banan-os/kernel/kernel/FS/Ext2/FileSystem.cpp

507 lines
16 KiB
C++

#include <BAN/ScopeGuard.h>
#include <kernel/FS/Ext2/FileSystem.h>
#include <kernel/Lock/LockGuard.h>
#define EXT2_DEBUG_PRINT 0
#define EXT2_VERIFY_INODE 0
#define EXT2_VERIFY_NO_BLOCKS 1
namespace Kernel
{
BAN::ErrorOr<Ext2FS*> Ext2FS::create(BAN::RefPtr<BlockDevice> block_device)
{
Ext2FS* ext2fs = new Ext2FS(block_device);
if (ext2fs == nullptr)
return BAN::Error::from_errno(ENOMEM);
BAN::ScopeGuard guard([ext2fs] { delete ext2fs; });
TRY(ext2fs->initialize_superblock());
TRY(ext2fs->initialize_root_inode());
guard.disable();
return ext2fs;
}
BAN::ErrorOr<void> Ext2FS::initialize_superblock()
{
// Read superblock from disk
{
const uint32_t sector_size = m_block_device->blksize();
ASSERT(1024 % sector_size == 0);
const uint32_t lba = 1024 / sector_size;
const uint32_t sector_count = BAN::Math::div_round_up<uint32_t>(sizeof(Ext2::Superblock), sector_size);
BAN::Vector<uint8_t> superblock_buffer;
TRY(superblock_buffer.resize(sector_count * sector_size));
TRY(m_block_device->read_blocks(lba, sector_count, BAN::ByteSpan(superblock_buffer.span())));
memcpy(&m_superblock, superblock_buffer.data(), sizeof(Ext2::Superblock));
}
if (m_superblock.magic != Ext2::Enum::SUPER_MAGIC)
return BAN::Error::from_error_code(ErrorCode::Ext2_Invalid);
if (m_superblock.rev_level == Ext2::Enum::GOOD_OLD_REV)
{
memset(m_superblock.__extension_start, 0, sizeof(Ext2::Superblock) - offsetof(Ext2::Superblock, Ext2::Superblock::__extension_start));
m_superblock.first_ino = Ext2::Enum::GOOD_OLD_FIRST_INO;
m_superblock.inode_size = Ext2::Enum::GOOD_OLD_INODE_SIZE;
}
const uint32_t number_of_block_groups = BAN::Math::div_round_up(superblock().inodes_count, superblock().inodes_per_group);
const uint32_t number_of_block_groups_check = BAN::Math::div_round_up(superblock().blocks_count, superblock().blocks_per_group);
if (number_of_block_groups != number_of_block_groups_check)
return BAN::Error::from_error_code(ErrorCode::Ext2_Corrupted);
if (!(m_superblock.feature_incompat & Ext2::Enum::FEATURE_INCOMPAT_FILETYPE))
{
dwarnln("Directory entries without filetype not supported");
return BAN::Error::from_errno(ENOTSUP);
}
if (m_superblock.feature_incompat & Ext2::Enum::FEATURE_INCOMPAT_COMPRESSION)
{
dwarnln("Required FEATURE_INCOMPAT_COMPRESSION");
return BAN::Error::from_errno(ENOTSUP);
}
if (m_superblock.feature_incompat & Ext2::Enum::FEATURE_INCOMPAT_JOURNAL_DEV)
{
dwarnln("Required FEATURE_INCOMPAT_JOURNAL_DEV");
return BAN::Error::from_errno(ENOTSUP);
}
if (m_superblock.feature_incompat & Ext2::Enum::FEATURE_INCOMPAT_META_BG)
{
dwarnln("Required FEATURE_INCOMPAT_META_BG");
return BAN::Error::from_errno(ENOTSUP);
}
if (m_superblock.feature_incompat & Ext2::Enum::FEATURE_INCOMPAT_RECOVER)
{
dwarnln("Required FEATURE_INCOMPAT_RECOVER");
return BAN::Error::from_errno(ENOTSUP);
}
#if EXT2_DEBUG_PRINT
dprintln("EXT2");
dprintln(" inodes {}", m_superblock.inodes_count);
dprintln(" blocks {}", m_superblock.blocks_count);
dprintln(" version {}.{}", m_superblock.rev_level, m_superblock.minor_rev_level);
dprintln(" first data at {}", m_superblock.first_data_block);
dprintln(" block size {}", 1024 << m_superblock.log_block_size);
dprintln(" inode size {}", m_superblock.inode_size);
dprintln(" inodes/group {}", m_superblock.inodes_per_group);
#endif
TRY(m_buffer_manager.initialize(block_size()));
{
auto block_buffer = m_buffer_manager.get_buffer();
if (superblock().rev_level == Ext2::Enum::GOOD_OLD_REV)
{
// In revision 0 all blockgroups contain superblock backup
TRY(m_superblock_backups.reserve(number_of_block_groups - 1));
for (uint32_t i = 1; i < number_of_block_groups; i++)
MUST(m_superblock_backups.push_back(i));
}
else
{
// In other revision superblock backups are on blocks 1 and powers of 3, 5 and 7
if (1 < number_of_block_groups)
TRY(m_superblock_backups.push_back(1));
for (uint32_t i = 3; i < number_of_block_groups; i *= 3)
TRY(m_superblock_backups.push_back(i));
for (uint32_t i = 5; i < number_of_block_groups; i *= 5)
TRY(m_superblock_backups.push_back(i));
for (uint32_t i = 7; i < number_of_block_groups; i *= 7)
TRY(m_superblock_backups.push_back(i));
// We don't really care if this succeeds or not
(void)m_superblock_backups.shrink_to_fit();
}
for (uint32_t bg : m_superblock_backups)
{
read_block(superblock().first_data_block + superblock().blocks_per_group * bg, block_buffer);
Ext2::Superblock& superblock_backup = *(Ext2::Superblock*)block_buffer.data();
if (superblock_backup.magic != Ext2::Enum::SUPER_MAGIC)
derrorln("superblock backup at block {} is invalid ({4H})", bg, superblock_backup.magic);
}
}
return {};
}
BAN::ErrorOr<void> Ext2FS::initialize_root_inode()
{
m_root_inode = TRY(Ext2Inode::create(*this, Ext2::Enum::ROOT_INO));
return {};
}
BAN::ErrorOr<uint32_t> Ext2FS::create_inode(const Ext2::Inode& ext2_inode)
{
LockGuard _(m_mutex);
ASSERT(ext2_inode.size == 0);
if (m_superblock.free_inodes_count == 0)
return BAN::Error::from_errno(ENOSPC);
const uint32_t block_size = this->block_size();
auto bgd_buffer = m_buffer_manager.get_buffer();
auto inode_bitmap = m_buffer_manager.get_buffer();
uint32_t current_group = -1;
BlockLocation bgd_location {};
Ext2::BlockGroupDescriptor* bgd = nullptr;
for (uint32_t ino = superblock().first_ino; ino <= superblock().inodes_count; ino++)
{
const uint32_t ino_group = (ino - 1) / superblock().inodes_per_group;
const uint32_t ino_index = (ino - 1) % superblock().inodes_per_group;
if (ino_group != current_group)
{
current_group = ino_group;
bgd_location = locate_block_group_descriptior(current_group);
read_block(bgd_location.block, bgd_buffer);
bgd = (Ext2::BlockGroupDescriptor*)(bgd_buffer.data() + bgd_location.offset);
if (bgd->free_inodes_count == 0)
{
ino = superblock().first_ino + (current_group + 1) * superblock().inodes_per_group - 1;
continue;
}
read_block(bgd->inode_bitmap, inode_bitmap);
}
const uint32_t ino_bitmap_byte = ino_index / 8;
const uint32_t ino_bitmap_bit = ino_index % 8;
if (inode_bitmap[ino_bitmap_byte] & (1 << ino_bitmap_bit))
continue;
inode_bitmap[ino_bitmap_byte] |= 1 << ino_bitmap_bit;
write_block(bgd->inode_bitmap, inode_bitmap);
bgd->free_inodes_count--;
if (Inode::Mode(ext2_inode.mode).ifdir())
bgd->used_dirs_count++;
write_block(bgd_location.block, bgd_buffer);
const uint32_t inode_table_offset = ino_index * superblock().inode_size;
const BlockLocation inode_location
{
.block = inode_table_offset / block_size + bgd->inode_table,
.offset = inode_table_offset % block_size
};
// NOTE: we don't need inode bitmap anymore, so we can reuse it
auto& inode_buffer = inode_bitmap;
read_block(inode_location.block, inode_buffer);
memcpy(inode_buffer.data() + inode_location.offset, &ext2_inode, sizeof(Ext2::Inode));
if (superblock().inode_size > sizeof(Ext2::Inode))
memset(inode_buffer.data() + inode_location.offset + sizeof(Ext2::Inode), 0, superblock().inode_size - sizeof(Ext2::Inode));
write_block(inode_location.block, inode_buffer);
m_superblock.free_inodes_count--;
sync_superblock();
return ino;
}
derrorln("Corrupted file system. Superblock indicates free inodes but none were found.");
return BAN::Error::from_error_code(ErrorCode::Ext2_Corrupted);
}
void Ext2FS::delete_inode(uint32_t ino)
{
LockGuard _(m_mutex);
ASSERT(ino >= superblock().first_ino);
ASSERT(ino <= superblock().inodes_count);
auto bgd_buffer = get_block_buffer();
auto bitmap_buffer = get_block_buffer();
auto inode_buffer = get_block_buffer();
const uint32_t inode_group = (ino - 1) / superblock().inodes_per_group;
const uint32_t inode_index = (ino - 1) % superblock().inodes_per_group;
auto bgd_location = locate_block_group_descriptior(inode_group);
read_block(bgd_location.block, bgd_buffer);
auto& bgd = bgd_buffer.span().slice(bgd_location.offset).as<Ext2::BlockGroupDescriptor>();
// update inode bitmap
read_block(bgd.inode_bitmap, bitmap_buffer);
const uint32_t byte = inode_index / 8;
const uint32_t bit = inode_index % 8;
ASSERT(bitmap_buffer[byte] & (1 << bit));
bitmap_buffer[byte] &= ~(1 << bit);
write_block(bgd.inode_bitmap, bitmap_buffer);
// memset inode to zero or fsck will complain
auto inode_location = locate_inode(ino);
read_block(inode_location.block, inode_buffer);
auto& inode = inode_buffer.span().slice(inode_location.offset).as<Ext2::Inode>();
#if EXT2_VERIFY_NO_BLOCKS
static const char zero_buffer[sizeof(inode.block)] {};
ASSERT(memcmp(inode.block, zero_buffer, sizeof(inode.block)) == 0);
#endif
bool is_directory = Inode::Mode(inode.mode).ifdir();
memset(&inode, 0x00, m_superblock.inode_size);
write_block(inode_location.block, inode_buffer);
// update bgd counts
bgd.free_inodes_count++;
if (is_directory)
bgd.used_dirs_count--;
write_block(bgd_location.block, bgd_buffer);
// update superblock inode count
m_superblock.free_inodes_count++;
sync_superblock();
// remove inode from cache
if (m_inode_cache.contains(ino))
m_inode_cache.remove(ino);
}
void Ext2FS::read_block(uint32_t block, BlockBufferWrapper& buffer)
{
LockGuard _(m_mutex);
const uint32_t sector_size = m_block_device->blksize();
const uint32_t block_size = this->block_size();
const uint32_t sectors_per_block = block_size / sector_size;
ASSERT(block >= superblock().first_data_block + 1);
ASSERT(buffer.size() >= block_size);
MUST(m_block_device->read_blocks(block * sectors_per_block, sectors_per_block, buffer.span()));
}
void Ext2FS::write_block(uint32_t block, const BlockBufferWrapper& buffer)
{
LockGuard _(m_mutex);
const uint32_t sector_size = m_block_device->blksize();
const uint32_t block_size = this->block_size();
const uint32_t sectors_per_block = block_size / sector_size;
ASSERT(block >= superblock().first_data_block + 1);
ASSERT(buffer.size() >= block_size);
MUST(m_block_device->write_blocks(block * sectors_per_block, sectors_per_block, buffer.span()));
}
void Ext2FS::sync_superblock()
{
LockGuard _(m_mutex);
const uint32_t sector_size = m_block_device->blksize();
ASSERT(1024 % sector_size == 0);
const uint32_t superblock_bytes =
(m_superblock.rev_level == Ext2::Enum::GOOD_OLD_REV)
? offsetof(Ext2::Superblock, __extension_start)
: sizeof(Ext2::Superblock);
const uint32_t lba = 1024 / sector_size;
const uint32_t sector_count = BAN::Math::div_round_up<uint32_t>(superblock_bytes, sector_size);
auto superblock_buffer = get_block_buffer();
MUST(m_block_device->read_blocks(lba, sector_count, superblock_buffer.span()));
if (memcmp(superblock_buffer.data(), &m_superblock, superblock_bytes))
{
memcpy(superblock_buffer.data(), &m_superblock, superblock_bytes);
MUST(m_block_device->write_blocks(lba, sector_count, superblock_buffer.span()));
}
}
Ext2FS::BlockBufferWrapper Ext2FS::get_block_buffer()
{
LockGuard _(m_mutex);
return m_buffer_manager.get_buffer();
}
BAN::ErrorOr<uint32_t> Ext2FS::reserve_free_block(uint32_t primary_bgd)
{
LockGuard _(m_mutex);
if (m_superblock.r_blocks_count >= m_superblock.free_blocks_count)
return BAN::Error::from_errno(ENOSPC);
auto bgd_buffer = m_buffer_manager.get_buffer();
auto block_bitmap = m_buffer_manager.get_buffer();
auto check_block_group =
[&](uint32_t block_group) -> uint32_t
{
auto bgd_location = locate_block_group_descriptior(block_group);
read_block(bgd_location.block, bgd_buffer);
auto& bgd = *(Ext2::BlockGroupDescriptor*)(bgd_buffer.data() + bgd_location.offset);
if (bgd.free_blocks_count == 0)
return 0;
read_block(bgd.block_bitmap, block_bitmap);
for (uint32_t block_offset = 0; block_offset < m_superblock.blocks_per_group; block_offset++)
{
const uint32_t fs_block_index = m_superblock.first_data_block + m_superblock.blocks_per_group * block_group + block_offset;
if (fs_block_index >= m_superblock.blocks_count)
break;
uint32_t byte = block_offset / 8;
uint32_t bit = block_offset % 8;
if (block_bitmap[byte] & (1 << bit))
continue;
block_bitmap[byte] |= 1 << bit;
write_block(bgd.block_bitmap, block_bitmap);
bgd.free_blocks_count--;
write_block(bgd_location.block, bgd_buffer);
m_superblock.free_blocks_count--;
sync_superblock();
return fs_block_index;
}
derrorln("Corrupted file system. Block group descriptor indicates free blocks but none were found");
return 0;
};
if (auto ret = check_block_group(primary_bgd))
return ret;
uint32_t number_of_block_groups = BAN::Math::div_round_up(m_superblock.blocks_count, m_superblock.blocks_per_group);
for (uint32_t block_group = 0; block_group < number_of_block_groups; block_group++)
if (block_group != primary_bgd)
if (auto ret = check_block_group(block_group))
return ret;
derrorln("Corrupted file system. Superblock indicates free blocks but none were found.");
return BAN::Error::from_error_code(ErrorCode::Ext2_Corrupted);
}
void Ext2FS::release_block(uint32_t block)
{
LockGuard _(m_mutex);
ASSERT(block >= m_superblock.first_data_block);
ASSERT(block < m_superblock.blocks_count);
const uint32_t block_group = (block - m_superblock.first_data_block) / m_superblock.blocks_per_group;
const uint32_t block_offset = (block - m_superblock.first_data_block) % m_superblock.blocks_per_group;
auto bgd_buffer = get_block_buffer();
auto bitmap_buffer = get_block_buffer();
auto bgd_location = locate_block_group_descriptior(block_group);
read_block(bgd_location.block, bgd_buffer);
auto& bgd = bgd_buffer.span().slice(bgd_location.offset).as<Ext2::BlockGroupDescriptor>();
read_block(bgd.block_bitmap, bitmap_buffer);
const uint32_t byte = block_offset / 8;
const uint32_t bit = block_offset % 8;
ASSERT(bitmap_buffer[byte] & (1 << bit));
bitmap_buffer[byte] &= ~(1 << bit);
write_block(bgd.block_bitmap, bitmap_buffer);
bgd.free_blocks_count++;
write_block(bgd_location.block, bgd_buffer);
m_superblock.free_blocks_count++;
sync_superblock();
}
Ext2FS::BlockLocation Ext2FS::locate_inode(uint32_t ino)
{
LockGuard _(m_mutex);
ASSERT(ino <= superblock().inodes_count);
const uint32_t block_size = this->block_size();
auto bgd_buffer = m_buffer_manager.get_buffer();
const uint32_t inode_group = (ino - 1) / superblock().inodes_per_group;
const uint32_t inode_index = (ino - 1) % superblock().inodes_per_group;
auto bgd_location = locate_block_group_descriptior(inode_group);
read_block(bgd_location.block, bgd_buffer);
auto& bgd = *(Ext2::BlockGroupDescriptor*)(bgd_buffer.data() + bgd_location.offset);
const uint32_t inode_byte_offset = inode_index * superblock().inode_size;
BlockLocation location
{
.block = inode_byte_offset / block_size + bgd.inode_table,
.offset = inode_byte_offset % block_size
};
#if EXT2_VERIFY_INODE
const uint32_t inode_bitmap_block = bgd.inode_bitmap;
// NOTE: we can reuse the bgd_buffer since it is not needed anymore
auto& inode_bitmap = bgd_buffer;
read_block(inode_bitmap_block, inode_bitmap.span());
const uint32_t byte = inode_index / 8;
const uint32_t bit = inode_index % 8;
ASSERT(inode_bitmap[byte] & (1 << bit));
#endif
return location;
}
Ext2FS::BlockLocation Ext2FS::locate_block_group_descriptior(uint32_t group_index)
{
LockGuard _(m_mutex);
const uint32_t block_size = this->block_size();
const uint32_t block_group_count = BAN::Math::div_round_up(superblock().inodes_count, superblock().inodes_per_group);
ASSERT(group_index < block_group_count);
// Block Group Descriptor table is in the block after superblock
const uint32_t bgd_byte_offset = (superblock().first_data_block + 1) * block_size + sizeof(Ext2::BlockGroupDescriptor) * group_index;
return
{
.block = bgd_byte_offset / block_size,
.offset = bgd_byte_offset % block_size
};
}
Ext2FS::BlockBufferWrapper Ext2FS::BlockBufferManager::get_buffer()
{
for (auto& buffer : m_buffers)
{
if (buffer.used)
continue;
buffer.used = true;
return Ext2FS::BlockBufferWrapper(buffer.buffer.span(), buffer.used);
}
ASSERT_NOT_REACHED();
}
BAN::ErrorOr<void> Ext2FS::BlockBufferManager::initialize(size_t block_size)
{
for (auto& buffer : m_buffers)
{
TRY(buffer.buffer.resize(block_size));
buffer.used = false;
}
return {};
}
}