github-mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-11-07 06:17:02 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity Graph

btrfs: remove the v1 space cache load path

Browse Source 
Nothing writes a v1 space cache any more, and since commit 545e560a5b
("btrfs: disable v1 space cache") the mount option can't be enabled to
read one either. Remove load_free_space_cache(), its io_ctl helpers and
struct btrfs_io_ctl. Drop the gfp constraint on the inode mapping as
well, it only covered the cache's page cache allocations.

Assisted-by: LLM
Signed-off-by: Tal Zussman <tz2294@columbia.edu>
Signed-off-by: David Sterba <dsterba@suse.com>
This commit is contained in:
Tal Zussman 2026-10-15 10:43:22 -04:00 committed by David Sterba
parent 8d598a2113
commit 324ad3a989
3 changed files with 1 additions and 597 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
fs/btrfs/block-group.c
View File

@ -904,22 +904,6 @@ static noinline void caching_thread(struct btrfs_work *work)
down_read(&fs_info->commit_root_sem);
load_block_group_size_class(caching_ctl);
if (btrfs_test_opt(fs_info, SPACE_CACHE)) {
ret = load_free_space_cache(block_group);
if (ret == 1) {
ret = 0;
goto done;
}
/*
* We failed to load the space cache, set ourselves to
* CACHE_STARTED and carry on.
*/
spin_lock(&block_group->lock);
block_group->cached = BTRFS_CACHE_STARTED;
spin_unlock(&block_group->lock);
wake_up(&caching_ctl->wait);
}
/*
* If we are in the transaction that populated the free space tree we
@ -933,7 +917,7 @@ static noinline void caching_thread(struct btrfs_work *work)
ret = btrfs_load_free_space_tree(caching_ctl);
else
ret = load_extent_tree_free(caching_ctl);
done:
spin_lock(&block_group->lock);
block_group->caching_ctl = NULL;
block_group->cached = ret ? BTRFS_CACHE_ERROR : BTRFS_CACHE_FINISHED;

565
fs/btrfs/free-space-cache.c
View File

@ -9,7 +9,6 @@
#include <linux/slab.h>
#include <linux/math64.h>
#include <linux/ratelimit.h>
#include <linux/error-injection.h>
#include <linux/sched/mm.h>
#include <linux/string_choices.h>
#include "extent-tree.h"
@ -23,7 +22,6 @@
#include "space-info.h"
#include "block-group.h"
#include "discard.h"
#include "subpage.h"
#include "inode-item.h"
#include "accessors.h"
#include "file-item.h"
@ -57,11 +55,6 @@ static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
struct btrfs_free_space *info, u64 offset,
u64 bytes, bool update_stats);
static void btrfs_crc32c_final(u32 crc, u8 *result)
{
put_unaligned_le32(~crc, result);
}
static void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
{
struct btrfs_free_space *info;
@ -123,10 +116,6 @@ static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
if (IS_ERR(inode))
return ERR_CAST(inode);
mapping_set_gfp_mask(inode->vfs_inode.i_mapping,
mapping_gfp_constraint(inode->vfs_inode.i_mapping,
~(__GFP_FS | __GFP_HIGHMEM)));
return &inode->vfs_inode;
}
@ -272,226 +261,6 @@ int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
return ret;
}
static void readahead_cache(struct inode *inode)
{
struct file_ra_state ra;
pgoff_t last_index;
file_ra_state_init(&ra, inode->i_mapping);
last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
page_cache_sync_readahead(inode->i_mapping, &ra, NULL, 0, last_index);
}
static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
int write)
{
int num_pages;
num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
/* Make sure we can fit our crcs and generation into the first page */
if (write && (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
return -ENOSPC;
memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
io_ctl->pages = kzalloc_objs(struct page *, num_pages, GFP_NOFS);
if (!io_ctl->pages)
return -ENOMEM;
io_ctl->num_pages = num_pages;
io_ctl->fs_info = inode_to_fs_info(inode);
io_ctl->inode = inode;
return 0;
}
ALLOW_ERROR_INJECTION(io_ctl_init, ERRNO);
static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
{
kfree(io_ctl->pages);
io_ctl->pages = NULL;
}
static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
{
if (io_ctl->cur) {
io_ctl->cur = NULL;
io_ctl->orig = NULL;
}
}
static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
{
ASSERT(io_ctl->index < io_ctl->num_pages);
io_ctl->page = io_ctl->pages[io_ctl->index++];
io_ctl->cur = page_address(io_ctl->page);
io_ctl->orig = io_ctl->cur;
io_ctl->size = PAGE_SIZE;
if (clear)
clear_page(io_ctl->cur);
}
static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
{
int i;
io_ctl_unmap_page(io_ctl);
for (i = 0; i < io_ctl->num_pages; i++) {
if (io_ctl->pages[i]) {
unlock_page(io_ctl->pages[i]);
put_page(io_ctl->pages[i]);
}
}
}
static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, bool uptodate)
{
struct folio *folio;
struct inode *inode = io_ctl->inode;
gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
int i;
for (i = 0; i < io_ctl->num_pages; i++) {
int ret;
folio = __filemap_get_folio(inode->i_mapping, i,
FGP_LOCK | FGP_ACCESSED | FGP_CREAT,
mask);
if (IS_ERR(folio)) {
io_ctl_drop_pages(io_ctl);
return PTR_ERR(folio);
}
ret = set_folio_extent_mapped(folio);
if (ret < 0) {
folio_unlock(folio);
folio_put(folio);
io_ctl_drop_pages(io_ctl);
return ret;
}
io_ctl->pages[i] = &folio->page;
if (uptodate && !folio_test_uptodate(folio)) {
btrfs_read_folio(NULL, folio);
folio_lock(folio);
if (folio->mapping != inode->i_mapping) {
btrfs_err(BTRFS_I(inode)->root->fs_info,
"free space cache page truncated");
io_ctl_drop_pages(io_ctl);
return -EIO;
}
if (!folio_test_uptodate(folio)) {
btrfs_err(BTRFS_I(inode)->root->fs_info,
"error reading free space cache");
io_ctl_drop_pages(io_ctl);
return -EIO;
}
}
}
for (i = 0; i < io_ctl->num_pages; i++)
clear_page_dirty_for_io(io_ctl->pages[i]);
return 0;
}
static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
{
u64 cache_gen;
/*
* Skip the crc area. If we don't check crcs then we just have a 64bit
* chunk at the front of the first page.
*/
io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
cache_gen = get_unaligned_le64(io_ctl->cur);
if (cache_gen != generation) {
btrfs_err_rl(io_ctl->fs_info,
"space cache generation (%llu) does not match inode (%llu)",
cache_gen, generation);
io_ctl_unmap_page(io_ctl);
return -EIO;
}
io_ctl->cur += sizeof(u64);
return 0;
}
static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
{
u32 *tmp, val;
u32 crc = ~(u32)0;
unsigned offset = 0;
if (index >= io_ctl->num_pages)
return -EIO;
if (index == 0)
offset = sizeof(u32) * io_ctl->num_pages;
tmp = page_address(io_ctl->pages[0]);
tmp += index;
val = *tmp;
io_ctl_map_page(io_ctl, 0);
crc = crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
btrfs_crc32c_final(crc, (u8 *)&crc);
if (val != crc) {
btrfs_err_rl(io_ctl->fs_info,
"csum mismatch on free space cache");
io_ctl_unmap_page(io_ctl);
return -EIO;
}
return 0;
}
static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
struct btrfs_free_space *entry, u8 *type)
{
struct btrfs_free_space_entry *e;
int ret;
if (!io_ctl->cur) {
ret = io_ctl_check_crc(io_ctl, io_ctl->index);
if (ret)
return ret;
}
e = io_ctl->cur;
entry->offset = get_unaligned_le64(&e->offset);
entry->bytes = get_unaligned_le64(&e->bytes);
*type = e->type;
io_ctl->cur += sizeof(struct btrfs_free_space_entry);
io_ctl->size -= sizeof(struct btrfs_free_space_entry);
if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
return 0;
io_ctl_unmap_page(io_ctl);
return 0;
}
static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
struct btrfs_free_space *entry)
{
int ret;
ret = io_ctl_check_crc(io_ctl, io_ctl->index);
if (ret)
return ret;
copy_page(entry->bitmap, io_ctl->cur);
io_ctl_unmap_page(io_ctl);
return 0;
}
static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
{
struct btrfs_block_group *block_group = ctl->block_group;
@ -537,340 +306,6 @@ static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
div_u64(extent_bytes, sizeof(struct btrfs_free_space));
}
static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
struct btrfs_free_space_ctl *ctl,
struct btrfs_path *path, u64 offset)
{
struct btrfs_fs_info *fs_info = root->fs_info;
struct btrfs_free_space_header *header;
struct extent_buffer *leaf;
struct btrfs_io_ctl io_ctl;
struct btrfs_key key;
struct btrfs_free_space *e, *n;
LIST_HEAD(bitmaps);
u64 num_entries;
u64 num_bitmaps;
u64 generation;
u8 type;
int ret = 0;
/* Nothing in the space cache, goodbye */
if (!i_size_read(inode))
return 0;
key.objectid = BTRFS_FREE_SPACE_OBJECTID;
key.type = 0;
key.offset = offset;
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
return 0;
else if (ret > 0) {
btrfs_release_path(path);
return 0;
}
ret = -1;
leaf = path->nodes[0];
header = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_free_space_header);
num_entries = btrfs_free_space_entries(leaf, header);
num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
generation = btrfs_free_space_generation(leaf, header);
btrfs_release_path(path);
if (!BTRFS_I(inode)->generation) {
btrfs_info(fs_info,
"the free space cache file (%llu) is invalid, skip it",
offset);
return 0;
}
if (BTRFS_I(inode)->generation != generation) {
btrfs_err(fs_info,
"free space inode generation (%llu) did not match free space cache generation (%llu)",
BTRFS_I(inode)->generation, generation);
return 0;
}
if (!num_entries)
return 0;
ret = io_ctl_init(&io_ctl, inode, 0);
if (ret)
return ret;
readahead_cache(inode);
ret = io_ctl_prepare_pages(&io_ctl, true);
if (ret)
goto out;
ret = io_ctl_check_crc(&io_ctl, 0);
if (ret)
goto free_cache;
ret = io_ctl_check_generation(&io_ctl, generation);
if (ret)
goto free_cache;
while (num_entries) {
e = kmem_cache_zalloc(btrfs_free_space_cachep,
GFP_NOFS);
if (!e) {
ret = -ENOMEM;
goto free_cache;
}
ret = io_ctl_read_entry(&io_ctl, e, &type);
if (ret) {
kmem_cache_free(btrfs_free_space_cachep, e);
goto free_cache;
}
if (!e->bytes) {
ret = -1;
kmem_cache_free(btrfs_free_space_cachep, e);
goto free_cache;
}
if (type == BTRFS_FREE_SPACE_EXTENT) {
spin_lock(&ctl->tree_lock);
ret = link_free_space(ctl, e);
spin_unlock(&ctl->tree_lock);
if (ret) {
btrfs_err(fs_info,
"Duplicate entries in free space cache, dumping");
kmem_cache_free(btrfs_free_space_cachep, e);
goto free_cache;
}
} else {
ASSERT(num_bitmaps);
num_bitmaps--;
e->bitmap = kmem_cache_zalloc(
btrfs_free_space_bitmap_cachep, GFP_NOFS);
if (!e->bitmap) {
ret = -ENOMEM;
kmem_cache_free(
btrfs_free_space_cachep, e);
goto free_cache;
}
spin_lock(&ctl->tree_lock);
ret = link_free_space(ctl, e);
if (ret) {
spin_unlock(&ctl->tree_lock);
btrfs_err(fs_info,
"Duplicate entries in free space cache, dumping");
kmem_cache_free(btrfs_free_space_bitmap_cachep, e->bitmap);
kmem_cache_free(btrfs_free_space_cachep, e);
goto free_cache;
}
ctl->total_bitmaps++;
recalculate_thresholds(ctl);
spin_unlock(&ctl->tree_lock);
list_add_tail(&e->list, &bitmaps);
}
num_entries--;
}
io_ctl_unmap_page(&io_ctl);
/*
* We add the bitmaps at the end of the entries in order that
* the bitmap entries are added to the cache.
*/
list_for_each_entry_safe(e, n, &bitmaps, list) {
list_del_init(&e->list);
ret = io_ctl_read_bitmap(&io_ctl, e);
if (ret)
goto free_cache;
}
io_ctl_drop_pages(&io_ctl);
ret = 1;
out:
io_ctl_free(&io_ctl);
return ret;
free_cache:
io_ctl_drop_pages(&io_ctl);
spin_lock(&ctl->tree_lock);
__btrfs_remove_free_space_cache(ctl);
spin_unlock(&ctl->tree_lock);
goto out;
}
static int copy_free_space_cache(struct btrfs_free_space_ctl *ctl)
{
struct btrfs_free_space *info;
struct rb_node *n;
int ret = 0;
while (!ret && (n = rb_first(&ctl->free_space_offset)) != NULL) {
info = rb_entry(n, struct btrfs_free_space, offset_index);
if (!info->bitmap) {
const u64 offset = info->offset;
const u64 bytes = info->bytes;
unlink_free_space(ctl, info, true);
spin_unlock(&ctl->tree_lock);
kmem_cache_free(btrfs_free_space_cachep, info);
ret = btrfs_add_free_space(ctl->block_group, offset, bytes);
spin_lock(&ctl->tree_lock);
} else {
u64 offset = info->offset;
u64 bytes = ctl->block_group->fs_info->sectorsize;
ret = search_bitmap(ctl, info, &offset, &bytes, false);
if (ret == 0) {
bitmap_clear_bits(ctl, info, offset, bytes, true);
spin_unlock(&ctl->tree_lock);
ret = btrfs_add_free_space(ctl->block_group, offset,
bytes);
spin_lock(&ctl->tree_lock);
} else {
free_bitmap(ctl, info);
ret = 0;
}
}
cond_resched_lock(&ctl->tree_lock);
}
return ret;
}
static struct lock_class_key btrfs_free_space_inode_key;
int load_free_space_cache(struct btrfs_block_group *block_group)
{
struct btrfs_fs_info *fs_info = block_group->fs_info;
struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
struct btrfs_free_space_ctl tmp_ctl = {};
struct inode *inode;
struct btrfs_path *path;
int ret = 0;
bool matched;
u64 used = block_group->used;
/*
* Because we could potentially discard our loaded free space, we want
* to load everything into a temporary structure first, and then if it's
* valid copy it all into the actual free space ctl.
*/
btrfs_init_free_space_ctl(block_group, &tmp_ctl);
/*
* If this block group has been marked to be cleared for one reason or
* another then we can't trust the on disk cache, so just return.
*/
spin_lock(&block_group->lock);
if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
spin_unlock(&block_group->lock);
return 0;
}
spin_unlock(&block_group->lock);
path = btrfs_alloc_path();
if (!path)
return 0;
path->search_commit_root = true;
path->skip_locking = true;
/*
* We must pass a path with search_commit_root set to btrfs_iget in
* order to avoid a deadlock when allocating extents for the tree root.
*
* When we are COWing an extent buffer from the tree root, when looking
* for a free extent, at extent-tree.c:find_free_extent(), we can find
* block group without its free space cache loaded. When we find one
* we must load its space cache which requires reading its free space
* cache's inode item from the root tree. If this inode item is located
* in the same leaf that we started COWing before, then we end up in
* deadlock on the extent buffer (trying to read lock it when we
* previously write locked it).
*
* It's safe to read the inode item using the commit root because
* block groups, once loaded, stay in memory forever (until they are
* removed) as well as their space caches once loaded. New block groups
* once created get their ->cached field set to BTRFS_CACHE_FINISHED so
* we will never try to read their inode item while the fs is mounted.
*/
inode = lookup_free_space_inode(block_group, path);
if (IS_ERR(inode)) {
btrfs_free_path(path);
return 0;
}
/* We may have converted the inode and made the cache invalid. */
spin_lock(&block_group->lock);
if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
spin_unlock(&block_group->lock);
btrfs_free_path(path);
goto out;
}
spin_unlock(&block_group->lock);
/*
* Reinitialize the class of struct inode's mapping->invalidate_lock for
* free space inodes to prevent false positives related to locks for normal
* inodes.
*/
lockdep_set_class(&(&inode->i_data)->invalidate_lock,
&btrfs_free_space_inode_key);
ret = __load_free_space_cache(fs_info->tree_root, inode, &tmp_ctl,
path, block_group->start);
btrfs_free_path(path);
if (ret <= 0)
goto out;
matched = (tmp_ctl.free_space == (block_group->length - used -
block_group->bytes_super));
if (matched) {
spin_lock(&tmp_ctl.tree_lock);
ret = copy_free_space_cache(&tmp_ctl);
spin_unlock(&tmp_ctl.tree_lock);
/*
* ret == 1 means we successfully loaded the free space cache,
* so we need to re-set it here.
*/
if (ret == 0)
ret = 1;
} else {
/*
* We need to call the _locked variant so we don't try to update
* the discard counters.
*/
spin_lock(&tmp_ctl.tree_lock);
__btrfs_remove_free_space_cache(&tmp_ctl);
spin_unlock(&tmp_ctl.tree_lock);
btrfs_warn(fs_info,
"block group %llu has wrong amount of free space",
block_group->start);
ret = -1;
}
out:
if (ret < 0) {
/* This cache is bogus, make sure it gets cleared */
spin_lock(&block_group->lock);
block_group->disk_cache_state = BTRFS_DC_CLEAR;
spin_unlock(&block_group->lock);
ret = 0;
btrfs_warn(fs_info,
"failed to load free space cache for block group %llu, rebuilding it now",
block_group->start);
}
spin_lock(&ctl->tree_lock);
btrfs_discard_update_discardable(block_group);
spin_unlock(&ctl->tree_lock);
iput(inode);
return ret;
}
static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
u64 offset)
{

15
fs/btrfs/free-space-cache.h
View File

@ -14,7 +14,6 @@
#include "fs.h"
struct inode;
struct page;
struct btrfs_fs_info;
struct btrfs_path;
struct btrfs_trans_handle;
@ -88,19 +87,6 @@ struct btrfs_free_space_ctl {
struct list_head trimming_ranges;
};
struct btrfs_io_ctl {
void *cur, *orig;
struct page *page;
struct page **pages;
struct btrfs_fs_info *fs_info;
struct inode *inode;
unsigned long size;
int index;
int num_pages;
int entries;
int bitmaps;
};
int __init btrfs_free_space_init(void);
void __cold btrfs_free_space_exit(void);
struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
@ -111,7 +97,6 @@ int btrfs_remove_free_space_inode(struct btrfs_trans_handle *trans,
int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
struct inode *inode);
int load_free_space_cache(struct btrfs_block_group *block_group);
void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group,
struct btrfs_free_space_ctl *ctl);