refactor: promote new architecture to repository root

Archive the deprecated implementation under lume-legacy, make the new kernel the default root-level build, and update repository entry points without changing kernel behavior.

3 files changed, 1432 insertions, 0 deletions

diff --git a/lume-legacy/src/fs/fat32.cc b/lume-legacy/src/fs/fat32.cc
new file mode 100644
index 0000000..e4089d4
--- /dev/null
+++ b/lume-legacy/src/fs/fat32.cc
@@ -0,0 +1,806 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Richard QIn
+ */
+#include "fs/fat32.h"
+#include "lib/string.h"
+#include "drivers/uart.h"
+#include "kernel/mm.h"
+#include "kernel/proc.h"
+
+FAT32FileSystem fat32_fs;
+
+// Helper function
+static void generate_short_name(char *dst, const char *src)
+{
+    memset(dst, ' ', 11);
+
+    int i = 0;
+    for (; *src && *src != '.' && i < 8; src++)
+    {
+        char c = *src;
+        if (c >= 'a' && c <= 'z')
+            c -= 32;
+        dst[i++] = c;
+    }
+
+    while (*src && *src != '.')
+        src++;
+    if (*src == '.')
+        src++;
+
+    i = 8;
+    for (; *src && i < 11; src++)
+    {
+        char c = *src;
+        if (c >= 'a' && c <= 'z')
+            c -= 32;
+        dst[i++] = c;
+    }
+}
+
+// FAT32 FileSystem Implementation
+
+void FAT32FileSystem::init()
+{
+    struct Buf *b = BufferCache::bread(0, 0);
+    uint8 *data = (uint8 *)b->data;
+    struct FAT32_BPB *bpb = (struct FAT32_BPB *)data;
+
+    uint32 partition_lba = 0;
+
+    bool is_valid_bpb = (bpb->bytesPerSector == 512) && (bpb->rootClus != 0);
+
+    if (!is_valid_bpb)
+    {
+        Drivers::uart_puts("[FAT32] Sector 0 is not a valid BPB. Checking MBR...\n");
+
+        if (data[510] != 0x55 || data[511] != 0xAA)
+        {
+            Drivers::uart_puts("[FAT32] Panic: Invalid Disk Signature!\n");
+            while (1)
+                ;
+        }
+
+        uint32 lba = *(uint32 *)(data + 446 + 8);
+
+        if (lba > 0)
+        {
+            Drivers::uart_puts("[FAT32] MBR Partition 1 found @ Sector ");
+            Drivers::uart_put_int(lba);
+            Drivers::uart_puts("\n");
+            partition_lba = lba;
+        }
+        else
+        {
+            Drivers::uart_puts("[FAT32] Panic: No partition found.\n");
+            while (1)
+                ;
+        }
+
+        BufferCache::brelse(b);
+
+        b = BufferCache::bread(0, partition_lba / 2);
+        bpb = (struct FAT32_BPB *)b->data;
+    }
+
+    if (bpb->bytesPerSector != 512)
+    {
+        Drivers::uart_puts("[FAT32] Panic: Invalid FAT32 BPB in partition.\n");
+        while (1)
+            ;
+    }
+
+    uint32 fat_sz = bpb->FATSz32;
+    this->fat_sz = fat_sz;
+    uint32 reserved_sec = bpb->reservedSecCnt;
+    uint8 num_fats = bpb->numFATs;
+
+    sectors_per_cluster = bpb->secPerClus;
+    bytes_per_cluster = sectors_per_cluster * 512;
+    root_cluster = bpb->rootClus;
+
+    fat_start_sector = partition_lba + reserved_sec;
+    data_start_sector = partition_lba + reserved_sec + (fat_sz * num_fats);
+
+    Drivers::uart_puts(ANSI_GREEN);
+    Drivers::uart_puts("[FAT32] Init Success: RootClus=");
+    Drivers::uart_put_int(root_cluster);
+    Drivers::uart_puts(", DataStart=");
+    Drivers::uart_put_int(data_start_sector);
+    Drivers::uart_puts(ANSI_RESET);
+    Drivers::uart_puts("\n");
+
+    BufferCache::brelse(b);
+}
+
+Inode *FAT32FileSystem::root()
+{
+    FAT32Inode *ip = new FAT32Inode();
+
+    ip->fs = this;
+    ip->dev = 0;
+    ip->inum = root_cluster;
+    ip->ref_cnt = 1;
+    ip->valid = 1;
+    ip->type = T_DIR;
+    ip->size = 0;
+    ip->nlink = 1;
+
+    ip->first_clus = root_cluster;
+    ip->parent_clus = 0;
+
+    return ip;
+}
+
+uint32 FAT32FileSystem::cluster_to_sector(uint32 cluster)
+{
+    return data_start_sector + (cluster - 2) * sectors_per_cluster;
+}
+
+uint32 FAT32FileSystem::next_cluster(uint32 cluster)
+{
+    uint32 fat_offset = cluster * 4;
+    uint32 fat_sector = fat_start_sector + (fat_offset / 512);
+    uint32 ent_offset = fat_offset % 512;
+
+    // block = sector / 2
+    struct Buf *b = BufferCache::bread(0, fat_sector / 2);
+
+    uint32 block_offset = (fat_sector % 2) * 512 + ent_offset;
+    uint32 next = *(uint32 *)(b->data + block_offset);
+
+    BufferCache::brelse(b);
+
+    return next & 0x0FFFFFFF;
+}
+
+// FAT32 Inode Implementation
+
+static bool match_name(const char *disk_name, const char *target)
+{
+    char temp[13];
+    int i, j;
+
+    for (i = 0; i < 8 && disk_name[i] != ' '; i++)
+        temp[i] = disk_name[i];
+    if (disk_name[8] != ' ')
+    {
+        temp[i++] = '.';
+        for (j = 0; j < 3 && disk_name[8 + j] != ' '; j++)
+            temp[i++] = disk_name[8 + j];
+    }
+    temp[i] = 0;
+    return strcasecmp(temp, target) == 0;
+}
+
+int FAT32Inode::read(char *dst, uint32 off, uint32 n, int isUser)
+{
+    if (first_clus == 0)
+        return 0;
+
+    // [FIX] 只有非目录文件才检查 size 边界
+    // 目录文件 size 可能为 0，应允许读取直到簇链结束
+    if (type != T_DIR)
+    {
+        if (off >= size)
+        {
+            return 0;
+        }
+        if (off + n > size)
+        {
+            n = size - off;
+        }
+    }
+
+    uint32 tot = 0;
+    uint32 clus = first_clus;
+    uint32 clus_sz = fs->bytes_per_cluster;
+
+    uint32 skip = off / clus_sz;
+    uint32 current_off = off % clus_sz;
+
+    while (skip > 0)
+    {
+        clus = fs->next_cluster(clus);
+        if (clus >= 0x0FFFFFF8)
+            return 0;
+        skip--;
+    }
+
+    while (tot < n)
+    {
+        uint32 sec_in_clus = current_off / 512;
+
+        if (sec_in_clus >= fs->sectors_per_cluster)
+        {
+            clus = fs->next_cluster(clus);
+            if (clus >= 0x0FFFFFF8)
+                break;
+            current_off = 0;
+            continue;
+        }
+
+        uint32 sec = fs->cluster_to_sector(clus) + sec_in_clus;
+        uint32 off_in_sec = current_off % 512;
+
+        uint32 m = 512 - off_in_sec;
+        if (n - tot < m)
+            m = n - tot;
+
+        struct Buf *b = BufferCache::bread(dev, sec / 2);
+        uint32 block_offset = (sec % 2) * 512 + off_in_sec;
+
+        if (isUser)
+        {
+            if (VM::copyout(myproc()->pagetable, (uint64)dst, (char *)(b->data + block_offset), m) < 0)
+            {
+                BufferCache::brelse(b);
+                return -1;
+            }
+        }
+        else
+        {
+            memmove(dst, b->data + block_offset, m);
+        }
+        BufferCache::brelse(b);
+
+        tot += m;
+        dst += m;
+        current_off += m;
+    }
+    return tot;
+}
+
+Inode *FAT32Inode::lookup(const char *name)
+{
+    if (type != T_DIR)
+        return nullptr;
+
+    uint32 clus = first_clus;
+    struct FAT32_DirEnt de;
+
+    while (clus < 0x0FFFFFF8)
+    {
+        uint32 sec = fs->cluster_to_sector(clus);
+
+        for (uint32 i = 0; i < fs->sectors_per_cluster; i++)
+        {
+            uint32 current_sec = sec + i;
+            struct Buf *b = BufferCache::bread(dev, current_sec / 2);
+            uint32 block_base = (current_sec % 2) * 512;
+
+            for (int off = 0; off < 512; off += 32)
+            {
+                memmove(&de, b->data + block_base + off, 32);
+
+                if (de.name[0] == 0x00)
+                {
+                    BufferCache::brelse(b);
+                    return nullptr;
+                }
+                if (de.name[0] == 0xE5)
+                    continue;
+                if (de.attr & ATTR_LONG_NAME)
+                    continue;
+
+                if (match_name(de.name, name))
+                {
+                    FAT32Inode *ip = new FAT32Inode();
+                    ip->fs = fs;
+                    ip->dev = dev;
+                    ip->type = (de.attr & ATTR_DIRECTORY) ? T_DIR : T_FILE;
+                    ip->size = de.fileSize;
+                    ip->first_clus = (de.fstClusHI << 16) | de.fstClusLO;
+                    ip->inum = ip->first_clus;
+                    ip->nlink = 1;
+                    ip->ref_cnt = 1;
+                    ip->valid = 1;
+                    ip->ent_sec = current_sec; // Log Sector
+                    ip->ent_off = off;         // Log Setoff
+
+                    BufferCache::brelse(b);
+                    return ip;
+                }
+            }
+            BufferCache::brelse(b);
+        }
+        clus = fs->next_cluster(clus);
+    }
+    return nullptr;
+}
+
+// Placeholder
+uint32 FAT32FileSystem::alloc_cluster()
+{
+    uint32 fat_entries_per_sector = 512 / 4;
+    for (uint32 i = 0; i < fat_sz; i++)
+    {
+        uint32 current_fat_sec = fat_start_sector + i;
+        struct Buf *b = BufferCache::bread(0, current_fat_sec / 2);
+        uint32 *table = (uint32 *)(b->data + (current_fat_sec % 2) * 512);
+        for (uint32 j = 0; j < fat_entries_per_sector; j++)
+        {
+            if (i == 0 && j < 2)
+                continue;
+
+            if ((table[j] & 0x0FFFFFFF) == 0)
+            {
+                // Find the free cluster and mark it as EOF (0x0FFFFFFF)
+                table[j] = 0x0FFFFFFF;
+                BufferCache::bwrite(b); // Write back to disk
+                BufferCache::brelse(b);
+
+                uint32 cluster_idx = i * fat_entries_per_sector + j;
+
+                // Clear the data content of the new cluster
+                uint32 data_sec = cluster_to_sector(cluster_idx);
+                for (uint32 k = 0; k < sectors_per_cluster; k++)
+                {
+                    struct Buf *db = BufferCache::bread(0, (data_sec + k) / 2);
+                    memset(db->data + ((data_sec + k) % 2) * 512, 0, 512);
+                    BufferCache::bwrite(db);
+                    BufferCache::brelse(db);
+                }
+
+                return cluster_idx;
+            }
+        }
+        BufferCache::brelse(b);
+    }
+    Drivers::uart_puts("[FAT32] Out of space!\n");
+    return 0;
+}
+
+void FAT32FileSystem::free_cluster_chain(uint32 cluster)
+{
+    while (cluster >= 2 && cluster < 0x0FFFFFF8)
+    {
+        uint32 next = next_cluster(cluster);
+
+        // Calculate the current cluster's position in the FAT table
+        uint32 fat_offset = cluster * 4;
+        uint32 fat_sector = fat_start_sector + (fat_offset / 512);
+        uint32 ent_offset = fat_offset % 512;
+
+        // Read FAT sector
+        struct Buf *b = BufferCache::bread(0, fat_sector / 2);
+
+        // Marked as idle (0x00000000)
+        uint32 *entry = (uint32 *)(b->data + (fat_sector % 2) * 512 + ent_offset);
+        *entry = 0;
+
+        BufferCache::bwrite(b);
+        BufferCache::brelse(b);
+
+        cluster = next;
+    }
+}
+
+void FAT32Inode::update()
+{
+    struct Buf *b = BufferCache::bread(dev, ent_sec / 2);
+    uint32 offset = (ent_sec % 2) * 512 + ent_off;
+
+    struct FAT32_DirEnt *de = (struct FAT32_DirEnt *)(b->data + offset);
+
+    // Update file size and starting cluster
+    de->fileSize = size;
+    de->fstClusHI = (uint16)((first_clus >> 16) & 0xFFFF);
+    de->fstClusLO = (uint16)(first_clus & 0xFFFF);
+
+    BufferCache::bwrite(b);
+    BufferCache::brelse(b);
+}
+
+int FAT32Inode::write(const char *src, uint32 off, uint32 n, int isUser)
+{
+    if (off > size)
+    {
+        return -1;
+    }
+
+    uint32 tot = 0;
+    uint32 clus = first_clus;
+    uint32 clus_sz = fs->bytes_per_cluster;
+    uint32 prev_clus = 0;
+
+    if (clus == 0)
+    {
+        clus = fs->alloc_cluster();
+        if (clus == 0)
+            return -1;
+        first_clus = clus;
+        // Update the metadata immediately to prevent data loss in case of a power outage.
+        update();
+    }
+
+    uint32 skip = off / clus_sz;
+    uint32 current_off = off % clus_sz;
+
+    while (skip > 0)
+    {
+        prev_clus = clus;
+        clus = fs->next_cluster(clus);
+        if (clus >= 0x0FFFFFF8)
+        {
+            return -1;
+        }
+        skip--;
+    }
+    while (tot < n)
+    {
+        if (current_off == clus_sz)
+        {
+            uint32 next = fs->next_cluster(clus);
+            if (next >= 0x0FFFFFF8)
+            {
+                uint32 new_clus = fs->alloc_cluster();
+                if (new_clus == 0)
+                    break;
+
+                // Point the current cluster to the new cluster (update the FAT table)
+                uint32 fat_offset = clus * 4;
+                uint32 fat_sec = fs->fat_start_sector + (fat_offset / 512);
+                struct Buf *fat_b = BufferCache::bread(dev, fat_sec / 2);
+                uint32 *entry = (uint32 *)(fat_b->data + (fat_sec % 2) * 512 + (fat_offset % 512));
+                *entry = new_clus; // Link
+                BufferCache::bwrite(fat_b);
+                BufferCache::brelse(fat_b);
+
+                clus = new_clus;
+            }
+            else
+            {
+                clus = next;
+            }
+            current_off = 0;
+        }
+
+        // Calculate the physical sector to write
+        uint32 sec_in_clus = current_off / 512;
+        uint32 sec = fs->cluster_to_sector(clus) + sec_in_clus;
+        uint32 off_in_sec = current_off % 512;
+
+        // Calculate how many bytes can be written this time
+        uint32 m = 512 - off_in_sec;
+        if (n - tot < m)
+            m = n - tot;
+
+        struct Buf *b = BufferCache::bread(dev, sec / 2);
+        uint32 block_offset = (sec % 2) * 512 + off_in_sec;
+
+        if (isUser)
+        {
+            if (VM::copyin(myproc()->pagetable, (char *)(b->data + block_offset), (uint64)src, m) < 0)
+            {
+                BufferCache::brelse(b);
+                return -1;
+            }
+        }
+        else
+        {
+            memmove(b->data + block_offset, src, m);
+        }
+
+        BufferCache::bwrite(b);
+        BufferCache::brelse(b);
+
+        tot += m;
+        src += m;
+        current_off += m;
+    }
+    if (off + n > size)
+    {
+        size = off + n;
+        update();
+    }
+
+    return tot;
+}
+
+Inode *FAT32Inode::create(const char *name, short type, short major, short minor)
+{
+    Inode *exist = lookup(name);
+    if (exist)
+    {
+        return exist;
+    }
+
+    struct FAT32_DirEnt de;
+    memset(&de, 0, sizeof(de));
+    generate_short_name(de.name, name);
+    de.attr = (type == T_DIR) ? ATTR_DIRECTORY : ATTR_ARCHIVE;
+
+    uint32 new_dir_clus = 0;
+    if (type == T_DIR)
+    {
+        new_dir_clus = fs->alloc_cluster();
+        if (new_dir_clus == 0)
+            return nullptr;
+
+        de.fstClusHI = (uint16)((new_dir_clus >> 16) & 0xFFFF);
+        de.fstClusLO = (uint16)(new_dir_clus & 0xFFFF);
+    }
+    else
+    {
+        de.fileSize = 0;
+    }
+
+    uint32 clus = first_clus;
+    bool written = false;
+
+    while (true)
+    {
+        uint32 sec = fs->cluster_to_sector(clus);
+
+        for (uint32 i = 0; i < fs->sectors_per_cluster; i++)
+        {
+            struct Buf *b = BufferCache::bread(dev, (sec + i) / 2);
+            uint8 *data = (uint8 *)b->data + ((sec + i) % 2) * 512;
+
+            for (int off = 0; off < 512; off += 32)
+            {
+                if (data[off] == 0x00 || data[off] == 0xE5)
+                {
+                    memmove(data + off, &de, 32);
+                    BufferCache::bwrite(b);
+                    BufferCache::brelse(b);
+                    written = true;
+                    break;
+                }
+            }
+
+            if (written)
+                break;
+            BufferCache::brelse(b);
+        }
+
+        if (written)
+            break;
+
+        uint32 next = fs->next_cluster(clus);
+        if (next >= 0x0FFFFFF8)
+        {
+            uint32 new_parent_clus = fs->alloc_cluster();
+            if (new_parent_clus == 0)
+                return nullptr;
+
+            uint32 fat_off = clus * 4;
+            uint32 fat_sec = fs->fat_start_sector + (fat_off / 512);
+            struct Buf *fb = BufferCache::bread(dev, fat_sec / 2);
+            *(uint32 *)(fb->data + (fat_sec % 2) * 512 + (fat_off % 512)) = new_parent_clus;
+            BufferCache::bwrite(fb);
+            BufferCache::brelse(fb);
+
+            uint32 s = fs->cluster_to_sector(new_parent_clus);
+            for (uint32 k = 0; k < fs->sectors_per_cluster; k++)
+            {
+                struct Buf *zb = BufferCache::bread(dev, (s + k) / 2);
+                uint8 *zdata = (uint8 *)zb->data + ((s + k) % 2) * 512;
+                memset(zdata, 0, 512);
+
+                if (k == 0)
+                {
+                    memmove(zdata, &de, 32);
+                }
+
+                BufferCache::bwrite(zb);
+                BufferCache::brelse(zb);
+            }
+
+            written = true;
+            break;
+        }
+        clus = next;
+    }
+
+    if (written && type == T_DIR)
+    {
+        struct FAT32_DirEnt dot;
+        memset(&dot, 0, 32);
+        memset(dot.name, ' ', 11);
+        dot.name[0] = '.';
+        dot.attr = ATTR_DIRECTORY;
+        dot.fstClusHI = (uint16)((new_dir_clus >> 16) & 0xFFFF);
+        dot.fstClusLO = (uint16)(new_dir_clus & 0xFFFF);
+
+        struct FAT32_DirEnt dotdot;
+        memset(&dotdot, 0, 32);
+        memset(dotdot.name, ' ', 11);
+        dotdot.name[0] = '.';
+        dotdot.name[1] = '.';
+        dotdot.attr = ATTR_DIRECTORY;
+
+        uint32 parent_clus_num = this->first_clus;
+        if (parent_clus_num == fs->root_cluster)
+        {
+            parent_clus_num = 0;
+        }
+        dotdot.fstClusHI = (uint16)((parent_clus_num >> 16) & 0xFFFF);
+        dotdot.fstClusLO = (uint16)(parent_clus_num & 0xFFFF);
+
+        uint32 s = fs->cluster_to_sector(new_dir_clus);
+
+        struct Buf *zb = BufferCache::bread(dev, s / 2);
+        uint8 *zdata = (uint8 *)zb->data + (s % 2) * 512;
+
+        memmove(zdata, &dot, 32);         // Offset 0
+        memmove(zdata + 32, &dotdot, 32); // Offset 32
+
+        BufferCache::bwrite(zb);
+        BufferCache::brelse(zb);
+    }
+
+    if (written)
+    {
+        return lookup(name);
+    }
+
+    return nullptr;
+}
+
+void FAT32Inode::truncate()
+{
+    if (first_clus == 0)
+    {
+        if (size != 0)
+        {
+            size = 0;
+            update();
+        }
+        return;
+    }
+
+    fs->free_cluster_chain(first_clus);
+
+    size = 0;
+    first_clus = 0;
+
+    update();
+}
+
+int FAT32Inode::unlink(const char *name)
+{
+    if (type != T_DIR)
+        return -1;
+    if (strcmp(name, ".") == 0 || strcmp(name, "..") == 0)
+        return -1;
+
+    uint32_t clus = first_clus;
+    bool found = false;
+
+    // [FIX] Deadlock prevention: Do not hold buffer while looking up inode
+    uint32_t target_block = 0;
+    uint32_t target_offset = 0;
+    struct FAT32_DirEnt de;
+
+    while (true)
+    {
+        uint32_t sec = fs->cluster_to_sector(clus);
+        for (uint32 i = 0; i < fs->sectors_per_cluster; ++i)
+        {
+            struct Buf *b = BufferCache::bread(dev, (sec + i) / 2);
+            uint8_t *data = (uint8_t *)b->data + ((sec + i) % 2) * 512;
+
+            for (int off = 0; off < 512; off += 32)
+            {
+                memmove(&de, data + off, 32);
+                if (de.name[0] == 0x00) // end of dir
+                {
+                    BufferCache::brelse(b);
+                    return -1; // Not found
+                }
+                if (de.name[0] == 0xE5)
+                    continue; // have been unlink
+                if (de.attr & ATTR_LONG_NAME)
+                    continue;
+
+                if (match_name(de.name, name))
+                {
+                    target_block = (sec + i) / 2;
+                    target_offset = ((sec + i) % 2) * 512 + off;
+                    found = true;
+                    break;
+                }
+            }
+
+            BufferCache::brelse(b); // Always release buffer
+            if (found)
+                break;
+        }
+
+        if (found)
+            break;
+
+        clus = fs->next_cluster(clus);
+        if (clus >= 0x0FFFFFF8)
+            return -1;
+    }
+
+    if (found)
+    {
+        Inode *target_ip = lookup(name);
+        if (target_ip)
+        {
+            VFS::ilock(target_ip);
+            if (target_ip->type == T_DIR)
+            {
+                VFS::iunlockput(target_ip);
+                return -1;
+            }
+
+            target_ip->nlink--;
+            target_ip->update();
+            VFS::iunlockput(target_ip);
+        }
+
+        // Re-read buffer to mark deletion
+        struct Buf *target_buf = BufferCache::bread(dev, target_block);
+        uint8_t *ptr = (uint8_t *)target_buf->data + target_offset;
+        *ptr = 0xE5;
+
+        BufferCache::bwrite(target_buf);
+        BufferCache::brelse(target_buf);
+        return 0;
+    }
+
+    return -1;
+}
+
+// Test Function
+void fat32_test()
+{
+    static bool inited = false;
+    if (!inited)
+    {
+        fat32_fs.init();
+        inited = true;
+    }
+
+    Drivers::uart_puts("\n[FAT32] Starting Test via Syscall...\n");
+
+    Inode *root = fat32_fs.root();
+    if (root)
+    {
+        Drivers::uart_puts("[FAT32] Root Inode retrieved. Inum: ");
+        Drivers::uart_put_int(root->inum);
+        Drivers::uart_puts("\n");
+
+        Drivers::uart_puts("[FAT32] Looking for TEST.TXT...\n");
+        Inode *node = root->lookup("test.txt");
+
+        if (node)
+        {
+            Drivers::uart_puts("[FAT32] Found TEST.TXT! Size: ");
+            Drivers::uart_put_int(node->size);
+            Drivers::uart_puts("\n");
+
+            char buf[128];
+            int n = node->read(buf, 0, 127, 0);
+            if (n > 0)
+            {
+                buf[n] = 0;
+                Drivers::uart_puts("[FAT32] Content:\n>> ");
+                Drivers::uart_puts(buf);
+                Drivers::uart_puts("\n");
+            }
+            else
+            {
+                Drivers::uart_puts("[FAT32] Read returned 0.\n");
+            }
+            VFS::iput(node);
+        }
+        else
+        {
+            Drivers::uart_puts("[FAT32] TEST.TXT not found.\n");
+        }
+        VFS::iput(root);
+    }
+    else
+    {
+        Drivers::uart_puts("[FAT32] Failed to get root inode.\n");
+    }
+
+    Drivers::uart_puts("[FAT32] Test Finished.\n");
+}
\ No newline at end of file
diff --git a/lume-legacy/src/fs/file.cc b/lume-legacy/src/fs/file.cc
new file mode 100644
index 0000000..d814332
--- /dev/null
+++ b/lume-legacy/src/fs/file.cc
@@ -0,0 +1,358 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Richard QIn
+ */
+#include "fs/file.h"
+#include "kernel/spinlock.h"
+#include "kernel/mm.h"
+#include "drivers/uart.h"
+#include "kernel/proc.h"
+#include "kernel/slab.h"
+
+#define NFILE 100
+#define SEEK_SET 0
+#define SEEK_CUR 1
+#define SEEK_END 2
+
+extern void *operator new(unsigned long, void *p);
+
+int Pipe::create_pair(file** f0, file** f1)
+{
+    void *mem = Slab::kmalloc(sizeof(Pipe));
+    if (!mem)
+        return -1;
+    Pipe *pi = new (mem) Pipe();
+
+    *f0 = FileTable::alloc();
+    *f1 = FileTable::alloc();
+
+    if (*f0 == nullptr || *f1 == nullptr)
+    {
+        if (*f0)
+            FileTable::close(*f0);
+        if (*f1)
+            FileTable::close(*f1);
+        pi->~Pipe();
+        Slab::kfree(pi);
+        return -1;
+    }
+
+    (*f0)->type = FD_PIPE;
+    (*f0)->readable = 1;
+    (*f0)->writable = 0;
+    (*f0)->pipe = pi;
+
+    (*f1)->type = FD_PIPE;
+    (*f1)->readable = 0;
+    (*f1)->writable = 1;
+    (*f1)->pipe = pi;
+
+    return 0;
+}
+
+void Pipe::close(bool is_writer)
+{
+    lock.acquire();
+    if (is_writer)
+    {
+        write_open = false;
+        ProcManager::wakeup(&nread); // Wake up the process that is currently reading
+    }
+    else
+    {
+        read_open = false;
+        ProcManager::wakeup(&nwrite); // Wake up the process that is writing
+    }
+
+    bool empty = !read_open && !write_open;
+    lock.release();
+
+    if (empty)
+    {
+        this->~Pipe();
+        Slab::kfree(this);
+    }
+}
+
+int Pipe::write(uint64_t addr, int n)
+{
+    int i = 0;
+    struct Proc *p = myproc();
+
+    lock.acquire();
+    while (i < n)
+    {
+        if (!read_open || p->xstate) // The read end is closed or the process is abnormal
+        {
+            lock.release();
+            return -1;
+        }
+
+        if (nwrite == nread + SIZE) // buffer is full
+        {
+            ProcManager::wakeup(&nread);
+            ProcManager::sleep(&nwrite, &lock);
+        }
+        else
+        {
+            char ch;
+            if (VM::copyin(p->pagetable, &ch, addr + i, 1) < 0)
+                break;
+            data[nwrite++ % SIZE] = ch;
+            i++;
+        }
+    }
+    ProcManager::wakeup(&nread);
+    lock.release();
+    return i;
+}
+
+int Pipe::read(uint64_t addr, int n)
+{
+    int i = 0;
+    struct Proc *p = myproc();
+
+    lock.acquire();
+    while (nread == nwrite && write_open) // The buffer is empty and the write end is not closed
+    {
+        if (p->xstate)
+        {
+            lock.release();
+            return -1;
+        }
+        ProcManager::sleep(&nread, &lock);
+    }
+
+    for (i = 0; i < n; ++i)
+    {
+        if (nread == nwrite)
+            break;
+        char ch = data[nread++ % SIZE];
+        if (VM::copyout(p->pagetable, addr + i, &ch, 1) < 0)
+            break;
+    }
+    ProcManager::wakeup(&nwrite);
+    lock.release();
+    return i;
+}
+
+namespace FileTable
+{
+    struct
+    {
+        Spinlock lock;
+        struct file files[NFILE];
+    } ftable;
+
+    void init()
+    {
+        ftable.lock.init("file_table");
+    }
+
+    struct file *alloc()
+    {
+        struct file *f;
+        ftable.lock.acquire();
+        for (f = ftable.files; f < ftable.files + NFILE; f++)
+        {
+            if (f->ref == 0)
+            {
+                f->ref = 1;
+                ftable.lock.release();
+                return f;
+            }
+        }
+        ftable.lock.release();
+        return nullptr;
+    }
+
+    struct file *dup(struct file *f)
+    {
+        ftable.lock.acquire();
+        if (f->ref < 1)
+        {
+            Drivers::uart_puts("panic: file dup\n");
+            while (1)
+                ;
+        }
+        f->ref++;
+        ftable.lock.release();
+        return f;
+    }
+
+    void close(struct file *f)
+    {
+        struct file ff;
+
+        ftable.lock.acquire();
+        if(f->ref < 1)
+        {
+            Drivers::uart_puts("panic: file close\n");
+            while (1)
+                ;
+        }
+        f->ref--;
+        if (f->ref > 0)
+        {
+            ftable.lock.release();
+            return;
+        }
+
+        ff = *f;
+        f->ref = 0;
+        f->type = FD_NONE;
+        ftable.lock.release();
+        if (ff.type == FD_INODE)
+        {
+            // Release the corresponding Inode
+            VFS::iput(ff.ip);
+        }
+        else if (ff.type == FD_PIPE)
+        {
+            if (ff.pipe)
+                ff.pipe->close(ff.writable);
+        }
+    }
+
+    int read(struct file *f, uint64 addr, int n)
+    {
+        if (f->readable == 0)
+            return -1;
+        if (f->type == FD_INODE)
+        {
+            VFS::ilock(f->ip);
+
+            if (f->ip->type == T_DEVICE)
+            {
+                if (f->ip->major == 1)
+                {
+                    VFS::iunlock(f->ip);
+                    return Drivers::console_read(addr, n);
+                }
+            }
+
+            int r = f->ip->read((char*)addr, f->off, n, 1);
+            if(r > 0)
+            {
+                f->off += r;
+            }
+            VFS::iunlock(f->ip);
+            return r;
+        }
+        else if (f->type == FD_PIPE)
+        {
+            if (f->pipe)
+                return f->pipe->read(addr, n);
+        }
+        return -1;
+    }
+    
+    int write(struct file *f, uint64 addr, int n)
+    {
+        if (f->writable == 0)
+            return -1;
+        if (f->type == FD_INODE)
+        {
+            VFS::ilock(f->ip);
+
+            if (f->ip->type == T_DEVICE)
+            {
+                if (f->ip->major == 1) // Console
+                {
+                    VFS::iunlock(f->ip);
+                    struct Proc *proc = myproc();
+                    constexpr int MAX_WRITE_BUF = 128;
+                    char buf[MAX_WRITE_BUF];
+                    int i = 0;
+                    while (i < n)
+                    {
+                        int len = n - i;
+                        if (len > MAX_WRITE_BUF)
+                            len = MAX_WRITE_BUF;
+
+                        if (VM::copyin(proc->pagetable, buf, addr + i, static_cast<uint64>(len)) < 0)
+                        {
+                            return -1;
+                        }
+
+                        for (int j = 0; j < len; j++)
+                        {
+                            Drivers::uart_putc(buf[j]);
+                        }
+                        i += len;
+                    }
+                    return n;
+                }
+            }
+
+            int r = f->ip->write((char *)addr, f->off, n, 1);
+            if (r > 0)
+            {
+                f->off += r;
+            }
+            VFS::iunlock(f->ip);
+            return r;
+        }
+        else if (f->type == FD_PIPE)
+        {
+            if (f->pipe)
+                return f->pipe->write(addr, n);
+        }
+
+        return -1;
+    }
+
+    int stat(struct file *f, uint64 addr)
+    {
+        if (f->type == FD_INODE)
+        {
+            struct kstat st;
+            VFS::ilock(f->ip);
+            f->ip->getattr(&st);
+            VFS::iunlock(f->ip);
+
+            struct Proc *p = myproc();
+            if (VM::copyout(p->pagetable, addr, (char *)&st, sizeof(st)) < 0)
+                return -1;
+            return 0;
+        }
+        return -1;
+    }
+
+    int lseek(struct file *f, int offset, int whence)
+    {
+        if (f->type != FD_INODE)
+            return -1;
+
+        VFS::ilock(f->ip);
+
+        int new_off = f->off;
+        int size = f->ip->size;
+
+        switch (whence)
+        {
+        case SEEK_SET:
+            new_off = offset;
+            break;
+        case SEEK_CUR:
+            new_off += offset;
+            break;
+        case SEEK_END:
+            new_off = size + offset;
+            break;
+        default:
+            VFS::iunlock(f->ip);
+            return -1;
+        }
+
+        if (new_off < 0)
+        {
+            VFS::iunlock(f->ip);
+            return -1;
+        }
+
+        f->off = new_off;
+        VFS::iunlock(f->ip);
+        return new_off;
+    }
+} // namespace FileTable
diff --git a/lume-legacy/src/fs/fs.cc b/lume-legacy/src/fs/fs.cc
new file mode 100644
index 0000000..fc8f555
--- /dev/null
+++ b/lume-legacy/src/fs/fs.cc
@@ -0,0 +1,268 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Richard QIn
+ */
+#include "fs/fs.h"
+#include "kernel/riscv.h"
+#include "drivers/uart.h"
+#include "kernel/spinlock.h"
+#include "kernel/proc.h"
+// #include "fs/fat32.h" // Decoupled
+#include "lib/string.h"
+
+static Inode *global_root_inode = nullptr;
+
+static const char *skipelem(const char *path, char *name)
+{
+    const char *s;
+    int len;
+
+    while (*path == '/')
+        path++;
+    if (*path == 0)
+        return nullptr;
+
+    s = path;
+    while (*path != '/' && *path != 0)
+        path++;
+
+    len = path - s;
+    if (len >= 128)
+        len = 127;
+
+    memmove(name, s, len);
+    name[len] = 0;
+
+    while (*path == '/')
+        path++;
+    return path;
+}
+
+void Inode::getattr(struct kstat *st)
+{
+    st->dev = dev;
+    st->ino = inum;
+    st->type = type;
+    st->nlink = nlink;
+    st->size = size;
+
+    st->uid = 0;
+    st->gid = 0;
+    st->mode = 0;
+}
+
+namespace VFS
+{
+    // File System Registry
+    static FileSystem *filesystems[8];
+    static int fs_count = 0;
+
+    void register_fs(FileSystem *fs)
+    {
+        if (fs_count < 8)
+        {
+            filesystems[fs_count++] = fs;
+        }
+    }
+
+    void init()
+    {
+        Drivers::uart_puts("[VFS] Inode subsystem initialized.\n");
+    }
+
+    void iput(Inode *ip)
+    {
+        if (ip == nullptr)
+            return;
+
+        ip->lock.acquire();
+        if(ip->ref_cnt < 1)
+        {
+            Drivers::uart_puts("panic: iput ref\n");
+            while (1)
+                ;
+        }
+        ip->ref_cnt--;
+        int ref = ip->ref_cnt;
+        ip->lock.release();
+
+        if (ref == 0)
+        {
+            delete ip;
+        }
+    }
+
+    Inode* idup(Inode *ip)
+    {
+        if (ip == nullptr)
+            return nullptr;
+
+        ip->lock.acquire();
+        ip->ref_cnt++;
+        ip->lock.release();
+        return ip;
+    }
+
+    void ilock(Inode *ip)
+    {
+        if (ip == 0 || ip->ref_cnt < 1)
+        {
+            Drivers::uart_puts("panic: ilock\n");
+            while (1)
+                ;
+        }
+
+        ip->lock.acquire();
+
+    }
+
+    void iunlock(Inode *ip)
+    {
+        if (ip == 0 || !ip->lock.holding() || ip->ref_cnt < 1)
+        {
+            Drivers::uart_puts("panic: iunlock\n");
+            while (1)
+                ;
+        }
+        ip->lock.release();
+    }
+
+    void iunlockput(Inode *ip)
+    {
+        if (!ip)
+            return;
+        iunlock(ip);
+        iput(ip);
+    }
+
+    void mount_root(Inode *root)
+    {
+        if (global_root_inode)
+        {
+            Drivers::uart_puts("VFS: Root already mounted!\n");
+            return;
+        }
+        global_root_inode = root;
+        Drivers::uart_puts("[VFS] Root filesystem mounted.\n");
+    }
+
+    Inode *get_root()
+    {
+        if (global_root_inode == nullptr)
+        {
+            // Try to mount the first registered FS
+            if (fs_count > 0)
+            {
+                Drivers::uart_puts("[VFS] Lazy init: Mounting root FS...\n");
+                filesystems[0]->init();
+                mount_root(filesystems[0]->root());
+            }
+            else
+            {
+                Drivers::uart_puts("[VFS] No filesystem registered!\n");
+            }
+        }
+        if (!global_root_inode)
+            return nullptr;
+        return idup(global_root_inode);
+    }
+
+    Inode *namei(const char *path)
+    {
+        char name[128];
+        Inode *ip, *next;
+        struct Proc *p = myproc();
+
+        // 1. 确定起点
+        if (*path == '/')
+        {
+            ip = get_root();
+        }
+        else
+        {
+            if (p->cwd == nullptr)
+            {
+                ip = get_root();
+            }
+            else
+            {
+                ip = idup(p->cwd);
+            }
+        }
+
+        if (ip == nullptr)
+            return nullptr;
+
+        while ((path = skipelem(path, name)) != nullptr)
+        {
+            if (strcmp(name, ".") == 0)
+            {
+                continue;
+            }
+
+            if (strcmp(name, "..") == 0)
+            {
+                Inode *root = get_root();
+                bool is_root = (ip->inum == root->inum);
+                iput(root);
+
+                if (is_root)
+                {
+                    continue;
+                }
+            }
+
+            ilock(ip);
+
+            if (ip->type != T_DIR)
+            {
+                iunlockput(ip);
+                return nullptr;
+            }
+
+            next = ip->lookup(name);
+            iunlockput(ip);
+
+            ip = next;
+            if (ip == nullptr)
+                return nullptr;
+        }
+
+        return ip;
+    }
+
+    Inode *nameiparent(const char* path, char* name)
+    {
+        Inode* ip;
+        Inode* next;
+
+        if (*path == '/')
+            ip = get_root();
+        else
+            ip = idup(myproc()->cwd ? myproc()->cwd : get_root());
+
+        while ((path = skipelem(path, name)) != nullptr)
+        {
+            if (*path == 0)
+            {
+                return ip;
+            }
+
+            ilock(ip);
+            if (ip->type != T_DIR)
+            {
+                iunlockput(ip);
+                return nullptr;
+            }
+
+            next = ip->lookup(name);
+            iunlockput(ip);
+
+            ip = next;
+            if (!ip)
+                return nullptr;
+        }
+        iput(ip);
+        return nullptr;
+    }
+} // namespace VFS