Add sysfs VFS driver, SYS_OPEN/CLOSE syscalls, cat app

Sysfs:
- New VFS driver mounted at /sys that lets kernel drivers expose
  virtual text files via namespace registration
- Drivers call sysfs_register(name, ops, ctx) with list/read/write
  callbacks for their namespace
- IDE driver registers 'ide' namespace exposing per-device attributes:
  model, type, channel, drive, sectors, sector_size
- Tested: ls /sys -> ide, ls /sys/ide -> hdd1 cd1,
  cat /sys/ide/hdd1/model -> QEMU HARDDISK

Syscalls:
- Added SYS_OPEN (11) and SYS_CLOSE (12) for file I/O from userspace
- Extended SYS_READ/SYS_WRITE to handle VFS file descriptors (fd >= 3)
- Updated userspace syscalls.h with open()/close() wrappers

Apps:
- New 'cat' app: reads and displays file contents via open/read/close
- Updated 'ls' to accept path argument via ARG1 env var
- Updated shell to pass ARG1 env var to external commands

README.md

@@ -57,11 +57,11 @@ Once a task is completed, it should be checked off.
 - [x] Implement the fork system call.
 - [x] Implement environment variables. Apps should be able to modify this, and it should be copied to new forks of an app.
 - [x] Create a basic shell program `sh`. This shell must be able to start the hello-world app. It must include `cd` as a built-in to change the current working directory.
-- [ ] Create an `ls` app. It should list the contents of the current working directory, via the environment variable.
-- [ ] Create a devicefs vfs driver. The devicefs subsystem should allow drivers to create new block devices devices.
-- [ ] Create an IDE driver. It should enumerate all IDE devices, and expose them to the vfs driver as `hddN` or `cdN`, where N is a number that starts at 1 and increases for each new device. The `N` value should be calucated by the devicefs subsystem, not the IDE driver.
-- [ ] Create an MBR driver. It should be able to automatically detect when new hard drives are detected, and automatically scan them for MBR partitions. Each MBR partition found should be listed as `hddNmbrY`, where Y is a number determined by the devicefs subsystem.
-- [ ] Create a `sysfs` vfs driver. It should allow drivers to expose text/config files to the VFS. Each driver can request a namespace in the sysfs. E.g.: the IDE driver could request `ide`. During this registration, the drive must provide a struct containing function callbacks. The callbacks must contain the function `list`, `read`, and `write`. These are executed when the user lists a directory, reads a file, or writes a file. It is expected that the contents of these files are extremely small and can simply be stored on the stack. It should be very easy for a driver to expose new information.
+- [x] Create an `ls` app. It should list the contents of the current working directory, via the environment variable.
+- [x] Create a devicefs vfs driver. The devicefs subsystem should allow drivers to create new block devices devices.
+- [x] Create an IDE driver. It should enumerate all IDE devices, and expose them to the vfs driver as `hddN` or `cdN`, where N is a number that starts at 1 and increases for each new device. The `N` value should be calucated by the devicefs subsystem, not the IDE driver.
+- [x] Create an MBR driver. It should be able to automatically detect when new hard drives are detected, and automatically scan them for MBR partitions. Each MBR partition found should be listed as `hddNmbrY`, where Y is a number determined by the devicefs subsystem.
+- [x] Create a `sysfs` vfs driver. It should allow drivers to expose text/config files to the VFS. Each driver can request a namespace in the sysfs. E.g.: the IDE driver could request `ide`. During this registration, the drive must provide a struct containing function callbacks. The callbacks must contain the function `list`, `read`, and `write`. These are executed when the user lists a directory, reads a file, or writes a file. It is expected that the contents of these files are extremely small and can simply be stored on the stack. It should be very easy for a driver to expose new information.
 - [ ] Create a FAT32 driver. It should allow reading and writing to and from a block device.
 - [ ] Create the `mount` app. It should allow on to mount a block device using the fat32 driver. Internally, it should use sysfs (which should be mounted automatically by the kernel to `/sys`) to setup a new mount.
 - [ ] Create a floppy driver. Each floppy device should be exposed as `/dev/floppyN`.

apps/cat/cat.c

@@ -0,0 +1,41 @@
+/**
+ * @file cat.c
+ * @brief Display file contents.
+ *
+ * Reads a file specified as the first argument (from shell)
+ * and prints its contents to stdout. If no argument is given,
+ * prints usage.
+ *
+ * Usage: cat <filepath>
+ */
+
+#include "syscalls.h"
+
+int main(void) {
+    /* Get the file path from the ARGS environment variable.
+     * The shell sets ARGS to the arguments after the command name. */
+    char path[128];
+    if (getenv("ARG1", path, sizeof(path)) < 0) {
+        puts("Usage: cat <file>\n");
+        return 1;
+    }
+
+    /* Open the file */
+    int32_t fd = open(path, 0);
+    if (fd < 0) {
+        puts("cat: ");
+        puts(path);
+        puts(": open failed\n");
+        return 1;
+    }
+
+    /* Read and print in chunks */
+    char buf[256];
+    int32_t n;
+    while ((n = read(fd, buf, sizeof(buf))) > 0) {
+        write(1, buf, (uint32_t)n);
+    }
+
+    close(fd);
+    return 0;
+}

apps/libc/syscalls.h

@@ -24,6 +24,8 @@ typedef int int32_t;
 #define SYS_GETENV  8
 #define SYS_SETENV  9
 #define SYS_READDIR 10
+#define SYS_OPEN    11
+#define SYS_CLOSE   12
 
 static inline int32_t syscall0(int num) {
     int32_t ret;
@@ -103,6 +105,25 @@ static inline int32_t readdir(const char *path, uint32_t idx, char *name) {
     return syscall3(SYS_READDIR, (uint32_t)path, idx, (uint32_t)name);
 }
 
+/**
+ * Open a file by path.
+ * @param path  File path.
+ * @param flags Open flags (currently unused, pass 0).
+ * @return File descriptor (>= 3) on success, -1 on failure.
+ */
+static inline int32_t open(const char *path, uint32_t flags) {
+    return syscall2(SYS_OPEN, (uint32_t)path, flags);
+}
+
+/**
+ * Close a file descriptor.
+ * @param fd File descriptor.
+ * @return 0 on success, -1 on failure.
+ */
+static inline int32_t close(int32_t fd) {
+    return syscall1(SYS_CLOSE, (uint32_t)fd);
+}
+
 /* Basic string operations for user-space */
 static inline uint32_t strlen(const char *s) {
     uint32_t len = 0;

apps/ls/ls.c

@@ -10,12 +10,14 @@
 #include "syscalls.h"
 
 int main(void) {
-    /* Get the current working directory */
+    /* Check for an explicit path argument first */
     char cwd[128];
-    if (getenv("CWD", cwd, sizeof(cwd)) < 0) {
-        /* Default to root if CWD not set */
-        cwd[0] = '/';
-        cwd[1] = '\0';
+    if (getenv("ARG1", cwd, sizeof(cwd)) < 0 || cwd[0] == '\0') {
+        /* No argument; use the current working directory */
+        if (getenv("CWD", cwd, sizeof(cwd)) < 0) {
+            cwd[0] = '/';
+            cwd[1] = '\0';
+        }
     }
 
     char name[128];

apps/sh/sh.c

@@ -105,7 +105,7 @@ static void builtin_help(void) {
 }
 
 /** Execute an external command via fork+exec. */
-static void run_command(const char *cmd) {
+static void run_command(const char *cmd, const char *arg) {
     int32_t pid = fork();
     if (pid < 0) {
         puts("sh: fork failed\n");
@@ -113,7 +113,14 @@ static void run_command(const char *cmd) {
     }
 
     if (pid == 0) {
-        /* Child: exec the command */
+        /* Child: set ARG1 if there's an argument */
+        if (arg && *arg) {
+            setenv("ARG1", arg);
+        } else {
+            setenv("ARG1", "");
+        }
+
+        /* exec the command */
         int32_t ret = exec(cmd);
         if (ret < 0) {
             puts("sh: ");
@@ -172,7 +179,7 @@ int main(void) {
             builtin_help();
         } else {
             /* External command */
-            run_command(line);
+            run_command(line, arg);
         }
     }
 

src/CMakeLists.txt

@@ -19,6 +19,10 @@ add_executable(kernel
     cpio.c
     vfs.c
     initrd_fs.c
+    devicefs.c
+    sysfs.c
+    ide.c
+    mbr.c
     env.c
     keyboard.c
     interrupts.S

src/devicefs.c

@@ -0,0 +1,350 @@
+/**
+ * @file devicefs.c
+ * @brief Device filesystem (devicefs) implementation.
+ *
+ * Provides a VFS driver mounted at /dev that exposes block and character
+ * devices. Kernel drivers register devices via devicefs_register_block()
+ * or devicefs_register_char(), and the devicefs assigns sequential numbers
+ * per device class (e.g., hdd1, hdd2, cd1).
+ *
+ * The VFS interface supports:
+ * - readdir: lists all registered devices
+ * - finddir: looks up a device by name
+ * - read/write: delegates to the device's block or char operations
+ */
+
+#include "devicefs.h"
+#include "vfs.h"
+#include <string.h>
+
+/* Debug print helpers defined in kernel.c */
+extern void offset_print(const char *str);
+extern void print_hex(uint32_t val);
+
+/** Device table. */
+static devicefs_device_t devices[DEVICEFS_MAX_DEVICES];
+
+/** Number of active devices. */
+static uint32_t device_count = 0;
+
+/**
+ * Find a free slot in the device table.
+ * @return Index of free slot, or -1 if full.
+ */
+static int find_free_slot(void) {
+    for (int i = 0; i < DEVICEFS_MAX_DEVICES; i++) {
+        if (!devices[i].active) {
+            return i;
+        }
+    }
+    return -1;
+}
+
+/* ================================================================
+ * VFS operations for /dev
+ * ================================================================ */
+
+/**
+ * Read from a device file.
+ *
+ * For block devices, translates byte offset/size to sector reads.
+ * For character devices, delegates directly to the char read op.
+ */
+static int32_t devfs_read(vfs_node_t *node, uint32_t offset,
+                          uint32_t size, void *buf) {
+    if (!node || !node->fs_data) return -1;
+
+    devicefs_device_t *dev = (devicefs_device_t *)node->fs_data;
+
+    if (dev->type == DEVICEFS_BLOCK && dev->block_ops) {
+        uint32_t sec_size = 512;
+        if (dev->block_ops->sector_size) {
+            sec_size = dev->block_ops->sector_size(dev->dev_data);
+        }
+        if (sec_size == 0) return -1;
+
+        uint32_t start_lba = offset / sec_size;
+        uint32_t end_byte = offset + size;
+        uint32_t end_lba = (end_byte + sec_size - 1) / sec_size;
+        uint32_t num_sectors = end_lba - start_lba;
+
+        /* For simplicity, require aligned reads for now */
+        if (offset % sec_size != 0 || size % sec_size != 0) {
+            /* Unaligned read: read full sectors, copy partial */
+            /* TODO: implement unaligned block reads with temp buffer */
+            return -1;
+        }
+
+        if (dev->block_ops->read_sectors) {
+            int ret = dev->block_ops->read_sectors(dev->dev_data,
+                                                    start_lba,
+                                                    num_sectors,
+                                                    buf);
+            return (ret == 0) ? (int32_t)size : -1;
+        }
+        return -1;
+    }
+
+    if (dev->type == DEVICEFS_CHAR && dev->char_ops) {
+        if (dev->char_ops->read) {
+            return dev->char_ops->read(dev->dev_data, size, buf);
+        }
+        return -1;
+    }
+
+    return -1;
+}
+
+/**
+ * Write to a device file.
+ *
+ * For block devices, translates byte offset/size to sector writes.
+ * For character devices, delegates directly to the char write op.
+ */
+static int32_t devfs_write(vfs_node_t *node, uint32_t offset,
+                           uint32_t size, const void *buf) {
+    if (!node || !node->fs_data) return -1;
+
+    devicefs_device_t *dev = (devicefs_device_t *)node->fs_data;
+
+    if (dev->type == DEVICEFS_BLOCK && dev->block_ops) {
+        uint32_t sec_size = 512;
+        if (dev->block_ops->sector_size) {
+            sec_size = dev->block_ops->sector_size(dev->dev_data);
+        }
+        if (sec_size == 0) return -1;
+
+        if (offset % sec_size != 0 || size % sec_size != 0) {
+            return -1; /* Require aligned writes */
+        }
+
+        uint32_t start_lba = offset / sec_size;
+        uint32_t num_sectors = size / sec_size;
+
+        if (dev->block_ops->write_sectors) {
+            int ret = dev->block_ops->write_sectors(dev->dev_data,
+                                                     start_lba,
+                                                     num_sectors,
+                                                     buf);
+            return (ret == 0) ? (int32_t)size : -1;
+        }
+        return -1;
+    }
+
+    if (dev->type == DEVICEFS_CHAR && dev->char_ops) {
+        if (dev->char_ops->write) {
+            return dev->char_ops->write(dev->dev_data, size, buf);
+        }
+        return -1;
+    }
+
+    return -1;
+}
+
+/**
+ * Read a directory entry from /dev.
+ * Lists all registered devices.
+ */
+static int devfs_readdir(vfs_node_t *dir, uint32_t idx, vfs_dirent_t *out) {
+    (void)dir;
+
+    uint32_t count = 0;
+    for (int i = 0; i < DEVICEFS_MAX_DEVICES; i++) {
+        if (!devices[i].active) continue;
+        if (count == idx) {
+            memset(out, 0, sizeof(vfs_dirent_t));
+            strncpy(out->name, devices[i].name, VFS_MAX_NAME - 1);
+            out->inode = (uint32_t)i;
+            out->type = (devices[i].type == DEVICEFS_BLOCK) ?
+                         VFS_BLOCKDEV : VFS_CHARDEV;
+            return 0;
+        }
+        count++;
+    }
+    return -1; /* No more entries */
+}
+
+/**
+ * Find a device by name within /dev.
+ */
+static int devfs_finddir(vfs_node_t *dir, const char *name, vfs_node_t *out) {
+    (void)dir;
+
+    for (int i = 0; i < DEVICEFS_MAX_DEVICES; i++) {
+        if (!devices[i].active) continue;
+        if (strcmp(devices[i].name, name) == 0) {
+            memset(out, 0, sizeof(vfs_node_t));
+            strncpy(out->name, devices[i].name, VFS_MAX_NAME - 1);
+            out->type = (devices[i].type == DEVICEFS_BLOCK) ?
+                         VFS_BLOCKDEV : VFS_CHARDEV;
+            out->inode = (uint32_t)i;
+            out->fs_data = &devices[i];
+
+            /* For block devices, compute size from sector count */
+            if (devices[i].type == DEVICEFS_BLOCK && devices[i].block_ops) {
+                uint32_t sec_size = 512;
+                uint32_t sec_count = 0;
+                if (devices[i].block_ops->sector_size) {
+                    sec_size = devices[i].block_ops->sector_size(devices[i].dev_data);
+                }
+                if (devices[i].block_ops->sector_count) {
+                    sec_count = devices[i].block_ops->sector_count(devices[i].dev_data);
+                }
+                out->size = sec_count * sec_size;
+            }
+
+            return 0;
+        }
+    }
+    return -1;
+}
+
+/** Filesystem operations for /dev. */
+static vfs_fs_ops_t devfs_ops = {
+    .open    = NULL,
+    .close   = NULL,
+    .read    = devfs_read,
+    .write   = devfs_write,
+    .readdir = devfs_readdir,
+    .finddir = devfs_finddir,
+};
+
+/* ================================================================
+ * Public API
+ * ================================================================ */
+
+uint32_t devicefs_next_number(const char *class_name) {
+    uint32_t max_num = 0;
+    for (int i = 0; i < DEVICEFS_MAX_DEVICES; i++) {
+        if (!devices[i].active) continue;
+        if (strcmp(devices[i].class_name, class_name) == 0) {
+            if (devices[i].number >= max_num) {
+                max_num = devices[i].number + 1;
+            }
+        }
+    }
+    /* Numbers start at 1 */
+    return (max_num == 0) ? 1 : max_num;
+}
+
+/**
+ * Format a uint32 as a decimal string into buf.
+ * Returns pointer to the start of the number within buf.
+ */
+static char *uint_to_str(uint32_t val, char *buf, int buf_size) {
+    buf[buf_size - 1] = '\0';
+    int pos = buf_size - 2;
+    if (val == 0) {
+        buf[pos] = '0';
+        return &buf[pos];
+    }
+    while (val > 0 && pos >= 0) {
+        buf[pos--] = (char)('0' + (val % 10));
+        val /= 10;
+    }
+    return &buf[pos + 1];
+}
+
+devicefs_device_t *devicefs_register_block(const char *class_name,
+                                           devicefs_block_ops_t *ops,
+                                           void *dev_data) {
+    int slot = find_free_slot();
+    if (slot < 0) {
+        offset_print("  DEVICEFS: no free device slots\n");
+        return NULL;
+    }
+
+    devicefs_device_t *dev = &devices[slot];
+    memset(dev, 0, sizeof(devicefs_device_t));
+
+    /* Copy class name */
+    strncpy(dev->class_name, class_name, DEVICEFS_MAX_CLASS_NAME - 1);
+
+    /* Assign sequential number */
+    dev->number = devicefs_next_number(class_name);
+
+    /* Build full device name: class_name + number */
+    char num_buf[12];
+    char *num_str = uint_to_str(dev->number, num_buf, sizeof(num_buf));
+    strncpy(dev->name, class_name, DEVICEFS_MAX_DEV_NAME - 12);
+    /* Append number string */
+    uint32_t nlen = strlen(dev->name);
+    uint32_t slen = strlen(num_str);
+    if (nlen + slen < DEVICEFS_MAX_DEV_NAME) {
+        memcpy(dev->name + nlen, num_str, slen + 1);
+    }
+
+    dev->type = DEVICEFS_BLOCK;
+    dev->block_ops = ops;
+    dev->dev_data = dev_data;
+    dev->active = 1;
+    device_count++;
+
+    offset_print("  DEVICEFS: registered block device /dev/");
+    offset_print(dev->name);
+    offset_print("\n");
+
+    return dev;
+}
+
+devicefs_device_t *devicefs_register_char(const char *class_name,
+                                          devicefs_char_ops_t *ops,
+                                          void *dev_data) {
+    int slot = find_free_slot();
+    if (slot < 0) {
+        offset_print("  DEVICEFS: no free device slots\n");
+        return NULL;
+    }
+
+    devicefs_device_t *dev = &devices[slot];
+    memset(dev, 0, sizeof(devicefs_device_t));
+
+    strncpy(dev->class_name, class_name, DEVICEFS_MAX_CLASS_NAME - 1);
+    dev->number = devicefs_next_number(class_name);
+
+    char num_buf[12];
+    char *num_str = uint_to_str(dev->number, num_buf, sizeof(num_buf));
+    strncpy(dev->name, class_name, DEVICEFS_MAX_DEV_NAME - 12);
+    /* Append number string */
+    uint32_t nlen2 = strlen(dev->name);
+    uint32_t slen2 = strlen(num_str);
+    if (nlen2 + slen2 < DEVICEFS_MAX_DEV_NAME) {
+        memcpy(dev->name + nlen2, num_str, slen2 + 1);
+    }
+
+    dev->type = DEVICEFS_CHAR;
+    dev->char_ops = ops;
+    dev->dev_data = dev_data;
+    dev->active = 1;
+    device_count++;
+
+    offset_print("  DEVICEFS: registered char device /dev/");
+    offset_print(dev->name);
+    offset_print("\n");
+
+    return dev;
+}
+
+devicefs_device_t *devicefs_find(const char *name) {
+    for (int i = 0; i < DEVICEFS_MAX_DEVICES; i++) {
+        if (!devices[i].active) continue;
+        if (strcmp(devices[i].name, name) == 0) {
+            return &devices[i];
+        }
+    }
+    return NULL;
+}
+
+int init_devicefs(void) {
+    memset(devices, 0, sizeof(devices));
+    device_count = 0;
+
+    int ret = vfs_mount("/dev", &devfs_ops, NULL);
+    if (ret != 0) {
+        offset_print("  DEVICEFS: failed to mount at /dev\n");
+        return -1;
+    }
+
+    offset_print("  DEVICEFS: mounted at /dev\n");
+    return 0;
+}

src/ide.c

@@ -0,0 +1,616 @@
+/**
+ * @file ide.c
+ * @brief IDE/ATA disk driver implementation.
+ *
+ * Probes the primary and secondary IDE channels for ATA hard drives and
+ * ATAPI CD/DVD drives using PIO-mode IDENTIFY commands. Detected devices
+ * are registered with the devicefs subsystem as block devices:
+ * - ATA drives → "hdd" class (hdd1, hdd2, ...)
+ * - ATAPI drives → "cd" class (cd1, cd2, ...)
+ *
+ * Supports PIO-mode sector reads and writes using 28-bit LBA addressing,
+ * which covers drives up to 128 GiB.
+ */
+
+#include "ide.h"
+#include "port_io.h"
+#include "devicefs.h"
+#include "sysfs.h"
+#include "driver.h"
+#include <string.h>
+
+/* Debug print helpers defined in kernel.c */
+extern void offset_print(const char *str);
+extern void print_hex(uint32_t val);
+
+/** All detected IDE devices. */
+static ide_device_t ide_devices[IDE_MAX_DEVICES];
+
+/* ================================================================
+ * Low-level IDE I/O helpers
+ * ================================================================ */
+
+/**
+ * Wait for the BSY flag to clear on the given channel.
+ * Returns the final status byte, or 0xFF on timeout.
+ *
+ * @param io_base Channel I/O base port.
+ * @return Status byte.
+ */
+static uint8_t ide_wait(uint16_t io_base) {
+    uint8_t status;
+    int timeout = 500000;
+    do {
+        status = inb(io_base + IDE_REG_STATUS);
+        if (status == 0xFF) return 0xFF; /* Floating bus */
+        if (--timeout == 0) return 0xFF;
+    } while (status & IDE_STATUS_BSY);
+    return status;
+}
+
+/**
+ * Wait for BSY to clear and DRQ to set (data ready).
+ * Returns 0 on success, -1 on timeout or error.
+ *
+ * @param io_base Channel I/O base port.
+ * @return 0 on success, -1 on error/timeout.
+ */
+static int ide_wait_drq(uint16_t io_base) {
+    uint8_t status;
+    int timeout = 500000;
+    do {
+        status = inb(io_base + IDE_REG_STATUS);
+        if (status == 0xFF) return -1;
+        if (status & (IDE_STATUS_ERR | IDE_STATUS_DF)) return -1;
+        if (--timeout == 0) return -1;
+    } while ((status & (IDE_STATUS_BSY | IDE_STATUS_DRQ)) != IDE_STATUS_DRQ);
+    return 0;
+}
+
+/**
+ * Read 256 16-bit words (512 bytes) from the data register.
+ *
+ * @param io_base Channel I/O base port.
+ * @param buf     Destination buffer (must be at least 512 bytes).
+ */
+static void ide_read_buffer(uint16_t io_base, uint16_t *buf) {
+    for (int i = 0; i < 256; i++) {
+        buf[i] = inw(io_base + IDE_REG_DATA);
+    }
+}
+
+/**
+ * Write 256 16-bit words (512 bytes) to the data register.
+ *
+ * @param io_base Channel I/O base port.
+ * @param buf     Source buffer (must be at least 512 bytes).
+ */
+static void ide_write_buffer(uint16_t io_base, const uint16_t *buf) {
+    for (int i = 0; i < 256; i++) {
+        outw(io_base + IDE_REG_DATA, buf[i]);
+    }
+}
+
+/**
+ * Perform a software reset on an IDE channel.
+ *
+ * @param ctrl_base Channel control port.
+ */
+static void ide_soft_reset(uint16_t ctrl_base) {
+    outb(ctrl_base, 0x04);  /* Set SRST bit */
+    /* Wait ~5 µs (several I/O reads) */
+    for (int i = 0; i < 4; i++) inb(ctrl_base);
+    outb(ctrl_base, 0x00);  /* Clear SRST bit */
+    /* Wait for BSY to clear */
+    for (int i = 0; i < 4; i++) inb(ctrl_base);
+}
+
+/**
+ * Select a drive on a channel (master=0, slave=1).
+ *
+ * @param io_base Channel I/O base port.
+ * @param drive   0 for master, 1 for slave.
+ */
+static void ide_select_drive(uint16_t io_base, uint8_t drive) {
+    outb(io_base + IDE_REG_DRIVE_HEAD, 0xA0 | (drive << 4));
+    /* Wait ~400 ns by reading status 4 times */
+    for (int i = 0; i < 4; i++) inb(io_base + IDE_REG_STATUS);
+}
+
+/* ================================================================
+ * IDENTIFY command
+ * ================================================================ */
+
+/**
+ * Send IDENTIFY (or IDENTIFY PACKET) to a drive and read the result.
+ *
+ * @param dev Pointer to the device descriptor to fill.
+ * @return 0 on success, -1 if no device or error.
+ */
+static int ide_identify(ide_device_t *dev) {
+    uint16_t io = dev->io_base;
+
+    /* Select the drive */
+    ide_select_drive(io, dev->drive);
+
+    /* Check for floating bus (no device) — read status, if 0xFF, no drive */
+    uint8_t check = inb(io + IDE_REG_STATUS);
+    if (check == 0xFF) return -1;
+
+    /* Clear sector count and LBA registers */
+    outb(io + IDE_REG_SECCOUNT, 0);
+    outb(io + IDE_REG_LBA_LO, 0);
+    outb(io + IDE_REG_LBA_MID, 0);
+    outb(io + IDE_REG_LBA_HI, 0);
+
+    /* Send IDENTIFY command */
+    outb(io + IDE_REG_COMMAND, IDE_CMD_IDENTIFY);
+
+    /* Read status — if 0 or 0xFF, no device */
+    uint8_t status = inb(io + IDE_REG_STATUS);
+    if (status == 0 || status == 0xFF) return -1;
+
+    /* Wait for BSY to clear */
+    status = ide_wait(io);
+    if (status == 0xFF) return -1;
+
+    /* Check if this is an ATAPI device (LBA_MID/HI will be non-zero) */
+    uint8_t lba_mid = inb(io + IDE_REG_LBA_MID);
+    uint8_t lba_hi  = inb(io + IDE_REG_LBA_HI);
+
+    if (lba_mid == 0x14 && lba_hi == 0xEB) {
+        /* ATAPI device — re-identify with IDENTIFY PACKET DEVICE */
+        dev->type = IDE_TYPE_ATAPI;
+        outb(io + IDE_REG_COMMAND, IDE_CMD_IDENTIFY_PKT);
+        status = ide_wait(io);
+        if (status == 0xFF) return -1;
+    } else if (lba_mid == 0 && lba_hi == 0) {
+        dev->type = IDE_TYPE_ATA;
+    } else {
+        /* Unknown device type */
+        return -1;
+    }
+
+    /* Wait for DRQ */
+    if (ide_wait_drq(io) != 0) return -1;
+
+    /* Read 256 words of identification data */
+    uint16_t identify_buf[256];
+    ide_read_buffer(io, identify_buf);
+
+    /* Parse model string (words 27-46, each word is big-endian) */
+    for (int i = 0; i < 20; i++) {
+        dev->model[i * 2]     = (char)(identify_buf[27 + i] >> 8);
+        dev->model[i * 2 + 1] = (char)(identify_buf[27 + i] & 0xFF);
+    }
+    dev->model[40] = '\0';
+    /* Trim trailing spaces */
+    for (int i = 39; i >= 0; i--) {
+        if (dev->model[i] == ' ') dev->model[i] = '\0';
+        else break;
+    }
+
+    /* Parse sector count (28-bit LBA: words 60-61) */
+    if (dev->type == IDE_TYPE_ATA) {
+        dev->sector_count = (uint32_t)identify_buf[60] |
+                            ((uint32_t)identify_buf[61] << 16);
+        dev->sector_size = 512;
+    } else {
+        /* ATAPI: sector count from READ CAPACITY, default to 0 */
+        dev->sector_count = 0;
+        dev->sector_size = 2048;  /* Standard CD sector size */
+    }
+
+    dev->present = 1;
+    return 0;
+}
+
+/* ================================================================
+ * Block device operations (for devicefs)
+ * ================================================================ */
+
+/**
+ * Read sectors from an ATA drive using PIO.
+ */
+static int ide_block_read(void *dev_data, uint32_t lba,
+                          uint32_t count, void *buf) {
+    ide_device_t *dev = (ide_device_t *)dev_data;
+    if (!dev || dev->type != IDE_TYPE_ATA) return -1;
+
+    uint16_t io = dev->io_base;
+    uint8_t *dest = (uint8_t *)buf;
+
+    for (uint32_t s = 0; s < count; s++) {
+        uint32_t cur_lba = lba + s;
+
+        /* Select drive with LBA mode and top 4 LBA bits */
+        outb(io + IDE_REG_DRIVE_HEAD,
+             0xE0 | (dev->drive << 4) | ((cur_lba >> 24) & 0x0F));
+
+        /* Set sector count = 1 and LBA */
+        outb(io + IDE_REG_SECCOUNT, 1);
+        outb(io + IDE_REG_LBA_LO, cur_lba & 0xFF);
+        outb(io + IDE_REG_LBA_MID, (cur_lba >> 8) & 0xFF);
+        outb(io + IDE_REG_LBA_HI, (cur_lba >> 16) & 0xFF);
+
+        /* Send READ SECTORS command */
+        outb(io + IDE_REG_COMMAND, IDE_CMD_READ_PIO);
+
+        /* Wait for data */
+        if (ide_wait_drq(io) != 0) return -1;
+
+        /* Read 256 words (512 bytes) */
+        ide_read_buffer(io, (uint16_t *)(dest + s * 512));
+    }
+
+    return 0;
+}
+
+/**
+ * Write sectors to an ATA drive using PIO.
+ */
+static int ide_block_write(void *dev_data, uint32_t lba,
+                           uint32_t count, const void *buf) {
+    ide_device_t *dev = (ide_device_t *)dev_data;
+    if (!dev || dev->type != IDE_TYPE_ATA) return -1;
+
+    uint16_t io = dev->io_base;
+    const uint8_t *src = (const uint8_t *)buf;
+
+    for (uint32_t s = 0; s < count; s++) {
+        uint32_t cur_lba = lba + s;
+
+        outb(io + IDE_REG_DRIVE_HEAD,
+             0xE0 | (dev->drive << 4) | ((cur_lba >> 24) & 0x0F));
+
+        outb(io + IDE_REG_SECCOUNT, 1);
+        outb(io + IDE_REG_LBA_LO, cur_lba & 0xFF);
+        outb(io + IDE_REG_LBA_MID, (cur_lba >> 8) & 0xFF);
+        outb(io + IDE_REG_LBA_HI, (cur_lba >> 16) & 0xFF);
+
+        outb(io + IDE_REG_COMMAND, IDE_CMD_WRITE_PIO);
+
+        if (ide_wait_drq(io) != 0) return -1;
+
+        ide_write_buffer(io, (const uint16_t *)(src + s * 512));
+
+        /* Flush cache: wait for BSY to clear after write */
+        ide_wait(io);
+    }
+
+    return 0;
+}
+
+/**
+ * Return sector size for a device.
+ */
+static uint32_t ide_block_sector_size(void *dev_data) {
+    ide_device_t *dev = (ide_device_t *)dev_data;
+    return dev ? dev->sector_size : 512;
+}
+
+/**
+ * Return total sector count for a device.
+ */
+static uint32_t ide_block_sector_count(void *dev_data) {
+    ide_device_t *dev = (ide_device_t *)dev_data;
+    return dev ? dev->sector_count : 0;
+}
+
+/** Block operations for ATA/ATAPI devices. */
+static devicefs_block_ops_t ide_block_ops = {
+    .read_sectors  = ide_block_read,
+    .write_sectors = ide_block_write,
+    .sector_size   = ide_block_sector_size,
+    .sector_count  = ide_block_sector_count,
+};
+
+/* ================================================================
+ * Sysfs interface for /sys/ide
+ * ================================================================
+ *
+ * Exposes per-device info:
+ *   /sys/ide/         → lists device names (hdd1, cd1, ...)
+ *   /sys/ide/hdd1/    → lists attributes (model, type, channel, drive,
+ *                        sectors, sector_size)
+ *   /sys/ide/hdd1/model   → "QEMU HARDDISK\n"
+ *   /sys/ide/hdd1/sectors → "0x00003800\n"
+ */
+
+/** Integer to hex string helper for sysfs output. */
+static int ide_sysfs_hex(uint32_t val, char *buf, uint32_t buf_size) {
+    static const char hex[] = "0123456789ABCDEF";
+    if (buf_size < 12) return -1;  /* "0x" + 8 hex + \n + \0 */
+    buf[0] = '0'; buf[1] = 'x';
+    for (int i = 7; i >= 0; i--) {
+        buf[2 + (7 - i)] = hex[(val >> (i * 4)) & 0xF];
+    }
+    buf[10] = '\n'; buf[11] = '\0';
+    return 11;
+}
+
+/**
+ * Split a path string into first component and remainder.
+ * "hdd1/model" → first="hdd1", rest="model"
+ */
+static void ide_split_path(const char *path, char *first, uint32_t fsize,
+                            const char **rest) {
+    while (*path == '/') path++;
+    const char *s = path;
+    while (*s && *s != '/') s++;
+    uint32_t len = (uint32_t)(s - path);
+    if (len >= fsize) len = fsize - 1;
+    memcpy(first, path, len);
+    first[len] = '\0';
+    if (*s == '/') s++;
+    *rest = s;
+}
+
+/** Attribute names exposed for each device. */
+static const char *ide_sysfs_attrs[] = {
+    "model", "type", "channel", "drive", "sectors", "sector_size"
+};
+#define IDE_SYSFS_NUM_ATTRS 6
+
+/**
+ * Find an IDE device by its devicefs name (hdd1, cd1, etc.).
+ * Returns the device pointer, or NULL.
+ */
+static ide_device_t *ide_find_by_name(const char *name) {
+    for (int i = 0; i < IDE_MAX_DEVICES; i++) {
+        if (!ide_devices[i].present) continue;
+
+        /* Reconstruct the devicefs name for this device */
+        const char *cls = (ide_devices[i].type == IDE_TYPE_ATA) ? "hdd" : "cd";
+        uint32_t cls_len = strlen(cls);
+        if (strncmp(name, cls, cls_len) != 0) continue;
+
+        /* Parse the number suffix */
+        const char *num_str = name + cls_len;
+        uint32_t num = 0;
+        while (*num_str >= '0' && *num_str <= '9') {
+            num = num * 10 + (*num_str - '0');
+            num_str++;
+        }
+        if (*num_str != '\0' || num == 0) continue;
+
+        /* Count how many devices of this class precede idx i */
+        uint32_t count = 0;
+        for (int j = 0; j <= i; j++) {
+            if (!ide_devices[j].present) continue;
+            const char *jcls = (ide_devices[j].type == IDE_TYPE_ATA) ? "hdd" : "cd";
+            if (strcmp(cls, jcls) == 0) count++;
+        }
+        if (count == num) return &ide_devices[i];
+    }
+    return NULL;
+}
+
+/**
+ * Sysfs list callback.
+ * path="" → list device names (hdd1, cd1, ...)
+ * path="hdd1" → list attributes
+ */
+static int ide_sysfs_list(void *ctx, const char *path, uint32_t idx,
+                            sysfs_entry_t *out) {
+    (void)ctx;
+
+    if (path[0] == '\0') {
+        /* List all present devices */
+        uint32_t count = 0;
+        for (int i = 0; i < IDE_MAX_DEVICES; i++) {
+            if (!ide_devices[i].present) continue;
+            if (count == idx) {
+                memset(out, 0, sizeof(sysfs_entry_t));
+                /* Build device name */
+                const char *cls = (ide_devices[i].type == IDE_TYPE_ATA)
+                                   ? "hdd" : "cd";
+                /* Count devices of this class up to and including i */
+                uint32_t cls_count = 0;
+                for (int j = 0; j <= i; j++) {
+                    if (!ide_devices[j].present) continue;
+                    const char *jcls = (ide_devices[j].type == IDE_TYPE_ATA)
+                                        ? "hdd" : "cd";
+                    if (strcmp(cls, jcls) == 0) cls_count++;
+                }
+                uint32_t clen = strlen(cls);
+                memcpy(out->name, cls, clen);
+                /* Append number as ASCII (max 1 digit for 4 devs) */
+                out->name[clen] = '0' + (char)cls_count;
+                out->name[clen + 1] = '\0';
+                out->is_dir = 1;
+                return 0;
+            }
+            count++;
+        }
+        return -1;
+    }
+
+    /* List attributes for a specific device */
+    ide_device_t *dev = ide_find_by_name(path);
+    if (!dev) return -1;
+
+    if (idx >= IDE_SYSFS_NUM_ATTRS) return -1;
+
+    memset(out, 0, sizeof(sysfs_entry_t));
+    strncpy(out->name, ide_sysfs_attrs[idx], SYSFS_MAX_NAME - 1);
+    out->is_dir = 0;
+    return 0;
+}
+
+/**
+ * Sysfs read callback.
+ * path="hdd1/model" → device model string
+ * path="hdd1/sectors" → hex sector count
+ */
+static int ide_sysfs_read(void *ctx, const char *path, char *buf,
+                            uint32_t buf_size) {
+    (void)ctx;
+
+    /* Split path into device name and attribute name */
+    char dev_name[SYSFS_MAX_NAME];
+    const char *attr;
+    ide_split_path(path, dev_name, sizeof(dev_name), &attr);
+
+    ide_device_t *dev = ide_find_by_name(dev_name);
+    if (!dev) return -1;
+    if (attr[0] == '\0') return -1;  /* no attribute specified */
+
+    if (strcmp(attr, "model") == 0) {
+        uint32_t mlen = strlen(dev->model);
+        if (mlen + 2 > buf_size) return -1;
+        memcpy(buf, dev->model, mlen);
+        buf[mlen] = '\n';
+        buf[mlen + 1] = '\0';
+        return (int)(mlen + 1);
+    }
+    if (strcmp(attr, "type") == 0) {
+        const char *t = (dev->type == IDE_TYPE_ATA) ? "ATA\n" :
+                        (dev->type == IDE_TYPE_ATAPI) ? "ATAPI\n" : "unknown\n";
+        uint32_t tlen = strlen(t);
+        if (tlen + 1 > buf_size) return -1;
+        memcpy(buf, t, tlen + 1);
+        return (int)tlen;
+    }
+    if (strcmp(attr, "channel") == 0) {
+        const char *c = (dev->channel == 0) ? "primary\n" : "secondary\n";
+        uint32_t clen = strlen(c);
+        if (clen + 1 > buf_size) return -1;
+        memcpy(buf, c, clen + 1);
+        return (int)clen;
+    }
+    if (strcmp(attr, "drive") == 0) {
+        const char *d = (dev->drive == 0) ? "master\n" : "slave\n";
+        uint32_t dlen = strlen(d);
+        if (dlen + 1 > buf_size) return -1;
+        memcpy(buf, d, dlen + 1);
+        return (int)dlen;
+    }
+    if (strcmp(attr, "sectors") == 0) {
+        return ide_sysfs_hex(dev->sector_count, buf, buf_size);
+    }
+    if (strcmp(attr, "sector_size") == 0) {
+        return ide_sysfs_hex(dev->sector_size, buf, buf_size);
+    }
+
+    return -1;
+}
+
+/**
+ * Sysfs write callback — IDE is read-only for now.
+ */
+static int ide_sysfs_write(void *ctx, const char *path, const char *buf,
+                             uint32_t size) {
+    (void)ctx; (void)path; (void)buf; (void)size;
+    return -1;  /* Read-only */
+}
+
+/** Sysfs operations for the IDE namespace. */
+static sysfs_ops_t ide_sysfs_ops = {
+    .list  = ide_sysfs_list,
+    .read  = ide_sysfs_read,
+    .write = ide_sysfs_write,
+};
+
+/* ================================================================
+ * Driver probe and init
+ * ================================================================ */
+
+/**
+ * Probe: always return OK since IDE ports are standard.
+ */
+static driver_probe_result_t ide_probe(void) {
+    return DRIVER_PROBE_OK;
+}
+
+/**
+ * Initialize the IDE driver: scan channels and register devices.
+ */
+static int ide_driver_init(void) {
+    memset(ide_devices, 0, sizeof(ide_devices));
+
+    /* Channel definitions: primary (0x1F0, 0x3F6), secondary (0x170, 0x376) */
+    static const uint16_t io_bases[2]   = { IDE_PRIMARY_IO, IDE_SECONDARY_IO };
+    static const uint16_t ctrl_bases[2] = { IDE_PRIMARY_CTRL, IDE_SECONDARY_CTRL };
+
+    int found = 0;
+
+    for (int ch = 0; ch < 2; ch++) {
+        /* Check if channel exists by reading status — 0xFF = floating bus */
+        uint8_t ch_status = inb(io_bases[ch] + IDE_REG_STATUS);
+        if (ch_status == 0xFF) {
+            offset_print("  IDE: ");
+            offset_print(ch == 0 ? "primary" : "secondary");
+            offset_print(" channel not present\n");
+            continue;
+        }
+
+        /* Software reset the channel */
+        ide_soft_reset(ctrl_bases[ch]);
+
+        for (int drv = 0; drv < 2; drv++) {
+            int idx = ch * 2 + drv;
+            ide_devices[idx].channel   = (uint8_t)ch;
+            ide_devices[idx].drive     = (uint8_t)drv;
+            ide_devices[idx].io_base   = io_bases[ch];
+            ide_devices[idx].ctrl_base = ctrl_bases[ch];
+            ide_devices[idx].present   = 0;
+
+            if (ide_identify(&ide_devices[idx]) == 0) {
+                found++;
+
+                const char *class_name = (ide_devices[idx].type == IDE_TYPE_ATA)
+                                          ? "hdd" : "cd";
+                const char *type_str   = (ide_devices[idx].type == IDE_TYPE_ATA)
+                                          ? "ATA" : "ATAPI";
+
+                offset_print("  IDE: ");
+                offset_print(type_str);
+                offset_print(" device on ");
+                offset_print(ch == 0 ? "primary" : "secondary");
+                offset_print(drv == 0 ? " master" : " slave");
+                offset_print(": ");
+                offset_print(ide_devices[idx].model);
+                offset_print(" (");
+                print_hex(ide_devices[idx].sector_count);
+                offset_print(" sectors)\n");
+
+                /* Register with devicefs */
+                devicefs_register_block(class_name, &ide_block_ops,
+                                        &ide_devices[idx]);
+            }
+        }
+    }
+
+    if (found == 0) {
+        offset_print("  IDE: no devices found\n");
+    } else {
+        offset_print("  IDE: ");
+        print_hex((uint32_t)found);
+        offset_print(" device(s) found\n");
+    }
+
+    /* Register sysfs namespace for IDE information */
+    sysfs_register("ide", &ide_sysfs_ops, NULL);
+
+    return 0;
+}
+
+int ide_init(void) {
+    return ide_driver_init();
+}
+
+ide_device_t *ide_get_device(int index) {
+    if (index < 0 || index >= IDE_MAX_DEVICES) return NULL;
+    if (!ide_devices[index].present) return NULL;
+    return &ide_devices[index];
+}
+
+/** IDE driver descriptor. */
+static const driver_t ide_driver = {
+    .name  = "ide",
+    .probe = ide_probe,
+    .init  = ide_driver_init,
+};
+
+REGISTER_DRIVER(ide_driver);

src/ide.h

@@ -0,0 +1,88 @@
+/**
+ * @file ide.h
+ * @brief IDE/ATA disk driver.
+ *
+ * Enumerates IDE devices on the primary and secondary channels, identifies
+ * ATA hard drives and ATAPI CD-ROMs, and registers them with the devicefs
+ * subsystem as block devices (hddN / cdN).
+ */
+
+#ifndef IDE_H
+#define IDE_H
+
+#include <stdint.h>
+
+/** Maximum number of IDE devices (2 channels × 2 drives). */
+#define IDE_MAX_DEVICES 4
+
+/** IDE channel I/O port bases. */
+#define IDE_PRIMARY_IO     0x1F0
+#define IDE_PRIMARY_CTRL   0x3F6
+#define IDE_SECONDARY_IO   0x170
+#define IDE_SECONDARY_CTRL 0x376
+
+/** IDE register offsets from I/O base. */
+#define IDE_REG_DATA       0x00
+#define IDE_REG_ERROR      0x01
+#define IDE_REG_FEATURES   0x01
+#define IDE_REG_SECCOUNT   0x02
+#define IDE_REG_LBA_LO     0x03
+#define IDE_REG_LBA_MID    0x04
+#define IDE_REG_LBA_HI     0x05
+#define IDE_REG_DRIVE_HEAD 0x06
+#define IDE_REG_STATUS     0x07
+#define IDE_REG_COMMAND    0x07
+
+/** IDE status register bits. */
+#define IDE_STATUS_ERR   0x01  /**< Error occurred. */
+#define IDE_STATUS_DRQ   0x08  /**< Data request ready. */
+#define IDE_STATUS_SRV   0x10  /**< Overlapped mode service request. */
+#define IDE_STATUS_DF    0x20  /**< Drive fault. */
+#define IDE_STATUS_DRDY  0x40  /**< Drive ready. */
+#define IDE_STATUS_BSY   0x80  /**< Drive busy. */
+
+/** IDE commands. */
+#define IDE_CMD_IDENTIFY     0xEC  /**< ATA IDENTIFY DEVICE. */
+#define IDE_CMD_IDENTIFY_PKT 0xA1  /**< ATAPI IDENTIFY PACKET DEVICE. */
+#define IDE_CMD_READ_PIO     0x20  /**< Read sectors (PIO, 28-bit LBA). */
+#define IDE_CMD_WRITE_PIO    0x30  /**< Write sectors (PIO, 28-bit LBA). */
+
+/** IDE device types. */
+#define IDE_TYPE_NONE  0  /**< No device present. */
+#define IDE_TYPE_ATA   1  /**< ATA hard disk. */
+#define IDE_TYPE_ATAPI 2  /**< ATAPI CD/DVD drive. */
+
+/**
+ * IDE device descriptor.
+ */
+typedef struct ide_device {
+    uint8_t present;            /**< 1 if device is present. */
+    uint8_t type;               /**< IDE_TYPE_ATA or IDE_TYPE_ATAPI. */
+    uint8_t channel;            /**< 0 = primary, 1 = secondary. */
+    uint8_t drive;              /**< 0 = master, 1 = slave. */
+    uint16_t io_base;           /**< I/O base port for this channel. */
+    uint16_t ctrl_base;         /**< Control port for this channel. */
+    uint32_t sector_count;      /**< Total sectors (28-bit LBA max). */
+    uint32_t sector_size;       /**< Sector size in bytes (usually 512). */
+    char model[41];             /**< Model string from IDENTIFY. */
+} ide_device_t;
+
+/**
+ * Initialize the IDE driver.
+ *
+ * Scans primary and secondary channels for ATA/ATAPI devices,
+ * reads their IDENTIFY data, and registers them with devicefs.
+ *
+ * @return Number of devices found.
+ */
+int ide_init(void);
+
+/**
+ * Get an IDE device by index (0–3).
+ *
+ * @param index Device index.
+ * @return Pointer to the device descriptor, or NULL if invalid/not present.
+ */
+ide_device_t *ide_get_device(int index);
+
+#endif /* IDE_H */

src/kernel.c

@@ -17,6 +17,9 @@
 #include "cpio.h"
 #include "vfs.h"
 #include "initrd_fs.h"
+#include "devicefs.h"
+#include "sysfs.h"
+#include "mbr.h"
 #include "keyboard.h"
 #include "framebuffer.h"
 
@@ -247,6 +250,14 @@ void kernel_main(uint32_t magic, uint32_t addr) {
         }
     }
 
+    init_devicefs();
+    EARLY_PRINT("DEV ");
+    offset_print("Devicefs initialized\n");
+
+    init_sysfs();
+    EARLY_PRINT("SYS ");
+    offset_print("Sysfs initialized\n");
+
     init_tss();
     EARLY_PRINT("TSS ");
     offset_print("TSS initialized\n");
@@ -281,6 +292,10 @@ void kernel_main(uint32_t magic, uint32_t addr) {
     EARLY_PRINT("DRV ");
     offset_print("Drivers initialized\n");
 
+    /* Scan for MBR partitions on detected hard drives */
+    init_mbr();
+    offset_print("MBR scan complete\n");
+
     /* At this point the VGA driver has been initialized and taken over
      * the display. The early VGA text is no longer visible. */
 

src/mbr.c

@@ -0,0 +1,270 @@
+/**
+ * @file mbr.c
+ * @brief MBR partition table driver implementation.
+ *
+ * Reads sector 0 from hard drives registered with the devicefs, parses
+ * the MBR partition table, and registers each partition as a sub-device.
+ * Partition devices are named using the parent device name + "mbr" class
+ * prefix, with the number assigned by the devicefs (e.g., hdd1mbr1).
+ *
+ * Each partition sub-device translates LBA offsets relative to its own
+ * start into absolute LBA addresses on the parent device.
+ */
+
+#include "mbr.h"
+#include "devicefs.h"
+#include <string.h>
+
+/* Debug print helpers defined in kernel.c */
+extern void offset_print(const char *str);
+extern void print_hex(uint32_t val);
+
+/** Maximum number of MBR partition sub-devices. */
+#define MBR_MAX_SUBS 16
+
+/**
+ * Per-partition context for sub-device read/write operations.
+ */
+typedef struct mbr_sub_device {
+    devicefs_device_t *parent;       /**< Parent block device. */
+    uint32_t lba_start;              /**< Partition start LBA on parent. */
+    uint32_t sector_count;           /**< Partition size in sectors. */
+    uint8_t type;                    /**< MBR partition type code. */
+    int active;                      /**< 1 if in use. */
+} mbr_sub_device_t;
+
+/** Pool of sub-device contexts. */
+static mbr_sub_device_t sub_devices[MBR_MAX_SUBS];
+
+/**
+ * Allocate a sub-device context.
+ * @return Pointer to a free sub-device, or NULL if exhausted.
+ */
+static mbr_sub_device_t *alloc_sub(void) {
+    for (int i = 0; i < MBR_MAX_SUBS; i++) {
+        if (!sub_devices[i].active) {
+            sub_devices[i].active = 1;
+            return &sub_devices[i];
+        }
+    }
+    return NULL;
+}
+
+/* ================================================================
+ * Block device ops for MBR partition sub-devices
+ * ================================================================ */
+
+/**
+ * Read sectors from a partition (translates LBA to parent device).
+ */
+static int mbr_sub_read(void *dev_data, uint32_t lba,
+                        uint32_t count, void *buf) {
+    mbr_sub_device_t *sub = (mbr_sub_device_t *)dev_data;
+    if (!sub || !sub->parent) return -1;
+
+    /* Bounds check */
+    if (lba + count > sub->sector_count) return -1;
+
+    /* Translate to absolute LBA on parent device */
+    uint32_t abs_lba = sub->lba_start + lba;
+
+    /* Read from parent device */
+    if (!sub->parent->block_ops || !sub->parent->block_ops->read_sectors) {
+        return -1;
+    }
+    return sub->parent->block_ops->read_sectors(sub->parent->dev_data,
+                                                 abs_lba, count, buf);
+}
+
+/**
+ * Write sectors to a partition (translates LBA to parent device).
+ */
+static int mbr_sub_write(void *dev_data, uint32_t lba,
+                         uint32_t count, const void *buf) {
+    mbr_sub_device_t *sub = (mbr_sub_device_t *)dev_data;
+    if (!sub || !sub->parent) return -1;
+
+    if (lba + count > sub->sector_count) return -1;
+
+    uint32_t abs_lba = sub->lba_start + lba;
+
+    if (!sub->parent->block_ops || !sub->parent->block_ops->write_sectors) {
+        return -1;
+    }
+    return sub->parent->block_ops->write_sectors(sub->parent->dev_data,
+                                                  abs_lba, count, buf);
+}
+
+/**
+ * Get sector size (same as parent device).
+ */
+static uint32_t mbr_sub_sector_size(void *dev_data) {
+    mbr_sub_device_t *sub = (mbr_sub_device_t *)dev_data;
+    if (!sub || !sub->parent || !sub->parent->block_ops) return 512;
+    if (sub->parent->block_ops->sector_size) {
+        return sub->parent->block_ops->sector_size(sub->parent->dev_data);
+    }
+    return 512;
+}
+
+/**
+ * Get total sector count of the partition.
+ */
+static uint32_t mbr_sub_sector_count(void *dev_data) {
+    mbr_sub_device_t *sub = (mbr_sub_device_t *)dev_data;
+    return sub ? sub->sector_count : 0;
+}
+
+/** Block operations for MBR partition sub-devices. */
+static devicefs_block_ops_t mbr_sub_ops = {
+    .read_sectors  = mbr_sub_read,
+    .write_sectors = mbr_sub_write,
+    .sector_size   = mbr_sub_sector_size,
+    .sector_count  = mbr_sub_sector_count,
+};
+
+/* ================================================================
+ * MBR scanning
+ * ================================================================ */
+
+int mbr_scan(const char *parent_name) {
+    devicefs_device_t *parent = devicefs_find(parent_name);
+    if (!parent) {
+        offset_print("  MBR: device not found: ");
+        offset_print(parent_name);
+        offset_print("\n");
+        return -1;
+    }
+
+    if (parent->type != DEVICEFS_BLOCK || !parent->block_ops) {
+        offset_print("  MBR: not a block device: ");
+        offset_print(parent_name);
+        offset_print("\n");
+        return -1;
+    }
+
+    if (!parent->block_ops->read_sectors) {
+        offset_print("  MBR: device has no read_sectors op\n");
+        return -1;
+    }
+
+    /* Read sector 0 (MBR) */
+    uint8_t sector[512];
+    if (parent->block_ops->read_sectors(parent->dev_data, 0, 1, sector) != 0) {
+        offset_print("  MBR: failed to read sector 0 of ");
+        offset_print(parent_name);
+        offset_print("\n");
+        return -1;
+    }
+
+    /* Check MBR signature */
+    uint16_t sig = (uint16_t)sector[510] | ((uint16_t)sector[511] << 8);
+    if (sig != MBR_SIGNATURE) {
+        offset_print("  MBR: no valid signature on ");
+        offset_print(parent_name);
+        offset_print(" (sig=");
+        print_hex(sig);
+        offset_print(")\n");
+        return 0; /* Not an error, just no MBR */
+    }
+
+    /* Build the class name for partition sub-devices: parentName + "mbr" */
+    char class_name[DEVICEFS_MAX_DEV_NAME];
+    memset(class_name, 0, sizeof(class_name));
+    strncpy(class_name, parent_name, DEVICEFS_MAX_DEV_NAME - 4);
+    /* Append "mbr" */
+    uint32_t clen = strlen(class_name);
+    if (clen + 3 < DEVICEFS_MAX_DEV_NAME) {
+        class_name[clen] = 'm';
+        class_name[clen + 1] = 'b';
+        class_name[clen + 2] = 'r';
+        class_name[clen + 3] = '\0';
+    }
+
+    /* Parse the 4 partition table entries (at offset 446) */
+    int found = 0;
+    for (int i = 0; i < MBR_MAX_PARTITIONS; i++) {
+        mbr_partition_entry_t *entry =
+            (mbr_partition_entry_t *)(sector + 446 + i * 16);
+
+        /* Skip empty partitions */
+        if (entry->type == MBR_TYPE_EMPTY) continue;
+        if (entry->sector_count == 0) continue;
+
+        /* Skip extended partitions for now */
+        if (entry->type == MBR_TYPE_EXTENDED) {
+            offset_print("  MBR: skipping extended partition on ");
+            offset_print(parent_name);
+            offset_print("\n");
+            continue;
+        }
+
+        mbr_sub_device_t *sub = alloc_sub();
+        if (!sub) {
+            offset_print("  MBR: no free sub-device slots\n");
+            break;
+        }
+
+        sub->parent = parent;
+        sub->lba_start = entry->lba_start;
+        sub->sector_count = entry->sector_count;
+        sub->type = entry->type;
+
+        /* Register with devicefs */
+        devicefs_device_t *dev = devicefs_register_block(class_name,
+                                                          &mbr_sub_ops,
+                                                          sub);
+        if (dev) {
+            offset_print("  MBR: ");
+            offset_print(dev->name);
+            offset_print(" type=");
+            print_hex(entry->type);
+            offset_print(" LBA=");
+            print_hex(entry->lba_start);
+            offset_print(" size=");
+            print_hex(entry->sector_count);
+            offset_print(" sectors\n");
+            found++;
+        }
+    }
+
+    if (found == 0) {
+        offset_print("  MBR: no partitions on ");
+        offset_print(parent_name);
+        offset_print("\n");
+    }
+
+    return found;
+}
+
+int init_mbr(void) {
+    memset(sub_devices, 0, sizeof(sub_devices));
+
+    /* Scan all "hdd" class devices for MBR partitions */
+    int total_partitions = 0;
+    for (uint32_t i = 1; i <= 16; i++) {
+        /* Build device name: hdd1, hdd2, ... */
+        char name[DEVICEFS_MAX_DEV_NAME];
+        memset(name, 0, sizeof(name));
+        name[0] = 'h'; name[1] = 'd'; name[2] = 'd';
+        /* Append number */
+        if (i < 10) {
+            name[3] = (char)('0' + i);
+        } else {
+            name[3] = (char)('0' + i / 10);
+            name[4] = (char)('0' + i % 10);
+        }
+
+        devicefs_device_t *dev = devicefs_find(name);
+        if (!dev) continue;
+
+        int n = mbr_scan(name);
+        if (n > 0) total_partitions += n;
+    }
+
+    offset_print("  MBR: ");
+    print_hex((uint32_t)total_partitions);
+    offset_print(" partition(s) found\n");
+
+    return 0;
+}

src/mbr.h

@@ -0,0 +1,67 @@
+/**
+ * @file mbr.h
+ * @brief MBR (Master Boot Record) partition table driver.
+ *
+ * Scans block devices for MBR partition tables and registers each
+ * discovered partition as a new block device in the devicefs.
+ * For example, hdd1 with two partitions becomes hdd1mbr1 and hdd1mbr2.
+ *
+ * The partition number (Y in hddNmbrY) is assigned by the devicefs
+ * subsystem using the device class "hddNmbr".
+ */
+
+#ifndef MBR_H
+#define MBR_H
+
+#include <stdint.h>
+
+/** MBR signature bytes (offset 510-511). */
+#define MBR_SIGNATURE 0xAA55
+
+/** Maximum partitions in a standard MBR. */
+#define MBR_MAX_PARTITIONS 4
+
+/** MBR partition table entry (16 bytes each). */
+typedef struct __attribute__((packed)) mbr_partition_entry {
+    uint8_t  status;          /**< 0x80 = bootable, 0x00 = not bootable. */
+    uint8_t  chs_first[3];   /**< CHS address of first sector. */
+    uint8_t  type;            /**< Partition type code. */
+    uint8_t  chs_last[3];    /**< CHS address of last sector. */
+    uint32_t lba_start;      /**< LBA of first sector. */
+    uint32_t sector_count;   /**< Number of sectors. */
+} mbr_partition_entry_t;
+
+/** Well-known MBR partition types. */
+#define MBR_TYPE_EMPTY      0x00
+#define MBR_TYPE_FAT12      0x01
+#define MBR_TYPE_FAT16_SM   0x04  /**< FAT16 < 32 MiB. */
+#define MBR_TYPE_EXTENDED   0x05
+#define MBR_TYPE_FAT16_LG   0x06  /**< FAT16 >= 32 MiB. */
+#define MBR_TYPE_FAT32      0x0B
+#define MBR_TYPE_FAT32_LBA  0x0C
+#define MBR_TYPE_FAT16_LBA  0x0E
+#define MBR_TYPE_LINUX      0x83
+#define MBR_TYPE_LINUX_SWAP 0x82
+
+/**
+ * Scan a block device for MBR partitions.
+ *
+ * Reads sector 0 of the given device, validates the MBR signature,
+ * and registers each non-empty partition entry with the devicefs
+ * as a sub-device.
+ *
+ * @param parent_name  Name of the parent device (e.g., "hdd1").
+ * @return Number of partitions found, or -1 on error.
+ */
+int mbr_scan(const char *parent_name);
+
+/**
+ * Initialize the MBR subsystem.
+ *
+ * Scans all currently registered "hdd" class devices for MBR partitions.
+ *
+ * @return 0 on success.
+ */
+int init_mbr(void);
+
+#endif /* MBR_H */

src/port_io.h

@@ -15,6 +15,18 @@ static inline uint8_t inb(uint16_t port)
     return ret;
 }
 
+static inline void outw(uint16_t port, uint16_t val)
+{
+    asm volatile ( "outw %w0, %w1" : : "a"(val), "Nd"(port) );
+}
+
+static inline uint16_t inw(uint16_t port)
+{
+    uint16_t ret;
+    asm volatile ( "inw %w1, %w0" : "=a"(ret) : "Nd"(port) );
+    return ret;
+}
+
 static inline void io_wait(void)
 {
     /* Port 0x80 is used for 'checkpoints' during POST. */

src/syscall.c

@@ -61,6 +61,11 @@ static int32_t sys_write(registers_t *regs) {
         return (int32_t)len;
     }
 
+    /* VFS file descriptors (fd >= 3) */
+    if (fd >= 3) {
+        return vfs_write(fd, buf, len);
+    }
+
     return -1; /* Invalid fd */
 }
 
@@ -83,6 +88,11 @@ static int32_t sys_read(registers_t *regs) {
         return 0; /* No data available */
     }
 
+    /* VFS file descriptors (fd >= 3) */
+    if (fd >= 3) {
+        return vfs_read(fd, buf, len);
+    }
+
     return -1; /* Invalid fd */
 }
 
@@ -282,6 +292,29 @@ static int32_t sys_setenv(registers_t *regs) {
     return env_set(&cur->env, key, value);
 }
 
+/**
+ * Handle SYS_OPEN: open a file by path.
+ * EBX = path string, ECX = flags.
+ * Returns file descriptor (>= 3) on success, -1 on failure.
+ */
+static int32_t sys_open(registers_t *regs) {
+    const char *path = (const char *)regs->ebx;
+    uint32_t flags = regs->ecx;
+    return (int32_t)vfs_open(path, flags);
+}
+
+/**
+ * Handle SYS_CLOSE: close a file descriptor.
+ * EBX = fd.
+ * Returns 0 on success, -1 on failure.
+ */
+static int32_t sys_close(registers_t *regs) {
+    int fd = (int)regs->ebx;
+    if (fd < 3) return -1;  /* Don't close stdin/stdout/stderr */
+    vfs_close(fd);
+    return 0;
+}
+
 /**
  * Handle SYS_READDIR: read a directory entry.
  * EBX = path, ECX = index, EDX = name buffer (128 bytes min).
@@ -320,6 +353,8 @@ static syscall_fn syscall_table[NUM_SYSCALLS] = {
     [SYS_GETENV]  = sys_getenv,
     [SYS_SETENV]  = sys_setenv,
     [SYS_READDIR] = sys_readdir,
+    [SYS_OPEN]    = sys_open,
+    [SYS_CLOSE]   = sys_close,
 };
 
 void syscall_handler(registers_t *regs) {

src/syscall.h

@@ -25,9 +25,11 @@
 #define SYS_GETENV  8   /**< Get environment variable. key=EBX, buf=ECX, bufsize=EDX. */
 #define SYS_SETENV  9   /**< Set environment variable. key=EBX, value=ECX. */
 #define SYS_READDIR 10  /**< Read directory entry. path=EBX, idx=ECX, buf=EDX. Returns type or -1. */
+#define SYS_OPEN    11  /**< Open a file. path=EBX, flags=ECX. Returns fd or -1. */
+#define SYS_CLOSE   12  /**< Close a file descriptor. fd=EBX. Returns 0 or -1. */
 
 /** Total number of system calls. */
-#define NUM_SYSCALLS 11
+#define NUM_SYSCALLS 13
 
 /**
  * Initialize the system call handler.

src/sysfs.c

@@ -0,0 +1,294 @@
+/**
+ * @file sysfs.c
+ * @brief System filesystem (sysfs) implementation.
+ *
+ * A VFS driver mounted at /sys that lets kernel drivers expose virtual
+ * text files.  Drivers register namespaces (e.g., "ide", "mem") and
+ * provide callbacks for listing entries and reading/writing file content.
+ *
+ * Path structure:
+ *   /sys/<namespace>/<subpath...>
+ *
+ * The VFS routes requests through sysfs_readdir, sysfs_finddir, and
+ * sysfs_read/write, which decompose the path and forward it to the
+ * appropriate driver's callbacks.
+ */
+
+#include "sysfs.h"
+#include "vfs.h"
+#include <string.h>
+
+/* Debug print helpers defined in kernel.c */
+extern void offset_print(const char *str);
+extern void print_hex(uint32_t val);
+
+/* ================================================================
+ * Namespace registry
+ * ================================================================ */
+
+/** A registered sysfs namespace. */
+typedef struct sysfs_ns {
+    char name[SYSFS_MAX_NAME];  /**< Namespace name. */
+    sysfs_ops_t *ops;           /**< Driver callbacks. */
+    void *ctx;                  /**< Driver context. */
+    int active;                 /**< 1 if registered. */
+} sysfs_ns_t;
+
+/** Namespace table. */
+static sysfs_ns_t namespaces[SYSFS_MAX_NAMESPACES];
+
+/**
+ * Find a namespace by name.
+ * @return Pointer to the namespace, or NULL.
+ */
+static sysfs_ns_t *find_ns(const char *name) {
+    for (int i = 0; i < SYSFS_MAX_NAMESPACES; i++) {
+        if (namespaces[i].active && strcmp(namespaces[i].name, name) == 0) {
+            return &namespaces[i];
+        }
+    }
+    return NULL;
+}
+
+/* ================================================================
+ * Path helpers
+ * ================================================================ */
+
+/**
+ * Extract the first path component from a relative path.
+ *
+ * Given "ide/hdd1/model", sets component="ide" and rest="hdd1/model".
+ * Given "ide", sets component="ide" and rest="".
+ *
+ * @param path      Input relative path.
+ * @param component Output buffer for the first component.
+ * @param comp_size Size of component buffer.
+ * @param rest      Output pointer to the remainder of the path.
+ */
+static void split_first(const char *path, char *component,
+                         uint32_t comp_size, const char **rest) {
+    /* Skip leading slashes */
+    while (*path == '/') path++;
+
+    const char *slash = path;
+    while (*slash && *slash != '/') slash++;
+
+    uint32_t len = (uint32_t)(slash - path);
+    if (len >= comp_size) len = comp_size - 1;
+    memcpy(component, path, len);
+    component[len] = '\0';
+
+    /* Skip the slash */
+    if (*slash == '/') slash++;
+    *rest = slash;
+}
+
+/* ================================================================
+ * VFS callbacks for /sys
+ * ================================================================ */
+
+/**
+ * Read from a sysfs virtual file.
+ *
+ * The node's fs_data points to the sysfs_ns_t.
+ * The node's name encodes the sub-path (set by finddir).
+ *
+ * We store the full sub-path (relative to namespace) in node->name.
+ */
+static int32_t sysfs_vfs_read(vfs_node_t *node, uint32_t offset,
+                                uint32_t size, void *buf) {
+    if (!node || !node->fs_data) return -1;
+
+    sysfs_ns_t *ns = (sysfs_ns_t *)node->fs_data;
+    if (!ns->ops || !ns->ops->read) return -1;
+
+    /* Read the full content into a stack buffer */
+    char content[SYSFS_MAX_CONTENT];
+    int total = ns->ops->read(ns->ctx, node->name, content,
+                               sizeof(content));
+    if (total < 0) return -1;
+
+    /* Apply offset */
+    if (offset >= (uint32_t)total) return 0;
+    uint32_t avail = (uint32_t)total - offset;
+    if (size > avail) size = avail;
+    memcpy(buf, content + offset, size);
+    return (int32_t)size;
+}
+
+/**
+ * Write to a sysfs virtual file.
+ */
+static int32_t sysfs_vfs_write(vfs_node_t *node, uint32_t offset,
+                                 uint32_t size, const void *buf) {
+    if (!node || !node->fs_data) return -1;
+
+    sysfs_ns_t *ns = (sysfs_ns_t *)node->fs_data;
+    if (!ns->ops || !ns->ops->write) return -1;
+
+    return ns->ops->write(ns->ctx, node->name, (const char *)buf, size);
+}
+
+/**
+ * Read directory entries under /sys or /sys/<namespace>/...
+ *
+ * When the directory node has no fs_data, we're at /sys root → list
+ * namespaces. When fs_data is a namespace, delegate to its list callback.
+ */
+static int sysfs_vfs_readdir(vfs_node_t *dir, uint32_t idx,
+                               vfs_dirent_t *out) {
+    if (!dir) return -1;
+
+    /* Root of /sys → list registered namespaces */
+    if (dir->fs_data == NULL) {
+        uint32_t count = 0;
+        for (int i = 0; i < SYSFS_MAX_NAMESPACES; i++) {
+            if (!namespaces[i].active) continue;
+            if (count == idx) {
+                memset(out, 0, sizeof(vfs_dirent_t));
+                strncpy(out->name, namespaces[i].name,
+                        VFS_MAX_NAME - 1);
+                out->type = VFS_DIRECTORY;
+                out->inode = (uint32_t)i;
+                return 0;
+            }
+            count++;
+        }
+        return -1;
+    }
+
+    /* Subdirectory within a namespace */
+    sysfs_ns_t *ns = (sysfs_ns_t *)dir->fs_data;
+    if (!ns->ops || !ns->ops->list) return -1;
+
+    sysfs_entry_t entry;
+    int ret = ns->ops->list(ns->ctx, dir->name, idx, &entry);
+    if (ret != 0) return -1;
+
+    memset(out, 0, sizeof(vfs_dirent_t));
+    strncpy(out->name, entry.name, VFS_MAX_NAME - 1);
+    out->type = entry.is_dir ? VFS_DIRECTORY : VFS_FILE;
+    out->inode = idx;
+    return 0;
+}
+
+/**
+ * Look up a child by name inside a sysfs directory.
+ *
+ * At the /sys root (fs_data == NULL), we look up namespace names.
+ * Inside a namespace, we ask the driver's list callback to check
+ * whether the child is a file or directory.
+ */
+static int sysfs_vfs_finddir(vfs_node_t *dir, const char *name,
+                               vfs_node_t *out) {
+    if (!dir) return -1;
+
+    memset(out, 0, sizeof(vfs_node_t));
+
+    /* Root of /sys → find a namespace */
+    if (dir->fs_data == NULL) {
+        sysfs_ns_t *ns = find_ns(name);
+        if (!ns) return -1;
+
+        strncpy(out->name, "", VFS_MAX_NAME - 1);  /* sub-path is "" */
+        out->type = VFS_DIRECTORY;
+        out->fs_data = ns;
+        return 0;
+    }
+
+    /* Inside a namespace: build sub-path */
+    sysfs_ns_t *ns = (sysfs_ns_t *)dir->fs_data;
+    if (!ns->ops || !ns->ops->list) return -1;
+
+    /* The dir->name contains the current sub-path (relative to namespace).
+     * The child path is dir->name + "/" + name (or just name if dir->name
+     * is empty). */
+    char child_path[VFS_MAX_NAME];
+    if (dir->name[0] == '\0') {
+        strncpy(child_path, name, VFS_MAX_NAME - 1);
+        child_path[VFS_MAX_NAME - 1] = '\0';
+    } else {
+        /* Manually concatenate dir->name + "/" + name */
+        uint32_t dlen = strlen(dir->name);
+        uint32_t nlen = strlen(name);
+        if (dlen + 1 + nlen >= VFS_MAX_NAME) return -1;
+        memcpy(child_path, dir->name, dlen);
+        child_path[dlen] = '/';
+        memcpy(child_path + dlen + 1, name, nlen);
+        child_path[dlen + 1 + nlen] = '\0';
+    }
+
+    /* Iterate the list callback to see if this child exists */
+    sysfs_entry_t entry;
+    for (uint32_t i = 0; ; i++) {
+        int ret = ns->ops->list(ns->ctx, dir->name, i, &entry);
+        if (ret != 0) break;  /* no more entries */
+        if (strcmp(entry.name, name) == 0) {
+            strncpy(out->name, child_path, VFS_MAX_NAME - 1);
+            out->type = entry.is_dir ? VFS_DIRECTORY : VFS_FILE;
+            out->fs_data = ns;
+
+            /* For files, get the content size */
+            if (!entry.is_dir && ns->ops->read) {
+                char tmp[SYSFS_MAX_CONTENT];
+                int sz = ns->ops->read(ns->ctx, child_path, tmp, sizeof(tmp));
+                out->size = (sz > 0) ? (uint32_t)sz : 0;
+            }
+
+            return 0;
+        }
+    }
+
+    return -1;  /* child not found */
+}
+
+/** VFS operations for /sys. */
+static vfs_fs_ops_t sysfs_vfs_ops = {
+    .open    = NULL,
+    .close   = NULL,
+    .read    = sysfs_vfs_read,
+    .write   = sysfs_vfs_write,
+    .readdir = sysfs_vfs_readdir,
+    .finddir = sysfs_vfs_finddir,
+};
+
+/* ================================================================
+ * Public API
+ * ================================================================ */
+
+void init_sysfs(void) {
+    memset(namespaces, 0, sizeof(namespaces));
+    vfs_mount("/sys", &sysfs_vfs_ops, NULL);
+    offset_print("  Sysfs mounted at /sys\n");
+}
+
+int sysfs_register(const char *name, sysfs_ops_t *ops, void *ctx) {
+    if (!name || !ops) return -1;
+
+    /* Check for duplicate */
+    if (find_ns(name)) {
+        offset_print("  Sysfs: namespace '");
+        offset_print(name);
+        offset_print("' already exists\n");
+        return -1;
+    }
+
+    /* Find a free slot */
+    for (int i = 0; i < SYSFS_MAX_NAMESPACES; i++) {
+        if (!namespaces[i].active) {
+            strncpy(namespaces[i].name, name, SYSFS_MAX_NAME - 1);
+            namespaces[i].name[SYSFS_MAX_NAME - 1] = '\0';
+            namespaces[i].ops = ops;
+            namespaces[i].ctx = ctx;
+            namespaces[i].active = 1;
+
+            offset_print("  Sysfs: registered namespace '");
+            offset_print(name);
+            offset_print("'\n");
+            return 0;
+        }
+    }
+
+    offset_print("  Sysfs: no free namespace slots\n");
+    return -1;
+}

src/sysfs.h

@@ -0,0 +1,100 @@
+/**
+ * @file sysfs.h
+ * @brief System filesystem (sysfs) interface.
+ *
+ * Allows kernel drivers to expose text/config files to userspace via the
+ * VFS at /sys.  Each driver registers a namespace (e.g., "ide") and
+ * provides callbacks for listing, reading, and writing virtual files.
+ *
+ * Usage:
+ *   1. Define a sysfs_ops_t with list/read/write callbacks.
+ *   2. Call sysfs_register("ide", &my_ops, my_context).
+ *   3. Users can then `ls /sys/ide`, `cat /sys/ide/model`, etc.
+ *
+ * File contents are small text strings stored on the stack; the read
+ * callback fills a caller-provided buffer.
+ */
+
+#ifndef SYSFS_H
+#define SYSFS_H
+
+#include <stdint.h>
+#include <stddef.h>
+
+/** Maximum number of sysfs namespaces. */
+#define SYSFS_MAX_NAMESPACES 32
+
+/** Maximum namespace name length. */
+#define SYSFS_MAX_NAME 64
+
+/** Maximum size for sysfs file content (text buffer). */
+#define SYSFS_MAX_CONTENT 512
+
+/**
+ * Sysfs file entry returned by the list callback.
+ */
+typedef struct sysfs_entry {
+    char name[SYSFS_MAX_NAME];  /**< File or directory name. */
+    uint8_t is_dir;              /**< 1 if directory, 0 if file. */
+} sysfs_entry_t;
+
+/**
+ * Callbacks provided by a driver to expose files under its namespace.
+ *
+ * All paths are relative to the namespace root. For a namespace "ide",
+ * a path of "hdd1/model" means the file "model" inside subdirectory "hdd1".
+ */
+typedef struct sysfs_ops {
+    /**
+     * List entries in a directory.
+     *
+     * @param ctx      Driver context pointer (from registration).
+     * @param path     Relative path within the namespace ("" for root).
+     * @param idx      Entry index (0-based).
+     * @param out      Output entry.
+     * @return 0 on success, -1 when no more entries.
+     */
+    int (*list)(void *ctx, const char *path, uint32_t idx,
+                sysfs_entry_t *out);
+
+    /**
+     * Read a file's content as a text string.
+     *
+     * @param ctx      Driver context pointer.
+     * @param path     Relative path to the file within the namespace.
+     * @param buf      Output buffer for file content.
+     * @param buf_size Buffer size.
+     * @return Number of bytes written to buf, or -1 on error.
+     */
+    int (*read)(void *ctx, const char *path, char *buf, uint32_t buf_size);
+
+    /**
+     * Write data to a file.
+     *
+     * @param ctx      Driver context pointer.
+     * @param path     Relative path to the file within the namespace.
+     * @param buf      Input data.
+     * @param size     Number of bytes to write.
+     * @return Number of bytes consumed, or -1 on error.
+     */
+    int (*write)(void *ctx, const char *path, const char *buf, uint32_t size);
+} sysfs_ops_t;
+
+/**
+ * Initialize the sysfs and mount it at /sys.
+ */
+void init_sysfs(void);
+
+/**
+ * Register a driver namespace in sysfs.
+ *
+ * After registration, files are accessible under /sys/<name>/.
+ *
+ * @param name  Namespace name (e.g., "ide").
+ * @param ops   Callback operations.
+ * @param ctx   Opaque driver context passed to all callbacks.
+ * @return 0 on success, -1 if the table is full or name is invalid.
+ */
+int sysfs_register(const char *name, sysfs_ops_t *ops, void *ctx);
+
+#endif /* SYSFS_H */