-i386
+QEMU= qemu-system-i386
-IMG_SIZE = 1440k
+OBJCOPY = i386-elf-objcopy
 
 BUILD_DIR = build
-
+CROSS_DIR = cross
-BOOT_SRC = bootloader/boot.asm
+DISK_IMG = $(BUILD_DIR)/disk.img
-BOOT_OBJ = $(BUILD_DIR)/boot.o
+STAGE2_SIZE = 2048
-BOOT_ELF = $(BUILD_DIR)/boot.elf
-BOOT_IMG = $(BUILD_DIR)/boot.img
 
 KERNEL_C_SRC = $(wildcard kernel/*.c)
 KERNEL_ASM_SRC = $(wildcard kernel/*.asm)
 KERNEL_OBJ = $(patsubst kernel/%.c, $(BUILD_DIR)/%.o, $(KERNEL_C_SRC))
 KERNEL_OBJ += $(patsubst kernel/%.asm, $(BUILD_DIR)/asm_%.o, $(KERNEL_ASM_SRC))
-KERNEL_OBJ += $(BUILD_DIR)/boot1.o
-KERNEL_ELF = $(BUILD_DIR)/kernel.elf
-KERNEL_BIN = $(BUILD_DIR)/kernel.bin
 
-DISK_IMG = $(BUILD_DIR)/disk.img
+KLIBC_SRC = $(wildcard klibc/src/*.c)
+KLIBC_OBJ = $(patsubst klibc/src/%.c, $(BUILD_DIR)/klibc/%.o, $(KLIBC_SRC))
 
-all: $(BOOT_IMG) $(KERNEL_BIN) $(DISK_IMG)
+.PHONY: all stage1 stage2 kernel compile-commands $(BUILD_DIR)/compile_commands.json run gdb clean clean-cross clean-all
+all: $(DISK_IMG)
+
+stage1: $(BUILD_DIR)
+	$(AS) $(ASFLAGS) -o $(BUILD_DIR)/$@.o bootloader/$@.asm
+	$(LD) -Ttext=0x7c00 -melf_i386 -o $(BUILD_DIR)/$@.elf $(BUILD_DIR)/$@.o
+	$(OBJCOPY) -O binary $(BUILD_DIR)/$@.elf $(BUILD_DIR)/$@.bin
+
+# NOTE: Stage2 final size should be checked against `$(STAGE2_SIZE)` by the build system to avoid an overflow.
+# Alternatively, convey the final stage2 size through other means to stage1.
+stage2: $(BUILD_DIR)
+	$(AS) $(ASFLAGS) -o $(BUILD_DIR)/stage2.o bootloader/stage2.asm
+	$(CC) -std=c11 -ffreestanding -nostdlib -nostdinc -fno-stack-protector -m32 -Iklibc/include -g -c -o $(BUILD_DIR)/stage2_load.o bootloader/stage2_load.c
+	$(LD) -Tbootloader/stage2.ld -melf_i386 -o $(BUILD_DIR)/$@.elf $(BUILD_DIR)/stage2.o $(BUILD_DIR)/stage2_load.o
+	$(OBJCOPY) -O binary $(BUILD_DIR)/$@.elf $(BUILD_DIR)/$@.bin
+	truncate -s $(STAGE2_SIZE) $(BUILD_DIR)/$@.bin
+
+$(BUILD_DIR)/asm_%.o: kernel/%.asm
+	$(AS) $(ASFLAGS) -o $@ $<
+
+$(BUILD_DIR)/%.o: kernel/%.c
+	$(CC) -std=c11 -ffreestanding -nostdlib -nostdinc -fno-stack-protector -m32 -Iklibc/include -g -c -o $@ $<
+
+$(BUILD_DIR)/klibc/%.o: klibc/src/%.c
+	$(CC) -std=c11 -ffreestanding -nostdlib -nostdinc -fno-stack-protector -m32 -Iklibc/include -g -c -o $@ $<
+
+kernel: $(KERNEL_OBJ) | $(BUILD_DIR) $(KLIBC_OBJ)
+	$(LD) -melf_i386 -Tkernel/linker.ld -o $(BUILD_DIR)/kernel.elf $(KERNEL_OBJ) $(KLIBC_OBJ)
+
+$(DISK_IMG): stage1 stage2 kernel
+	dd if=$(BUILD_DIR)/stage1.bin of=$@
+	dd if=$(BUILD_DIR)/stage2.bin of=$@ oflag=append conv=notrunc
+	dd if=$(BUILD_DIR)/kernel.elf of=$@ oflag=append conv=notrunc
+	truncate -s 1M $@
 
 $(BUILD_DIR):
 	mkdir -p $@
+	mkdir -p $(BUILD_DIR)/klibc
 
-$(BOOT_OBJ): $(BOOT_SRC) | $(BUILD_DIR)
+compile-commands: $(BUILD_DIR)/compile_commands.json
-	$(AS) -f elf32 -g -F dwarf -o $@ $<
+$(BUILD_DIR)/compile_commands.json: $(BUILD_DIR)
+	bear --output $@ -- make -B
 
-$(BOOT_ELF): $(BOOT_OBJ)
+run:
-	$(LD) -Ttext=0x7c00 -melf_i386 -o $@ $<
+	qemu-system-i386 -s -S $(DISK_IMG)
 
-$(BOOT_IMG): $(BOOT_ELF)
+gdb:
-	objcopy -O binary $< $@
+	gdb -x gdb.txt
-	truncate -s $(IMG_SIZE) $@
-
-$(BUILD_DIR)/boot1.o: bootloader/boot1.asm
-	$(AS) -f elf32 -o $@ $<
-
-$(BUILD_DIR)/asm_%.o: kernel/%.asm
-	$(AS) -f elf32 -o $@ $<
-
-$(BUILD_DIR)/%.o: kernel/%.c
-	$(CC) $(CFLAGS) $< -o $@
-
-$(KERNEL_BIN): $(KERNEL_OBJ) | $(BUILD_DIR)
-	$(LD) -melf_i386 --oformat binary -T bootloader/linker.ld -o $@ $(KERNEL_OBJ)
-
-$(DISK_IMG): $(BOOT_IMG) $(KERNEL_BIN)
-	dd if=$(BOOT_IMG) of=$@ bs=512 seek=4
-	dd if=$(KERNEL_BIN) of=$@ bs=512 seek=200
-
-run: $(DISK_IMG)
-	$(QEMU) -drive file=$<,format=raw,if=floppy
-
-.PHONY: stage1 clean
-
-stage1: $(BOOT_IMG)
 
 clean:
 	rm -rf $(BUILD_DIR)
+
+clean-cross:
+	rm -rf $(CROSS_DIR)
+	rm -rf .build.env
+
+clean-all: clean clean-cross

README.md

@@ -35,6 +35,7 @@ You’ll need the following tools installed:
 - `qemu-system-i386`
 
 Optional:
+
 - `gdb`
 - `vncviewer` (TigerVNC or similar)
 
@@ -42,13 +43,27 @@ Optional:
 
 ## 🛠️ Building ClassicOS
 
-Clone and build:
+Clone repository:
 
-```bash
+```sh
 git clone https://github.com/gbowne1/ClassicOS.git
 cd ClassicOS
-make
 ```
 
-build kernel
+Run `configure` script to build a cross-compiler toolchain for `i386-elf`:
-for %f in (*.c) do gcc -m32 -O0 -Wall -Wextra -Werror -pedantic -ffreestanding -nostdlib -fno-pic -fno-stack-protector -fno-pie -march=i386 -mtune=i386 -c "%f" -o "%f.o"
+
+```sh
+./configure
+```
+
+Source the `.build.env` file to add the cross-compiler toolchain to your PATH:
+
+```sh
+source .build.env
+```
+
+Build the kernel:
+
+```sh
+make
+```

bootloader/README.md

@@ -0,0 +1,27 @@
+# ClassicOS 2-stage bootloader
+
+Bootloader documentation for ClassicOS
+
+## Disk image organization:
+
+```
+[  512 B  ] [      2048 B      ] [    Unspecified    ]
+  Stage 1         Stage 2               Kernel
+```
+
+## Stage 1 (`stage1.asm`)
+
+Responsible for loading the second stage using BIOS routines, and switching to protected mode.
+
+- Queries CHS parameters from BIOS
+- Loads the second stage bootloader (2048 B) to `0x7c00`
+- Sets up a GDT with descriptor entries for code and data both covering the whole 32-bit address space
+- Enables A20
+- Set CR0.PE (enable protected mode) and jump to stage 2
+
+## Stage 2 (`stage2.asm, stage2_load.c`)
+
+- Set up segment registers
+- Load the kernel ELF header
+- Parse the program headers, and load all `PT_LOAD` segments from disk
+- Jump to the kernel entry

bootloader/boot1.asm

@@ -1,203 +0,0 @@
-; ==============================================================================
-; boot1.asm - Second Stage Bootloader (Fixed Real Mode Transition)
-; ==============================================================================
-
-[BITS 32]
-global _start
-extern kmain
-
-_start:
-    ; Set up segments
-    mov ax, 0x10
-    mov ds, ax
-    mov es, ax
-    mov fs, ax
-    mov gs, ax
-    mov ss, ax
-
-    ; Stack (must be identity-mapped)
-    mov esp, 0x90000
-
-    ; CPU Feature Detection: check CPUID support
-    pushfd                      ; Save flags
-    pop eax
-    mov ecx, eax
-    xor eax, 1 << 21            ; Flip ID bit
-    push eax
-    popfd
-    pushfd
-    pop eax
-    xor eax, ecx
-    jz .no_cpuid                ; CPUID unsupported if no change
-
-    ; CPUID supported, verify features
-    mov eax, 1
-    cpuid
-    ; Check for paging support (bit 31 of edx)
-    test edx, 1 << 31
-    jz .no_paging_support
-
-    ; Additional CPU feature checks could be added here
-
-    jmp .cpuid_check_done
-
-.no_cpuid:
-    mov si, no_cpuid_msg
-    call print_string_16
-    jmp halt
-
-.no_paging_support:
-    mov si, no_paging_msg
-    call print_string_16
-    jmp halt
-
-.cpuid_check_done:
-
-    ; Temporarily switch back to real mode
-    cli
-    mov eax, cr0
-    and eax, 0x7FFFFFFE        ; Clear PE & PG bits
-    mov cr0, eax
-    jmp 0x18:real_mode_entry
-
-; ----------------------------------------------------------------
-[BITS 16]
-real_mode_entry:
-    ; Real mode for BIOS access (E820, VESA)
-    xor ax, ax
-    mov es, ax
-
-    ; VESA call
-    mov di, VbeControllerInfo
-    mov ax, 0x4F00
-    int 0x10
-    jc vesa_error
-
-    ; E820 memory map
-    xor ebx, ebx
-    mov edx, 0x534D4150
-    mov di, MemoryMapBuffer
-    mov [MemoryMapEntries], dword 0
-
-.e820_loop:
-    mov eax, 0xE820
-    mov ecx, 24
-    int 0x15
-    jc e820_error
-    add di, 24
-    inc dword [MemoryMapEntries]
-    test ebx, ebx
-    jnz .e820_loop
-    jmp e820_done
-
-e820_error:
-    mov si, e820_error_msg
-    call print_string_16
-    jmp halt
-
-vesa_error:
-    mov si, vesa_error_msg
-    call print_string_16
-
-    ; Fallback: set VGA text mode 3 and continue
-    mov ah, 0x00           ; BIOS Set Video Mode function
-    mov al, 0x03           ; VGA 80x25 text mode
-    int 0x10
-
-    ; Clear screen
-    mov ah, 0x06           ; Scroll up function
-    mov al, 0              ; Clear entire screen
-    mov bh, 0x07           ; Text attribute (gray on black)
-    mov cx, 0              ; Upper-left corner
-    mov dx, 0x184F         ; Lower-right corner
-    int 0x10
-
-    jmp e820_done         ; Continue booting without VESA graphics
-
-e820_done:
-    ; Back to protected mode
-    cli
-    mov eax, cr0
-    or eax, 1
-    mov cr0, eax
-    jmp 0x08:protected_entry
-
-; ----------------------------------------------------------------
-[BITS 16]
-print_string_16:
-.loop:
-    lodsb
-    or al, al
-    jz .done
-    mov ah, 0x0E
-    int 0x10
-    jmp .loop
-.done:
-    ret
-
-e820_error_msg db "E820 Failed!", 0
-vesa_error_msg db "VESA Failed!", 0
-no_cpuid_msg db "No CPUID support detected!", 0
-no_paging_msg db "CPU lacks paging support!", 0
-
-; ----------------------------------------------------------------
-[BITS 32]
-protected_entry:
-    ; Paging setup
-    xor eax, eax
-    mov edi, page_directory
-    mov ecx, 1024
-    rep stosd
-    mov edi, page_table
-    rep stosd
-    mov eax, page_table
-    or eax, 0x3
-    mov [page_directory], eax
-    mov ecx, 1024
-    mov edi, page_table
-    mov eax, 0x00000003
-.fill_pages:
-    mov [edi], eax
-    add eax, 0x1000
-    add edi, 4
-    loop .fill_pages
-
-    mov eax, page_directory
-    mov cr3, eax
-    mov eax, cr0
-    or eax, 0x80000000
-    mov cr0, eax
-
-    jmp kmain
-
-halt:
-    cli
-
-.hang:
-    hlt
-    jmp .hang
-
-; ----------------------------------------------------------------
-; Data buffers and variables must be appropriately defined in your data section
-MemoryMapBuffer times 128 db 0          ; 128*24 bytes reserved for E820 memory map (adjust size as needed)
-MemoryMapEntries dd 0
-VbeControllerInfo times 512 db 0        ; Buffer for VESA controller info (adjust size as needed)
-
-; Define page directory and page table aligned as needed (in your data section)
-align 4096
-page_directory times 1024 dd 0
-align 4096
-page_table times 1024 dd 0
-
-%assign pad_size 4096
-%ifdef __SIZE__
-  %define size_current __SIZE__
-%else
-  %define size_current ($ - $$)
-%endif
-
-%if size_current < pad_size
-  times pad_size - size_current db 0
-%endif
-
-checksum_byte db 0

bootloader/linker.ld

@@ -1,26 +0,0 @@
-ENTRY(kmain)
-
-SECTIONS {
-    . = 1M;
-
-    .text : {
-        *(.text*)
-    }
-
-    .rodata : { *(.rodata*) }
-    .data : { *(.data*) }
-    .bss : {
-        *(.bss*)
-        *(COMMON)
-    }
-
-    .stack (NOLOAD) : {
-        . = ALIGN(4);
-        . = . + 0x1000;
-    }
-
-    .heap (NOLOAD) : {
-        . = ALIGN(4);
-        . = . + 0x10000;
-    }
-}

bootloader/stage1.asm

@@ -2,10 +2,16 @@
 ; boot.asm - First Stage Bootloader (CHS Based)
 ; ==============================================================================
 
-[BITS 16]
+; Params for stage2
-; [ORG 0x7C00]
+%define s2_addr 1               ; stage2 disk offset, in sectors
+%define s2_laddr 0x7e00         ; stage2 load address
+%define s2_size 2048            ; stage2 size
+%define s2_nsect s2_size / 512  ; stage2 size in sectors
 
-start:
+[BITS 16]
+global _start
+
+_start:
     cli                         ; Disable interrupts
 
     mov [bootdev], dl           ; Save boot device number (from BIOS in DL)
@@ -20,29 +26,16 @@ start:
     mov ds, ax
     mov es, ax
 
+    ; Query bios for disk parameters
+    call get_disk_params
+
     ; Load second-stage bootloader (boot1.asm) to 0x7E00
     mov ax, 1                   ; LBA of boot1.asm (starts at sector 1)
     call lba_to_chs
-    mov al, 4                   ; Number of sectors to read
+    mov al, s2_nsect            ; Number of sectors to read
     mov bx, 0x7E00              ; Destination address offset ES = 0 (0x0000:0x7E00)
     call read_chs
 
-    ; Load kernel to 0x100000 (1 MB)
-    mov ax, 0xffff
-    mov es, ax                  ; Set ES for destination address (0xffff << 4 == 0xffff0)
-    mov ax, 5                   ; LBA of kernel start (boot1 is 4 sectors: LBA 1–4 → kernel at LBA 5)
-    call lba_to_chs
-    mov al, 16                  ; Number of sectors for kernel
-    mov bx, 0x10                ; Destination address offset (0xffff:0x10 = 0xffff << 4 + 0x10 = 0x100000)
-    call read_chs
-    jc disk_error
-
-    ; Memory Validation: Verify checksum of second stage bootloader
-    mov si, 0x7E00             ; Start of second stage
-    mov cx, 512 * 4            ; Size in bytes (adjust if more sectors loaded)
-    call verify_checksum
-    jc disk_error              ; Jump if checksum fails
-
     ; Enable A20 line
     call enable_a20
     jc a20_error                ; Jump if A20 enable fails
@@ -95,17 +88,32 @@ verify_checksum:
     pop ax
     ret
 
+get_disk_params:
+    mov  ah, 08h          ; BIOS: Get Drive Parameters
+    int  13h
+    ; TODO: error checking
+
+    ; CL bits 0–5 contain sectors per track
+    mov  al, cl
+    and  al, 3Fh           ; mask bits 0–5
+    mov  ah, 0
+    mov  [sectors_per_track], ax
+
+    ; DH = maximum head number (0-based)
+    mov  al, dh
+    inc  ax                ; convert to count (heads = maxhead + 1)
+    mov  [heads_per_cylinder], ax
+    ret
+
 ; ----------------------------------------------------------------
 ; CHS Disk Read Routine
 ; AL = number of sectors
 ; CL = starting sector (1-based)
-; SI = destination offset (Segment:ES already set)
 ; Inputs:
 ; AL = sector count
 ; CH = cylinder
 ; DH = head
 ; CL = sector (1–63, with top 2 bits as high cylinder bits)
-; SI = destination offset (segment ES must be set)
 
 ; ----------------------------------------------------------------
 ; Convert LBA to CHS
@@ -117,36 +125,29 @@ verify_checksum:
 ;   CL = sector (1-63, top 2 bits are upper cylinder bits)
 
 lba_to_chs:
-    pusha
+    ; Sector
-
     xor dx, dx
-    mov bx, [sectors_per_track]
+    mov bx, ax
-    div bx                     ; AX = LBA / sectors_per_track, DX = remainder (sector number)
+    div word [sectors_per_track] ; divide lba with max sectors
-    mov si, ax                 ; SI = temp quotient (track index)
+    add dl, 1 ; take the remainder, sectors start at 1
-    mov cx, [heads_per_cylinder]
+    mov cl, dl ; sector is in cl
-    xor dx, dx
-    div cx                     ; AX = cylinder, DX = head
-    mov ch, al                 ; CH = cylinder low byte
-    mov dh, dl                 ; DH = head
 
-    ; Now take sector number from earlier remainder
+    ; Head
-    mov cx, si                 ; Copy track index to CX to access CL
+    mov ax, bx
-    and cl, 0x3F               ; Mask to 6 bits (sector number)
+    mov dx, 0
-    inc cl                     ; Sector numbers are 1-based
+    div word [sectors_per_track] ; divide lba with max sectors
+    mov dx, 0
+    div word [heads_per_cylinder] ; divide quotient with heads
+    mov dh, dl ; take the remainder, head is in dh
 
-    ; Insert upper 2 bits of cylinder into CL
+    ; Cylinder
-    mov ah, al                 ; AH = cylinder again
+    mov ch, al ; take the quotient, cylinder is in ch
-    and ah, 0xC0               ; Get top 2 bits of cylinder
-    or cl, ah                  ; OR them into sector byte
-
-    popa
     ret
 
 read_chs:
     pusha
     push dx
 
-    mov cx, 5
 .retry:
     mov ah, 0x02         ; BIOS: Read sectors
     mov dl, [bootdev]    ; Boot device
@@ -262,7 +263,7 @@ switch_to_pm:
     mov eax, cr0
     or eax, 1
     mov cr0, eax
-    jmp 0x08:0x7E00
+    jmp 0x08:0x7E00             ; jump to S2
 
 ; ----------------------------------------------------------------
 print_string_16:
@@ -284,8 +285,8 @@ halt:
     hlt
 
 bootdev db 0
-sectors_per_track dw 63
+sectors_per_track dw 0
-heads_per_cylinder dw 255
+heads_per_cylinder dw 0
 
 times 510 - ($ - $$) db 0
 dw 0xAA55

bootloader/stage2.asm

@@ -0,0 +1,26 @@
+[BITS 32]
+global _start
+global ata_lba_read
+
+extern load_kernel
+
+_start:
+    ; Set up segments
+    ; Data segments
+    mov ax, 0x10
+    mov ds, ax
+    mov es, ax
+    mov fs, ax
+    mov gs, ax
+    mov ss, ax
+
+    ; Code segment
+    mov ax, 0x08
+    mov cs, ax
+
+    ; Stack (must be identity-mapped)
+    mov esp, 0x90000
+
+    call load_kernel
+
+    jmp eax

bootloader/stage2.ld

@@ -0,0 +1,12 @@
+SECTIONS {
+    . = 0x7e00;
+
+    .text : { *(.text*) }
+    .rodata : { *(.rodata*) }
+    .data : { *(.data*) }
+
+    .bss : {
+        *(.bss*)
+        *(COMMON)
+    }
+}

bootloader/stage2_load.c

@@ -0,0 +1,235 @@
+#include <stddef.h>
+#include <stdint.h>
+
+// ATA IO Ports
+#define ATA_PRIMARY_DATA            0x1F0
+#define ATA_PRIMARY_ERR_FEATURES    0x1F1
+#define ATA_PRIMARY_SEC_COUNT       0x1F2
+#define ATA_PRIMARY_LBA_LOW         0x1F3
+#define ATA_PRIMARY_LBA_MID         0x1F4
+#define ATA_PRIMARY_LBA_HIGH        0x1F5
+#define ATA_PRIMARY_DRIVE_SEL       0x1F6
+#define ATA_PRIMARY_COMM_STAT       0x1F7
+
+// ATA Commands
+#define ATA_CMD_READ_PIO            0x20
+#define ATA_CMD_WRITE_PIO           0x30
+
+// ELF Ident indexes
+#define EI_NIDENT                   16
+
+// Program header types
+#define PT_NULL                     0
+#define PT_LOAD                     1
+
+// Disk sector size
+#define SECTOR_SIZE                 512
+#define PH_PER_SECTOR               (SECTOR_SIZE / sizeof(Elf32_Phdr))
+
+// Kernel start LBA
+#define KERN_START_SECT             5
+
+// VGA
+// Expects bios initialization for text mode (3), buffer at 0xb8000
+#define VGA_ADDRESS                 0xB8000
+#define VGA_COLS                    80
+#define VGA_ROWS                    25
+
+// ELF Header (32-bit)
+typedef struct {
+    uint8_t  e_ident[EI_NIDENT];
+    uint16_t e_type;
+    uint16_t e_machine;
+    uint32_t e_version;
+    uint32_t e_entry;     // Entry point
+    uint32_t e_phoff;     // Program header table offset
+    uint32_t e_shoff;     // Section header table offset
+    uint32_t e_flags;
+    uint16_t e_ehsize;
+    uint16_t e_phentsize;
+    uint16_t e_phnum;
+    uint16_t e_shentsize;
+    uint16_t e_shnum;
+    uint16_t e_shstrndx;
+} __attribute__((packed)) Elf32_Ehdr;
+
+// Program Header (32-bit)
+typedef struct {
+    uint32_t p_type;
+    uint32_t p_offset;
+    uint32_t p_vaddr;
+    uint32_t p_paddr;
+    uint32_t p_filesz;
+    uint32_t p_memsz;
+    uint32_t p_flags;
+    uint32_t p_align;
+} __attribute__((packed)) Elf32_Phdr;
+
+static inline uint8_t inb(uint16_t port)
+{
+    uint8_t ret;
+    __asm__ volatile ("inb %1, %0"
+                      : "=a"(ret)
+                      : "Nd"(port));
+    return ret;
+}
+
+static inline void outb(uint16_t port, uint8_t val)
+{
+    __asm__ volatile ("outb %0, %1"
+                      :
+                      : "a"(val), "Nd"(port));
+}
+
+static inline uint16_t inw(uint16_t port)
+{
+    uint16_t ret;
+    __asm__ volatile ("inw %1, %0"
+                      : "=a"(ret)
+                      : "Nd"(port));
+    return ret;
+}
+
+static inline void ata_wait_bsy() {
+    while (inb(ATA_PRIMARY_COMM_STAT) & 0x80);
+}
+
+static inline void ata_wait_drq() {
+    while (!(inb(ATA_PRIMARY_COMM_STAT) & 0x08));
+}
+
+static void ata_read_sector(void *addr, uint32_t lba) {
+    ata_wait_bsy();
+
+    outb(ATA_PRIMARY_DRIVE_SEL, 0xE0 | ((lba >> 24) & 0x0F));
+    outb(ATA_PRIMARY_SEC_COUNT, 1);
+    outb(ATA_PRIMARY_LBA_LOW, (uint8_t)lba);
+    outb(ATA_PRIMARY_LBA_MID, (uint8_t)(lba >> 8));
+    outb(ATA_PRIMARY_LBA_HIGH, (uint8_t)(lba >> 16));
+    outb(ATA_PRIMARY_COMM_STAT, ATA_CMD_READ_PIO);
+
+    uint16_t* ptr = (uint16_t*)addr;
+    ata_wait_bsy();
+    ata_wait_drq();
+    for (int i = 0; i < 256; i++) {
+        *ptr++ = inw(ATA_PRIMARY_DATA);
+    }
+}
+
+static void ata_read_sectors(uint8_t *addr, uint32_t offset, uint32_t size)
+{
+    // Reads are offset from the starting sector of the kernel
+    uint32_t lba = KERN_START_SECT + offset / SECTOR_SIZE;
+    uint32_t off = offset % 512;
+    uint8_t data[512];
+
+    while (size > 0) {
+        ata_read_sector(data, lba);
+
+        uint32_t copy = 512 - off;
+        if (copy > size) {
+            copy = size;
+        }
+
+        for (uint32_t i = 0; i < copy; i++) {
+            addr[i] = data[off + i];
+        }
+
+        addr += copy;
+        size -= copy;
+        lba++;
+        off = 0;
+    }
+}
+
+static void on_error(const char *msg)
+{
+    uint16_t *ptr = (uint16_t *)VGA_ADDRESS;
+
+    // Clear
+    uint16_t val = 0x0f00 | (uint8_t)' ';
+    for (size_t i = 0; i < VGA_COLS * VGA_ROWS; i++) {
+        ptr[i] = val;
+    }
+
+    // Print error
+    for (size_t i = 0; msg[i]; i++) {
+        ptr[i] = 0xf00 | (uint8_t)msg[i];
+    }
+
+    // Halt
+    while (1) {
+        __asm__("hlt");
+    }
+}
+
+// Load an ELF executable into memory.
+// NOTE: Only 32-byte program headers are supported.
+// Returns the entry point to the program.
+static void *elf_load(const void *data) {
+    const Elf32_Ehdr *header = (const Elf32_Ehdr*)data;
+
+    if (header->e_phentsize != sizeof(Elf32_Phdr)) {
+        // The bootloader only handles 32-byte program header entries
+        on_error("ERROR: Unsupported program header entry size, halting...");
+    }
+
+    // Buffer for the program headers
+    uint8_t file_buf[SECTOR_SIZE];
+
+    // Current file offset to the next program header
+    uint32_t file_offset = header->e_phoff;
+
+    for (int i = 0; i < header->e_phnum; i++) {
+        // Check for sector boundary.
+        // Program headers are read in a sector at a time
+        // 512 / 32 = 16 PH per sector
+        if (i % PH_PER_SECTOR == 0) {
+            uint32_t count = (header->e_phnum - i) * sizeof(Elf32_Phdr);
+            if (count > SECTOR_SIZE) {
+                count = SECTOR_SIZE;
+            }
+
+            // Reads
+            ata_read_sectors(file_buf, file_offset, count);
+            file_offset += count;
+        }
+
+        // PH being processed currently, index mod 16 as headers
+        // are being loaded in sector by sector.
+        const Elf32_Phdr *ph = (const Elf32_Phdr *)file_buf + (i % PH_PER_SECTOR);
+
+        // Discard non-load segments
+        if (ph->p_type != PT_LOAD)
+            continue;
+
+        // Load in the segment
+        uint32_t offset = ph->p_offset;
+        uint32_t filesz = ph->p_filesz;
+        uint32_t memsz = ph->p_memsz;
+        uint8_t *vaddr = (uint8_t *)ph->p_vaddr;
+        ata_read_sectors(vaddr, offset, filesz);
+
+        // Zero remaining BSS (if any)
+        if (memsz > filesz) {
+            uint8_t* bss_start = vaddr + filesz;
+            for (uint32_t j = 0; j < memsz - filesz; j++) {
+                bss_start[j] = 0;
+            }
+        }
+    }
+
+    // Return the entry point
+    return (void *)header->e_entry;
+}
+
+void *load_kernel(void) {
+    // ELF header buffer
+    uint8_t header_buf[SECTOR_SIZE];
+
+    // Read the first sector (contains the ELF header)
+    ata_read_sector(header_buf, KERN_START_SECT);
+
+    // `elf_load()` returns the entry point
+    return elf_load(header_buf);
+}

configure

@@ -0,0 +1,169 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Configuration
+TARGET="i386-elf"
+BINUTILS_VERSION="2.45"
+GCC_VERSION="15.2.0"
+
+# Paths
+SCRIPT_PATH="$(realpath "${BASH_SOURCE[0]}")"
+SCRIPT_DIR="$(dirname "$SCRIPT_PATH")"
+PREFIX="$SCRIPT_DIR/cross"
+SRC_DIR="$PREFIX/src"
+
+BINUTILS_SRC="$SRC_DIR/binutils-$BINUTILS_VERSION"
+BINUTILS_BUILD="$PREFIX/build-binutils"
+GCC_SRC="$SRC_DIR/gcc-$GCC_VERSION"
+GCC_BUILD="$PREFIX/build-gcc"
+
+# Flags
+DEBUG=0
+HELP=0
+
+# Parse arguments
+for arg in "$@"; do
+  case "$arg" in
+    -h|--help)
+      HELP=1
+      ;;
+    -d|--debug)
+      DEBUG=1
+      ;;
+    *)
+      echo "Unknown option: $arg"
+      echo "Use -h or --help for usage information"
+      exit 1
+      ;;
+  esac
+done
+
+# Show help
+if [[ "$HELP" -eq 1 ]]; then
+  cat << EOF
+Usage: $0 [OPTIONS]
+
+Build a cross-compiler toolchain for $TARGET.
+
+OPTIONS:
+  -h, --help     Show this help message
+  -d, --debug    Enable debug mode (set -x)
+
+This script will:
+  1. Download binutils $BINUTILS_VERSION and GCC $GCC_VERSION
+  2. Build and install them to: $PREFIX
+
+EOF
+  exit 0
+fi
+
+# Enable debug mode
+if [[ "$DEBUG" -eq 1 ]]; then
+  set -x
+fi
+
+# Print configuration
+cat << EOF
+
+=== Build Configuration ===
+Target       : $TARGET
+Prefix       : $PREFIX
+Binutils     : $BINUTILS_VERSION
+GCC          : $GCC_VERSION
+===========================
+
+EOF
+
+# Create directory structure
+echo "Setting up directories..."
+mkdir -p "$SRC_DIR"
+
+# Download sources
+cd "$SRC_DIR"
+
+if [[ ! -d "$BINUTILS_SRC" ]]; then
+  echo "Downloading binutils $BINUTILS_VERSION..."
+  wget "https://ftp.gnu.org/gnu/binutils/binutils-$BINUTILS_VERSION.tar.gz"
+  echo "Extracting binutils..."
+  tar xf "binutils-$BINUTILS_VERSION.tar.gz"
+  rm "binutils-$BINUTILS_VERSION.tar.gz"
+else
+  echo "Binutils source already exists, skipping download"
+fi
+
+if [[ ! -d "$GCC_SRC" ]]; then
+  echo "Downloading GCC $GCC_VERSION..."
+  wget "https://ftp.gnu.org/gnu/gcc/gcc-$GCC_VERSION/gcc-$GCC_VERSION.tar.gz"
+  echo "Extracting GCC..."
+  tar xf "gcc-$GCC_VERSION.tar.gz"
+  rm "gcc-$GCC_VERSION.tar.gz"
+else
+  echo "GCC source already exists, skipping download"
+fi
+
+# Download GCC prerequisites
+if [[ ! -d "$GCC_SRC/gmp" ]]; then
+  echo "Downloading GCC prerequisites..."
+  cd "$GCC_SRC"
+  ./contrib/download_prerequisites
+  cd "$SRC_DIR"
+else
+  echo "GCC prerequisites already downloaded, skipping"
+fi
+
+# Build binutils
+if [[ ! -f "$PREFIX/bin/$TARGET-ld" ]]; then
+  echo "Building binutils..."
+  mkdir -p "$BINUTILS_BUILD"
+  cd "$BINUTILS_BUILD"
+
+  "$BINUTILS_SRC/configure" \
+    --target="$TARGET" \
+    --prefix="$PREFIX" \
+    --with-sysroot \
+    --disable-nls \
+    --disable-werror
+
+  make -j"$(nproc)"
+  make install
+else
+  echo "Binutils already installed, skipping build"
+fi
+
+# Build GCC
+if [[ ! -f "$PREFIX/bin/$TARGET-gcc" ]]; then
+  echo "Building GCC..."
+  mkdir -p "$GCC_BUILD"
+  cd "$GCC_BUILD"
+
+  "$GCC_SRC/configure" \
+    --target="$TARGET" \
+    --prefix="$PREFIX" \
+    --disable-nls \
+    --enable-languages=c \
+    --without-headers
+
+  make all-gcc -j"$(nproc)"
+  make all-target-libgcc -j"$(nproc)"
+  make install-gcc
+  make install-target-libgcc
+else
+  echo "GCC already installed, skipping build"
+fi
+
+cd "$SCRIPT_DIR"
+
+# Generate .build.env file
+cat > .build.env << EOF
+# Generated by configure on $(date)
+# Source this file to add the cross-compiler toolchain to your PATH
+export PATH="$PREFIX/bin:\$PATH"
+EOF
+
+echo ""
+echo "=== Build Complete ==="
+echo "Toolchain installed to: $PREFIX"
+echo ""
+echo "To use the toolchain, run:"
+echo "  source .build.env"
+echo "======================"

gdb.txt

@@ -0,0 +1,6 @@
+target remote :1234
+add-symbol-file build/stage1.elf
+add-symbol-file build/stage2.elf
+add-symbol-file build/kernel.elf
+hbreak *0x7c00
+c

kernel/ata.c

@@ -0,0 +1,91 @@
+#include "ata.h"
+#include "io.h"
+#include "print.h"
+
+#define ATA_TIMEOUT 100000
+
+static inline void ata_delay(void) {
+    /* 400ns delay by reading alternate status */
+    inb(ATA_PRIMARY_CTRL);
+    inb(ATA_PRIMARY_CTRL);
+    inb(ATA_PRIMARY_CTRL);
+    inb(ATA_PRIMARY_CTRL);
+}
+
+static bool ata_wait(uint8_t mask) {
+    for (int i = 0; i < ATA_TIMEOUT; i++) {
+        uint8_t status = inb(ATA_PRIMARY_IO + ATA_REG_STATUS);
+        if (!(status & ATA_SR_BSY) && (status & mask))
+            return true;
+    }
+    return false;
+}
+
+bool ata_init(void) {
+    outb(ATA_PRIMARY_IO + ATA_REG_HDDEVSEL, 0xA0); // master
+    ata_delay();
+
+    outb(ATA_PRIMARY_IO + ATA_REG_COMMAND, ATA_CMD_IDENTIFY);
+    ata_delay();
+
+    uint8_t status = inb(ATA_PRIMARY_IO + ATA_REG_STATUS);
+    if (status == 0)
+        return false; // no drive
+
+    if (!ata_wait(ATA_SR_DRQ))
+        return false;
+
+    uint16_t identify[256];
+    for (int i = 0; i < 256; i++)
+        identify[i] = inw(ATA_PRIMARY_IO);
+
+    print_string("[ATA] Primary master detected\n");
+    return true;
+}
+
+bool ata_read_sector(uint32_t lba, uint8_t* buffer) {
+    if (!buffer) return false;
+
+    outb(ATA_PRIMARY_IO + ATA_REG_HDDEVSEL, 0xE0 | ((lba >> 24) & 0x0F));
+    outb(ATA_PRIMARY_IO + ATA_REG_SECCOUNT0, 1);
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA0, (uint8_t)(lba));
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA1, (uint8_t)(lba >> 8));
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA2, (uint8_t)(lba >> 16));
+    outb(ATA_PRIMARY_IO + ATA_REG_COMMAND, ATA_CMD_READ_PIO);
+
+    if (!ata_wait(ATA_SR_DRQ))
+        return false;
+
+    for (int i = 0; i < 256; i++) {
+        uint16_t data = inw(ATA_PRIMARY_IO);
+        buffer[i * 2]     = data & 0xFF;
+        buffer[i * 2 + 1] = data >> 8;
+    }
+
+    ata_delay();
+    return true;
+}
+
+bool ata_write_sector(uint32_t lba, const uint8_t* buffer) {
+    if (!buffer) return false;
+
+    outb(ATA_PRIMARY_IO + ATA_REG_HDDEVSEL, 0xE0 | ((lba >> 24) & 0x0F));
+    outb(ATA_PRIMARY_IO + ATA_REG_SECCOUNT0, 1);
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA0, (uint8_t)(lba));
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA1, (uint8_t)(lba >> 8));
+    outb(ATA_PRIMARY_IO + ATA_REG_LBA2, (uint8_t)(lba >> 16));
+    outb(ATA_PRIMARY_IO + ATA_REG_COMMAND, ATA_CMD_WRITE_PIO);
+
+    if (!ata_wait(ATA_SR_DRQ))
+        return false;
+
+    for (int i = 0; i < 256; i++) {
+        uint16_t word =
+            buffer[i * 2] |
+            (buffer[i * 2 + 1] << 8);
+        outw(ATA_PRIMARY_IO, word);
+    }
+
+    ata_delay();
+    return true;
+}

kernel/ata.h

@@ -0,0 +1,43 @@
+#ifndef ATA_H
+#define ATA_H
+
+#include <stdint.h>
+#include <stdbool.h>
+
+/* ATA I/O ports */
+#define ATA_PRIMARY_IO     0x1F0
+#define ATA_PRIMARY_CTRL   0x3F6
+
+/* ATA registers */
+#define ATA_REG_DATA       0x00
+#define ATA_REG_ERROR      0x01
+#define ATA_REG_FEATURES   0x01
+#define ATA_REG_SECCOUNT0  0x02
+#define ATA_REG_LBA0       0x03
+#define ATA_REG_LBA1       0x04
+#define ATA_REG_LBA2       0x05
+#define ATA_REG_HDDEVSEL   0x06
+#define ATA_REG_COMMAND    0x07
+#define ATA_REG_STATUS     0x07
+
+/* ATA commands */
+#define ATA_CMD_READ_PIO   0x20
+#define ATA_CMD_WRITE_PIO  0x30
+#define ATA_CMD_IDENTIFY   0xEC
+
+/* Status flags */
+#define ATA_SR_BSY         0x80
+#define ATA_SR_DRDY        0x40
+#define ATA_SR_DRQ         0x08
+#define ATA_SR_ERR         0x01
+
+/* Drive select */
+#define ATA_MASTER         0x00
+#define ATA_SLAVE          0x10
+
+/* Public API */
+bool ata_init(void);
+bool ata_read_sector(uint32_t lba, uint8_t* buffer);
+bool ata_write_sector(uint32_t lba, const uint8_t* buffer);
+
+#endif

kernel/context_switch.asm

@@ -0,0 +1,25 @@
+global ctx_switch
+
+; void ctx_switch(uint32_t **old_sp_ptr, uint32_t *new_sp);
+; Arguments on stack (cdecl convention):
+; [ESP + 4] -> old_sp_ptr (pointer to the 'stack_ptr' field of current task)
+; [ESP + 8] -> new_sp     (value of 'stack_ptr' of the next task)
+
+ctx_switch:
+    ; 1. Save the context of the CURRENT task
+    pushf               ; Save EFLAGS (CPU status flags)
+    pusha               ; Save all General Purpose Regs (EAX, ECX, EDX, EBX, ESP, EBP, ESI, EDI)
+
+    ; 2. Save the current stack pointer (ESP) into the pointer passed as 1st arg
+    mov eax, [esp + 40] ; Get 1st argument (old_sp_ptr). Offset 40 = 36 (regs) + 4 (ret addr)
+    mov [eax], esp      ; *old_sp_ptr = ESP
+
+    ; 3. Load the stack pointer of the NEW task
+    mov esp, [esp + 44] ; Get 2nd argument (new_sp). Offset 44 = 40 + 4
+
+    ; 4. Restore the context of the NEW task
+    popa                ; Restore all General Purpose Regs
+    popf                ; Restore EFLAGS
+
+    ; 5. Jump to the new task (The 'ret' pops EIP from the new stack)
+    ret

kernel/cpu.c

@@ -2,36 +2,106 @@
 #include "serial.h"
 #include "terminal.h"
 #include "utils.h"
-#include "print.h"
 
-void cpuid(uint32_t function, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) {
+void cpuid(uint32_t leaf, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx) {
     __asm__(
         "cpuid"
         : "=a"(*eax), "=b"(*ebx), "=c"(*ecx), "=d"(*edx)
-        : "a"(function)
+        : "a"(leaf)
     );
 }
 
+// Helper to print a labeled decimal value
+void print_val(const char* label, uint32_t val) {
+    char buf[12];
+    utoa(val, buf, 10);
+    terminal_write(label);
+    terminal_write(buf);
+    terminal_write(" ");
+}
+
+// Safely check if CPUID is supported by attempting to flip bit 21 of EFLAGS
+int check_cpuid_supported() {
+    uint32_t f1, f2;
+    __asm__ volatile (
+        "pushfl\n\t"
+        "pushfl\n\t"
+        "popl %0\n\t"
+        "movl %0, %1\n\t"
+        "xorl $0x200000, %0\n\t"
+        "pushl %0\n\t"
+        "popfl\n\t"
+        "pushfl\n\t"
+        "popl %0\n\t"
+        "popfl\n\t"
+        : "=&r" (f1), "=&r" (f2));
+    return ((f1 ^ f2) & 0x200000) != 0;
+}
+
 void identify_cpu() {
+    if (!check_cpuid_supported()) {
+        terminal_write("CPUID not supported. Likely a 386 or early 486.\n");
+        return;
+    }
+
     uint32_t eax, ebx, ecx, edx;
     char vendor[13];
 
+    // Leaf 0: Vendor String & Max Leaf
     cpuid(0, &eax, &ebx, &ecx, &edx);
-
+    uint32_t max_leaf = eax;
     *(uint32_t *)&vendor[0] = ebx;
     *(uint32_t *)&vendor[4] = edx;
     *(uint32_t *)&vendor[8] = ecx;
     vendor[12] = '\0';
 
-    terminal_write("CPU Vendor: ");
+    terminal_write("Vendor: ");
     terminal_write(vendor);
     terminal_write("\n");
 
-    serial_write("CPU Vendor: ");
+    // Leaf 1: Family, Model, Stepping
-    serial_write(vendor);
+    if (max_leaf >= 1) {
-    serial_write("\n");
+        cpuid(1, &eax, &ebx, &ecx, &edx);
         
-    terminal_write("CPUID max leaf: ");
+        uint32_t stepping = eax & 0xF;
-    print_hex(eax, false, false);  // You must implement this (see below)
+        uint32_t model = (eax >> 4) & 0xF;
-    terminal_write("\n");
+        uint32_t family = (eax >> 8) & 0xF;
+        uint32_t type = (eax >> 12) & 0x3;
+
+        // Handle Extended Family/Model (Required for Pentium 4 and newer)
+        if (family == 0xF) {
+            family += (eax >> 20) & 0xFF;
+            model += ((eax >> 16) & 0xF) << 4;
+        }
+
+        print_val("Family:", family);
+        print_val("Model:", model);
+        print_val("Step:", stepping);
+        terminal_write("\n");
+    }
+
+    // Leaf 2: Cache Descriptors
+    if (max_leaf >= 2) {
+        cpuid(2, &eax, &ebx, &ecx, &edx);
+        
+        terminal_write("Cache Descriptors: ");
+        // Note: Leaf 2 returns a list of 1-byte descriptors in the registers.
+        // We look for common Intel ones:
+        uint32_t regs[4] = {eax, ebx, ecx, edx};
+        for (int i = 0; i < 4; i++) {
+            if (regs[i] & 0x80000000) continue; // Reserved bit
+            for (int j = 0; j < 4; j++) {
+                uint8_t desc = (regs[i] >> (j * 8)) & 0xFF;
+                if (desc == 0) continue;
+                
+                // Example decoding for specific chips you mentioned:
+                if (desc == 0x06) terminal_write("8KB L1 I-Cache ");
+                if (desc == 0x0A) terminal_write("8KB L1 D-Cache ");
+                if (desc == 0x41) terminal_write("128KB L2 ");
+                if (desc == 0x43) terminal_write("512KB L2 ");
+                if (desc == 0x2C) terminal_write("32KB L1 D-Cache ");
+            }
+        }
+        terminal_write("\n");
+    }
 }

kernel/cpu.h

@@ -2,8 +2,42 @@
 #define CPU_H
 
 #include <stdint.h>
+#include <stdbool.h>
 
-void cpuid(uint32_t function, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx);
+// Specific Intel Model Definitions for your targets
+#define INTEL_FAM4_486_DX      0x00 // Also 0x01
+#define INTEL_FAM4_486_SX      0x02
+#define INTEL_FAM4_486_DX2     0x03
+#define INTEL_FAM4_486_DX4     0x08
+#define INTEL_FAM5_PENTIUM     0x01 // P5
+#define INTEL_FAM5_PENTIUM_MMX 0x04 // P55C
+#define INTEL_FAM6_PENTIUM_PRO 0x01 // P6
+#define INTEL_FAM6_PENTIUM_II  0x05 // Deschutes
+#define INTEL_FAM6_PENTIUM_III 0x07 // Katmai/Coppermine
+#define INTEL_FAM15_P4_WILLY   0x00 // Willamette
+#define INTEL_FAM15_P4_NORTH   0x02 // Northwood
+#define INTEL_FAM15_P4_PRES    0x03 // Prescott
+
+typedef struct {
+    char vendor[13];
+    uint32_t family;
+    uint32_t model;
+    uint32_t stepping;
+    uint32_t type;
+    uint32_t max_leaf;
+    
+    // Feature flags (optional, but very helpful later)
+    bool has_fpu;
+    bool has_mmx;
+    bool has_sse;
+} cpu_info_t;
+
+// Function Prototypes
+void cpuid(uint32_t leaf, uint32_t *eax, uint32_t *ebx, uint32_t *ecx, uint32_t *edx);
+bool cpu_check_cpuid_support(void);
 void identify_cpu(void);
 
+// Helper to get the current CPU info after identification
+cpu_info_t* cpu_get_info(void);
+
 #endif // CPU_H

kernel/fat12.c

@@ -1,5 +1,196 @@
 #include "fat12.h"
+#include <stddef.h> // for NULL
+
+// --- Globals for Filesystem State ---
+static fat12_bpb_t bpb;
+static uint32_t fat_start_lba;
+static uint32_t root_dir_lba;
+static uint32_t data_start_lba;
+static uint32_t root_dir_sectors;
+
+// Scratch buffer to read sectors (avoids large stack usage)
+static uint8_t g_sector_buffer[FAT12_SECTOR_SIZE];
+
+// --- Utils (Since we don't have string.h) ---
+static int k_memcmp(const void *s1, const void *s2, uint32_t n) {
+    const uint8_t *p1 = (const uint8_t *)s1;
+    const uint8_t *p2 = (const uint8_t *)s2;
+
+    for (uint32_t i = 0; i < n; i++) {
+        if (p1[i] != p2[i]) {
+            // Correct way to return the difference:
+            // If p1[i] > p2[i], returns positive.
+            // If p1[i] < p2[i], returns negative.
+            return (int)p1[i] - (int)p2[i];
+        }
+    }
+
+    return 0;
+}
+
+// Converts "file.txt" to "FILE    TXT" for comparison
+static void to_fat_name(const char *src, char *dest) {
+    // Initialize with spaces
+    for(int i=0; i<11; i++) dest[i] = ' ';
+
+    int i = 0, j = 0;
+    // Copy Name
+    while (src[i] != '\0' && src[i] != '.' && j < 8) {
+        // Convert to uppercase (simple version)
+        char c = src[i];
+        if (c >= 'a' && c <= 'z') c -= 32; 
+        dest[j++] = c;
+        i++;
+    }
+    
+    // Skip extension dot
+    if (src[i] == '.') i++;
+
+    // Copy Extension
+    j = 8;
+    while (src[i] != '\0' && j < 11) {
+        char c = src[i];
+        if (c >= 'a' && c <= 'z') c -= 32;
+        dest[j++] = c;
+        i++;
+    }
+}
+
+// --- Core Logic ---
 
 void fat12_init() {
-    // Filesystem initialization code
+    // 1. Read Boot Sector (LBA 0)
+    disk_read_sector(0, g_sector_buffer);
+
+    // 2. Copy BPB data safely
+    // We cast the buffer to our struct
+    fat12_bpb_t *boot_sector = (fat12_bpb_t*)g_sector_buffer;
+    bpb = *boot_sector;
+
+    // 3. Calculate System Offsets
+    fat_start_lba  = bpb.reserved_sectors;
+    
+    // Root Dir starts after FATs
+    // LBA = Reserved + (FatCount * SectorsPerFat)
+    root_dir_lba   = fat_start_lba + (bpb.fat_count * bpb.sectors_per_fat);
+    
+    // Calculate size of Root Directory in sectors
+    // (Entries * 32 bytes) / 512
+    root_dir_sectors = (bpb.dir_entries_count * 32 + FAT12_SECTOR_SIZE - 1) / FAT12_SECTOR_SIZE;
+
+    // Data starts after Root Directory
+    data_start_lba = root_dir_lba + root_dir_sectors;
+}
+
+// Helper: Read the FAT table to find the NEXT cluster
+static uint16_t fat12_get_next_cluster(uint16_t current_cluster) {
+    // FAT12 Offset Calculation:
+    // Offset = Cluster + (Cluster / 2)
+    uint32_t fat_offset = current_cluster + (current_cluster / 2);
+    
+    uint32_t fat_sector = fat_start_lba + (fat_offset / FAT12_SECTOR_SIZE);
+    uint32_t ent_offset = fat_offset % FAT12_SECTOR_SIZE;
+
+    // Read the sector containing the FAT entry
+    disk_read_sector(fat_sector, g_sector_buffer);
+
+    // Read 16 bits (2 bytes)
+    // Note: If ent_offset == 511, the entry spans two sectors. 
+    // For simplicity in this snippet, we ignore that edge case (rare).
+    // A robust kernel would check if(ent_offset == 511) and read next sector.
+    
+    uint16_t val = *(uint16_t*)&g_sector_buffer[ent_offset];
+
+    if (current_cluster & 1) {
+        return val >> 4;      // Odd: High 12 bits
+    } else {
+        return val & 0x0FFF;  // Even: Low 12 bits
+    }
+}
+
+file_t fat12_open(const char *filename) {
+    file_t file = {0};
+    char target_name[11];
+    to_fat_name(filename, target_name);
+
+    // Search Root Directory
+    for (uint32_t i = 0; i < root_dir_sectors; i++) {
+        disk_read_sector(root_dir_lba + i, g_sector_buffer);
+
+        fat12_entry_t *entry = (fat12_entry_t*)g_sector_buffer;
+
+        // Check all 16 entries in this sector (512 / 32 = 16)
+        for (int j = 0; j < 16; j++) {
+            if (entry[j].filename[0] == 0x00) return file; // End of Dir
+            
+            // Check if filename matches
+            if (k_memcmp(entry[j].filename, target_name, 11) == 0) {
+                // Found it!
+                file.start_cluster = entry[j].low_cluster_num;
+                file.size = entry[j].file_size;
+                
+                // Initialize file cursor
+                file.current_cluster = file.start_cluster;
+                file.bytes_read = 0;
+                return file;
+            }
+        }
+    }
+    
+    // Not found (file.start_cluster will be 0)
+    return file;
+}
+
+uint32_t fat12_read(file_t *file, uint8_t *buffer, uint32_t bytes_to_read) {
+    if (file->start_cluster == 0) return 0; // File not open
+
+    uint32_t total_read = 0;
+
+    while (bytes_to_read > 0) {
+        // Check for EOF marker in FAT12 (>= 0xFF8)
+        if (file->current_cluster >= 0xFF8) break;
+
+        // Calculate Physical LBA of current cluster
+        // LBA = DataStart + ((Cluster - 2) * SectorsPerCluster)
+        uint32_t lba = data_start_lba + ((file->current_cluster - 2) * bpb.sectors_per_cluster);
+
+        // Read the cluster
+        // NOTE: Assumes SectorsPerCluster = 1 (Standard Floppy)
+        disk_read_sector(lba, g_sector_buffer);
+
+        // Determine how much to copy from this sector
+        uint32_t chunk_size = FAT12_SECTOR_SIZE;
+        
+        // If the file is smaller than a sector, or we are at the end
+        if (chunk_size > bytes_to_read) chunk_size = bytes_to_read;
+        
+        // Check if we are reading past file size
+        if (file->bytes_read + chunk_size > file->size) {
+            chunk_size = file->size - file->bytes_read;
+        }
+
+        // Copy to user buffer
+        for (uint32_t i = 0; i < chunk_size; i++) {
+            buffer[total_read + i] = g_sector_buffer[i];
+        }
+
+        total_read += chunk_size;
+        file->bytes_read += chunk_size;
+        bytes_to_read -= chunk_size;
+
+        // If we finished this cluster, move to the next one
+        if (chunk_size == FAT12_SECTOR_SIZE) { // Or strictly logic based on position
+             file->current_cluster = fat12_get_next_cluster(file->current_cluster);
+        } else {
+            // We finished the file or the request
+            break;
+        }
+    }
+
+    return total_read;
+}
+
+int disk_read_sector(uint32_t lba, uint8_t *buffer) {
+    // For now, do nothing and return success
+    return 0; 
 }

kernel/fat12.h

@@ -1,47 +1,67 @@
 #ifndef FAT12_H
 #define FAT12_H
 
-#include <stdint.h> /* Include standard integer types */
+#include <stdint.h>
-#include <stdio.h>  /* Include standard I/O library */
-#include <stdlib.h> /* Include standard library */
 
-#define FAT12_SECTOR_SIZE 512 /* Sector size for FAT12 */
+// --- Configuration ---
-#define FAT12_MAX_FILES 128 /* Maximum number of files in root directory */
+#define FAT12_SECTOR_SIZE 512
-#define FAT12_ROOT_DIR_SECTORS 1 /* Number of sectors for root directory */
 
+// --- On-Disk Structures (Must be Packed) ---
+
+// BIOS Parameter Block (Start of Boot Sector)
 typedef struct {
-    uint8_t jump[3]; /* Jump instruction for boot */
+    uint8_t  jump[3];
-    char oem[8]; /* OEM name */
+    char     oem[8];
-    uint16_t bytes_per_sector; /* Bytes per sector */
+    uint16_t bytes_per_sector;    // 512
-    uint8_t sectors_per_cluster; /* Sectors per cluster */
+    uint8_t  sectors_per_cluster; // 1
-    uint16_t reserved_sectors; /* Reserved sectors count */
+    uint16_t reserved_sectors;    // 1 (Boot sector)
-    uint8_t num_fats; /* Number of FATs */
+    uint8_t  fat_count;           // 2
-    uint16_t max_root_dir_entries; /* Max entries in root directory */
+    uint16_t dir_entries_count;   // 224
-    uint16_t total_sectors; /* Total sectors */
+    uint16_t total_sectors;       // 2880
-    uint8_t media_descriptor; /* Media descriptor */
+    uint8_t  media_descriptor;    // 0xF0
-    uint16_t fat_size; /* Size of each FAT */
+    uint16_t sectors_per_fat;     // 9
-    uint16_t sectors_per_track; /* Sectors per track */
+    uint16_t sectors_per_track;   // 18
-    uint16_t num_heads; /* Number of heads */
+    uint16_t heads;               // 2
-    uint32_t hidden_sectors; /* Hidden sectors count */
+    uint32_t hidden_sectors;
-    uint32_t total_sectors_large; /* Total sectors for large disks */
+    uint32_t total_sectors_large;
-} __attribute__((packed)) FAT12_BootSector; /* Packed structure for boot sector */
+} __attribute__((packed)) fat12_bpb_t;
 
+// Directory Entry (32 bytes)
 typedef struct {
-    char name[11]; /* File name (8.3 format) */
+    char     filename[8];
-    uint8_t attr; /* File attributes */
+    char     ext[3];
-    uint16_t reserved; /* Reserved */
+    uint8_t  attributes;
-    uint16_t time; /* Time of last write */
+    uint8_t  reserved;
-    uint16_t date; /* Date of last write */
+    uint8_t  creation_ms;
-    uint16_t start_cluster; /* Starting cluster number */
+    uint16_t creation_time;
-    uint32_t file_size; /* File size in bytes */
+    uint16_t creation_date;
-} __attribute__((packed)) FAT12_DirEntry; /* Directory entry structure */
+    uint16_t last_access_date;
+    uint16_t high_cluster_num; // Always 0 in FAT12
+    uint16_t last_mod_time;
+    uint16_t last_mod_date;
+    uint16_t low_cluster_num;  // The starting cluster
+    uint32_t file_size;        // Size in bytes
+} __attribute__((packed)) fat12_entry_t;
 
-void initialize_fat12(const char *disk_image); /* Function to initialize FAT12 */
+// --- Kernel File Handle ---
-void read_fat12(const char *disk_image); /* Function to read FAT12 */
+// This is what your kernel uses to track an open file
-void write_fat12(const char *disk_image); /* Function to write FAT12 */
+typedef struct {
-void list_files(const char *disk_image); /* Function to list files in root directory */
+    char     name[11];
-void read_file(const char *disk_image, const char *filename); /* Function to read a file */
+    uint32_t size;
-void write_file(const char *disk_image, const char *filename, const uint8_t *data, size_t size); /* Function to write a file */
+    uint16_t start_cluster;
+    uint16_t current_cluster;
+    uint32_t current_sector_in_cluster;
+    uint32_t bytes_read;
+} file_t;
 
-#endif 
+// --- Public API ---
-/* FAT12_H */
+
+// You must implement this in your disk driver (e.g., floppy.c)
+// Returns 0 on success, non-zero on error.
+int disk_read_sector(uint32_t lba, uint8_t *buffer);
+
+void fat12_init();
+file_t fat12_open(const char *filename);
+uint32_t fat12_read(file_t *file, uint8_t *buffer, uint32_t bytes_to_read);
+
+#endif // FAT12_H

kernel/framebuffer.c

@@ -0,0 +1,92 @@
+#include "framebuffer.h"
+#include <stddef.h>
+#include <stdint.h>
+
+// Simple init
+void framebuffer_init(framebuffer_t *fb, void *base, uint32_t width, uint32_t height, uint32_t pitch, uint8_t bpp) {
+    fb->base = base;
+    fb->width = width;
+    fb->height = height;
+    fb->pitch = pitch;
+    fb->bpp = bpp;
+    fb->initialized = true;
+}
+
+// Pack color into 32-bit value. Format: 0xAARRGGBB
+uint32_t framebuffer_pack_color(uint8_t r, uint8_t g, uint8_t b, uint8_t a) {
+    return ((uint32_t)a << 24) | ((uint32_t)r << 16) | ((uint32_t)g << 8) | (uint32_t)b;
+}
+
+void framebuffer_put_pixel(framebuffer_t *fb, uint32_t x, uint32_t y, uint8_t r, uint8_t g, uint8_t b, uint8_t a) {
+    if (!fb->initialized) return;
+    if (x >= fb->width || y >= fb->height) return;
+    if (fb->bpp != 32) return; // only 32bpp implemented here
+
+    uint8_t *line = (uint8_t*)fb->base + (size_t)y * fb->pitch;
+    uint32_t *pixel = (uint32_t*)(line + x * 4);
+    *pixel = framebuffer_pack_color(r, g, b, a);
+}
+
+void framebuffer_clear(framebuffer_t *fb, uint8_t r, uint8_t g, uint8_t b) {
+    if (!fb->initialized) return;
+    if (fb->bpp != 32) return;
+
+    uint32_t color = framebuffer_pack_color(r,g,b,0xFF);
+    for (uint32_t y = 0; y < fb->height; ++y) {
+        uint32_t *row = (uint32_t*)((uint8_t*)fb->base + (size_t)y * fb->pitch);
+        for (uint32_t x = 0; x < fb->width; ++x) {
+            row[x] = color;
+        }
+    }
+}
+
+void framebuffer_fill_rect(framebuffer_t *fb, uint32_t x, uint32_t y, uint32_t w, uint32_t h, uint8_t r, uint8_t g, uint8_t b) {
+    if (!fb->initialized) return;
+    if (fb->bpp != 32) return;
+
+    if (x >= fb->width || y >= fb->height) return;
+    if (x + w > fb->width) w = fb->width - x;
+    if (y + h > fb->height) h = fb->height - y;
+
+    uint32_t color = framebuffer_pack_color(r,g,b,0xFF);
+    for (uint32_t yy = 0; yy < h; ++yy) {
+        uint32_t *row = (uint32_t*)((uint8_t*)fb->base + (size_t)(y + yy) * fb->pitch) + x;
+        for (uint32_t xx = 0; xx < w; ++xx) {
+            row[xx] = color;
+        }
+    }
+}
+
+// Simple blit from a source buffer with 32bpp pixels and given pitch (bytes per line)
+void framebuffer_blit(framebuffer_t *fb, uint32_t dst_x, uint32_t dst_y, const void *src, uint32_t src_w, uint32_t src_h, uint32_t src_pitch) {
+    if (!fb->initialized) return;
+    if (fb->bpp != 32) return;
+
+    if (dst_x >= fb->width || dst_y >= fb->height) return;
+
+    uint32_t copy_w = src_w;
+    uint32_t copy_h = src_h;
+    if (dst_x + copy_w > fb->width) copy_w = fb->width - dst_x;
+    if (dst_y + copy_h > fb->height) copy_h = fb->height - dst_y;
+
+    const uint8_t *s = (const uint8_t*)src;
+    for (uint32_t yy = 0; yy < copy_h; ++yy) {
+        uint32_t *dst_row = (uint32_t*)((uint8_t*)fb->base + (size_t)(dst_y + yy) * fb->pitch) + dst_x;
+        const uint32_t *src_row = (const uint32_t*)(s + (size_t)yy * src_pitch);
+        for (uint32_t xx = 0; xx < copy_w; ++xx) {
+            dst_row[xx] = src_row[xx];
+        }
+    }
+}
+
+void framebuffer_test_pattern(framebuffer_t *fb) {
+    if (!fb->initialized) return;
+    // simple color bars
+    uint32_t band_h = fb->height / 6;
+    framebuffer_fill_rect(fb, 0, 0, fb->width, band_h, 0xFF, 0x00, 0x00); // red
+    framebuffer_fill_rect(fb, 0, band_h, fb->width, band_h, 0x00, 0xFF, 0x00); // green
+    framebuffer_fill_rect(fb, 0, band_h*2, fb->width, band_h, 0x00, 0x00, 0xFF); // blue
+    framebuffer_fill_rect(fb, 0, band_h*3, fb->width, band_h, 0xFF, 0xFF, 0x00); // yellow
+    framebuffer_fill_rect(fb, 0, band_h*4, fb->width, band_h, 0xFF, 0x00, 0xFF); // magenta
+    framebuffer_fill_rect(fb, 0, band_h*5, fb->width, fb->height - band_h*5, 0x00, 0xFF, 0xFF); // cyan
+}

kernel/framebuffer.h

@@ -0,0 +1,25 @@
+#ifndef FRAMEBUFFER_H
+#define FRAMEBUFFER_H
+
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+typedef struct {
+    void *base;
+    uint32_t width;
+    uint32_t height;
+    uint32_t pitch;
+    uint8_t bpp;
+    bool initialized;
+} framebuffer_t;
+
+void framebuffer_init(framebuffer_t *fb, void *base, uint32_t width, uint32_t height, uint32_t pitch, uint8_t bpp);
+uint32_t framebuffer_pack_color(uint8_t r, uint8_t g, uint8_t b, uint8_t a);
+void framebuffer_put_pixel(framebuffer_t *fb, uint32_t x, uint32_t y, uint8_t r, uint8_t g, uint8_t b, uint8_t a);
+void framebuffer_clear(framebuffer_t *fb, uint8_t r, uint8_t g, uint8_t b);
+void framebuffer_fill_rect(framebuffer_t *fb, uint32_t x, uint32_t y, uint32_t w, uint32_t h, uint8_t r, uint8_t g, uint8_t b);
+void framebuffer_blit(framebuffer_t *fb, uint32_t dst_x, uint32_t dst_y, const void *src, uint32_t src_w, uint32_t src_h, uint32_t src_pitch);
+void framebuffer_test_pattern(framebuffer_t *fb);
+
+#endif /* FRAMEBUFFER_H */

kernel/io.h

@@ -13,4 +13,24 @@ static inline uint8_t inb(uint16_t port) {
     return ret;
 }
 
+static inline void outw(uint16_t port, uint16_t val) {
+    __asm__("outw %0, %1" : : "a"(val), "Nd"(port));
+}
+
+static inline uint16_t inw(uint16_t port) {
+    uint16_t ret;
+    __asm__("inw %1, %0" : "=a"(ret) : "Nd"(port));
+    return ret;
+}
+
+static inline void outl(uint16_t port, uint32_t val) {
+    __asm__("outl %0, %1" : : "a"(val), "Nd"(port));
+}
+
+static inline uint32_t inl(uint16_t port) {
+    uint32_t ret;
+    __asm__("inl %1, %0" : "=a"(ret) : "Nd"(port));
+    return ret;
+}
+
 #endif

kernel/irq.c

@@ -1,5 +1,76 @@
+#include "idt.h"
 #include "irq.h"
+#include "io.h"
+#include "isr.h"
 
-void irq_init() {
+#define PIC1_CMD   0x20
-    // IRQ initialization code
+#define PIC1_DATA  0x21
+#define PIC2_CMD   0xA0
+#define PIC2_DATA  0xA1
+
+// FIXME: stubs
+void irq0() {}
+void irq1() {}
+void irq2() {}
+void irq3() {}
+void irq4() {}
+void irq5() {}
+void irq6() {}
+void irq7() {}
+void irq8() {}
+void irq9() {}
+void irq10() {}
+void irq11() {}
+void irq12() {}
+void irq13() {}
+void irq14() {}
+void irq15() {}
+// --- stubs end
+
+void irq_remap(void)
+{
+    outb(PIC1_CMD, 0x11);   // ICW1 – edge triggered, cascade, need ICW4
+    outb(PIC2_CMD, 0x11);
+
+    outb(PIC1_DATA, 0x20);  // ICW2 – master base vector
+    outb(PIC2_DATA, 0x28);  // ICW2 – slave base vector
+
+    outb(PIC1_DATA, 0x04);  // ICW3 – slave on IRQ2
+    outb(PIC2_DATA, 0x02);  // ICW3 – cascade identity
+
+    outb(PIC1_DATA, 0x01);  // ICW4 – 8086 mode
+    outb(PIC2_DATA, 0x01);
+
+    // Mask everything except IRQ0 (timer) and IRQ1 (keyboard) for now
+    outb(PIC1_DATA, 0b11111001);
+    outb(PIC2_DATA, 0xFF);
+}
+
+void irq_install(void)
+{
+    irq_remap();
+
+    /* Fill IRQ entries in the IDT (0x20 … 0x2F) */
+    //extern void irq0(), irq1(), irq2(), irq3(), irq4(), irq5(), irq6(), irq7();
+    //extern void irq8(), irq9(), irq10(), irq11(), irq12(), irq13(), irq14(), irq15();
+
+    idt_set_gate(0x20, (uint32_t)irq0);
+    idt_set_gate(0x21, (uint32_t)irq1);
+    /* … repeat for the rest or loop … */
+    for (int i = 2; i < 16; ++i)
+        idt_set_gate(0x20 + i, (uint32_t)irq0 + i * 8); // crude but works
+}
+
+/* Called from the assembly stubs (see irq.asm below) */
+void irq_handler(uint32_t int_num)
+{
+    /* int_num is the *remapped* vector, e.g. 0x21 for keyboard */
+    if (interrupt_handlers[int_num]) {
+        interrupt_handlers[int_num]();
+    }
+
+    /* ---- EOI ---- */
+    if (int_num >= 0x28)          // slave PIC
+        outb(PIC2_CMD, 0x20);
+    outb(PIC1_CMD, 0x20);         // always master
 }

kernel/irq.h

@@ -1,6 +1,10 @@
 #ifndef IRQ_H
 #define IRQ_H
 
-void irq_init();
+#include <stdint.h>
 
-#endif // IRQ_H
+void irq_remap(void);
+void irq_install(void);
+void irq_handler(uint32_t int_num);
+
+#endif

kernel/isr.c

@@ -1,10 +1,11 @@
+#include <stdbool.h>
 #include "terminal.h"
 #include "serial.h"
 #include "isr.h"
 #include "io.h"
 #include "print.h"
 
-static isr_callback_t interrupt_handlers[MAX_INTERRUPTS] = { 0 };
+isr_callback_t interrupt_handlers[MAX_INTERRUPTS] = { 0 };
 
 void isr_handler(uint32_t int_num, uint32_t err_code) {
     terminal_write("Interrupt occurred: ");

kernel/isr.h

@@ -6,6 +6,7 @@
 #define MAX_INTERRUPTS 256
 
 typedef void (*isr_callback_t)(void);
+extern isr_callback_t interrupt_handlers[MAX_INTERRUPTS];
 
 void isr_handler(uint32_t int_num, uint32_t err_code);
 void register_interrupt_handler(uint8_t n, isr_callback_t handler);

kernel/keyboard.c

@@ -4,8 +4,12 @@
 #include "terminal.h"
 
 #define KEYBOARD_DATA_PORT 0x60
+#define KEY_BUFFER_SIZE 256
 
-static char key_buffer[256];
+static char key_buffer[KEY_BUFFER_SIZE];
+static uint8_t buffer_head = 0;  // Write position (interrupt)
+static uint8_t buffer_tail = 0;  // Read position (get_char)
+static uint8_t buffer_count = 0;
 static uint8_t buffer_index = 0;
 
 // Basic US QWERTY keymap (scancode to ASCII)
@@ -21,39 +25,41 @@ static const char scancode_map[128] = {
 
 // Interrupt handler for IRQ1
 void keyboard_callback(void) {
-    uint8_t scancode = inb(0x60);
+    uint8_t scancode = inb(KEYBOARD_DATA_PORT);
 
-    // Only handle key press (ignore key release)
+    if (scancode & 0x80) return;  // Ignore key release
-    if (!(scancode & 0x80)) {
-        char c = scancode_map[scancode];
-        if (c && buffer_index < sizeof(key_buffer) - 1) {
-            key_buffer[buffer_index++] = c;
-            terminal_putchar(c);
-        }
-    }
 
-    // Send End of Interrupt (EOI) to the PIC
+    char c = scancode_map[scancode];
-    outb(0x20, 0x20);
+    if (!c) return;
+
+    uint8_t next_head = (buffer_head + 1) % KEY_BUFFER_SIZE;
+
+    // Drop key if buffer full
+    if (next_head == buffer_tail) return;
+
+    key_buffer[buffer_head] = c;
+    buffer_head = next_head;
+    buffer_count++;
+
+    terminal_putchar(c);
 }
 
-
 void keyboard_init() {
     register_interrupt_handler(33, keyboard_callback); // IRQ1 = int 33 (0x21)
 }
 
 // Blocking read (returns one char)
-char keyboard_get_char() {
+char keyboard_get_char(void) {
-    while (buffer_index == 0); // Busy wait
+    while (buffer_count == 0) {
+        __asm__ __volatile__("hlt");  // Better than busy loop
+    }
 
     char c;
     __asm__ __volatile__("cli");
-    c = key_buffer[0];
+    c = key_buffer[buffer_tail];
-    for (uint8_t i = 1; i < buffer_index; i++) {
+    buffer_tail = (buffer_tail + 1) % KEY_BUFFER_SIZE;
-        key_buffer[i - 1] = key_buffer[i];
+    buffer_count--;
-    }
-    buffer_index--;
     __asm__ __volatile__("sti");
 
     return c;
 }
-

kernel/kmain.c

@@ -12,6 +12,7 @@
 #include "timer.h"
 #include "utils.h"
 #include "keyboard.h"
+#include "irq.h"
 
 #define LPT1 0x378
 
@@ -41,6 +42,9 @@ void kmain(void) {
     idt_init();
     serial_write("IDT initialized.\n");
 
+    irq_install();
+    __asm__ __volatile__ ("sti");
+
     terminal_write("Enabling paging...\n");
     paging_init();
     serial_write("Paging initialized.\n");

kernel/kmalloc.c

@@ -1,7 +1,7 @@
 #include "kmalloc.h"
 #include "terminal.h"  // Optional: for debug output
 
-#define HEAP_END 0xC0100000
+#define HEAP_END 0xC0500000
 
 static uint32_t current_heap = 0;
 

kernel/linker.ld

@@ -0,0 +1,38 @@
+ENTRY(kmain)
+
+PHDRS {
+    text PT_LOAD FLAGS(5);    /* Read + Execute */
+    rodata PT_LOAD FLAGS(4);  /* Read only */
+    data PT_LOAD FLAGS(6);    /* Read + Write */
+}
+
+SECTIONS {
+    . = 1M;
+
+    .text : {
+        *(.text*)
+    } :text
+
+    .rodata : {
+        *(.rodata*)
+    } :rodata
+
+    .data : {
+        *(.data*)
+    } :data
+
+    .bss : {
+        *(.bss*)
+        *(COMMON)
+    } :data
+
+    .stack (NOLOAD) : {
+        . = ALIGN(4);
+        . = . + 0x1000;
+    }
+
+    .heap (NOLOAD) : {
+        . = ALIGN(4);
+        . = . + 0x10000;
+    }
+}

kernel/memory.c

@@ -0,0 +1,17 @@
+#include "memory.h"
+
+/* note: this is a stub, please use care as theres duplicate functions in utils implementation
+/* --------------------------------------------------------------------- *
+ *  Helper: copy a single byte (used by both memcpy and memmove)
+ * --------------------------------------------------------------------- */
+static inline void byte_copy_forward(uint8_t *dst, const uint8_t *src, size_t n)
+{
+    while (n--) *dst++ = *src++;
+}
+
+static inline void byte_copy_backward(uint8_t *dst, const uint8_t *src, size_t n)
+{
+    dst += n; src += n;
+    while (n--) *--dst = *--src;
+}
+

kernel/memory.h

@@ -0,0 +1,25 @@
+#ifndef MEMORY_H
+#define MEMORY_H
+
+#include <stddef.h>   /* size_t, NULL */
+#include <stdint.h>   /* uint8_t */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* C11 / POSIX-2004 signatures */
+void *memcpy(void *restrict dst, const void *restrict src, size_t n);
+void *memmove(void *dst, const void *src, size_t n);
+int   memcmp(const void *s1, const void *s2, size_t n);
+
+/* Optional fast-path using 32-bit loads (x86 only) */
+#if defined(__i386__) && !defined(MEMORY_NO_OPT)
+# define MEMORY_OPTIMIZED 1
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* MEMORY_H */

kernel/paging.c

@@ -1,21 +1,17 @@
-#include "paging.h"
-#include "io.h"
 #include <stdint.h>
-#include <stddef.h>
+#include <string.h>
+#include "io.h"
+#include "paging.h"
 
-page_directory_entry_t *page_directory = (page_directory_entry_t *)0x100000;
+page_directory_entry_t *page_directory = (page_directory_entry_t *)0x200000;
-page_table_entry_t *page_table = (page_table_entry_t *)0x101000;
+page_table_entry_t     *page_table     = (page_table_entry_t *)0x201000;
-page_table_entry_t *heap_page_table = (page_table_entry_t *)0x102000; // Located right after the page directory
 
 // Helper function to set up the page directory entry
 void set_page_directory(page_directory_entry_t *dir) {
-    for (int i = 0; i < PAGE_DIRECTORY_SIZE; i++) {
+    // Set first PDE
-        dir[i].present = 0;
-    }
     dir[0].present = 1;
     dir[0].rw = 1;
-    dir[0].user = 0;
+    dir[0].addr = (uint32_t)page_table >> 12;
-    dir[0].frame = (uint32_t)page_table >> 12;
 }
 
 // Helper function to set up the page table entry
@@ -23,12 +19,8 @@ void set_page_table(page_table_entry_t *table) {
     for (int i = 0; i < PAGE_TABLE_SIZE; i++) {
         // Set up page table entries with identity mapping
         table[i].present = 1;
-        table[i].rw = 1;          // Read/Write
+        table[i].rw = 1;        // Read/Write
-        table[i].user = 0;        // Kernel mode
+        table[i].addr = i;      // Identity mapping
-        table[i].write_through = 0;
-        table[i].cache_disabled = 0;
-        table[i].accessed = 0;
-        table[i].frame = i;       // Identity mapping
     }
 }
 
@@ -47,26 +39,13 @@ void enable_paging() {
 
 // Initialize paging: set up the page directory and enable paging
 void paging_init() {
+    // Zero out the tables
+    memset(page_directory, 0x00, PAGE_DIRECTORY_SIZE * sizeof *page_directory);
+    memset(page_table, 0x00, PAGE_TABLE_SIZE * sizeof *page_table);
+
     // Set up identity-mapped page directory + table
     set_page_directory(page_directory);
     set_page_table(page_table);
 
-    // === Set up heap mapping at 0xC0100000 ===
-    for (int i = 0; i < PAGE_TABLE_SIZE; i++) {
-        heap_page_table[i].present = 1;
-        heap_page_table[i].rw = 1;
-        heap_page_table[i].user = 0;
-        heap_page_table[i].write_through = 0;
-        heap_page_table[i].cache_disabled = 0;
-        heap_page_table[i].accessed = 0;
-        heap_page_table[i].frame = (256 + i); // Start physical heap at 1MB (256*4KB = 1MB)
-    }
-
-    // Index 772 = 0xC0100000 / 4MB
-    page_directory[772].present = 1;
-    page_directory[772].rw = 1;
-    page_directory[772].user = 0;
-    page_directory[772].frame = (uint32_t)heap_page_table >> 12;
-
     enable_paging();
 }

kernel/paging.h

@@ -10,31 +10,31 @@
 
 // Page Directory and Page Table structure
 typedef struct {
-    uint32_t present : 1;  // Present bit (1: page is present in memory)
+    uint32_t present : 1;           // Present bit (1: page is present in memory)
-    uint32_t rw : 1;       // Read-Write bit (1: page is read-write)
+    uint32_t rw : 1;                // Read-Write bit (1: page is read-write)
-    uint32_t user : 1;     // User-supervisor bit (1: user mode access)
+    uint32_t user : 1;              // User-supervisor bit (1: user mode access)
-    uint32_t write_through : 1;  // Write-through cache
+    uint32_t write_through : 1;     // Write-through cache
-    uint32_t cache_disabled : 1; // Cache disabled
+    uint32_t cache_disabled : 1;    // Cache disabled
-    uint32_t accessed : 1;        // Accessed bit
+    uint32_t accessed : 1;          // Accessed bit
-    uint32_t reserved : 1;        // Reserved bit
+    uint32_t dirty : 1;             // Dirty bit
-    uint32_t page_size : 1;       // Page size (0: 4KB, 1: 4MB)
+    uint32_t attribute : 1;         // Page size (0: 4KB, 1: 4MB)
-    uint32_t global : 1;           // Global page (can be used across different processes)
+    uint32_t global : 1;            // Global page (can be used across different processes)
-    uint32_t available : 3;        // Available bits for the system
+    uint32_t reserved : 3;          // Unused
-    uint32_t frame : 20;           // Frame address (physical address)
+    uint32_t addr : 20;             // Page frame address (physical address)
 } __attribute__((packed)) page_table_entry_t;
 
 // Define page directory entry
 typedef struct {
-    uint32_t present : 1;
+    uint32_t present : 1;           // Present bit (1: PTE is present in memory)
-    uint32_t rw : 1;
+    uint32_t rw : 1;                // Read-Write bit (1: pages are read-write)
-    uint32_t user : 1;
+    uint32_t user : 1;              // User-supervisor bit (1: user mode access)
-    uint32_t write_through : 1;
+    uint32_t write_through : 1;     // Write-through cache
-    uint32_t cache_disabled : 1;
+    uint32_t cache_disabled : 1;    // Cache disabled
-    uint32_t accessed : 1;
+    uint32_t accessed : 1;          // Accessed bit
-    uint32_t reserved : 1;
+    uint32_t available : 1;         // Unused
-    uint32_t zero : 5;           // Must be zero for page directory
+    uint32_t page_size : 1;         // Page size (0: 4KB, 1: 4MB)
-    uint32_t reserved_2 : 7;     // Reserved bits
+    uint32_t available_2 : 4;       // Unused
-    uint32_t frame : 20;         // Frame address of the page table
+    uint32_t addr : 20;             // Page table address
 } __attribute__((packed)) page_directory_entry_t;
 
 extern page_directory_entry_t *page_directory;

kernel/print.h

@@ -1,7 +1,7 @@
 #ifndef PRINT_H
 #define PRINT_H
 
-#include "types.h"
+#include <stdint.h>
 
 void print_string(const char *str);
 void my_printf(const char *format, ...);

kernel/scheduler.c

@@ -1,7 +1,12 @@
 #include "scheduler.h"
 #include <stddef.h>
 
+// Defined in context_switch.s
+extern void ctx_switch(uint32_t **old_sp_ptr, uint32_t *new_sp);
+
 static task_t tasks[MAX_TASKS];
+
+// Stack memory area. Note: x86 Stacks grow DOWN from high to low addresses.
 static uint32_t task_stacks[MAX_TASKS][STACK_SIZE / sizeof(uint32_t)];
 
 static int task_count = 0;
@@ -9,7 +14,6 @@ static task_t *task_list = NULL;
 static task_t *current_task = NULL;
 
 void scheduler_init() {
-    // Initialize task list, etc.
     task_list = NULL;
     current_task = NULL;
     task_count = 0;
@@ -20,16 +24,42 @@ void scheduler_add_task(void (*entry)(void)) {
 
     task_t *new_task = &tasks[task_count];
     new_task->id = task_count;
-    new_task->entry = entry;
 
-    // Simulate a stack pointer pointing to the "top" of the stack
+    // 1. Calculate the top of the stack (High Address)
-    new_task->stack_ptr = &task_stacks[task_count][STACK_SIZE / sizeof(uint32_t) - 1];
+    // We point to the very end of the array.
+    uint32_t *sp = &task_stacks[task_count][STACK_SIZE / sizeof(uint32_t)];
 
+    // 2. "Forge" the stack frame to look like ctx_switch saved it.
+    // We push values onto the stack by decrementing the pointer and writing.
+
+    // --- Return Address (EIP) ---
+    sp--; 
+    *sp = (uint32_t)entry; // When ctx_switch does 'ret', it pops this and jumps to 'entry'
+
+    // --- EFLAGS ---
+    sp--;
+    *sp = 0x00000202; // Reserved bit set, Interrupts Enabled (IF=1). Important!
+
+    // --- General Purpose Registers (PUSHA/POPA layout) ---
+    // Order: EAX, ECX, EDX, EBX, ESP, EBP, ESI, EDI
+    // We initialize them to 0 or meaningful values.
+    sp--; *sp = 0; // EAX
+    sp--; *sp = 0; // ECX
+    sp--; *sp = 0; // EDX
+    sp--; *sp = 0; // EBX
+    sp--; *sp = 0; // ESP (Ignored by POPA)
+    sp--; *sp = 0; // EBP
+    sp--; *sp = 0; // ESI
+    sp--; *sp = 0; // EDI
+
+    // Save this final stack location to the TCB
+    new_task->stack_ptr = sp;
     new_task->next = NULL;
 
-    // Add to task list
+    // 3. Add to linked list
     if (task_list == NULL) {
         task_list = new_task;
+        current_task = new_task; // Make sure we have a current task to start
     } else {
         task_t *tail = task_list;
         while (tail->next) {
@@ -42,21 +72,25 @@ void scheduler_add_task(void (*entry)(void)) {
 }
 
 void scheduler_schedule() {
-    // Very basic round-robin switch
+    if (!current_task) return;
-    if (current_task && current_task->next) {
+
+    task_t *prev = current_task;
+    
+    // Round-robin logic
+    if (current_task->next) {
         current_task = current_task->next;
     } else {
-        current_task = task_list; // Loop back
+        current_task = task_list;
     }
 
-    // Call context switch or simulate yielding to current_task
+    // Perform the ACTUAL context switch
-    // In real system: context_switch_to(current_task)
+    // We pass the address of the previous task's stack pointer storage
-    if (current_task && current_task->entry) {
+    // and the value of the new task's stack pointer.
-        current_task->entry();  // Simulate switching by calling
+    if (prev != current_task) {
+        ctx_switch(&prev->stack_ptr, current_task->stack_ptr);
     }
 }
 
 void scheduler_yield() {
-    // Stub: manually call schedule for cooperative multitasking
     scheduler_schedule();
 }

kernel/scheduler.h

@@ -4,18 +4,21 @@
 #include <stdint.h>
 
 #define MAX_TASKS 8
-#define STACK_SIZE 1024
+#define STACK_SIZE 1024 // in bytes
 
 typedef struct task {
     uint32_t id;
-    void (*entry)(void);
+    
+    // The most important field: 
+    // Where was the stack pointer when we last left this task?
     uint32_t *stack_ptr; 
+    
     struct task *next;
 } task_t;
 
 void scheduler_init();
 void scheduler_add_task(void (*entry)(void));
 void scheduler_schedule();
-void scheduler_yield();  // Optional for cooperative scheduling
+void scheduler_yield();
 
 #endif // SCHEDULER_H

kernel/terminal.c

@@ -12,6 +12,7 @@ static uint16_t* const vga_buffer = (uint16_t*) VGA_ADDRESS;
 static uint8_t cursor_x = 0;
 static uint8_t cursor_y = 0;
 static uint8_t current_color = WHITE_ON_BLACK;
+static uint16_t last_cursor_pos = 0xFFFF;
 
 void terminal_initialize(void) {
     for (uint16_t y = 0; y < VGA_HEIGHT; y++) {
@@ -96,8 +97,10 @@ void terminal_clear(void) {
     update_cursor();
 }
 
-void update_cursor() {
+void update_cursor(void) {
     uint16_t pos = cursor_y * VGA_WIDTH + cursor_x;
+    if (pos == last_cursor_pos) return;
+    last_cursor_pos = pos;
 
     outb(0x3D4, 0x0F);
     outb(0x3D5, (uint8_t)(pos & 0xFF));

kernel/threading.c

@@ -1,9 +1,8 @@
+#include <stdbool.h>
+#include <string.h>
 #include "malloc.h"
 #include "print.h"
 #include "threading.h"
-#include "types.h"
-#include "utils.h"
-#include <stdint.h>
 
 #define MAX_THREADS 16           // Maximum number of threads
 #define THREAD_STACK_SIZE 8192   // Stack size for each thread

kernel/types.c

@@ -1 +0,0 @@
-#include "types.h"

kernel/types.h

@@ -1,62 +0,0 @@
-#ifndef TYPES_H
-#define TYPES_H
-
-// ----------------------------
-// Fixed-width integer types
-// ----------------------------
-typedef unsigned char      uint8_t;
-typedef signed char        int8_t;
-typedef unsigned short     uint16_t;
-typedef signed short       int16_t;
-typedef unsigned int       uint32_t;
-typedef signed int         int32_t;
-typedef unsigned long long uint64_t;
-typedef signed long long   int64_t;
-
-// ----------------------------
-// Boolean & NULL definitions
-// ----------------------------
-#ifndef __cplusplus
-typedef enum { false = 0, true = 1 } bool;
-#endif
-
-#ifndef NULL
-#define NULL ((void*)0)
-#endif
-
-// ----------------------------
-// OS subsystem types
-// ----------------------------
-typedef uint32_t size_t;
-typedef int32_t  ssize_t;
-
-typedef uint32_t phys_addr_t; // Physical address
-typedef uint32_t virt_addr_t; // Virtual address
-
-typedef uint32_t pid_t;       // Process ID
-typedef uint32_t tid_t;       // Thread ID
-
-// ----------------------------
-// Bitfield & utility macros
-// ----------------------------
-#define BIT(n) (1U << (n))
-#define BITS(m, n) (((1U << ((n) - (m) + 1)) - 1) << (m))
-
-// Align value to next multiple of alignment
-#define ALIGN_UP(val, align) (((val) + ((align)-1)) & ~((align)-1))
-#define ALIGN_DOWN(val, align) ((val) & ~((align)-1))
-
-// ----------------------------
-// Attributes for structures
-// ----------------------------
-#define PACKED      __attribute__((packed))
-#define ALIGN(x)    __attribute__((aligned(x)))
-
-// ----------------------------
-// Likely/unlikely branch hints
-// (for future optimization use)
-// ----------------------------
-#define likely(x)   __builtin_expect(!!(x), 1)
-#define unlikely(x) __builtin_expect(!!(x), 0)
-
-#endif // TYPES_H

kernel/utils.c

@@ -76,20 +76,3 @@ char* utoa(unsigned int value, char* str, int base) {
     reverse(str, i);
     return str;
 }
-
-int memcmp(const void *ptr1, const void *ptr2, size_t num) {
-    const uint8_t *p1 = ptr1, *p2 = ptr2;
-    for (size_t i = 0; i < num; i++) {
-        if (p1[i] != p2[i]) {
-            return p1[i] < p2[i] ? -1 : 1;
-        }
-    }
-    return 0;
-}
-
-void *memset(void *dest, int value, size_t len) {
-    unsigned char *ptr = (unsigned char *)dest;
-    while (len-- > 0)
-        *ptr++ = (unsigned char)value;
-    return dest;
-}

kernel/utils.h

@@ -1,7 +1,7 @@
 #ifndef UTILS_H
 #define UTILS_H
 
-#include "types.h"
+#include <stddef.h>
 
 // Convert integer to string (base is typically 10, 16, etc.)
 char* itoa(int value, char* str, int base);
@@ -9,7 +9,4 @@ char* itoa(int value, char* str, int base);
 // Convert unsigned integer to string (base is typically 10, 16, etc.)
 char* utoa(unsigned int value, char* str, int base);
 
-int memcmp(const void *ptr1, const void *ptr2, size_t num);
-void *memset(void *dest, int value, size_t len);
-
 #endif // UTILS_H

klibc/include/stdarg.h

@@ -0,0 +1,14 @@
+#ifndef CLASSICOS_KLIBC_STDARG_H
+#define CLASSICOS_KLIBC_STDARG_H
+
+typedef __builtin_va_list va_list;
+
+#ifndef va_start
+#define va_start(ap, param) __builtin_va_start(ap, param)
+#endif
+
+#define va_end(ap) __builtin_va_end(ap)
+#define va_arg(ap, type) __builtin_va_arg(ap, type)
+#define va_copy(dest, src) __builtin_va_copy(dest, src)
+
+#endif  // CLASSICOS_KLIBC_STDARG_H

klibc/include/stdbool.h

@@ -0,0 +1,12 @@
+#ifndef CLASSICOS_KLIBC_STDBOOL_H
+#define CLASSICOS_KLIBC_STDBOOL_H
+
+#ifndef __cplusplus
+#define bool _Bool
+#define true 1
+#define false 0
+#endif
+
+#define __bool_true_false_are_defined 1
+
+#endif  // CLASSICOS_KLIBC_STDBOOL_H

klibc/include/stddef.h

@@ -0,0 +1,10 @@
+#ifndef CLASSICOS_KLIBC_STDDEF_H
+#define CLASSICOS_KLIBC_STDDEF_H
+
+typedef __SIZE_TYPE__ size_t;
+typedef __PTRDIFF_TYPE__ ptrdiff_t;
+
+#undef NULL
+#define NULL ((void*)0)
+
+#endif  // CLASSICOS_KLIBC_STDDEF_H

klibc/include/stdint.h

@@ -0,0 +1,16 @@
+#ifndef CLASSICOS_KLIBC_STDINT_H
+#define CLASSICOS_KLIBC_STDINT_H
+
+typedef signed char int8_t;
+typedef short int int16_t;
+typedef int int32_t;
+typedef long long int int64_t;
+
+typedef unsigned char uint8_t;
+typedef unsigned short int uint16_t;
+typedef unsigned int uint32_t;
+typedef unsigned long long int uint64_t;
+
+typedef unsigned int uintptr_t;
+
+#endif  // CLASSICOS_KLIBC_STDINT_H

klibc/include/stdio.h

@@ -0,0 +1,4 @@
+#ifndef CLASSICOS_KLIBC_STDIO_H
+#define CLASSICOS_KLIBC_STDIO_H
+
+#endif  // CLASSICOS_KLIBC_STDIO_H

klibc/include/stdlib.h

@@ -0,0 +1,4 @@
+#ifndef CLASSICOS_KLIBC_STDLIB_H
+#define CLASSICOS_KLIBC_STDLIB_H
+
+#endif  // CLASSICOS_KLIBC_STDLIB_H

klibc/include/string.h

@@ -0,0 +1,14 @@
+#ifndef CLASSICOS_KLIBC_STRING_H
+#define CLASSICOS_KLIBC_STRING_H
+
+#include <stddef.h>
+
+extern int memcmp(const void* s1, const void* s2, size_t n);
+extern void* memmove(void* dst, const void* src, size_t n);
+extern void* memcpy(void* dst, const void* src, size_t n);
+extern void* memset(void* dst, int c, size_t n);
+
+extern size_t strlen(const char* s);
+extern int strcmp(const char* s1, const char* s2);
+
+#endif  // CLASSICOS_KLIBC_STRING_H

klibc/src/string.c

@@ -0,0 +1,107 @@
+#include <string.h>
+
+int memcmp(const void* s1, const void* s2, size_t n) {
+    const unsigned char* c1 = s1;
+    const unsigned char* c2 = s2;
+    int d = 0;
+
+    while (n--) {
+        d = (int)*c1++ - (int)*c2++;
+        if (d) break;
+    }
+
+    return d;
+}
+
+void* memmove(void* dst, const void* src, size_t n) {
+    const char* p = src;
+    char* q = dst;
+#if defined(__i386__) || defined(__x86_64__)
+    if (q < p) {
+        __asm__ volatile("cld; rep; movsb" : "+c"(n), "+S"(p), "+D"(q));
+    } else {
+        p += (n - 1);
+        q += (n - 1);
+        __asm__ volatile("std; rep; movsb; cld" : "+c"(n), "+S"(p), "+D"(q));
+    }
+#else
+    if (q < p) {
+        while (n--) {
+            *q++ = *p++;
+        }
+    } else {
+        p += n;
+        q += n;
+        while (n--) {
+            *--q = *--p;
+        }
+    }
+#endif
+
+    return dst;
+}
+
+void* memcpy(void* dst, const void* src, size_t n) {
+    const char* p = src;
+    char* q = dst;
+#if defined(__i386__)
+    size_t nl = n >> 2;
+    __asm__ volatile("cld ; rep ; movsl ; movl %3,%0 ; rep ; movsb"
+                    : "+c"(nl), "+S"(p), "+D"(q)
+                    : "r"(n & 3));
+#elif defined(__x86_64__)
+    size_t nq = n >> 3;
+    __asm__ volatile("cld ; rep ; movsq ; movl %3,%%ecx ; rep ; movsb"
+                    : "+c"(nq), "+S"(p), "+D"(q)
+                    : "r"((uint32_t)(n & 7)));
+#else
+    while (n--) {
+        *q++ = *p++;
+    }
+#endif
+
+    return dst;
+}
+
+void* memset(void* dst, int c, size_t n) {
+    char* q = dst;
+
+#if defined(__i386__)
+    size_t nl = n >> 2;
+    __asm__ volatile("cld ; rep ; stosl ; movl %3,%0 ; rep ; stosb"
+                    : "+c"(nl), "+D"(q)
+                    : "a"((unsigned char)c * 0x01010101U), "r"(n & 3));
+#elif defined(__x86_64__)
+    size_t nq = n >> 3;
+    __asm__ volatile("cld ; rep ; stosq ; movl %3,%%ecx ; rep ; stosb"
+                    : "+c"(nq), "+D"(q)
+                    : "a"((unsigned char)c * 0x0101010101010101U),
+                      "r"((uint32_t)n & 7));
+#else
+    while (n--) {
+        *q++ = c;
+    }
+#endif
+
+    return dst;
+}
+
+size_t strlen(const char* s) {
+    const char* ss = s;
+    while (*ss) ss++;
+    return ss - s;
+}
+
+int strcmp(const char* s1, const char* s2) {
+    const unsigned char* c1 = (const unsigned char*)s1;
+    const unsigned char* c2 = (const unsigned char*)s2;
+    unsigned char ch;
+    int d = 0;
+
+    while (1) {
+        d = (int)(ch = *c1++) - (int)*c2++;
+        if (d || !ch) break;
+    }
+
+    return d;
+}