Merge pull request #90 from gbowne1/gbowne1-patch-2

Fix extern declaration for disk_read_sector function

.gitignore

@@ -1,3 +1,5 @@
 .build.env
 build
 cross
+.cache/
+compile_commands.json

Makefile

@@ -6,6 +6,7 @@ QEMU= qemu-system-i386
 OBJCOPY = i386-elf-objcopy
 
 BUILD_DIR = build
+CROSS_DIR = cross
 DISK_IMG = $(BUILD_DIR)/disk.img
 STAGE2_SIZE = 2048
 
@@ -14,6 +15,9 @@ 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))
 
+KLIBC_SRC = $(wildcard klibc/src/*.c)
+KLIBC_OBJ = $(patsubst klibc/src/%.c, $(BUILD_DIR)/klibc/%.o, $(KLIBC_SRC))
+
 all: $(DISK_IMG)
 
 .PHONY: stage1 stage2 kernel run gdb clean
@@ -26,7 +30,7 @@ stage1: $(BUILD_DIR)
 # 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 -fno-stack-protector -m32 -g -c -o $(BUILD_DIR)/stage2_load.o bootloader/stage2_load.c
+	$(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
@@ -35,10 +39,13 @@ $(BUILD_DIR)/asm_%.o: kernel/%.asm
 	$(AS) $(ASFLAGS) -o $@ $<
 
 $(BUILD_DIR)/%.o: kernel/%.c
-	$(CC) -std=c11 -ffreestanding -nostdlib -fno-stack-protector -m32 -g -c -o $@ $<
+	$(CC) -std=c11 -ffreestanding -nostdlib -nostdinc -fno-stack-protector -m32 -Iklibc/include -g -c -o $@ $<
 
-kernel: $(KERNEL_OBJ) | $(BUILD_DIR)
-	$(LD) -melf_i386 -Tkernel/linker.ld -o $(BUILD_DIR)/kernel.elf $(KERNEL_OBJ)
+$(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=$@
@@ -48,6 +55,7 @@ $(DISK_IMG): stage1 stage2 kernel
 
 $(BUILD_DIR):
 	mkdir -p $@
+	mkdir -p $(BUILD_DIR)/klibc
 
 run:
 	qemu-system-i386 -s -S $(DISK_IMG)
@@ -57,4 +65,9 @@ gdb:
 
 clean:
 	rm -rf $(BUILD_DIR)
+
+clean-cross:
 	rm -rf $(CROSS_DIR)
+	rm -rf .build.env
+
+clean-all: clean clean-cross

bootloader/stage2.asm

@@ -24,75 +24,3 @@ _start:
     call load_kernel
 
     jmp eax
-
-; ----------------------------------------------------------------------------
-; ATA read sectors (LBA mode) 
-;
-; sysv32 abi signature:
-; void ata_lba_read(uint32_t lba, uint8_t nsect, void *addr);
-; ----------------------------------------------------------------------------
-ata_lba_read:
-    push ebp
-    mov ebp, esp
-
-    push ebx
-    push ecx
-    push edx
-    push edi
-
-    mov eax, [ebp+8]     ; arg #1 = LBA
-    mov cl, [ebp+12]     ; arg #2 = # of sectors
-    mov edi, [ebp+16]    ; arg #3 = buffer address
-    and eax, 0x0FFFFFFF
-
-    mov ebx, eax         ; Save LBA in RBX
-
-    mov edx, 0x01F6      ; Port to send drive and bit 24 - 27 of LBA
-    shr eax, 24          ; Get bit 24 - 27 in al
-    or al, 11100000b     ; Set bit 6 in al for LBA mode
-    out dx, al
-
-    mov edx, 0x01F2      ; Port to send number of sectors
-    mov al, cl           ; Get number of sectors from CL
-    out dx, al
-
-    mov edx, 0x1F3       ; Port to send bit 0 - 7 of LBA
-    mov eax, ebx         ; Get LBA from EBX
-    out dx, al
-
-    mov edx, 0x1F4       ; Port to send bit 8 - 15 of LBA
-    mov eax, ebx         ; Get LBA from EBX
-    shr eax, 8           ; Get bit 8 - 15 in AL
-    out dx, al
-
-    mov edx, 0x1F5       ; Port to send bit 16 - 23 of LBA
-    mov eax, ebx         ; Get LBA from EBX
-    shr eax, 16          ; Get bit 16 - 23 in AL
-    out dx, al
-
-    mov edx, 0x1F7       ; Command port
-    mov al, 0x20         ; Read with retry.
-    out dx, al
-
-    mov bl, cl          ; Save # of sectors in BL
-
-.wait_drq:
-    mov edx, 0x1F7
-.do_wait_drq:
-    in al, dx
-    test al, 8           ; the sector buffer requires servicing.
-    jz .do_wait_drq      ; keep polling until the sector buffer is ready.
-
-    mov edx, 0x1F0       ; Data port, in and out
-    mov ecx, 256
-    rep insw             ; in to [RDI]
-
-    dec bl               ; are we...
-    jnz .wait_drq        ; ...done?
-
-    pop edi
-    pop edx
-    pop ecx
-    pop ebx
-    pop ebp
-    ret

bootloader/stage2.ld

@@ -9,6 +9,4 @@ SECTIONS {
         *(.bss*)
         *(COMMON)
     }
-
-    read_buf = .;
 }

bootloader/stage2_load.c

@@ -1,5 +1,20 @@
+#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
 
@@ -7,6 +22,19 @@
 #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];
@@ -37,82 +65,171 @@ typedef struct {
     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.
-static int elf_load(const void* data, void (*load_segment)(uint8_t *vaddr, uint32_t src, uint32_t size)) {
+// 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;
-    const Elf32_Phdr* ph = (const Elf32_Phdr*)((uint8_t*)data + header->e_phoff);
+
+    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++) {
-        if (ph[i].p_type != PT_LOAD)
+        // 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;
 
-        uint32_t offset = ph[i].p_offset;
-        uint32_t vaddr = ph[i].p_vaddr;
-        uint32_t filesz = ph[i].p_filesz;
-        uint32_t memsz = ph[i].p_memsz;
-
-        // Copy data segment
-        //load_segment((uint8_t *)vaddr, offset, filesz);
-        load_segment((uint8_t *)vaddr, offset, filesz);
+        // 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 = (uint8_t*)(vaddr + filesz);
+            uint8_t* bss_start = vaddr + filesz;
             for (uint32_t j = 0; j < memsz - filesz; j++) {
                 bss_start[j] = 0;
             }
         }
     }
 
-    return header->e_entry;
-}
-
-#define KERN_START_SECT 5
-#define MAX(a, b) ((a)>(b) ? (a) : (b))
-
-extern void ata_lba_read(uint32_t lba, uint8_t nsect, void *addr);
-extern uint8_t read_buf[];
-
-static uint32_t
-total_header_size(const Elf32_Ehdr *header) {
-    uint32_t phend = header->e_phoff + header->e_phentsize*header->e_phnum;
-
-    // Align to 512
-    return (phend + 511) & ~511;
-}
-
-static void read_sectors(uint8_t *vaddr, uint32_t offset, uint32_t size) {
-    // # of sectors to read
-    uint32_t rem_nsect = ((size + 511) & ~511) / 512;
-
-    // Current lba address, offset by the first sector already read
-    uint32_t lba = KERN_START_SECT + offset / 512;
-
-    // Max 255 sectors at a time
-    while (rem_nsect) {
-        uint8_t nsect = rem_nsect > 255 ? 255 : rem_nsect;
-        ata_lba_read(lba, nsect, vaddr);
-
-        vaddr += nsect * 512;
-        rem_nsect -= nsect;
-        lba += nsect;
-    }
+    // Return the entry point
+    return (void *)header->e_entry;
 }
 
 void *load_kernel(void) {
-    // Read the first sector
-    ata_lba_read(KERN_START_SECT, 1, read_buf);
+    // ELF header buffer
+    uint8_t header_buf[SECTOR_SIZE];
 
-    const Elf32_Ehdr* header = (const Elf32_Ehdr*)read_buf;
+    // Read the first sector (contains the ELF header)
+    ata_read_sector(header_buf, KERN_START_SECT);
 
-    // Remaining data size, subtract the first 512B already read
-    uint32_t rem = total_header_size(header) - 512;
-
-    // Read the rest if necessary
-    if (rem)
-        read_sectors(read_buf+512, 512, rem);
-
-    elf_load(read_buf, read_sectors);
-
-    return (void *)header->e_entry;
+    // `elf_load()` returns the entry point
+    return elf_load(header_buf);
 }

kernel/context_switch.asm

@@ -1,4 +1,4 @@
-.global ctx_switch
+global ctx_switch
 
 ; void ctx_switch(uint32_t **old_sp_ptr, uint32_t *new_sp);
 ; Arguments on stack (cdecl convention):

kernel/cpu.c

@@ -2,36 +2,98 @@
 #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) {
-    __asm__(
-        "cpuid"
-        : "=a"(*eax), "=b"(*ebx), "=c"(*ecx), "=d"(*edx)
-        : "a"(function)
-    );
+// 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: ");
-    serial_write(vendor);
-    serial_write("\n");
+    // Leaf 1: Family, Model, Stepping
+    if (max_leaf >= 1) {
+        cpuid(1, &eax, &ebx, &ecx, &edx);
         
-    terminal_write("CPUID max leaf: ");
-    print_hex(eax, false, false);  // You must implement this (see below)
+        uint32_t stepping = eax & 0xF;
+        uint32_t model = (eax >> 4) & 0xF;
+        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

@@ -15,9 +15,16 @@ static uint8_t g_sector_buffer[FAT12_SECTOR_SIZE];
 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]) return p1[i] - p2[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;
 }
 
@@ -182,3 +189,8 @@ uint32_t fat12_read(file_t *file, uint8_t *buffer, uint32_t bytes_to_read) {
 
     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

@@ -58,7 +58,7 @@ typedef struct {
 
 // You must implement this in your disk driver (e.g., floppy.c)
 // Returns 0 on success, non-zero on error.
-extern int disk_read_sector(uint32_t lba, uint8_t *buffer);
+int disk_read_sector(uint32_t lba, uint8_t *buffer);
 
 void fat12_init();
 file_t fat12_open(const char *filename);

kernel/linker.ld

@@ -1,18 +1,30 @@
 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
 
-    .rodata : { *(.rodata*) }
-    .data : { *(.data*) }
     .bss : {
         *(.bss*)
         *(COMMON)
-    }
+    } :data
 
     .stack (NOLOAD) : {
         . = ALIGN(4);

kernel/memory.c

@@ -15,124 +15,3 @@ static inline void byte_copy_backward(uint8_t *dst, const uint8_t *src, size_t n
     while (n--) *--dst = *--src;
 }
 
-/* --------------------------------------------------------------------- *
- *  memcpy  –  no overlap allowed (behaviour undefined if overlap)
- * --------------------------------------------------------------------- */
-void *memcpy(void *restrict dst, const void *restrict src, size_t n)
-{
-    uint8_t *d = (uint8_t *)dst;
-    const uint8_t *s = (const uint8_t *)src;
-
-#if defined(MEMORY_OPTIMIZED)
-    /* Align destination to 4-byte boundary */
-    size_t align = (uintptr_t)d & 3U;
-    if (align) {
-        size_t head = 4 - align;
-        if (head > n) head = n;
-        byte_copy_forward(d, s, head);
-        d += head; s += head; n -= head;
-    }
-
-    /* 32-bit word copy – safe because we already aligned dst */
-    {
-        uint32_t *d32 = (uint32_t *)d;
-        const uint32_t *s32 = (const uint32_t *)s;
-        size_t words = n / 4;
-        while (words--) *d32++ = *s32++;
-        d = (uint8_t *)d32;
-        s = (const uint8_t *)s32;
-        n &= 3;
-    }
-#endif
-
-    byte_copy_forward(d, s, n);
-    return dst;
-}
-
-/* --------------------------------------------------------------------- *
- *  memmove – handles overlapping regions correctly
- * --------------------------------------------------------------------- */
-void *memmove(void *dst, const void *src, size_t n)
-{
-    uint8_t *d = (uint8_t *)dst;
-    const uint8_t *s = (const uint8_t *)src;
-
-    if (n == 0 || dst == src)
-        return dst;
-
-    if (d < s) {                     /* copy forward */
-#if defined(MEMORY_OPTIMIZED)
-        /* Same fast path as memcpy when no overlap */
-        size_t align = (uintptr_t)d & 3U;
-        if (align) {
-            size_t head = 4 - align;
-            if (head > n) head = n;
-            byte_copy_forward(d, s, head);
-            d += head; s += head; n -= head;
-        }
-        {
-            uint32_t *d32 = (uint32_t *)d;
-            const uint32_t *s32 = (const uint32_t *)s;
-            size_t words = n / 4;
-            while (words--) *d32++ = *s32++;
-            d = (uint8_t *)d32;
-            s = (const uint8_t *)s32;
-            n &= 3;
-        }
-#endif
-        byte_copy_forward(d, s, n);
-    } else {                         /* copy backward */
-        byte_copy_backward(d, s, n);
-    }
-    return dst;
-}
-
-/* --------------------------------------------------------------------- *
- *  memcmp  – lexicographical compare
- * --------------------------------------------------------------------- */
-int memcmp(const void *s1, const void *s2, size_t n)
-{
-    const uint8_t *a = (const uint8_t *)s1;
-    const uint8_t *b = (const uint8_t *)s2;
-
-#if defined(MEMORY_OPTIMIZED)
-    /* Align to 4-byte boundary */
-    size_t align = (uintptr_t)a & 3U;
-    if (align && align == ((uintptr_t)b & 3U)) {
-        size_t head = 4 - align;
-        if (head > n) head = n;
-        while (head--) {
-            int diff = *a++ - *b++;
-            if (diff) return diff;
-        }
-        n -= head;
-    }
-
-    {
-        const uint32_t *a32 = (const uint32_t *)a;
-        const uint32_t *b32 = (const uint32_t *)b;
-        size_t words = n / 4;
-        while (words--) {
-            uint32_t va = *a32++, vb = *b32++;
-            if (va != vb) {
-                /* byte-wise fallback for the differing word */
-                const uint8_t *pa = (const uint8_t *)(a32 - 1);
-                const uint8_t *pb = (const uint8_t *)(b32 - 1);
-                for (int i = 0; i < 4; ++i) {
-                    int diff = pa[i] - pb[i];
-                    if (diff) return diff;
-                }
-            }
-        }
-        a = (const uint8_t *)a32;
-        b = (const uint8_t *)b32;
-        n &= 3;
-    }
-#endif
-
-    while (n--) {
-        int diff = *a++ - *b++;
-        if (diff) return diff;
-    }
-    return 0;
-}

kernel/types.h

@@ -27,7 +27,6 @@ typedef enum { false = 0, true = 1 } bool;
 // ----------------------------
 // OS subsystem types
 // ----------------------------
-typedef uint32_t size_t;
 typedef int32_t  ssize_t;
 
 typedef uint32_t phys_addr_t; // Physical address

kernel/utils.c

@@ -76,10 +76,3 @@ char* utoa(unsigned int value, char* str, int base) {
     reverse(str, i);
     return str;
 }
-
-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

@@ -1,6 +1,8 @@
 #ifndef UTILS_H
 #define UTILS_H
 
+#include <stddef.h>
+
 #include "types.h"
 
 // Convert integer to string (base is typically 10, 16, etc.)

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 +1,6 @@
+#ifndef CLASSICOS_KLIBC_STDBOOL_H
+#define CLASSICOS_KLIBC_STDBOOL_H
+
+typedef enum { false = 0, true = 1 } bool;
+
+#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;
+}