page = cb->compressed_pages[i];
csum = ~(u32)0;
- kaddr = kmap_atomic(page, KM_USER0);
+ kaddr = kmap_atomic(page);
csum = btrfs_csum_data(root, kaddr, csum, PAGE_CACHE_SIZE);
btrfs_csum_final(csum, (char *)&csum);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
if (csum != *cb_sum) {
printk(KERN_INFO "btrfs csum failed ino %llu "
* Clear the writeback bits on all of the file
* pages for a compressed write
*/
- static noinline int end_compressed_writeback(struct inode *inode, u64 start,
- unsigned long ram_size)
+ static noinline void end_compressed_writeback(struct inode *inode, u64 start,
+ unsigned long ram_size)
{
unsigned long index = start >> PAGE_CACHE_SHIFT;
unsigned long end_index = (start + ram_size - 1) >> PAGE_CACHE_SHIFT;
index += ret;
}
/* the inode may be gone now */
- return 0;
}
/*
*/
atomic_inc(&cb->pending_bios);
ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
if (!skip_sum) {
ret = btrfs_csum_one_bio(root, inode, bio,
start, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
}
ret = btrfs_map_bio(root, WRITE, bio, 0, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
bio_put(bio);
bio_get(bio);
ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
if (!skip_sum) {
ret = btrfs_csum_one_bio(root, inode, bio, start, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
}
ret = btrfs_map_bio(root, WRITE, bio, 0, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
bio_put(bio);
return 0;
* sure they map to this compressed extent on disk.
*/
set_page_extent_mapped(page);
- lock_extent(tree, last_offset, end, GFP_NOFS);
+ lock_extent(tree, last_offset, end);
read_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, last_offset,
PAGE_CACHE_SIZE);
(last_offset + PAGE_CACHE_SIZE > extent_map_end(em)) ||
(em->block_start >> 9) != cb->orig_bio->bi_sector) {
free_extent_map(em);
- unlock_extent(tree, last_offset, end, GFP_NOFS);
+ unlock_extent(tree, last_offset, end);
unlock_page(page);
page_cache_release(page);
break;
if (zero_offset) {
int zeros;
zeros = PAGE_CACHE_SIZE - zero_offset;
- userpage = kmap_atomic(page, KM_USER0);
+ userpage = kmap_atomic(page);
memset(userpage + zero_offset, 0, zeros);
flush_dcache_page(page);
- kunmap_atomic(userpage, KM_USER0);
+ kunmap_atomic(userpage);
}
}
nr_pages++;
page_cache_release(page);
} else {
- unlock_extent(tree, last_offset, end, GFP_NOFS);
+ unlock_extent(tree, last_offset, end);
unlock_page(page);
page_cache_release(page);
break;
bio_get(comp_bio);
ret = btrfs_bio_wq_end_io(root->fs_info, comp_bio, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
/*
* inc the count before we submit the bio so
if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
ret = btrfs_lookup_bio_sums(root, inode,
comp_bio, sums);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
}
sums += (comp_bio->bi_size + root->sectorsize - 1) /
root->sectorsize;
ret = btrfs_map_bio(root, READ, comp_bio,
mirror_num, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
bio_put(comp_bio);
bio_get(comp_bio);
ret = btrfs_bio_wq_end_io(root->fs_info, comp_bio, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
ret = btrfs_lookup_bio_sums(root, inode, comp_bio, sums);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
}
ret = btrfs_map_bio(root, READ, comp_bio, mirror_num, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
bio_put(comp_bio);
return 0;
&btrfs_lzo_compress,
};
- int __init btrfs_init_compress(void)
+ void __init btrfs_init_compress(void)
{
int i;
atomic_set(&comp_alloc_workspace[i], 0);
init_waitqueue_head(&comp_workspace_wait[i]);
}
- return 0;
}
/*
bytes = min(PAGE_CACHE_SIZE - *pg_offset,
PAGE_CACHE_SIZE - buf_offset);
bytes = min(bytes, working_bytes);
- kaddr = kmap_atomic(page_out, KM_USER0);
+ kaddr = kmap_atomic(page_out);
memcpy(kaddr + *pg_offset, buf + buf_offset, bytes);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
flush_dcache_page(page_out);
*pg_offset += bytes;
#include "btrfs_inode.h"
#include "volumes.h"
#include "check-integrity.h"
+ #include "locking.h"
static struct kmem_cache *extent_state_cache;
static struct kmem_cache *extent_buffer_cache;
unsigned int sync_io:1;
};
+ static noinline void flush_write_bio(void *data);
+ static inline struct btrfs_fs_info *
+ tree_fs_info(struct extent_io_tree *tree)
+ {
+ return btrfs_sb(tree->mapping->host->i_sb);
+ }
+
int __init extent_io_init(void)
{
extent_state_cache = kmem_cache_create("extent_state",
#endif
atomic_set(&state->refs, 1);
init_waitqueue_head(&state->wq);
+ trace_alloc_extent_state(state, mask, _RET_IP_);
return state;
}
list_del(&state->leak_list);
spin_unlock_irqrestore(&leak_lock, flags);
#endif
+ trace_free_extent_state(state, _RET_IP_);
kmem_cache_free(extent_state_cache, state);
}
}
return prealloc;
}
+ void extent_io_tree_panic(struct extent_io_tree *tree, int err)
+ {
+ btrfs_panic(tree_fs_info(tree), err, "Locking error: "
+ "Extent tree was modified by another "
+ "thread while locked.");
+ }
+
/*
* clear some bits on a range in the tree. This may require splitting
* or inserting elements in the tree, so the gfp mask is used to
*
* the range [start, end] is inclusive.
*
- * This takes the tree lock, and returns < 0 on error, > 0 if any of the
- * bits were already set, or zero if none of the bits were already set.
+ * This takes the tree lock, and returns 0 on success and < 0 on error.
*/
int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
int bits, int wake, int delete,
struct rb_node *node;
u64 last_end;
int err;
- int set = 0;
int clear = 0;
if (delete)
prealloc = alloc_extent_state_atomic(prealloc);
BUG_ON(!prealloc);
err = split_state(tree, state, prealloc, start);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
+
prealloc = NULL;
if (err)
goto out;
if (state->end <= end) {
- set |= clear_state_bit(tree, state, &bits, wake);
+ clear_state_bit(tree, state, &bits, wake);
if (last_end == (u64)-1)
goto out;
start = last_end + 1;
prealloc = alloc_extent_state_atomic(prealloc);
BUG_ON(!prealloc);
err = split_state(tree, state, prealloc, end + 1);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
+
if (wake)
wake_up(&state->wq);
- set |= clear_state_bit(tree, prealloc, &bits, wake);
+ clear_state_bit(tree, prealloc, &bits, wake);
prealloc = NULL;
goto out;
}
- set |= clear_state_bit(tree, state, &bits, wake);
+ clear_state_bit(tree, state, &bits, wake);
next:
if (last_end == (u64)-1)
goto out;
if (prealloc)
free_extent_state(prealloc);
- return set;
+ return 0;
search_again:
if (start > end)
goto again;
}
- static int wait_on_state(struct extent_io_tree *tree,
- struct extent_state *state)
+ static void wait_on_state(struct extent_io_tree *tree,
+ struct extent_state *state)
__releases(tree->lock)
__acquires(tree->lock)
{
schedule();
spin_lock(&tree->lock);
finish_wait(&state->wq, &wait);
- return 0;
}
/*
* The range [start, end] is inclusive.
* The tree lock is taken by this function
*/
- int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
+ void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
{
struct extent_state *state;
struct rb_node *node;
}
out:
spin_unlock(&tree->lock);
- return 0;
}
static void set_state_bits(struct extent_io_tree *tree,
* [start, end] is inclusive This takes the tree lock.
*/
- int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
- int bits, int exclusive_bits, u64 *failed_start,
- struct extent_state **cached_state, gfp_t mask)
+ static int __must_check
+ __set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+ int bits, int exclusive_bits, u64 *failed_start,
+ struct extent_state **cached_state, gfp_t mask)
{
struct extent_state *state;
struct extent_state *prealloc = NULL;
prealloc = alloc_extent_state_atomic(prealloc);
BUG_ON(!prealloc);
err = insert_state(tree, prealloc, start, end, &bits);
+ if (err)
+ extent_io_tree_panic(tree, err);
+
prealloc = NULL;
- BUG_ON(err == -EEXIST);
goto out;
}
state = rb_entry(node, struct extent_state, rb_node);
prealloc = alloc_extent_state_atomic(prealloc);
BUG_ON(!prealloc);
err = split_state(tree, state, prealloc, start);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
+
prealloc = NULL;
if (err)
goto out;
*/
err = insert_state(tree, prealloc, start, this_end,
&bits);
- BUG_ON(err == -EEXIST);
- if (err) {
- free_extent_state(prealloc);
- prealloc = NULL;
- goto out;
- }
+ if (err)
+ extent_io_tree_panic(tree, err);
+
cache_state(prealloc, cached_state);
prealloc = NULL;
start = this_end + 1;
prealloc = alloc_extent_state_atomic(prealloc);
BUG_ON(!prealloc);
err = split_state(tree, state, prealloc, end + 1);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
set_state_bits(tree, prealloc, &bits);
cache_state(prealloc, cached_state);
goto again;
}
+ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits,
+ u64 *failed_start, struct extent_state **cached_state,
+ gfp_t mask)
+ {
+ return __set_extent_bit(tree, start, end, bits, 0, failed_start,
+ cached_state, mask);
+ }
+
+
/**
* convert_extent - convert all bits in a given range from one bit to another
* @tree: the io tree to search
}
err = insert_state(tree, prealloc, start, end, &bits);
prealloc = NULL;
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
goto out;
}
state = rb_entry(node, struct extent_state, rb_node);
goto out;
}
err = split_state(tree, state, prealloc, start);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
prealloc = NULL;
if (err)
goto out;
*/
err = insert_state(tree, prealloc, start, this_end,
&bits);
- BUG_ON(err == -EEXIST);
- if (err) {
- free_extent_state(prealloc);
- prealloc = NULL;
- goto out;
- }
+ if (err)
+ extent_io_tree_panic(tree, err);
prealloc = NULL;
start = this_end + 1;
goto search_again;
}
err = split_state(tree, state, prealloc, end + 1);
- BUG_ON(err == -EEXIST);
+ if (err)
+ extent_io_tree_panic(tree, err);
set_state_bits(tree, prealloc, &bits);
clear_state_bit(tree, prealloc, &clear_bits, 0);
int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask)
{
- return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
+ return set_extent_bit(tree, start, end, EXTENT_DIRTY, NULL,
NULL, mask);
}
int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
int bits, gfp_t mask)
{
- return set_extent_bit(tree, start, end, bits, 0, NULL,
+ return set_extent_bit(tree, start, end, bits, NULL,
NULL, mask);
}
{
return set_extent_bit(tree, start, end,
EXTENT_DELALLOC | EXTENT_UPTODATE,
- 0, NULL, cached_state, mask);
+ NULL, cached_state, mask);
}
int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask)
{
- return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
+ return set_extent_bit(tree, start, end, EXTENT_NEW, NULL,
NULL, mask);
}
struct extent_state **cached_state, gfp_t mask)
{
return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0,
- NULL, cached_state, mask);
+ cached_state, mask);
}
static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
* us if waiting is desired.
*/
int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
- int bits, struct extent_state **cached_state, gfp_t mask)
+ int bits, struct extent_state **cached_state)
{
int err;
u64 failed_start;
while (1) {
- err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
- EXTENT_LOCKED, &failed_start,
- cached_state, mask);
- if (err == -EEXIST && (mask & __GFP_WAIT)) {
+ err = __set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
+ EXTENT_LOCKED, &failed_start,
+ cached_state, GFP_NOFS);
+ if (err == -EEXIST) {
wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
start = failed_start;
- } else {
+ } else
break;
- }
WARN_ON(start > end);
}
return err;
}
- int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+ int lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
{
- return lock_extent_bits(tree, start, end, 0, NULL, mask);
+ return lock_extent_bits(tree, start, end, 0, NULL);
}
- int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
- gfp_t mask)
+ int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
{
int err;
u64 failed_start;
- err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
- &failed_start, NULL, mask);
+ err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
+ &failed_start, NULL, GFP_NOFS);
if (err == -EEXIST) {
if (failed_start > start)
clear_extent_bit(tree, start, failed_start - 1,
- EXTENT_LOCKED, 1, 0, NULL, mask);
+ EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS);
return 0;
}
return 1;
mask);
}
- int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+ int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end)
{
return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL,
- mask);
+ GFP_NOFS);
}
/*
while (index <= end_index) {
page = find_get_page(tree->mapping, index);
- BUG_ON(!page);
+ BUG_ON(!page); /* Pages should be in the extent_io_tree */
set_page_writeback(page);
page_cache_release(page);
index++;
return found;
}
- static noinline int __unlock_for_delalloc(struct inode *inode,
- struct page *locked_page,
- u64 start, u64 end)
+ static noinline void __unlock_for_delalloc(struct inode *inode,
+ struct page *locked_page,
+ u64 start, u64 end)
{
int ret;
struct page *pages[16];
int i;
if (index == locked_page->index && end_index == index)
- return 0;
+ return;
while (nr_pages > 0) {
ret = find_get_pages_contig(inode->i_mapping, index,
index += ret;
cond_resched();
}
- return 0;
}
static noinline int lock_delalloc_pages(struct inode *inode,
goto out_failed;
}
}
- BUG_ON(ret);
+ BUG_ON(ret); /* Only valid values are 0 and -EAGAIN */
/* step three, lock the state bits for the whole range */
- lock_extent_bits(tree, delalloc_start, delalloc_end,
- 0, &cached_state, GFP_NOFS);
+ lock_extent_bits(tree, delalloc_start, delalloc_end, 0, &cached_state);
/* then test to make sure it is all still delalloc */
ret = test_range_bit(tree, delalloc_start, delalloc_end,
* helper function to set a given page up to date if all the
* extents in the tree for that page are up to date
*/
- static int check_page_uptodate(struct extent_io_tree *tree,
- struct page *page)
+ static void check_page_uptodate(struct extent_io_tree *tree, struct page *page)
{
u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
u64 end = start + PAGE_CACHE_SIZE - 1;
if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
SetPageUptodate(page);
- return 0;
}
/*
* helper function to unlock a page if all the extents in the tree
* for that page are unlocked
*/
- static int check_page_locked(struct extent_io_tree *tree,
- struct page *page)
+ static void check_page_locked(struct extent_io_tree *tree, struct page *page)
{
u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
u64 end = start + PAGE_CACHE_SIZE - 1;
if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
unlock_page(page);
- return 0;
}
/*
* helper function to end page writeback if all the extents
* in the tree for that page are done with writeback
*/
- static int check_page_writeback(struct extent_io_tree *tree,
- struct page *page)
+ static void check_page_writeback(struct extent_io_tree *tree,
+ struct page *page)
{
end_page_writeback(page);
- return 0;
}
/*
return 0;
}
+ int repair_eb_io_failure(struct btrfs_root *root, struct extent_buffer *eb,
+ int mirror_num)
+ {
+ struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
+ u64 start = eb->start;
+ unsigned long i, num_pages = num_extent_pages(eb->start, eb->len);
+ int ret;
+
+ for (i = 0; i < num_pages; i++) {
+ struct page *p = extent_buffer_page(eb, i);
+ ret = repair_io_failure(map_tree, start, PAGE_CACHE_SIZE,
+ start, p, mirror_num);
+ if (ret)
+ break;
+ start += PAGE_CACHE_SIZE;
+ }
+
+ return ret;
+ }
+
/*
* each time an IO finishes, we do a fast check in the IO failure tree
* to see if we need to process or clean up an io_failure_record
u64 start;
u64 end;
int whole_page;
+ int failed_mirror;
int ret;
if (err)
else
clean_io_failure(start, page);
}
- if (!uptodate) {
- int failed_mirror;
+
+ if (!uptodate)
failed_mirror = (int)(unsigned long)bio->bi_bdev;
+
+ if (!uptodate && tree->ops && tree->ops->readpage_io_failed_hook) {
+ ret = tree->ops->readpage_io_failed_hook(page, failed_mirror);
+ if (!ret && !err &&
+ test_bit(BIO_UPTODATE, &bio->bi_flags))
+ uptodate = 1;
+ } else if (!uptodate) {
/*
* The generic bio_readpage_error handles errors the
* following way: If possible, new read requests are
ret = bio_readpage_error(bio, page, start, end,
failed_mirror, NULL);
if (ret == 0) {
- error_handled:
uptodate =
test_bit(BIO_UPTODATE, &bio->bi_flags);
if (err)
uncache_state(&cached);
continue;
}
- if (tree->ops && tree->ops->readpage_io_failed_hook) {
- ret = tree->ops->readpage_io_failed_hook(
- bio, page, start, end,
- failed_mirror, state);
- if (ret == 0)
- goto error_handled;
- }
}
- if (uptodate) {
+ if (uptodate && tree->track_uptodate) {
set_extent_uptodate(tree, start, end, &cached,
GFP_ATOMIC);
}
return bio;
}
- static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
- unsigned long bio_flags)
+ /*
+ * Since writes are async, they will only return -ENOMEM.
+ * Reads can return the full range of I/O error conditions.
+ */
+ static int __must_check submit_one_bio(int rw, struct bio *bio,
+ int mirror_num, unsigned long bio_flags)
{
int ret = 0;
struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
return ret;
}
+ static int merge_bio(struct extent_io_tree *tree, struct page *page,
+ unsigned long offset, size_t size, struct bio *bio,
+ unsigned long bio_flags)
+ {
+ int ret = 0;
+ if (tree->ops && tree->ops->merge_bio_hook)
+ ret = tree->ops->merge_bio_hook(page, offset, size, bio,
+ bio_flags);
+ BUG_ON(ret < 0);
+ return ret;
+
+ }
+
static int submit_extent_page(int rw, struct extent_io_tree *tree,
struct page *page, sector_t sector,
size_t size, unsigned long offset,
sector;
if (prev_bio_flags != bio_flags || !contig ||
- (tree->ops && tree->ops->merge_bio_hook &&
- tree->ops->merge_bio_hook(page, offset, page_size, bio,
- bio_flags)) ||
+ merge_bio(tree, page, offset, page_size, bio, bio_flags) ||
bio_add_page(bio, page, page_size, offset) < page_size) {
ret = submit_one_bio(rw, bio, mirror_num,
prev_bio_flags);
+ if (ret < 0)
+ return ret;
bio = NULL;
} else {
return 0;
return ret;
}
- void set_page_extent_mapped(struct page *page)
+ void attach_extent_buffer_page(struct extent_buffer *eb, struct page *page)
{
if (!PagePrivate(page)) {
SetPagePrivate(page);
page_cache_get(page);
- set_page_private(page, EXTENT_PAGE_PRIVATE);
+ set_page_private(page, (unsigned long)eb);
+ } else {
+ WARN_ON(page->private != (unsigned long)eb);
}
}
- static void set_page_extent_head(struct page *page, unsigned long len)
+ void set_page_extent_mapped(struct page *page)
{
- WARN_ON(!PagePrivate(page));
- set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2);
+ if (!PagePrivate(page)) {
+ SetPagePrivate(page);
+ page_cache_get(page);
+ set_page_private(page, EXTENT_PAGE_PRIVATE);
+ }
}
/*
* basic readpage implementation. Locked extent state structs are inserted
* into the tree that are removed when the IO is done (by the end_io
* handlers)
+ * XXX JDM: This needs looking at to ensure proper page locking
*/
static int __extent_read_full_page(struct extent_io_tree *tree,
struct page *page,
end = page_end;
while (1) {
- lock_extent(tree, start, end, GFP_NOFS);
+ lock_extent(tree, start, end);
ordered = btrfs_lookup_ordered_extent(inode, start);
if (!ordered)
break;
- unlock_extent(tree, start, end, GFP_NOFS);
+ unlock_extent(tree, start, end);
btrfs_start_ordered_extent(inode, ordered, 1);
btrfs_put_ordered_extent(ordered);
}
if (zero_offset) {
iosize = PAGE_CACHE_SIZE - zero_offset;
- userpage = kmap_atomic(page, KM_USER0);
+ userpage = kmap_atomic(page);
memset(userpage + zero_offset, 0, iosize);
flush_dcache_page(page);
- kunmap_atomic(userpage, KM_USER0);
+ kunmap_atomic(userpage);
}
}
while (cur <= end) {
struct extent_state *cached = NULL;
iosize = PAGE_CACHE_SIZE - pg_offset;
- userpage = kmap_atomic(page, KM_USER0);
+ userpage = kmap_atomic(page);
memset(userpage + pg_offset, 0, iosize);
flush_dcache_page(page);
- kunmap_atomic(userpage, KM_USER0);
+ kunmap_atomic(userpage);
set_extent_uptodate(tree, cur, cur + iosize - 1,
&cached, GFP_NOFS);
unlock_extent_cached(tree, cur, cur + iosize - 1,
end - cur + 1, 0);
if (IS_ERR_OR_NULL(em)) {
SetPageError(page);
- unlock_extent(tree, cur, end, GFP_NOFS);
+ unlock_extent(tree, cur, end);
break;
}
extent_offset = cur - em->start;
char *userpage;
struct extent_state *cached = NULL;
- userpage = kmap_atomic(page, KM_USER0);
+ userpage = kmap_atomic(page);
memset(userpage + pg_offset, 0, iosize);
flush_dcache_page(page);
- kunmap_atomic(userpage, KM_USER0);
+ kunmap_atomic(userpage);
set_extent_uptodate(tree, cur, cur + iosize - 1,
&cached, GFP_NOFS);
if (test_range_bit(tree, cur, cur_end,
EXTENT_UPTODATE, 1, NULL)) {
check_page_uptodate(tree, page);
- unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
+ unlock_extent(tree, cur, cur + iosize - 1);
cur = cur + iosize;
pg_offset += iosize;
continue;
*/
if (block_start == EXTENT_MAP_INLINE) {
SetPageError(page);
- unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
+ unlock_extent(tree, cur, cur + iosize - 1);
cur = cur + iosize;
pg_offset += iosize;
continue;
end_bio_extent_readpage, mirror_num,
*bio_flags,
this_bio_flag);
+ BUG_ON(ret == -ENOMEM);
nr++;
*bio_flags = this_bio_flag;
}
if (page->index == end_index) {
char *userpage;
- userpage = kmap_atomic(page, KM_USER0);
+ userpage = kmap_atomic(page);
memset(userpage + pg_offset, 0,
PAGE_CACHE_SIZE - pg_offset);
- kunmap_atomic(userpage, KM_USER0);
+ kunmap_atomic(userpage);
flush_dcache_page(page);
}
pg_offset = 0;
delalloc_end,
&page_started,
&nr_written);
- BUG_ON(ret);
+ /* File system has been set read-only */
+ if (ret) {
+ SetPageError(page);
+ goto done;
+ }
/*
* delalloc_end is already one less than the total
* length, so we don't subtract one from
return 0;
}
+ static int eb_wait(void *word)
+ {
+ io_schedule();
+ return 0;
+ }
+
+ static void wait_on_extent_buffer_writeback(struct extent_buffer *eb)
+ {
+ wait_on_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK, eb_wait,
+ TASK_UNINTERRUPTIBLE);
+ }
+
+ static int lock_extent_buffer_for_io(struct extent_buffer *eb,
+ struct btrfs_fs_info *fs_info,
+ struct extent_page_data *epd)
+ {
+ unsigned long i, num_pages;
+ int flush = 0;
+ int ret = 0;
+
+ if (!btrfs_try_tree_write_lock(eb)) {
+ flush = 1;
+ flush_write_bio(epd);
+ btrfs_tree_lock(eb);
+ }
+
+ if (test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags)) {
+ btrfs_tree_unlock(eb);
+ if (!epd->sync_io)
+ return 0;
+ if (!flush) {
+ flush_write_bio(epd);
+ flush = 1;
+ }
+ while (1) {
+ wait_on_extent_buffer_writeback(eb);
+ btrfs_tree_lock(eb);
+ if (!test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags))
+ break;
+ btrfs_tree_unlock(eb);
+ }
+ }
+
+ if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
+ set_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+ btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
+ spin_lock(&fs_info->delalloc_lock);
+ if (fs_info->dirty_metadata_bytes >= eb->len)
+ fs_info->dirty_metadata_bytes -= eb->len;
+ else
+ WARN_ON(1);
+ spin_unlock(&fs_info->delalloc_lock);
+ ret = 1;
+ }
+
+ btrfs_tree_unlock(eb);
+
+ if (!ret)
+ return ret;
+
+ num_pages = num_extent_pages(eb->start, eb->len);
+ for (i = 0; i < num_pages; i++) {
+ struct page *p = extent_buffer_page(eb, i);
+
+ if (!trylock_page(p)) {
+ if (!flush) {
+ flush_write_bio(epd);
+ flush = 1;
+ }
+ lock_page(p);
+ }
+ }
+
+ return ret;
+ }
+
+ static void end_extent_buffer_writeback(struct extent_buffer *eb)
+ {
+ clear_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+ smp_mb__after_clear_bit();
+ wake_up_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK);
+ }
+
+ static void end_bio_extent_buffer_writepage(struct bio *bio, int err)
+ {
+ int uptodate = err == 0;
+ struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
+ struct extent_buffer *eb;
+ int done;
+
+ do {
+ struct page *page = bvec->bv_page;
+
+ bvec--;
+ eb = (struct extent_buffer *)page->private;
+ BUG_ON(!eb);
+ done = atomic_dec_and_test(&eb->io_pages);
+
+ if (!uptodate || test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
+ set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+ ClearPageUptodate(page);
+ SetPageError(page);
+ }
+
+ end_page_writeback(page);
+
+ if (!done)
+ continue;
+
+ end_extent_buffer_writeback(eb);
+ } while (bvec >= bio->bi_io_vec);
+
+ bio_put(bio);
+
+ }
+
+ static int write_one_eb(struct extent_buffer *eb,
+ struct btrfs_fs_info *fs_info,
+ struct writeback_control *wbc,
+ struct extent_page_data *epd)
+ {
+ struct block_device *bdev = fs_info->fs_devices->latest_bdev;
+ u64 offset = eb->start;
+ unsigned long i, num_pages;
+ int rw = (epd->sync_io ? WRITE_SYNC : WRITE);
+ int ret;
+
+ clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+ num_pages = num_extent_pages(eb->start, eb->len);
+ atomic_set(&eb->io_pages, num_pages);
+ for (i = 0; i < num_pages; i++) {
+ struct page *p = extent_buffer_page(eb, i);
+
+ clear_page_dirty_for_io(p);
+ set_page_writeback(p);
+ ret = submit_extent_page(rw, eb->tree, p, offset >> 9,
+ PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
+ -1, end_bio_extent_buffer_writepage,
+ 0, 0, 0);
+ if (ret) {
+ set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+ SetPageError(p);
+ if (atomic_sub_and_test(num_pages - i, &eb->io_pages))
+ end_extent_buffer_writeback(eb);
+ ret = -EIO;
+ break;
+ }
+ offset += PAGE_CACHE_SIZE;
+ update_nr_written(p, wbc, 1);
+ unlock_page(p);
+ }
+
+ if (unlikely(ret)) {
+ for (; i < num_pages; i++) {
+ struct page *p = extent_buffer_page(eb, i);
+ unlock_page(p);
+ }
+ }
+
+ return ret;
+ }
+
+ int btree_write_cache_pages(struct address_space *mapping,
+ struct writeback_control *wbc)
+ {
+ struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
+ struct btrfs_fs_info *fs_info = BTRFS_I(mapping->host)->root->fs_info;
+ struct extent_buffer *eb, *prev_eb = NULL;
+ struct extent_page_data epd = {
+ .bio = NULL,
+ .tree = tree,
+ .extent_locked = 0,
+ .sync_io = wbc->sync_mode == WB_SYNC_ALL,
+ };
+ int ret = 0;
+ int done = 0;
+ int nr_to_write_done = 0;
+ struct pagevec pvec;
+ int nr_pages;
+ pgoff_t index;
+ pgoff_t end; /* Inclusive */
+ int scanned = 0;
+ int tag;
+
+ pagevec_init(&pvec, 0);
+ if (wbc->range_cyclic) {
+ index = mapping->writeback_index; /* Start from prev offset */
+ end = -1;
+ } else {
+ index = wbc->range_start >> PAGE_CACHE_SHIFT;
+ end = wbc->range_end >> PAGE_CACHE_SHIFT;
+ scanned = 1;
+ }
+ if (wbc->sync_mode == WB_SYNC_ALL)
+ tag = PAGECACHE_TAG_TOWRITE;
+ else
+ tag = PAGECACHE_TAG_DIRTY;
+ retry:
+ if (wbc->sync_mode == WB_SYNC_ALL)
+ tag_pages_for_writeback(mapping, index, end);
+ while (!done && !nr_to_write_done && (index <= end) &&
+ (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
+ min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+ unsigned i;
+
+ scanned = 1;
+ for (i = 0; i < nr_pages; i++) {
+ struct page *page = pvec.pages[i];
+
+ if (!PagePrivate(page))
+ continue;
+
+ if (!wbc->range_cyclic && page->index > end) {
+ done = 1;
+ break;
+ }
+
+ eb = (struct extent_buffer *)page->private;
+ if (!eb) {
+ WARN_ON(1);
+ continue;
+ }
+
+ if (eb == prev_eb)
+ continue;
+
+ if (!atomic_inc_not_zero(&eb->refs)) {
+ WARN_ON(1);
+ continue;
+ }
+
+ prev_eb = eb;
+ ret = lock_extent_buffer_for_io(eb, fs_info, &epd);
+ if (!ret) {
+ free_extent_buffer(eb);
+ continue;
+ }
+
+ ret = write_one_eb(eb, fs_info, wbc, &epd);
+ if (ret) {
+ done = 1;
+ free_extent_buffer(eb);
+ break;
+ }
+ free_extent_buffer(eb);
+
+ /*
+ * the filesystem may choose to bump up nr_to_write.
+ * We have to make sure to honor the new nr_to_write
+ * at any time
+ */
+ nr_to_write_done = wbc->nr_to_write <= 0;
+ }
+ pagevec_release(&pvec);
+ cond_resched();
+ }
+ if (!scanned && !done) {
+ /*
+ * We hit the last page and there is more work to be done: wrap
+ * back to the start of the file
+ */
+ scanned = 1;
+ index = 0;
+ goto retry;
+ }
+ flush_write_bio(&epd);
+ return ret;
+ }
+
/**
* write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
* @mapping: address space structure to write
static void flush_epd_write_bio(struct extent_page_data *epd)
{
if (epd->bio) {
+ int rw = WRITE;
+ int ret;
+
if (epd->sync_io)
- submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
- else
- submit_one_bio(WRITE, epd->bio, 0, 0);
+ rw = WRITE_SYNC;
+
+ ret = submit_one_bio(rw, epd->bio, 0, 0);
+ BUG_ON(ret < 0); /* -ENOMEM */
epd->bio = NULL;
}
}
}
BUG_ON(!list_empty(pages));
if (bio)
- submit_one_bio(READ, bio, 0, bio_flags);
+ return submit_one_bio(READ, bio, 0, bio_flags);
return 0;
}
if (start > end)
return 0;
- lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS);
+ lock_extent_bits(tree, start, end, 0, &cached_state);
wait_on_page_writeback(page);
clear_extent_bit(tree, start, end,
EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
}
lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
- &cached_state, GFP_NOFS);
+ &cached_state);
em = get_extent_skip_holes(inode, start, last_for_get_extent,
get_extent);
inline struct page *extent_buffer_page(struct extent_buffer *eb,
unsigned long i)
{
- struct page *p;
- struct address_space *mapping;
-
- if (i == 0)
- return eb->first_page;
- i += eb->start >> PAGE_CACHE_SHIFT;
- mapping = eb->first_page->mapping;
- if (!mapping)
- return NULL;
-
- /*
- * extent_buffer_page is only called after pinning the page
- * by increasing the reference count. So we know the page must
- * be in the radix tree.
- */
- rcu_read_lock();
- p = radix_tree_lookup(&mapping->page_tree, i);
- rcu_read_unlock();
-
- return p;
+ return eb->pages[i];
}
inline unsigned long num_extent_pages(u64 start, u64 len)
(start >> PAGE_CACHE_SHIFT);
}
+ static void __free_extent_buffer(struct extent_buffer *eb)
+ {
+ #if LEAK_DEBUG
+ unsigned long flags;
+ spin_lock_irqsave(&leak_lock, flags);
+ list_del(&eb->leak_list);
+ spin_unlock_irqrestore(&leak_lock, flags);
+ #endif
+ if (eb->pages && eb->pages != eb->inline_pages)
+ kfree(eb->pages);
+ kmem_cache_free(extent_buffer_cache, eb);
+ }
+
static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
u64 start,
unsigned long len,
return NULL;
eb->start = start;
eb->len = len;
+ eb->tree = tree;
rwlock_init(&eb->lock);
atomic_set(&eb->write_locks, 0);
atomic_set(&eb->read_locks, 0);
list_add(&eb->leak_list, &buffers);
spin_unlock_irqrestore(&leak_lock, flags);
#endif
+ spin_lock_init(&eb->refs_lock);
atomic_set(&eb->refs, 1);
+ atomic_set(&eb->io_pages, 0);
+
+ if (len > MAX_INLINE_EXTENT_BUFFER_SIZE) {
+ struct page **pages;
+ int num_pages = (len + PAGE_CACHE_SIZE - 1) >>
+ PAGE_CACHE_SHIFT;
+ pages = kzalloc(num_pages, mask);
+ if (!pages) {
+ __free_extent_buffer(eb);
+ return NULL;
+ }
+ eb->pages = pages;
+ } else {
+ eb->pages = eb->inline_pages;
+ }
return eb;
}
- static void __free_extent_buffer(struct extent_buffer *eb)
+ static int extent_buffer_under_io(struct extent_buffer *eb)
{
- #if LEAK_DEBUG
- unsigned long flags;
- spin_lock_irqsave(&leak_lock, flags);
- list_del(&eb->leak_list);
- spin_unlock_irqrestore(&leak_lock, flags);
- #endif
- kmem_cache_free(extent_buffer_cache, eb);
+ return (atomic_read(&eb->io_pages) ||
+ test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags) ||
+ test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
}
/*
unsigned long index;
struct page *page;
- if (!eb->first_page)
- return;
+ BUG_ON(extent_buffer_under_io(eb));
index = num_extent_pages(eb->start, eb->len);
if (start_idx >= index)
do {
index--;
page = extent_buffer_page(eb, index);
- if (page)
+ if (page) {
+ spin_lock(&page->mapping->private_lock);
+ /*
+ * We do this since we'll remove the pages after we've
+ * removed the eb from the radix tree, so we could race
+ * and have this page now attached to the new eb. So
+ * only clear page_private if it's still connected to
+ * this eb.
+ */
+ if (PagePrivate(page) &&
+ page->private == (unsigned long)eb) {
+ BUG_ON(test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+ BUG_ON(PageDirty(page));
+ BUG_ON(PageWriteback(page));
+ /*
+ * We need to make sure we haven't be attached
+ * to a new eb.
+ */
+ ClearPagePrivate(page);
+ set_page_private(page, 0);
+ /* One for the page private */
+ page_cache_release(page);
+ }
+ spin_unlock(&page->mapping->private_lock);
+
+ /* One for when we alloced the page */
page_cache_release(page);
+ }
} while (index != start_idx);
}
__free_extent_buffer(eb);
}
+ static void check_buffer_tree_ref(struct extent_buffer *eb)
+ {
+ /* the ref bit is tricky. We have to make sure it is set
+ * if we have the buffer dirty. Otherwise the
+ * code to free a buffer can end up dropping a dirty
+ * page
+ *
+ * Once the ref bit is set, it won't go away while the
+ * buffer is dirty or in writeback, and it also won't
+ * go away while we have the reference count on the
+ * eb bumped.
+ *
+ * We can't just set the ref bit without bumping the
+ * ref on the eb because free_extent_buffer might
+ * see the ref bit and try to clear it. If this happens
+ * free_extent_buffer might end up dropping our original
+ * ref by mistake and freeing the page before we are able
+ * to add one more ref.
+ *
+ * So bump the ref count first, then set the bit. If someone
+ * beat us to it, drop the ref we added.
+ */
+ if (!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+ atomic_inc(&eb->refs);
+ if (test_and_set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+ atomic_dec(&eb->refs);
+ }
+ }
+
+ static void mark_extent_buffer_accessed(struct extent_buffer *eb)
+ {
+ unsigned long num_pages, i;
+
+ check_buffer_tree_ref(eb);
+
+ num_pages = num_extent_pages(eb->start, eb->len);
+ for (i = 0; i < num_pages; i++) {
+ struct page *p = extent_buffer_page(eb, i);
+ mark_page_accessed(p);
+ }
+ }
+
struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
- u64 start, unsigned long len,
- struct page *page0)
+ u64 start, unsigned long len)
{
unsigned long num_pages = num_extent_pages(start, len);
unsigned long i;
eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
if (eb && atomic_inc_not_zero(&eb->refs)) {
rcu_read_unlock();
- mark_page_accessed(eb->first_page);
+ mark_extent_buffer_accessed(eb);
return eb;
}
rcu_read_unlock();
if (!eb)
return NULL;
- if (page0) {
- eb->first_page = page0;
- i = 1;
- index++;
- page_cache_get(page0);
- mark_page_accessed(page0);
- set_page_extent_mapped(page0);
- set_page_extent_head(page0, len);
- uptodate = PageUptodate(page0);
- } else {
- i = 0;
- }
- for (; i < num_pages; i++, index++) {
+ for (i = 0; i < num_pages; i++, index++) {
p = find_or_create_page(mapping, index, GFP_NOFS);
if (!p) {
WARN_ON(1);
goto free_eb;
}
- set_page_extent_mapped(p);
- mark_page_accessed(p);
- if (i == 0) {
- eb->first_page = p;
- set_page_extent_head(p, len);
- } else {
- set_page_private(p, EXTENT_PAGE_PRIVATE);
+
+ spin_lock(&mapping->private_lock);
+ if (PagePrivate(p)) {
+ /*
+ * We could have already allocated an eb for this page
+ * and attached one so lets see if we can get a ref on
+ * the existing eb, and if we can we know it's good and
+ * we can just return that one, else we know we can just
+ * overwrite page->private.
+ */
+ exists = (struct extent_buffer *)p->private;
+ if (atomic_inc_not_zero(&exists->refs)) {
+ spin_unlock(&mapping->private_lock);
+ unlock_page(p);
+ mark_extent_buffer_accessed(exists);
+ goto free_eb;
+ }
+
+ /*
+ * Do this so attach doesn't complain and we need to
+ * drop the ref the old guy had.
+ */
+ ClearPagePrivate(p);
+ WARN_ON(PageDirty(p));
+ page_cache_release(p);
}
+ attach_extent_buffer_page(eb, p);
+ spin_unlock(&mapping->private_lock);
+ WARN_ON(PageDirty(p));
+ mark_page_accessed(p);
+ eb->pages[i] = p;
if (!PageUptodate(p))
uptodate = 0;
* see below about how we avoid a nasty race with release page
* and why we unlock later
*/
- if (i != 0)
- unlock_page(p);
}
if (uptodate)
set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-
+ again:
ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
if (ret)
goto free_eb;
if (ret == -EEXIST) {
exists = radix_tree_lookup(&tree->buffer,
start >> PAGE_CACHE_SHIFT);
- /* add one reference for the caller */
- atomic_inc(&exists->refs);
+ if (!atomic_inc_not_zero(&exists->refs)) {
+ spin_unlock(&tree->buffer_lock);
+ radix_tree_preload_end();
+ exists = NULL;
+ goto again;
+ }
spin_unlock(&tree->buffer_lock);
radix_tree_preload_end();
+ mark_extent_buffer_accessed(exists);
goto free_eb;
}
/* add one reference for the tree */
- atomic_inc(&eb->refs);
+ spin_lock(&eb->refs_lock);
+ check_buffer_tree_ref(eb);
+ spin_unlock(&eb->refs_lock);
spin_unlock(&tree->buffer_lock);
radix_tree_preload_end();
* after the extent buffer is in the radix tree so
* it doesn't get lost
*/
- set_page_extent_mapped(eb->first_page);
- set_page_extent_head(eb->first_page, eb->len);
- if (!page0)
- unlock_page(eb->first_page);
+ SetPageChecked(eb->pages[0]);
+ for (i = 1; i < num_pages; i++) {
+ p = extent_buffer_page(eb, i);
+ ClearPageChecked(p);
+ unlock_page(p);
+ }
+ unlock_page(eb->pages[0]);
return eb;
free_eb:
- if (eb->first_page && !page0)
- unlock_page(eb->first_page);
+ for (i = 0; i < num_pages; i++) {
+ if (eb->pages[i])
+ unlock_page(eb->pages[i]);
+ }
if (!atomic_dec_and_test(&eb->refs))
return exists;
eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
if (eb && atomic_inc_not_zero(&eb->refs)) {
rcu_read_unlock();
- mark_page_accessed(eb->first_page);
+ mark_extent_buffer_accessed(eb);
return eb;
}
rcu_read_unlock();
return NULL;
}
+ static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
+ {
+ struct extent_buffer *eb =
+ container_of(head, struct extent_buffer, rcu_head);
+
+ __free_extent_buffer(eb);
+ }
+
+ /* Expects to have eb->eb_lock already held */
+ static void release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
+ {
+ WARN_ON(atomic_read(&eb->refs) == 0);
+ if (atomic_dec_and_test(&eb->refs)) {
+ struct extent_io_tree *tree = eb->tree;
+
+ spin_unlock(&eb->refs_lock);
+
+ spin_lock(&tree->buffer_lock);
+ radix_tree_delete(&tree->buffer,
+ eb->start >> PAGE_CACHE_SHIFT);
+ spin_unlock(&tree->buffer_lock);
+
+ /* Should be safe to release our pages at this point */
+ btrfs_release_extent_buffer_page(eb, 0);
+
+ call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
+ return;
+ }
+ spin_unlock(&eb->refs_lock);
+ }
+
void free_extent_buffer(struct extent_buffer *eb)
{
if (!eb)
return;
- if (!atomic_dec_and_test(&eb->refs))
+ spin_lock(&eb->refs_lock);
+ if (atomic_read(&eb->refs) == 2 &&
+ test_bit(EXTENT_BUFFER_STALE, &eb->bflags) &&
+ !extent_buffer_under_io(eb) &&
+ test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+ atomic_dec(&eb->refs);
+
+ /*
+ * I know this is terrible, but it's temporary until we stop tracking
+ * the uptodate bits and such for the extent buffers.
+ */
+ release_extent_buffer(eb, GFP_ATOMIC);
+ }
+
+ void free_extent_buffer_stale(struct extent_buffer *eb)
+ {
+ if (!eb)
return;
- WARN_ON(1);
+ spin_lock(&eb->refs_lock);
+ set_bit(EXTENT_BUFFER_STALE, &eb->bflags);
+
+ if (atomic_read(&eb->refs) == 2 && !extent_buffer_under_io(eb) &&
+ test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+ atomic_dec(&eb->refs);
+ release_extent_buffer(eb, GFP_NOFS);
}
- int clear_extent_buffer_dirty(struct extent_io_tree *tree,
- struct extent_buffer *eb)
+ void clear_extent_buffer_dirty(struct extent_buffer *eb)
{
unsigned long i;
unsigned long num_pages;
lock_page(page);
WARN_ON(!PagePrivate(page));
- set_page_extent_mapped(page);
- if (i == 0)
- set_page_extent_head(page, eb->len);
-
clear_page_dirty_for_io(page);
spin_lock_irq(&page->mapping->tree_lock);
if (!PageDirty(page)) {
ClearPageError(page);
unlock_page(page);
}
- return 0;
+ WARN_ON(atomic_read(&eb->refs) == 0);
}
- int set_extent_buffer_dirty(struct extent_io_tree *tree,
- struct extent_buffer *eb)
+ int set_extent_buffer_dirty(struct extent_buffer *eb)
{
unsigned long i;
unsigned long num_pages;
int was_dirty = 0;
+ check_buffer_tree_ref(eb);
+
was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
+
num_pages = num_extent_pages(eb->start, eb->len);
+ WARN_ON(atomic_read(&eb->refs) == 0);
+ WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));
+
for (i = 0; i < num_pages; i++)
- __set_page_dirty_nobuffers(extent_buffer_page(eb, i));
+ set_page_dirty(extent_buffer_page(eb, i));
return was_dirty;
}
- static int __eb_straddles_pages(u64 start, u64 len)
+ static int range_straddles_pages(u64 start, u64 len)
{
if (len < PAGE_CACHE_SIZE)
return 1;
return 0;
}
- static int eb_straddles_pages(struct extent_buffer *eb)
- {
- return __eb_straddles_pages(eb->start, eb->len);
- }
-
- int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
- struct extent_buffer *eb,
- struct extent_state **cached_state)
+ int clear_extent_buffer_uptodate(struct extent_buffer *eb)
{
unsigned long i;
struct page *page;
unsigned long num_pages;
- num_pages = num_extent_pages(eb->start, eb->len);
clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-
- clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
- cached_state, GFP_NOFS);
-
+ num_pages = num_extent_pages(eb->start, eb->len);
for (i = 0; i < num_pages; i++) {
page = extent_buffer_page(eb, i);
if (page)
return 0;
}
- int set_extent_buffer_uptodate(struct extent_io_tree *tree,
- struct extent_buffer *eb)
+ int set_extent_buffer_uptodate(struct extent_buffer *eb)
{
unsigned long i;
struct page *page;
unsigned long num_pages;
+ set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
num_pages = num_extent_pages(eb->start, eb->len);
-
- if (eb_straddles_pages(eb)) {
- set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
- NULL, GFP_NOFS);
- }
for (i = 0; i < num_pages; i++) {
page = extent_buffer_page(eb, i);
- if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
- ((i == num_pages - 1) &&
- ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
- check_page_uptodate(tree, page);
- continue;
- }
SetPageUptodate(page);
}
return 0;
int uptodate;
unsigned long index;
- if (__eb_straddles_pages(start, end - start + 1)) {
+ if (range_straddles_pages(start, end - start + 1)) {
ret = test_range_bit(tree, start, end,
EXTENT_UPTODATE, 1, NULL);
if (ret)
return pg_uptodate;
}
- int extent_buffer_uptodate(struct extent_io_tree *tree,
- struct extent_buffer *eb,
- struct extent_state *cached_state)
+ int extent_buffer_uptodate(struct extent_buffer *eb)
{
- int ret = 0;
- unsigned long num_pages;
- unsigned long i;
- struct page *page;
- int pg_uptodate = 1;
-
- if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
- return 1;
-
- if (eb_straddles_pages(eb)) {
- ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
- EXTENT_UPTODATE, 1, cached_state);
- if (ret)
- return ret;
- }
-
- num_pages = num_extent_pages(eb->start, eb->len);
- for (i = 0; i < num_pages; i++) {
- page = extent_buffer_page(eb, i);
- if (!PageUptodate(page)) {
- pg_uptodate = 0;
- break;
- }
- }
- return pg_uptodate;
+ return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
}
int read_extent_buffer_pages(struct extent_io_tree *tree,
int ret = 0;
int locked_pages = 0;
int all_uptodate = 1;
- int inc_all_pages = 0;
unsigned long num_pages;
+ unsigned long num_reads = 0;
struct bio *bio = NULL;
unsigned long bio_flags = 0;
if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
return 0;
- if (eb_straddles_pages(eb)) {
- if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
- EXTENT_UPTODATE, 1, NULL)) {
- return 0;
- }
- }
-
if (start) {
WARN_ON(start < eb->start);
start_i = (start >> PAGE_CACHE_SHIFT) -
lock_page(page);
}
locked_pages++;
- if (!PageUptodate(page))
+ if (!PageUptodate(page)) {
+ num_reads++;
all_uptodate = 0;
+ }
}
if (all_uptodate) {
if (start_i == 0)
goto unlock_exit;
}
+ clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+ eb->failed_mirror = 0;
+ atomic_set(&eb->io_pages, num_reads);
for (i = start_i; i < num_pages; i++) {
page = extent_buffer_page(eb, i);
-
- WARN_ON(!PagePrivate(page));
-
- set_page_extent_mapped(page);
- if (i == 0)
- set_page_extent_head(page, eb->len);
-
- if (inc_all_pages)
- page_cache_get(page);
if (!PageUptodate(page)) {
- if (start_i == 0)
- inc_all_pages = 1;
ClearPageError(page);
err = __extent_read_full_page(tree, page,
get_extent, &bio,
}
}
- if (bio)
- submit_one_bio(READ, bio, mirror_num, bio_flags);
+ if (bio) {
+ err = submit_one_bio(READ, bio, mirror_num, bio_flags);
+ if (err)
+ return err;
+ }
if (ret || wait != WAIT_COMPLETE)
return ret;
ret = -EIO;
}
- if (!ret)
- set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
return ret;
unlock_exit:
{
char *dst_kaddr = page_address(dst_page);
char *src_kaddr;
+ int must_memmove = 0;
if (dst_page != src_page) {
src_kaddr = page_address(src_page);
} else {
src_kaddr = dst_kaddr;
- BUG_ON(areas_overlap(src_off, dst_off, len));
+ if (areas_overlap(src_off, dst_off, len))
+ must_memmove = 1;
}
- memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
+ if (must_memmove)
+ memmove(dst_kaddr + dst_off, src_kaddr + src_off, len);
+ else
+ memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
}
void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
"len %lu len %lu\n", dst_offset, len, dst->len);
BUG_ON(1);
}
- if (!areas_overlap(src_offset, dst_offset, len)) {
+ if (dst_offset < src_offset) {
memcpy_extent_buffer(dst, dst_offset, src_offset, len);
return;
}
}
}
- static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
+ int try_release_extent_buffer(struct page *page, gfp_t mask)
{
- struct extent_buffer *eb =
- container_of(head, struct extent_buffer, rcu_head);
-
- btrfs_release_extent_buffer(eb);
- }
-
- int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page)
- {
- u64 start = page_offset(page);
struct extent_buffer *eb;
- int ret = 1;
- spin_lock(&tree->buffer_lock);
- eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
- if (!eb) {
- spin_unlock(&tree->buffer_lock);
- return ret;
+ /*
+ * We need to make sure noboody is attaching this page to an eb right
+ * now.
+ */
+ spin_lock(&page->mapping->private_lock);
+ if (!PagePrivate(page)) {
+ spin_unlock(&page->mapping->private_lock);
+ return 1;
}
- if (test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
- ret = 0;
- goto out;
- }
+ eb = (struct extent_buffer *)page->private;
+ BUG_ON(!eb);
/*
- * set @eb->refs to 0 if it is already 1, and then release the @eb.
- * Or go back.
+ * This is a little awful but should be ok, we need to make sure that
+ * the eb doesn't disappear out from under us while we're looking at
+ * this page.
*/
- if (atomic_cmpxchg(&eb->refs, 1, 0) != 1) {
- ret = 0;
- goto out;
+ spin_lock(&eb->refs_lock);
+ if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb)) {
+ spin_unlock(&eb->refs_lock);
+ spin_unlock(&page->mapping->private_lock);
+ return 0;
}
+ spin_unlock(&page->mapping->private_lock);
- radix_tree_delete(&tree->buffer, start >> PAGE_CACHE_SHIFT);
- out:
- spin_unlock(&tree->buffer_lock);
+ if ((mask & GFP_NOFS) == GFP_NOFS)
+ mask = GFP_NOFS;
- /* at this point we can safely release the extent buffer */
- if (atomic_read(&eb->refs) == 0)
- call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
- return ret;
+ /*
+ * If tree ref isn't set then we know the ref on this eb is a real ref,
+ * so just return, this page will likely be freed soon anyway.
+ */
+ if (!test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+ spin_unlock(&eb->refs_lock);
+ return 0;
+ }
+ release_extent_buffer(eb, mask);
+
+ return 1;
}
#include "transaction.h"
#include "print-tree.h"
- #define MAX_CSUM_ITEMS(r, size) ((((BTRFS_LEAF_DATA_SIZE(r) - \
+ #define __MAX_CSUM_ITEMS(r, size) ((((BTRFS_LEAF_DATA_SIZE(r) - \
sizeof(struct btrfs_item) * 2) / \
size) - 1))
+ #define MAX_CSUM_ITEMS(r, size) (min(__MAX_CSUM_ITEMS(r, size), PAGE_CACHE_SIZE))
+
#define MAX_ORDERED_SUM_BYTES(r) ((PAGE_SIZE - \
sizeof(struct btrfs_ordered_sum)) / \
sizeof(struct btrfs_sector_sum) * \
sizeof(*item));
if (ret < 0)
goto out;
- BUG_ON(ret);
+ BUG_ON(ret); /* Can't happen */
leaf = path->nodes[0];
item = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
struct btrfs_ordered_sum *sums;
struct btrfs_sector_sum *sector_sum;
struct btrfs_csum_item *item;
+ LIST_HEAD(tmplist);
unsigned long offset;
int ret;
size_t size;
MAX_ORDERED_SUM_BYTES(root));
sums = kzalloc(btrfs_ordered_sum_size(root, size),
GFP_NOFS);
- BUG_ON(!sums);
+ if (!sums) {
+ ret = -ENOMEM;
+ goto fail;
+ }
sector_sum = sums->sums;
sums->bytenr = start;
offset += csum_size;
sector_sum++;
}
- list_add_tail(&sums->list, list);
+ list_add_tail(&sums->list, &tmplist);
}
path->slots[0]++;
}
ret = 0;
fail:
+ while (ret < 0 && !list_empty(&tmplist)) {
+ sums = list_entry(&tmplist, struct btrfs_ordered_sum, list);
+ list_del(&sums->list);
+ kfree(sums);
+ }
+ list_splice_tail(&tmplist, list);
+
btrfs_free_path(path);
return ret;
}
offset = page_offset(bvec->bv_page) + bvec->bv_offset;
ordered = btrfs_lookup_ordered_extent(inode, offset);
- BUG_ON(!ordered);
+ BUG_ON(!ordered); /* Logic error */
sums->bytenr = ordered->start;
while (bio_index < bio->bi_vcnt) {
sums = kzalloc(btrfs_ordered_sum_size(root, bytes_left),
GFP_NOFS);
- BUG_ON(!sums);
+ BUG_ON(!sums); /* -ENOMEM */
sector_sum = sums->sums;
sums->len = bytes_left;
ordered = btrfs_lookup_ordered_extent(inode, offset);
- BUG_ON(!ordered);
+ BUG_ON(!ordered); /* Logic error */
sums->bytenr = ordered->start;
}
- data = kmap_atomic(bvec->bv_page, KM_USER0);
+ data = kmap_atomic(bvec->bv_page);
sector_sum->sum = ~(u32)0;
sector_sum->sum = btrfs_csum_data(root,
data + bvec->bv_offset,
sector_sum->sum,
bvec->bv_len);
- kunmap_atomic(data, KM_USER0);
+ kunmap_atomic(data);
btrfs_csum_final(sector_sum->sum,
(char *)§or_sum->sum);
sector_sum->bytenr = disk_bytenr;
* This calls btrfs_truncate_item with the correct args based on the
* overlap, and fixes up the key as required.
*/
- static noinline int truncate_one_csum(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_path *path,
- struct btrfs_key *key,
- u64 bytenr, u64 len)
+ static noinline void truncate_one_csum(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct btrfs_key *key,
+ u64 bytenr, u64 len)
{
struct extent_buffer *leaf;
u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
u64 csum_end;
u64 end_byte = bytenr + len;
u32 blocksize_bits = root->fs_info->sb->s_blocksize_bits;
- int ret;
leaf = path->nodes[0];
csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size;
*/
u32 new_size = (bytenr - key->offset) >> blocksize_bits;
new_size *= csum_size;
- ret = btrfs_truncate_item(trans, root, path, new_size, 1);
+ btrfs_truncate_item(trans, root, path, new_size, 1);
} else if (key->offset >= bytenr && csum_end > end_byte &&
end_byte > key->offset) {
/*
u32 new_size = (csum_end - end_byte) >> blocksize_bits;
new_size *= csum_size;
- ret = btrfs_truncate_item(trans, root, path, new_size, 0);
+ btrfs_truncate_item(trans, root, path, new_size, 0);
key->offset = end_byte;
- ret = btrfs_set_item_key_safe(trans, root, path, key);
- BUG_ON(ret);
+ btrfs_set_item_key_safe(trans, root, path, key);
} else {
BUG();
}
- return 0;
}
/*
* item changed size or key
*/
ret = btrfs_split_item(trans, root, path, &key, offset);
- BUG_ON(ret && ret != -EAGAIN);
+ if (ret && ret != -EAGAIN) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
key.offset = end_byte - 1;
} else {
- ret = truncate_one_csum(trans, root, path,
- &key, bytenr, len);
- BUG_ON(ret);
+ truncate_one_csum(trans, root, path, &key, bytenr, len);
if (key.offset < bytenr)
break;
}
if (diff != csum_size)
goto insert;
- ret = btrfs_extend_item(trans, root, path, diff);
+ btrfs_extend_item(trans, root, path, diff);
goto csum;
}
if (ret) {
trans->block_rsv = rsv;
- WARN_ON(1);
+ btrfs_abort_transaction(trans, root, ret);
return ret;
}
ret = btrfs_update_inode(trans, root, inode);
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
trans->block_rsv = rsv;
return ret;
io_ctl_prepare_pages(&io_ctl, inode, 0);
lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
- 0, &cached_state, GFP_NOFS);
+ 0, &cached_state);
node = rb_first(&ctl->free_space_offset);
if (!node && cluster) {
spin_unlock(&block_group->lock);
ret = 0;
#ifdef DEBUG
- printk(KERN_ERR "btrfs: failed to write free space cace "
+ printk(KERN_ERR "btrfs: failed to write free space cache "
"for block group %llu\n", block_group->key.objectid);
#endif
}
*/
ret = btrfs_add_free_space(block_group, old_start,
offset - old_start);
- WARN_ON(ret);
+ WARN_ON(ret); /* -ENOMEM */
goto out;
}
ret = remove_from_bitmap(ctl, info, &offset, &bytes);
if (ret == -EAGAIN)
goto again;
- BUG_ON(ret);
+ BUG_ON(ret); /* logic error */
out_lock:
spin_unlock(&ctl->tree_lock);
out:
rb_erase(&entry->offset_index, &ctl->free_space_offset);
ret = tree_insert_offset(&cluster->root, entry->offset,
&entry->offset_index, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -EEXIST; Logic error */
trace_btrfs_setup_cluster(block_group, cluster,
total_found * block_group->sectorsize, 1);
ret = tree_insert_offset(&cluster->root, entry->offset,
&entry->offset_index, 0);
total_size += entry->bytes;
- BUG_ON(ret);
+ BUG_ON(ret); /* -EEXIST; Logic error */
} while (node && entry != last);
cluster->max_size = max_extent;
int ret;
ret = search_bitmap(ctl, entry, &offset, &count);
+ /* Logic error; Should be empty if it can't find anything */
BUG_ON(ret);
ino = offset;
inode_add_bytes(inode, size);
ret = btrfs_insert_empty_item(trans, root, path, &key,
datasize);
- BUG_ON(ret);
if (ret) {
err = ret;
goto fail;
cur_size = min_t(unsigned long, compressed_size,
PAGE_CACHE_SIZE);
- kaddr = kmap_atomic(cpage, KM_USER0);
+ kaddr = kmap_atomic(cpage);
write_extent_buffer(leaf, kaddr, ptr, cur_size);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
i++;
ptr += cur_size;
page = find_get_page(inode->i_mapping,
start >> PAGE_CACHE_SHIFT);
btrfs_set_file_extent_compression(leaf, ei, 0);
- kaddr = kmap_atomic(page, KM_USER0);
+ kaddr = kmap_atomic(page);
offset = start & (PAGE_CACHE_SIZE - 1);
write_extent_buffer(leaf, kaddr + offset, ptr, size);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
page_cache_release(page);
}
btrfs_mark_buffer_dirty(leaf);
* could end up racing with unlink.
*/
BTRFS_I(inode)->disk_i_size = inode->i_size;
- btrfs_update_inode(trans, root, inode);
+ ret = btrfs_update_inode(trans, root, inode);
- return 0;
+ return ret;
fail:
btrfs_free_path(path);
return err;
ret = btrfs_drop_extents(trans, inode, start, aligned_end,
&hint_byte, 1);
- BUG_ON(ret);
+ if (ret)
+ return ret;
if (isize > actual_end)
inline_len = min_t(u64, isize, actual_end);
ret = insert_inline_extent(trans, root, inode, start,
inline_len, compressed_size,
compress_type, compressed_pages);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ return ret;
+ }
btrfs_delalloc_release_metadata(inode, end + 1 - start);
btrfs_drop_extent_cache(inode, start, aligned_end - 1, 0);
return 0;
struct async_extent *async_extent;
async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS);
- BUG_ON(!async_extent);
+ BUG_ON(!async_extent); /* -ENOMEM */
async_extent->start = start;
async_extent->ram_size = ram_size;
async_extent->compressed_size = compressed_size;
int will_compress;
int compress_type = root->fs_info->compress_type;
- /* if this is a small write inside eof, kick off a defragbot */
- if (end <= BTRFS_I(inode)->disk_i_size && (end - start + 1) < 16 * 1024)
+ /* if this is a small write inside eof, kick off a defrag */
+ if ((end - start + 1) < 16 * 1024 &&
+ (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
btrfs_add_inode_defrag(NULL, inode);
actual_end = min_t(u64, isize, end + 1);
* sending it down to disk
*/
if (offset) {
- kaddr = kmap_atomic(page, KM_USER0);
+ kaddr = kmap_atomic(page);
memset(kaddr + offset, 0,
PAGE_CACHE_SIZE - offset);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
}
will_compress = 1;
}
cont:
if (start == 0) {
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ trans = NULL;
+ goto cleanup_and_out;
+ }
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
/* lets try to make an inline extent */
total_compressed,
compress_type, pages);
}
- if (ret == 0) {
+ if (ret <= 0) {
/*
- * inline extent creation worked, we don't need
- * to create any more async work items. Unlock
- * and free up our temp pages.
+ * inline extent creation worked or returned error,
+ * we don't need to create any more async work items.
+ * Unlock and free up our temp pages.
*/
extent_clear_unlock_delalloc(inode,
&BTRFS_I(inode)->io_tree,
}
out:
- return 0;
+ return ret;
free_pages_out:
for (i = 0; i < nr_pages_ret; i++) {
kfree(pages);
goto out;
+
+ cleanup_and_out:
+ extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_DIRTY |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_SET_WRITEBACK |
+ EXTENT_END_WRITEBACK);
+ if (!trans || IS_ERR(trans))
+ btrfs_error(root->fs_info, ret, "Failed to join transaction");
+ else
+ btrfs_abort_transaction(trans, root, ret);
+ goto free_pages_out;
}
/*
lock_extent(io_tree, async_extent->start,
async_extent->start +
- async_extent->ram_size - 1, GFP_NOFS);
+ async_extent->ram_size - 1);
/* allocate blocks */
ret = cow_file_range(inode, async_cow->locked_page,
async_extent->ram_size - 1,
&page_started, &nr_written, 0);
+ /* JDM XXX */
+
/*
* if page_started, cow_file_range inserted an
* inline extent and took care of all the unlocking
}
lock_extent(io_tree, async_extent->start,
- async_extent->start + async_extent->ram_size - 1,
- GFP_NOFS);
+ async_extent->start + async_extent->ram_size - 1);
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
- trans->block_rsv = &root->fs_info->delalloc_block_rsv;
- ret = btrfs_reserve_extent(trans, root,
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ } else {
+ trans->block_rsv = &root->fs_info->delalloc_block_rsv;
+ ret = btrfs_reserve_extent(trans, root,
async_extent->compressed_size,
async_extent->compressed_size,
- 0, alloc_hint,
- (u64)-1, &ins, 1);
- btrfs_end_transaction(trans, root);
+ 0, alloc_hint, &ins, 1);
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
+ btrfs_end_transaction(trans, root);
+ }
if (ret) {
int i;
async_extent->pages = NULL;
unlock_extent(io_tree, async_extent->start,
async_extent->start +
- async_extent->ram_size - 1, GFP_NOFS);
- goto retry;
+ async_extent->ram_size - 1);
+ if (ret == -ENOSPC)
+ goto retry;
+ goto out_free; /* JDM: Requeue? */
}
/*
async_extent->ram_size - 1, 0);
em = alloc_extent_map();
- BUG_ON(!em);
+ BUG_ON(!em); /* -ENOMEM */
em->start = async_extent->start;
em->len = async_extent->ram_size;
em->orig_start = em->start;
ins.offset,
BTRFS_ORDERED_COMPRESSED,
async_extent->compress_type);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
/*
* clear dirty, set writeback and unlock the pages.
ins.offset, async_extent->pages,
async_extent->nr_pages);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
alloc_hint = ins.objectid + ins.offset;
kfree(async_extent);
cond_resched();
}
-
- return 0;
+ ret = 0;
+ out:
+ return ret;
+ out_free:
+ kfree(async_extent);
+ goto out;
}
static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
BUG_ON(btrfs_is_free_space_inode(root, inode));
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
+ if (IS_ERR(trans)) {
+ extent_clear_unlock_delalloc(inode,
+ &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_UNLOCK |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_CLEAR_DIRTY |
+ EXTENT_SET_WRITEBACK |
+ EXTENT_END_WRITEBACK);
+ return PTR_ERR(trans);
+ }
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
num_bytes = (end - start + blocksize) & ~(blocksize - 1);
ret = 0;
/* if this is a small write inside eof, kick off defrag */
- if (end <= BTRFS_I(inode)->disk_i_size && num_bytes < 64 * 1024)
+ if (num_bytes < 64 * 1024 &&
+ (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
btrfs_add_inode_defrag(trans, inode);
if (start == 0) {
*nr_written = *nr_written +
(end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
*page_started = 1;
- ret = 0;
goto out;
+ } else if (ret < 0) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_unlock;
}
}
cur_alloc_size = disk_num_bytes;
ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
root->sectorsize, 0, alloc_hint,
- (u64)-1, &ins, 1);
- BUG_ON(ret);
+ &ins, 1);
+ if (ret < 0) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_unlock;
+ }
em = alloc_extent_map();
- BUG_ON(!em);
+ BUG_ON(!em); /* -ENOMEM */
em->start = start;
em->orig_start = em->start;
ram_size = ins.offset;
cur_alloc_size = ins.offset;
ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
ram_size, cur_alloc_size, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
if (root->root_key.objectid ==
BTRFS_DATA_RELOC_TREE_OBJECTID) {
ret = btrfs_reloc_clone_csums(inode, start,
cur_alloc_size);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_unlock;
+ }
}
if (disk_num_bytes < cur_alloc_size)
alloc_hint = ins.objectid + ins.offset;
start += cur_alloc_size;
}
- out:
ret = 0;
+ out:
btrfs_end_transaction(trans, root);
return ret;
+ out_unlock:
+ extent_clear_unlock_delalloc(inode,
+ &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_UNLOCK |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_CLEAR_DIRTY |
+ EXTENT_SET_WRITEBACK |
+ EXTENT_END_WRITEBACK);
+
+ goto out;
}
/*
1, 0, NULL, GFP_NOFS);
while (start < end) {
async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS);
- BUG_ON(!async_cow);
+ BUG_ON(!async_cow); /* -ENOMEM */
async_cow->inode = inode;
async_cow->root = root;
async_cow->locked_page = locked_page;
u64 disk_bytenr;
u64 num_bytes;
int extent_type;
- int ret;
+ int ret, err;
int type;
int nocow;
int check_prev = 1;
else
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
+ if (IS_ERR(trans)) {
+ btrfs_free_path(path);
+ return PTR_ERR(trans);
+ }
+
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
cow_start = (u64)-1;
while (1) {
ret = btrfs_lookup_file_extent(trans, root, path, ino,
cur_offset, 0);
- BUG_ON(ret < 0);
+ if (ret < 0) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto error;
+ }
if (ret > 0 && path->slots[0] > 0 && check_prev) {
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &found_key,
leaf = path->nodes[0];
if (path->slots[0] >= btrfs_header_nritems(leaf)) {
ret = btrfs_next_leaf(root, path);
- if (ret < 0)
- BUG_ON(1);
+ if (ret < 0) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto error;
+ }
if (ret > 0)
break;
leaf = path->nodes[0];
ret = cow_file_range(inode, locked_page, cow_start,
found_key.offset - 1, page_started,
nr_written, 1);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto error;
+ }
cow_start = (u64)-1;
}
struct extent_map_tree *em_tree;
em_tree = &BTRFS_I(inode)->extent_tree;
em = alloc_extent_map();
- BUG_ON(!em);
+ BUG_ON(!em); /* -ENOMEM */
em->start = cur_offset;
em->orig_start = em->start;
em->len = num_bytes;
ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr,
num_bytes, num_bytes, type);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
if (root->root_key.objectid ==
BTRFS_DATA_RELOC_TREE_OBJECTID) {
ret = btrfs_reloc_clone_csums(inode, cur_offset,
num_bytes);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto error;
+ }
}
extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
if (cow_start != (u64)-1) {
ret = cow_file_range(inode, locked_page, cow_start, end,
page_started, nr_written, 1);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto error;
+ }
}
+ error:
if (nolock) {
- ret = btrfs_end_transaction_nolock(trans, root);
- BUG_ON(ret);
+ err = btrfs_end_transaction_nolock(trans, root);
} else {
- ret = btrfs_end_transaction(trans, root);
- BUG_ON(ret);
+ err = btrfs_end_transaction(trans, root);
}
+ if (!ret)
+ ret = err;
+
btrfs_free_path(path);
- return 0;
+ return ret;
}
/*
map_length = length;
ret = btrfs_map_block(map_tree, READ, logical,
&map_length, NULL, 0);
-
+ /* Will always return 0 or 1 with map_multi == NULL */
+ BUG_ON(ret < 0);
if (map_length < length + size)
return 1;
- return ret;
+ return 0;
}
/*
int ret = 0;
ret = btrfs_csum_one_bio(root, inode, bio, 0, 0);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
return 0;
}
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
int skip_sum;
+ int metadata = 0;
skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
if (btrfs_is_free_space_inode(root, inode))
- ret = btrfs_bio_wq_end_io(root->fs_info, bio, 2);
- else
- ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
- BUG_ON(ret);
+ metadata = 2;
+
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, metadata);
+ if (ret)
+ return ret;
if (!(rw & REQ_WRITE)) {
if (bio_flags & EXTENT_BIO_COMPRESSED) {
page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;
lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0,
- &cached_state, GFP_NOFS);
+ &cached_state);
/* already ordered? We're done */
if (PagePrivate2(page))
*/
ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
&hint, 0);
- BUG_ON(ret);
+ if (ret)
+ goto out;
ins.objectid = btrfs_ino(inode);
ins.offset = file_pos;
ins.type = BTRFS_EXTENT_DATA_KEY;
ret = btrfs_insert_empty_item(trans, root, path, &ins, sizeof(*fi));
- BUG_ON(ret);
+ if (ret)
+ goto out;
leaf = path->nodes[0];
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
ret = btrfs_alloc_reserved_file_extent(trans, root,
root->root_key.objectid,
btrfs_ino(inode), file_pos, &ins);
- BUG_ON(ret);
+ out:
btrfs_free_path(path);
- return 0;
+ return ret;
}
/*
end - start + 1);
if (!ret)
return 0;
- BUG_ON(!ordered_extent);
+ BUG_ON(!ordered_extent); /* Logic error */
nolock = btrfs_is_free_space_inode(root, inode);
if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
- BUG_ON(!list_empty(&ordered_extent->list));
+ BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */
ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
if (!ret) {
if (nolock)
trans = btrfs_join_transaction_nolock(root);
else
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
ret = btrfs_update_inode_fallback(trans, root, inode);
- BUG_ON(ret);
+ if (ret) /* -ENOMEM or corruption */
+ btrfs_abort_transaction(trans, root, ret);
}
goto out;
}
lock_extent_bits(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset + ordered_extent->len - 1,
- 0, &cached_state, GFP_NOFS);
+ 0, &cached_state);
if (nolock)
trans = btrfs_join_transaction_nolock(root);
else
trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ trans = NULL;
+ goto out_unlock;
+ }
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
ordered_extent->file_offset,
ordered_extent->file_offset +
ordered_extent->len);
- BUG_ON(ret);
} else {
BUG_ON(root == root->fs_info->tree_root);
ret = insert_reserved_file_extent(trans, inode,
unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
ordered_extent->file_offset,
ordered_extent->len);
- BUG_ON(ret);
}
unlock_extent_cached(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset +
ordered_extent->len - 1, &cached_state, GFP_NOFS);
+ if (ret < 0) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
add_pending_csums(trans, inode, ordered_extent->file_offset,
&ordered_extent->list);
ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
ret = btrfs_update_inode_fallback(trans, root, inode);
- BUG_ON(ret);
+ if (ret) { /* -ENOMEM or corruption */
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
}
ret = 0;
out:
btrfs_put_ordered_extent(ordered_extent);
return 0;
+ out_unlock:
+ unlock_extent_cached(io_tree, ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len - 1, &cached_state, GFP_NOFS);
+ goto out;
}
static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
} else {
ret = get_state_private(io_tree, start, &private);
}
- kaddr = kmap_atomic(page, KM_USER0);
+ kaddr = kmap_atomic(page);
if (ret)
goto zeroit;
if (csum != private)
goto zeroit;
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
good:
return 0;
(unsigned long long)private);
memset(kaddr + offset, 1, end - start + 1);
flush_dcache_page(page);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
if (private == 0)
return 0;
return -EIO;
struct inode *inode;
};
+ /* JDM: If this is fs-wide, why can't we add a pointer to
+ * btrfs_inode instead and avoid the allocation? */
void btrfs_add_delayed_iput(struct inode *inode)
{
struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
/* grab metadata reservation from transaction handle */
if (reserve) {
ret = btrfs_orphan_reserve_metadata(trans, inode);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOSPC in reservation; Logic error? JDM */
}
/* insert an orphan item to track this unlinked/truncated file */
if (insert >= 1) {
ret = btrfs_insert_orphan_item(trans, root, btrfs_ino(inode));
- BUG_ON(ret && ret != -EEXIST);
+ if (ret && ret != -EEXIST) {
+ btrfs_abort_transaction(trans, root, ret);
+ return ret;
+ }
+ ret = 0;
}
/* insert an orphan item to track subvolume contains orphan files */
if (insert >= 2) {
ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
root->root_key.objectid);
- BUG_ON(ret);
+ if (ret && ret != -EEXIST) {
+ btrfs_abort_transaction(trans, root, ret);
+ return ret;
+ }
}
return 0;
}
if (trans && delete_item) {
ret = btrfs_del_orphan_item(trans, root, btrfs_ino(inode));
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */
}
if (release_rsv)
}
ret = btrfs_del_orphan_item(trans, root,
found_key.objectid);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */
btrfs_end_transaction(trans, root);
continue;
}
printk(KERN_INFO "btrfs failed to delete reference to %.*s, "
"inode %llu parent %llu\n", name_len, name,
(unsigned long long)ino, (unsigned long long)dir_ino);
+ btrfs_abort_transaction(trans, root, ret);
goto err;
}
ret = btrfs_delete_delayed_dir_index(trans, root, dir, index);
- if (ret)
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
goto err;
+ }
ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len,
inode, dir_ino);
- BUG_ON(ret != 0 && ret != -ENOENT);
+ if (ret != 0 && ret != -ENOENT) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto err;
+ }
ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len,
dir, index);
err = ret;
goto out;
}
- BUG_ON(ret == 0);
+ BUG_ON(ret == 0); /* Corruption */
if (check_path_shared(root, path))
goto out;
btrfs_release_path(path);
err = PTR_ERR(ref);
goto out;
}
- BUG_ON(!ref);
+ BUG_ON(!ref); /* Logic error */
if (check_path_shared(root, path))
goto out;
index = btrfs_inode_ref_index(path->nodes[0], ref);
di = btrfs_lookup_dir_item(trans, root, path, dir_ino,
name, name_len, -1);
- BUG_ON(IS_ERR_OR_NULL(di));
+ if (IS_ERR_OR_NULL(di)) {
+ if (!di)
+ ret = -ENOENT;
+ else
+ ret = PTR_ERR(di);
+ goto out;
+ }
leaf = path->nodes[0];
btrfs_dir_item_key_to_cpu(leaf, di, &key);
WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
ret = btrfs_delete_one_dir_name(trans, root, path, di);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
btrfs_release_path(path);
ret = btrfs_del_root_ref(trans, root->fs_info->tree_root,
objectid, root->root_key.objectid,
dir_ino, &index, name, name_len);
if (ret < 0) {
- BUG_ON(ret != -ENOENT);
+ if (ret != -ENOENT) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
di = btrfs_search_dir_index_item(root, path, dir_ino,
name, name_len);
- BUG_ON(IS_ERR_OR_NULL(di));
+ if (IS_ERR_OR_NULL(di)) {
+ if (!di)
+ ret = -ENOENT;
+ else
+ ret = PTR_ERR(di);
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
btrfs_release_path(path);
ret = btrfs_delete_delayed_dir_index(trans, root, dir, index);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out;
+ }
btrfs_i_size_write(dir, dir->i_size - name_len * 2);
dir->i_mtime = dir->i_ctime = CURRENT_TIME;
ret = btrfs_update_inode(trans, root, dir);
- BUG_ON(ret);
-
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
+ out:
btrfs_free_path(path);
- return 0;
+ return ret;
}
static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
}
size =
btrfs_file_extent_calc_inline_size(size);
- ret = btrfs_truncate_item(trans, root, path,
- size, 1);
+ btrfs_truncate_item(trans, root, path,
+ size, 1);
} else if (root->ref_cows) {
inode_sub_bytes(inode, item_end + 1 -
found_key.offset);
ret = btrfs_del_items(trans, root, path,
pending_del_slot,
pending_del_nr);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans,
+ root, ret);
+ goto error;
+ }
pending_del_nr = 0;
}
btrfs_release_path(path);
if (pending_del_nr) {
ret = btrfs_del_items(trans, root, path, pending_del_slot,
pending_del_nr);
- BUG_ON(ret);
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
}
+ error:
btrfs_free_path(path);
return err;
}
}
wait_on_page_writeback(page);
- lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state,
- GFP_NOFS);
+ lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
set_page_extent_mapped(page);
ordered = btrfs_lookup_ordered_extent(inode, page_start);
btrfs_wait_ordered_range(inode, hole_start,
block_end - hole_start);
lock_extent_bits(io_tree, hole_start, block_end - 1, 0,
- &cached_state, GFP_NOFS);
+ &cached_state);
ordered = btrfs_lookup_ordered_extent(inode, hole_start);
if (!ordered)
break;
while (1) {
em = btrfs_get_extent(inode, NULL, 0, cur_offset,
block_end - cur_offset, 0);
- BUG_ON(IS_ERR_OR_NULL(em));
+ if (IS_ERR(em)) {
+ err = PTR_ERR(em);
+ break;
+ }
last_byte = min(extent_map_end(em), block_end);
last_byte = (last_byte + mask) & ~mask;
if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
cur_offset + hole_size,
&hint_byte, 1);
if (err) {
- btrfs_update_inode(trans, root, inode);
+ btrfs_abort_transaction(trans, root, err);
btrfs_end_transaction(trans, root);
break;
}
0, hole_size, 0, hole_size,
0, 0, 0);
if (err) {
- btrfs_update_inode(trans, root, inode);
+ btrfs_abort_transaction(trans, root, err);
btrfs_end_transaction(trans, root);
break;
}
}
}
- int btrfs_invalidate_inodes(struct btrfs_root *root)
+ void btrfs_invalidate_inodes(struct btrfs_root *root)
{
struct rb_node *node;
struct rb_node *prev;
node = rb_next(node);
}
spin_unlock(&root->inode_lock);
- return 0;
}
static int btrfs_init_locked_inode(struct inode *inode, void *p)
parent_ino, index);
}
- if (ret == 0) {
- ret = btrfs_insert_dir_item(trans, root, name, name_len,
- parent_inode, &key,
- btrfs_inode_type(inode), index);
- if (ret)
- goto fail_dir_item;
+ /* Nothing to clean up yet */
+ if (ret)
+ return ret;
- btrfs_i_size_write(parent_inode, parent_inode->i_size +
- name_len * 2);
- parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
- ret = btrfs_update_inode(trans, root, parent_inode);
+ ret = btrfs_insert_dir_item(trans, root, name, name_len,
+ parent_inode, &key,
+ btrfs_inode_type(inode), index);
+ if (ret == -EEXIST)
+ goto fail_dir_item;
+ else if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ return ret;
}
+
+ btrfs_i_size_write(parent_inode, parent_inode->i_size +
+ name_len * 2);
+ parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
+ ret = btrfs_update_inode(trans, root, parent_inode);
+ if (ret)
+ btrfs_abort_transaction(trans, root, ret);
return ret;
fail_dir_item:
} else {
struct dentry *parent = dentry->d_parent;
err = btrfs_update_inode(trans, root, inode);
- BUG_ON(err);
+ if (err)
+ goto fail;
d_instantiate(dentry, inode);
btrfs_log_new_name(trans, inode, NULL, parent);
}
ret = btrfs_decompress(compress_type, tmp, page,
extent_offset, inline_size, max_size);
if (ret) {
- char *kaddr = kmap_atomic(page, KM_USER0);
+ char *kaddr = kmap_atomic(page);
unsigned long copy_size = min_t(u64,
PAGE_CACHE_SIZE - pg_offset,
max_size - extent_offset);
memset(kaddr + pg_offset, 0, copy_size);
- kunmap_atomic(kaddr, KM_USER0);
+ kunmap_atomic(kaddr);
}
kfree(tmp);
return 0;
ret = uncompress_inline(path, inode, page,
pg_offset,
extent_offset, item);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
} else {
map = kmap(page);
read_extent_buffer(leaf, map + pg_offset, ptr,
free_extent_map(em);
return ERR_PTR(err);
}
+ BUG_ON(!em); /* Error is always set */
return em;
}
alloc_hint = get_extent_allocation_hint(inode, start, len);
ret = btrfs_reserve_extent(trans, root, len, root->sectorsize, 0,
- alloc_hint, (u64)-1, &ins, 1);
+ alloc_hint, &ins, 1);
if (ret) {
em = ERR_PTR(ret);
goto out;
free_extent_map(em);
/* DIO will do one hole at a time, so just unlock a sector */
unlock_extent(&BTRFS_I(inode)->io_tree, start,
- start + root->sectorsize - 1, GFP_NOFS);
+ start + root->sectorsize - 1);
return 0;
}
unsigned long flags;
local_irq_save(flags);
- kaddr = kmap_atomic(page, KM_IRQ0);
+ kaddr = kmap_atomic(page);
csum = btrfs_csum_data(root, kaddr + bvec->bv_offset,
csum, bvec->bv_len);
btrfs_csum_final(csum, (char *)&csum);
- kunmap_atomic(kaddr, KM_IRQ0);
+ kunmap_atomic(kaddr);
local_irq_restore(flags);
flush_dcache_page(bvec->bv_page);
} while (bvec <= bvec_end);
unlock_extent(&BTRFS_I(inode)->io_tree, dip->logical_offset,
- dip->logical_offset + dip->bytes - 1, GFP_NOFS);
+ dip->logical_offset + dip->bytes - 1);
bio->bi_private = dip->private;
kfree(dip->csums);
lock_extent_bits(&BTRFS_I(inode)->io_tree, ordered->file_offset,
ordered->file_offset + ordered->len - 1, 0,
- &cached_state, GFP_NOFS);
+ &cached_state);
if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) {
ret = btrfs_mark_extent_written(trans, inode,
int ret;
struct btrfs_root *root = BTRFS_I(inode)->root;
ret = btrfs_csum_one_bio(root, inode, bio, offset, 1);
- BUG_ON(ret);
+ BUG_ON(ret); /* -ENOMEM */
return 0;
}
while (1) {
lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
- 0, &cached_state, GFP_NOFS);
+ 0, &cached_state);
/*
* We're concerned with the entire range that we're going to be
* doing DIO to, so we need to make sure theres no ordered
if (writing) {
write_bits = EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING;
ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
- EXTENT_DELALLOC, 0, NULL, &cached_state,
+ EXTENT_DELALLOC, NULL, &cached_state,
GFP_NOFS);
if (ret) {
clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
btrfs_releasepage(page, GFP_NOFS);
return;
}
- lock_extent_bits(tree, page_start, page_end, 0, &cached_state,
- GFP_NOFS);
+ lock_extent_bits(tree, page_start, page_end, 0, &cached_state);
ordered = btrfs_lookup_ordered_extent(page->mapping->host,
page_offset(page));
if (ordered) {
}
btrfs_put_ordered_extent(ordered);
cached_state = NULL;
- lock_extent_bits(tree, page_start, page_end, 0, &cached_state,
- GFP_NOFS);
+ lock_extent_bits(tree, page_start, page_end, 0, &cached_state);
}
clear_extent_bit(tree, page_start, page_end,
EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
}
wait_on_page_writeback(page);
- lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state,
- GFP_NOFS);
+ lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
set_page_extent_mapped(page);
/*
btrfs_i_size_write(inode, 0);
err = btrfs_update_inode(trans, new_root, inode);
- BUG_ON(err);
iput(inode);
- return 0;
+ return err;
}
struct inode *btrfs_alloc_inode(struct super_block *sb)
extent_map_tree_init(&ei->extent_tree);
extent_io_tree_init(&ei->io_tree, &inode->i_data);
extent_io_tree_init(&ei->io_failure_tree, &inode->i_data);
+ ei->io_tree.track_uptodate = 1;
+ ei->io_failure_tree.track_uptodate = 1;
mutex_init(&ei->log_mutex);
mutex_init(&ei->delalloc_mutex);
btrfs_ordered_inode_tree_init(&ei->ordered_tree);
if (!ret)
ret = btrfs_update_inode(trans, root, old_inode);
}
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_fail;
+ }
if (new_inode) {
new_inode->i_ctime = CURRENT_TIME;
new_dentry->d_name.name,
new_dentry->d_name.len);
}
- BUG_ON(ret);
- if (new_inode->i_nlink == 0) {
+ if (!ret && new_inode->i_nlink == 0) {
ret = btrfs_orphan_add(trans, new_dentry->d_inode);
BUG_ON(ret);
}
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_fail;
+ }
}
fixup_inode_flags(new_dir, old_inode);
ret = btrfs_add_link(trans, new_dir, old_inode,
new_dentry->d_name.name,
new_dentry->d_name.len, 0, index);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ goto out_fail;
+ }
if (old_ino != BTRFS_FIRST_FREE_OBJECTID) {
struct dentry *parent = new_dentry->d_parent;
}
ret = btrfs_reserve_extent(trans, root, num_bytes, min_size,
- 0, *alloc_hint, (u64)-1, &ins, 1);
+ 0, *alloc_hint, &ins, 1);
if (ret) {
if (own_trans)
btrfs_end_transaction(trans, root);
ins.offset, ins.offset,
ins.offset, 0, 0, 0,
BTRFS_FILE_EXTENT_PREALLOC);
- BUG_ON(ret);
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ if (own_trans)
+ btrfs_end_transaction(trans, root);
+ break;
+ }
btrfs_drop_extent_cache(inode, cur_offset,
cur_offset + ins.offset -1, 0);
}
ret = btrfs_update_inode(trans, root, inode);
- BUG_ON(ret);
+
+ if (ret) {
+ btrfs_abort_transaction(trans, root, ret);
+ if (own_trans)
+ btrfs_end_transaction(trans, root);
+ break;
+ }
if (own_trans)
btrfs_end_transaction(trans, root);
* Future enhancements:
* - In case an unrepairable extent is encountered, track which files are
* affected and report them
- * - In case of a read error on files with nodatasum, map the file and read
- * the extent to trigger a writeback of the good copy
* - track and record media errors, throw out bad devices
* - add a mode to also read unallocated space
*/
- struct scrub_bio;
- struct scrub_page;
+ struct scrub_block;
struct scrub_dev;
- static void scrub_bio_end_io(struct bio *bio, int err);
- static void scrub_checksum(struct btrfs_work *work);
- static int scrub_checksum_data(struct scrub_dev *sdev,
- struct scrub_page *spag, void *buffer);
- static int scrub_checksum_tree_block(struct scrub_dev *sdev,
- struct scrub_page *spag, u64 logical,
- void *buffer);
- static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer);
- static int scrub_fixup_check(struct scrub_bio *sbio, int ix);
- static void scrub_fixup_end_io(struct bio *bio, int err);
- static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
- struct page *page);
- static void scrub_fixup(struct scrub_bio *sbio, int ix);
#define SCRUB_PAGES_PER_BIO 16 /* 64k per bio */
#define SCRUB_BIOS_PER_DEV 16 /* 1 MB per device in flight */
+ #define SCRUB_MAX_PAGES_PER_BLOCK 16 /* 64k per node/leaf/sector */
struct scrub_page {
+ struct scrub_block *sblock;
+ struct page *page;
+ struct block_device *bdev;
u64 flags; /* extent flags */
u64 generation;
- int mirror_num;
- int have_csum;
+ u64 logical;
+ u64 physical;
+ struct {
+ unsigned int mirror_num:8;
+ unsigned int have_csum:1;
+ unsigned int io_error:1;
+ };
u8 csum[BTRFS_CSUM_SIZE];
};
int err;
u64 logical;
u64 physical;
- struct scrub_page spag[SCRUB_PAGES_PER_BIO];
- u64 count;
+ struct scrub_page *pagev[SCRUB_PAGES_PER_BIO];
+ int page_count;
int next_free;
struct btrfs_work work;
};
+ struct scrub_block {
+ struct scrub_page pagev[SCRUB_MAX_PAGES_PER_BLOCK];
+ int page_count;
+ atomic_t outstanding_pages;
+ atomic_t ref_count; /* free mem on transition to zero */
+ struct scrub_dev *sdev;
+ struct {
+ unsigned int header_error:1;
+ unsigned int checksum_error:1;
+ unsigned int no_io_error_seen:1;
+ };
+ };
+
struct scrub_dev {
struct scrub_bio *bios[SCRUB_BIOS_PER_DEV];
struct btrfs_device *dev;
struct list_head csum_list;
atomic_t cancel_req;
int readonly;
+ int pages_per_bio; /* <= SCRUB_PAGES_PER_BIO */
+ u32 sectorsize;
+ u32 nodesize;
+ u32 leafsize;
/*
* statistics
*/
int scratch_bufsize;
};
+
+ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check);
+ static int scrub_setup_recheck_block(struct scrub_dev *sdev,
+ struct btrfs_mapping_tree *map_tree,
+ u64 length, u64 logical,
+ struct scrub_block *sblock);
+ static int scrub_recheck_block(struct btrfs_fs_info *fs_info,
+ struct scrub_block *sblock, int is_metadata,
+ int have_csum, u8 *csum, u64 generation,
+ u16 csum_size);
+ static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info,
+ struct scrub_block *sblock,
+ int is_metadata, int have_csum,
+ const u8 *csum, u64 generation,
+ u16 csum_size);
+ static void scrub_complete_bio_end_io(struct bio *bio, int err);
+ static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
+ struct scrub_block *sblock_good,
+ int force_write);
+ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
+ struct scrub_block *sblock_good,
+ int page_num, int force_write);
+ static int scrub_checksum_data(struct scrub_block *sblock);
+ static int scrub_checksum_tree_block(struct scrub_block *sblock);
+ static int scrub_checksum_super(struct scrub_block *sblock);
+ static void scrub_block_get(struct scrub_block *sblock);
+ static void scrub_block_put(struct scrub_block *sblock);
+ static int scrub_add_page_to_bio(struct scrub_dev *sdev,
+ struct scrub_page *spage);
+ static int scrub_pages(struct scrub_dev *sdev, u64 logical, u64 len,
+ u64 physical, u64 flags, u64 gen, int mirror_num,
+ u8 *csum, int force);
+ static void scrub_bio_end_io(struct bio *bio, int err);
+ static void scrub_bio_end_io_worker(struct btrfs_work *work);
+ static void scrub_block_complete(struct scrub_block *sblock);
+
+
static void scrub_free_csums(struct scrub_dev *sdev)
{
while (!list_empty(&sdev->csum_list)) {
}
}
- static void scrub_free_bio(struct bio *bio)
- {
- int i;
- struct page *last_page = NULL;
-
- if (!bio)
- return;
-
- for (i = 0; i < bio->bi_vcnt; ++i) {
- if (bio->bi_io_vec[i].bv_page == last_page)
- continue;
- last_page = bio->bi_io_vec[i].bv_page;
- __free_page(last_page);
- }
- bio_put(bio);
- }
-
static noinline_for_stack void scrub_free_dev(struct scrub_dev *sdev)
{
int i;
if (!sdev)
return;
+ /* this can happen when scrub is cancelled */
+ if (sdev->curr != -1) {
+ struct scrub_bio *sbio = sdev->bios[sdev->curr];
+
+ for (i = 0; i < sbio->page_count; i++) {
+ BUG_ON(!sbio->pagev[i]);
+ BUG_ON(!sbio->pagev[i]->page);
+ scrub_block_put(sbio->pagev[i]->sblock);
+ }
+ bio_put(sbio->bio);
+ }
+
for (i = 0; i < SCRUB_BIOS_PER_DEV; ++i) {
struct scrub_bio *sbio = sdev->bios[i];
if (!sbio)
break;
-
- scrub_free_bio(sbio->bio);
kfree(sbio);
}
struct scrub_dev *sdev;
int i;
struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
+ int pages_per_bio;
+ pages_per_bio = min_t(int, SCRUB_PAGES_PER_BIO,
+ bio_get_nr_vecs(dev->bdev));
sdev = kzalloc(sizeof(*sdev), GFP_NOFS);
if (!sdev)
goto nomem;
sdev->dev = dev;
+ sdev->pages_per_bio = pages_per_bio;
+ sdev->curr = -1;
for (i = 0; i < SCRUB_BIOS_PER_DEV; ++i) {
struct scrub_bio *sbio;
sbio->index = i;
sbio->sdev = sdev;
- sbio->count = 0;
- sbio->work.func = scrub_checksum;
+ sbio->page_count = 0;
+ sbio->work.func = scrub_bio_end_io_worker;
if (i != SCRUB_BIOS_PER_DEV-1)
sdev->bios[i]->next_free = i + 1;
sdev->bios[i]->next_free = -1;
}
sdev->first_free = 0;
- sdev->curr = -1;
+ sdev->nodesize = dev->dev_root->nodesize;
+ sdev->leafsize = dev->dev_root->leafsize;
+ sdev->sectorsize = dev->dev_root->sectorsize;
atomic_set(&sdev->in_flight, 0);
atomic_set(&sdev->fixup_cnt, 0);
atomic_set(&sdev->cancel_req, 0);
return 0;
}
- static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
- int ix)
+ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
{
- struct btrfs_device *dev = sbio->sdev->dev;
+ struct btrfs_device *dev = sblock->sdev->dev;
struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
struct btrfs_path *path;
struct btrfs_key found_key;
swarn.scratch_buf = kmalloc(bufsize, GFP_NOFS);
swarn.msg_buf = kmalloc(bufsize, GFP_NOFS);
- swarn.sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
- swarn.logical = sbio->logical + ix * PAGE_SIZE;
+ BUG_ON(sblock->page_count < 1);
+ swarn.sector = (sblock->pagev[0].physical) >> 9;
+ swarn.logical = sblock->pagev[0].logical;
swarn.errstr = errstr;
swarn.dev = dev;
swarn.msg_bufsize = bufsize;
do {
ret = tree_backref_for_extent(&ptr, eb, ei, item_size,
&ref_root, &ref_level);
- printk(KERN_WARNING "%s at logical %llu on dev %s, "
+ printk(KERN_WARNING
+ "btrfs: %s at logical %llu on dev %s, "
"sector %llu: metadata %s (level %d) in tree "
"%llu\n", errstr, swarn.logical, dev->name,
(unsigned long long)swarn.sector,
} while (ret != 1);
} else {
swarn.path = path;
- iterate_extent_inodes(fs_info, path, found_key.objectid,
- extent_item_pos,
+ iterate_extent_inodes(fs_info, found_key.objectid,
+ extent_item_pos, 1,
scrub_print_warning_inode, &swarn);
}
spin_lock(&sdev->stat_lock);
++sdev->stat.uncorrectable_errors;
spin_unlock(&sdev->stat_lock);
- printk_ratelimited(KERN_ERR "btrfs: unable to fixup "
- "(nodatasum) error at logical %llu\n",
- fixup->logical);
+ printk_ratelimited(KERN_ERR
+ "btrfs: unable to fixup (nodatasum) error at logical %llu on dev %s\n",
+ (unsigned long long)fixup->logical, sdev->dev->name);
}
btrfs_free_path(path);
}
/*
- * scrub_recheck_error gets called when either verification of the page
- * failed or the bio failed to read, e.g. with EIO. In the latter case,
- * recheck_error gets called for every page in the bio, even though only
- * one may be bad
+ * scrub_handle_errored_block gets called when either verification of the
+ * pages failed or the bio failed to read, e.g. with EIO. In the latter
+ * case, this function handles all pages in the bio, even though only one
+ * may be bad.
+ * The goal of this function is to repair the errored block by using the
+ * contents of one of the mirrors.
*/
- static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
+ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
{
- struct scrub_dev *sdev = sbio->sdev;
- u64 sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+ struct scrub_dev *sdev = sblock_to_check->sdev;
+ struct btrfs_fs_info *fs_info;
+ u64 length;
+ u64 logical;
+ u64 generation;
+ unsigned int failed_mirror_index;
+ unsigned int is_metadata;
+ unsigned int have_csum;
+ u8 *csum;
+ struct scrub_block *sblocks_for_recheck; /* holds one for each mirror */
+ struct scrub_block *sblock_bad;
+ int ret;
+ int mirror_index;
+ int page_num;
+ int success;
static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
- DEFAULT_RATELIMIT_BURST);
+ DEFAULT_RATELIMIT_BURST);
+
+ BUG_ON(sblock_to_check->page_count < 1);
+ fs_info = sdev->dev->dev_root->fs_info;
+ length = sblock_to_check->page_count * PAGE_SIZE;
+ logical = sblock_to_check->pagev[0].logical;
+ generation = sblock_to_check->pagev[0].generation;
+ BUG_ON(sblock_to_check->pagev[0].mirror_num < 1);
+ failed_mirror_index = sblock_to_check->pagev[0].mirror_num - 1;
+ is_metadata = !(sblock_to_check->pagev[0].flags &
+ BTRFS_EXTENT_FLAG_DATA);
+ have_csum = sblock_to_check->pagev[0].have_csum;
+ csum = sblock_to_check->pagev[0].csum;
- if (sbio->err) {
- if (scrub_fixup_io(READ, sbio->sdev->dev->bdev, sector,
- sbio->bio->bi_io_vec[ix].bv_page) == 0) {
- if (scrub_fixup_check(sbio, ix) == 0)
- return 0;
- }
- if (__ratelimit(&_rs))
- scrub_print_warning("i/o error", sbio, ix);
- } else {
- if (__ratelimit(&_rs))
- scrub_print_warning("checksum error", sbio, ix);
+ /*
+ * read all mirrors one after the other. This includes to
+ * re-read the extent or metadata block that failed (that was
+ * the cause that this fixup code is called) another time,
+ * page by page this time in order to know which pages
+ * caused I/O errors and which ones are good (for all mirrors).
+ * It is the goal to handle the situation when more than one
+ * mirror contains I/O errors, but the errors do not
+ * overlap, i.e. the data can be repaired by selecting the
+ * pages from those mirrors without I/O error on the
+ * particular pages. One example (with blocks >= 2 * PAGE_SIZE)
+ * would be that mirror #1 has an I/O error on the first page,
+ * the second page is good, and mirror #2 has an I/O error on
+ * the second page, but the first page is good.
+ * Then the first page of the first mirror can be repaired by
+ * taking the first page of the second mirror, and the
+ * second page of the second mirror can be repaired by
+ * copying the contents of the 2nd page of the 1st mirror.
+ * One more note: if the pages of one mirror contain I/O
+ * errors, the checksum cannot be verified. In order to get
+ * the best data for repairing, the first attempt is to find
+ * a mirror without I/O errors and with a validated checksum.
+ * Only if this is not possible, the pages are picked from
+ * mirrors with I/O errors without considering the checksum.
+ * If the latter is the case, at the end, the checksum of the
+ * repaired area is verified in order to correctly maintain
+ * the statistics.
+ */
+
+ sblocks_for_recheck = kzalloc(BTRFS_MAX_MIRRORS *
+ sizeof(*sblocks_for_recheck),
+ GFP_NOFS);
+ if (!sblocks_for_recheck) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.malloc_errors++;
+ sdev->stat.read_errors++;
+ sdev->stat.uncorrectable_errors++;
+ spin_unlock(&sdev->stat_lock);
+ goto out;
}
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.read_errors;
- spin_unlock(&sdev->stat_lock);
+ /* setup the context, map the logical blocks and alloc the pages */
+ ret = scrub_setup_recheck_block(sdev, &fs_info->mapping_tree, length,
+ logical, sblocks_for_recheck);
+ if (ret) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.read_errors++;
+ sdev->stat.uncorrectable_errors++;
+ spin_unlock(&sdev->stat_lock);
+ goto out;
+ }
+ BUG_ON(failed_mirror_index >= BTRFS_MAX_MIRRORS);
+ sblock_bad = sblocks_for_recheck + failed_mirror_index;
- scrub_fixup(sbio, ix);
- return 1;
- }
+ /* build and submit the bios for the failed mirror, check checksums */
+ ret = scrub_recheck_block(fs_info, sblock_bad, is_metadata, have_csum,
+ csum, generation, sdev->csum_size);
+ if (ret) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.read_errors++;
+ sdev->stat.uncorrectable_errors++;
+ spin_unlock(&sdev->stat_lock);
+ goto out;
+ }
- static int scrub_fixup_check(struct scrub_bio *sbio, int ix)
- {
- int ret = 1;
- struct page *page;
- void *buffer;
- u64 flags = sbio->spag[ix].flags;
+ if (!sblock_bad->header_error && !sblock_bad->checksum_error &&
+ sblock_bad->no_io_error_seen) {
+ /*
+ * the error disappeared after reading page by page, or
+ * the area was part of a huge bio and other parts of the
+ * bio caused I/O errors, or the block layer merged several
+ * read requests into one and the error is caused by a
+ * different bio (usually one of the two latter cases is
+ * the cause)
+ */
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.unverified_errors++;
+ spin_unlock(&sdev->stat_lock);
- page = sbio->bio->bi_io_vec[ix].bv_page;
- buffer = kmap_atomic(page);
- if (flags & BTRFS_EXTENT_FLAG_DATA) {
- ret = scrub_checksum_data(sbio->sdev,
- sbio->spag + ix, buffer);
- } else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
- ret = scrub_checksum_tree_block(sbio->sdev,
- sbio->spag + ix,
- sbio->logical + ix * PAGE_SIZE,
- buffer);
- } else {
- WARN_ON(1);
+ goto out;
}
- kunmap_atomic(buffer);
- return ret;
- }
+ if (!sblock_bad->no_io_error_seen) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.read_errors++;
+ spin_unlock(&sdev->stat_lock);
+ if (__ratelimit(&_rs))
+ scrub_print_warning("i/o error", sblock_to_check);
+ } else if (sblock_bad->checksum_error) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.csum_errors++;
+ spin_unlock(&sdev->stat_lock);
+ if (__ratelimit(&_rs))
+ scrub_print_warning("checksum error", sblock_to_check);
+ } else if (sblock_bad->header_error) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.verify_errors++;
+ spin_unlock(&sdev->stat_lock);
+ if (__ratelimit(&_rs))
+ scrub_print_warning("checksum/header error",
+ sblock_to_check);
+ }
- static void scrub_fixup_end_io(struct bio *bio, int err)
- {
- complete((struct completion *)bio->bi_private);
- }
+ if (sdev->readonly)
+ goto did_not_correct_error;
- static void scrub_fixup(struct scrub_bio *sbio, int ix)
- {
- struct scrub_dev *sdev = sbio->sdev;
- struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
- struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
- struct btrfs_bio *bbio = NULL;
- struct scrub_fixup_nodatasum *fixup;
- u64 logical = sbio->logical + ix * PAGE_SIZE;
- u64 length;
- int i;
- int ret;
- DECLARE_COMPLETION_ONSTACK(complete);
-
- if ((sbio->spag[ix].flags & BTRFS_EXTENT_FLAG_DATA) &&
- (sbio->spag[ix].have_csum == 0)) {
- fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
- if (!fixup)
- goto uncorrectable;
- fixup->sdev = sdev;
- fixup->logical = logical;
- fixup->root = fs_info->extent_root;
- fixup->mirror_num = sbio->spag[ix].mirror_num;
+ if (!is_metadata && !have_csum) {
+ struct scrub_fixup_nodatasum *fixup_nodatasum;
+
+ /*
+ * !is_metadata and !have_csum, this means that the data
+ * might not be COW'ed, that it might be modified
+ * concurrently. The general strategy to work on the
+ * commit root does not help in the case when COW is not
+ * used.
+ */
+ fixup_nodatasum = kzalloc(sizeof(*fixup_nodatasum), GFP_NOFS);
+ if (!fixup_nodatasum)
+ goto did_not_correct_error;
+ fixup_nodatasum->sdev = sdev;
+ fixup_nodatasum->logical = logical;
+ fixup_nodatasum->root = fs_info->extent_root;
+ fixup_nodatasum->mirror_num = failed_mirror_index + 1;
/*
* increment scrubs_running to prevent cancel requests from
* completing as long as a fixup worker is running. we must also
atomic_inc(&fs_info->scrubs_paused);
mutex_unlock(&fs_info->scrub_lock);
atomic_inc(&sdev->fixup_cnt);
- fixup->work.func = scrub_fixup_nodatasum;
- btrfs_queue_worker(&fs_info->scrub_workers, &fixup->work);
- return;
+ fixup_nodatasum->work.func = scrub_fixup_nodatasum;
+ btrfs_queue_worker(&fs_info->scrub_workers,
+ &fixup_nodatasum->work);
+ goto out;
}
- length = PAGE_SIZE;
- ret = btrfs_map_block(map_tree, REQ_WRITE, logical, &length,
- &bbio, 0);
- if (ret || !bbio || length < PAGE_SIZE) {
- printk(KERN_ERR
- "scrub_fixup: btrfs_map_block failed us for %llu\n",
- (unsigned long long)logical);
- WARN_ON(1);
- kfree(bbio);
- return;
+ /*
+ * now build and submit the bios for the other mirrors, check
+ * checksums
+ */
+ for (mirror_index = 0;
+ mirror_index < BTRFS_MAX_MIRRORS &&
+ sblocks_for_recheck[mirror_index].page_count > 0;
+ mirror_index++) {
+ if (mirror_index == failed_mirror_index)
+ continue;
+
+ /* build and submit the bios, check checksums */
+ ret = scrub_recheck_block(fs_info,
+ sblocks_for_recheck + mirror_index,
+ is_metadata, have_csum, csum,
+ generation, sdev->csum_size);
+ if (ret)
+ goto did_not_correct_error;
}
- if (bbio->num_stripes == 1)
- /* there aren't any replicas */
- goto uncorrectable;
+ /*
+ * first try to pick the mirror which is completely without I/O
+ * errors and also does not have a checksum error.
+ * If one is found, and if a checksum is present, the full block
+ * that is known to contain an error is rewritten. Afterwards
+ * the block is known to be corrected.
+ * If a mirror is found which is completely correct, and no
+ * checksum is present, only those pages are rewritten that had
+ * an I/O error in the block to be repaired, since it cannot be
+ * determined, which copy of the other pages is better (and it
+ * could happen otherwise that a correct page would be
+ * overwritten by a bad one).
+ */
+ for (mirror_index = 0;
+ mirror_index < BTRFS_MAX_MIRRORS &&
+ sblocks_for_recheck[mirror_index].page_count > 0;
+ mirror_index++) {
+ struct scrub_block *sblock_other = sblocks_for_recheck +
+ mirror_index;
+
+ if (!sblock_other->header_error &&
+ !sblock_other->checksum_error &&
+ sblock_other->no_io_error_seen) {
+ int force_write = is_metadata || have_csum;
+
+ ret = scrub_repair_block_from_good_copy(sblock_bad,
+ sblock_other,
+ force_write);
+ if (0 == ret)
+ goto corrected_error;
+ }
+ }
/*
- * first find a good copy
+ * in case of I/O errors in the area that is supposed to be
+ * repaired, continue by picking good copies of those pages.
+ * Select the good pages from mirrors to rewrite bad pages from
+ * the area to fix. Afterwards verify the checksum of the block
+ * that is supposed to be repaired. This verification step is
+ * only done for the purpose of statistic counting and for the
+ * final scrub report, whether errors remain.
+ * A perfect algorithm could make use of the checksum and try
+ * all possible combinations of pages from the different mirrors
+ * until the checksum verification succeeds. For example, when
+ * the 2nd page of mirror #1 faces I/O errors, and the 2nd page
+ * of mirror #2 is readable but the final checksum test fails,
+ * then the 2nd page of mirror #3 could be tried, whether now
+ * the final checksum succeedes. But this would be a rare
+ * exception and is therefore not implemented. At least it is
+ * avoided that the good copy is overwritten.
+ * A more useful improvement would be to pick the sectors
+ * without I/O error based on sector sizes (512 bytes on legacy
+ * disks) instead of on PAGE_SIZE. Then maybe 512 byte of one
+ * mirror could be repaired by taking 512 byte of a different
+ * mirror, even if other 512 byte sectors in the same PAGE_SIZE
+ * area are unreadable.
*/
- for (i = 0; i < bbio->num_stripes; ++i) {
- if (i + 1 == sbio->spag[ix].mirror_num)
- continue;
- if (scrub_fixup_io(READ, bbio->stripes[i].dev->bdev,
- bbio->stripes[i].physical >> 9,
- sbio->bio->bi_io_vec[ix].bv_page)) {
- /* I/O-error, this is not a good copy */
+ /* can only fix I/O errors from here on */
+ if (sblock_bad->no_io_error_seen)
+ goto did_not_correct_error;
+
+ success = 1;
+ for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
+ struct scrub_page *page_bad = sblock_bad->pagev + page_num;
+
+ if (!page_bad->io_error)
continue;
+
+ for (mirror_index = 0;
+ mirror_index < BTRFS_MAX_MIRRORS &&
+ sblocks_for_recheck[mirror_index].page_count > 0;
+ mirror_index++) {
+ struct scrub_block *sblock_other = sblocks_for_recheck +
+ mirror_index;
+ struct scrub_page *page_other = sblock_other->pagev +
+ page_num;
+
+ if (!page_other->io_error) {
+ ret = scrub_repair_page_from_good_copy(
+ sblock_bad, sblock_other, page_num, 0);
+ if (0 == ret) {
+ page_bad->io_error = 0;
+ break; /* succeeded for this page */
+ }
+ }
}
- if (scrub_fixup_check(sbio, ix) == 0)
- break;
+ if (page_bad->io_error) {
+ /* did not find a mirror to copy the page from */
+ success = 0;
+ }
}
- if (i == bbio->num_stripes)
- goto uncorrectable;
- if (!sdev->readonly) {
- /*
- * bi_io_vec[ix].bv_page now contains good data, write it back
- */
- if (scrub_fixup_io(WRITE, sdev->dev->bdev,
- (sbio->physical + ix * PAGE_SIZE) >> 9,
- sbio->bio->bi_io_vec[ix].bv_page)) {
- /* I/O-error, writeback failed, give up */
- goto uncorrectable;
+ if (success) {
+ if (is_metadata || have_csum) {
+ /*
+ * need to verify the checksum now that all
+ * sectors on disk are repaired (the write
+ * request for data to be repaired is on its way).
+ * Just be lazy and use scrub_recheck_block()
+ * which re-reads the data before the checksum
+ * is verified, but most likely the data comes out
+ * of the page cache.
+ */
+ ret = scrub_recheck_block(fs_info, sblock_bad,
+ is_metadata, have_csum, csum,
+ generation, sdev->csum_size);
+ if (!ret && !sblock_bad->header_error &&
+ !sblock_bad->checksum_error &&
+ sblock_bad->no_io_error_seen)
+ goto corrected_error;
+ else
+ goto did_not_correct_error;
+ } else {
+ corrected_error:
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.corrected_errors++;
+ spin_unlock(&sdev->stat_lock);
+ printk_ratelimited(KERN_ERR
+ "btrfs: fixed up error at logical %llu on dev %s\n",
+ (unsigned long long)logical, sdev->dev->name);
}
+ } else {
+ did_not_correct_error:
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.uncorrectable_errors++;
+ spin_unlock(&sdev->stat_lock);
+ printk_ratelimited(KERN_ERR
+ "btrfs: unable to fixup (regular) error at logical %llu on dev %s\n",
+ (unsigned long long)logical, sdev->dev->name);
}
- kfree(bbio);
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.corrected_errors;
- spin_unlock(&sdev->stat_lock);
+ out:
+ if (sblocks_for_recheck) {
+ for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS;
+ mirror_index++) {
+ struct scrub_block *sblock = sblocks_for_recheck +
+ mirror_index;
+ int page_index;
+
+ for (page_index = 0; page_index < SCRUB_PAGES_PER_BIO;
+ page_index++)
+ if (sblock->pagev[page_index].page)
+ __free_page(
+ sblock->pagev[page_index].page);
+ }
+ kfree(sblocks_for_recheck);
+ }
- printk_ratelimited(KERN_ERR "btrfs: fixed up error at logical %llu\n",
- (unsigned long long)logical);
- return;
+ return 0;
+ }
- uncorrectable:
- kfree(bbio);
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.uncorrectable_errors;
- spin_unlock(&sdev->stat_lock);
+ static int scrub_setup_recheck_block(struct scrub_dev *sdev,
+ struct btrfs_mapping_tree *map_tree,
+ u64 length, u64 logical,
+ struct scrub_block *sblocks_for_recheck)
+ {
+ int page_index;
+ int mirror_index;
+ int ret;
+
+ /*
+ * note: the three members sdev, ref_count and outstanding_pages
+ * are not used (and not set) in the blocks that are used for
+ * the recheck procedure
+ */
+
+ page_index = 0;
+ while (length > 0) {
+ u64 sublen = min_t(u64, length, PAGE_SIZE);
+ u64 mapped_length = sublen;
+ struct btrfs_bio *bbio = NULL;
+
+ /*
+ * with a length of PAGE_SIZE, each returned stripe
+ * represents one mirror
+ */
+ ret = btrfs_map_block(map_tree, WRITE, logical, &mapped_length,
+ &bbio, 0);
+ if (ret || !bbio || mapped_length < sublen) {
+ kfree(bbio);
+ return -EIO;
+ }
- printk_ratelimited(KERN_ERR "btrfs: unable to fixup (regular) error at "
- "logical %llu\n", (unsigned long long)logical);
+ BUG_ON(page_index >= SCRUB_PAGES_PER_BIO);
+ for (mirror_index = 0; mirror_index < (int)bbio->num_stripes;
+ mirror_index++) {
+ struct scrub_block *sblock;
+ struct scrub_page *page;
+
+ if (mirror_index >= BTRFS_MAX_MIRRORS)
+ continue;
+
+ sblock = sblocks_for_recheck + mirror_index;
+ page = sblock->pagev + page_index;
+ page->logical = logical;
+ page->physical = bbio->stripes[mirror_index].physical;
+ page->bdev = bbio->stripes[mirror_index].dev->bdev;
+ page->mirror_num = mirror_index + 1;
+ page->page = alloc_page(GFP_NOFS);
+ if (!page->page) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.malloc_errors++;
+ spin_unlock(&sdev->stat_lock);
+ return -ENOMEM;
+ }
+ sblock->page_count++;
+ }
+ kfree(bbio);
+ length -= sublen;
+ logical += sublen;
+ page_index++;
+ }
+
+ return 0;
}
- static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
- struct page *page)
+ /*
+ * this function will check the on disk data for checksum errors, header
+ * errors and read I/O errors. If any I/O errors happen, the exact pages
+ * which are errored are marked as being bad. The goal is to enable scrub
+ * to take those pages that are not errored from all the mirrors so that
+ * the pages that are errored in the just handled mirror can be repaired.
+ */
+ static int scrub_recheck_block(struct btrfs_fs_info *fs_info,
+ struct scrub_block *sblock, int is_metadata,
+ int have_csum, u8 *csum, u64 generation,
+ u16 csum_size)
{
- struct bio *bio = NULL;
- int ret;
- DECLARE_COMPLETION_ONSTACK(complete);
+ int page_num;
+
+ sblock->no_io_error_seen = 1;
+ sblock->header_error = 0;
+ sblock->checksum_error = 0;
+
+ for (page_num = 0; page_num < sblock->page_count; page_num++) {
+ struct bio *bio;
+ int ret;
+ struct scrub_page *page = sblock->pagev + page_num;
+ DECLARE_COMPLETION_ONSTACK(complete);
+
+ BUG_ON(!page->page);
+ bio = bio_alloc(GFP_NOFS, 1);
+ bio->bi_bdev = page->bdev;
+ bio->bi_sector = page->physical >> 9;
+ bio->bi_end_io = scrub_complete_bio_end_io;
+ bio->bi_private = &complete;
+
+ ret = bio_add_page(bio, page->page, PAGE_SIZE, 0);
+ if (PAGE_SIZE != ret) {
+ bio_put(bio);
+ return -EIO;
+ }
+ btrfsic_submit_bio(READ, bio);
- bio = bio_alloc(GFP_NOFS, 1);
- bio->bi_bdev = bdev;
- bio->bi_sector = sector;
- bio_add_page(bio, page, PAGE_SIZE, 0);
- bio->bi_end_io = scrub_fixup_end_io;
- bio->bi_private = &complete;
- btrfsic_submit_bio(rw, bio);
+ /* this will also unplug the queue */
+ wait_for_completion(&complete);
- /* this will also unplug the queue */
- wait_for_completion(&complete);
+ page->io_error = !test_bit(BIO_UPTODATE, &bio->bi_flags);
+ if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
+ sblock->no_io_error_seen = 0;
+ bio_put(bio);
+ }
- ret = !test_bit(BIO_UPTODATE, &bio->bi_flags);
- bio_put(bio);
- return ret;
+ if (sblock->no_io_error_seen)
+ scrub_recheck_block_checksum(fs_info, sblock, is_metadata,
+ have_csum, csum, generation,
+ csum_size);
+
+ return 0;
}
- static void scrub_bio_end_io(struct bio *bio, int err)
+ static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info,
+ struct scrub_block *sblock,
+ int is_metadata, int have_csum,
+ const u8 *csum, u64 generation,
+ u16 csum_size)
{
- struct scrub_bio *sbio = bio->bi_private;
- struct scrub_dev *sdev = sbio->sdev;
- struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
+ int page_num;
+ u8 calculated_csum[BTRFS_CSUM_SIZE];
+ u32 crc = ~(u32)0;
+ struct btrfs_root *root = fs_info->extent_root;
+ void *mapped_buffer;
+
+ BUG_ON(!sblock->pagev[0].page);
+ if (is_metadata) {
+ struct btrfs_header *h;
+
- mapped_buffer = kmap_atomic(sblock->pagev[0].page, KM_USER0);
++ mapped_buffer = kmap_atomic(sblock->pagev[0].page);
+ h = (struct btrfs_header *)mapped_buffer;
+
+ if (sblock->pagev[0].logical != le64_to_cpu(h->bytenr) ||
+ generation != le64_to_cpu(h->generation) ||
+ memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE) ||
+ memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid,
+ BTRFS_UUID_SIZE))
+ sblock->header_error = 1;
+ csum = h->csum;
+ } else {
+ if (!have_csum)
+ return;
- sbio->err = err;
- sbio->bio = bio;
- mapped_buffer = kmap_atomic(sblock->pagev[0].page, KM_USER0);
++ mapped_buffer = kmap_atomic(sblock->pagev[0].page);
+ }
- btrfs_queue_worker(&fs_info->scrub_workers, &sbio->work);
+ for (page_num = 0;;) {
+ if (page_num == 0 && is_metadata)
+ crc = btrfs_csum_data(root,
+ ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE,
+ crc, PAGE_SIZE - BTRFS_CSUM_SIZE);
+ else
+ crc = btrfs_csum_data(root, mapped_buffer, crc,
+ PAGE_SIZE);
+
- kunmap_atomic(mapped_buffer, KM_USER0);
++ kunmap_atomic(mapped_buffer);
+ page_num++;
+ if (page_num >= sblock->page_count)
+ break;
+ BUG_ON(!sblock->pagev[page_num].page);
+
- mapped_buffer = kmap_atomic(sblock->pagev[page_num].page,
- KM_USER0);
++ mapped_buffer = kmap_atomic(sblock->pagev[page_num].page);
+ }
+
+ btrfs_csum_final(crc, calculated_csum);
+ if (memcmp(calculated_csum, csum, csum_size))
+ sblock->checksum_error = 1;
}
- static void scrub_checksum(struct btrfs_work *work)
+ static void scrub_complete_bio_end_io(struct bio *bio, int err)
{
- struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
- struct scrub_dev *sdev = sbio->sdev;
- struct page *page;
- void *buffer;
- int i;
- u64 flags;
- u64 logical;
- int ret;
+ complete((struct completion *)bio->bi_private);
+ }
- if (sbio->err) {
- ret = 0;
- for (i = 0; i < sbio->count; ++i)
- ret |= scrub_recheck_error(sbio, i);
- if (!ret) {
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.unverified_errors;
- spin_unlock(&sdev->stat_lock);
- }
+ static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
+ struct scrub_block *sblock_good,
+ int force_write)
+ {
+ int page_num;
+ int ret = 0;
- sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
- sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
- sbio->bio->bi_phys_segments = 0;
- sbio->bio->bi_idx = 0;
+ for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
+ int ret_sub;
- for (i = 0; i < sbio->count; i++) {
- struct bio_vec *bi;
- bi = &sbio->bio->bi_io_vec[i];
- bi->bv_offset = 0;
- bi->bv_len = PAGE_SIZE;
- }
- goto out;
+ ret_sub = scrub_repair_page_from_good_copy(sblock_bad,
+ sblock_good,
+ page_num,
+ force_write);
+ if (ret_sub)
+ ret = ret_sub;
}
- for (i = 0; i < sbio->count; ++i) {
- page = sbio->bio->bi_io_vec[i].bv_page;
- buffer = kmap_atomic(page);
- flags = sbio->spag[i].flags;
- logical = sbio->logical + i * PAGE_SIZE;
- ret = 0;
- if (flags & BTRFS_EXTENT_FLAG_DATA) {
- ret = scrub_checksum_data(sdev, sbio->spag + i, buffer);
- } else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
- ret = scrub_checksum_tree_block(sdev, sbio->spag + i,
- logical, buffer);
- } else if (flags & BTRFS_EXTENT_FLAG_SUPER) {
- BUG_ON(i);
- (void)scrub_checksum_super(sbio, buffer);
- } else {
- WARN_ON(1);
- }
- kunmap_atomic(buffer);
- if (ret) {
- ret = scrub_recheck_error(sbio, i);
- if (!ret) {
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.unverified_errors;
- spin_unlock(&sdev->stat_lock);
- }
+
+ return ret;
+ }
+
+ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
+ struct scrub_block *sblock_good,
+ int page_num, int force_write)
+ {
+ struct scrub_page *page_bad = sblock_bad->pagev + page_num;
+ struct scrub_page *page_good = sblock_good->pagev + page_num;
+
+ BUG_ON(sblock_bad->pagev[page_num].page == NULL);
+ BUG_ON(sblock_good->pagev[page_num].page == NULL);
+ if (force_write || sblock_bad->header_error ||
+ sblock_bad->checksum_error || page_bad->io_error) {
+ struct bio *bio;
+ int ret;
+ DECLARE_COMPLETION_ONSTACK(complete);
+
+ bio = bio_alloc(GFP_NOFS, 1);
+ bio->bi_bdev = page_bad->bdev;
+ bio->bi_sector = page_bad->physical >> 9;
+ bio->bi_end_io = scrub_complete_bio_end_io;
+ bio->bi_private = &complete;
+
+ ret = bio_add_page(bio, page_good->page, PAGE_SIZE, 0);
+ if (PAGE_SIZE != ret) {
+ bio_put(bio);
+ return -EIO;
}
+ btrfsic_submit_bio(WRITE, bio);
+
+ /* this will also unplug the queue */
+ wait_for_completion(&complete);
+ bio_put(bio);
}
- out:
- scrub_free_bio(sbio->bio);
- sbio->bio = NULL;
- spin_lock(&sdev->list_lock);
- sbio->next_free = sdev->first_free;
- sdev->first_free = sbio->index;
- spin_unlock(&sdev->list_lock);
- atomic_dec(&sdev->in_flight);
- wake_up(&sdev->list_wait);
+ return 0;
+ }
+
+ static void scrub_checksum(struct scrub_block *sblock)
+ {
+ u64 flags;
+ int ret;
+
+ BUG_ON(sblock->page_count < 1);
+ flags = sblock->pagev[0].flags;
+ ret = 0;
+ if (flags & BTRFS_EXTENT_FLAG_DATA)
+ ret = scrub_checksum_data(sblock);
+ else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+ ret = scrub_checksum_tree_block(sblock);
+ else if (flags & BTRFS_EXTENT_FLAG_SUPER)
+ (void)scrub_checksum_super(sblock);
+ else
+ WARN_ON(1);
+ if (ret)
+ scrub_handle_errored_block(sblock);
}
- static int scrub_checksum_data(struct scrub_dev *sdev,
- struct scrub_page *spag, void *buffer)
+ static int scrub_checksum_data(struct scrub_block *sblock)
{
+ struct scrub_dev *sdev = sblock->sdev;
u8 csum[BTRFS_CSUM_SIZE];
+ u8 *on_disk_csum;
+ struct page *page;
+ void *buffer;
u32 crc = ~(u32)0;
int fail = 0;
struct btrfs_root *root = sdev->dev->dev_root;
+ u64 len;
+ int index;
- if (!spag->have_csum)
+ BUG_ON(sblock->page_count < 1);
+ if (!sblock->pagev[0].have_csum)
return 0;
- crc = btrfs_csum_data(root, buffer, crc, PAGE_SIZE);
+ on_disk_csum = sblock->pagev[0].csum;
+ page = sblock->pagev[0].page;
- buffer = kmap_atomic(page, KM_USER0);
++ buffer = kmap_atomic(page);
+
+ len = sdev->sectorsize;
+ index = 0;
+ for (;;) {
+ u64 l = min_t(u64, len, PAGE_SIZE);
+
+ crc = btrfs_csum_data(root, buffer, crc, l);
- kunmap_atomic(buffer, KM_USER0);
++ kunmap_atomic(buffer);
+ len -= l;
+ if (len == 0)
+ break;
+ index++;
+ BUG_ON(index >= sblock->page_count);
+ BUG_ON(!sblock->pagev[index].page);
+ page = sblock->pagev[index].page;
- buffer = kmap_atomic(page, KM_USER0);
++ buffer = kmap_atomic(page);
+ }
+
btrfs_csum_final(crc, csum);
- if (memcmp(csum, spag->csum, sdev->csum_size))
+ if (memcmp(csum, on_disk_csum, sdev->csum_size))
fail = 1;
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.data_extents_scrubbed;
- sdev->stat.data_bytes_scrubbed += PAGE_SIZE;
- if (fail)
+ if (fail) {
+ spin_lock(&sdev->stat_lock);
++sdev->stat.csum_errors;
- spin_unlock(&sdev->stat_lock);
+ spin_unlock(&sdev->stat_lock);
+ }
return fail;
}
- static int scrub_checksum_tree_block(struct scrub_dev *sdev,
- struct scrub_page *spag, u64 logical,
- void *buffer)
+ static int scrub_checksum_tree_block(struct scrub_block *sblock)
{
+ struct scrub_dev *sdev = sblock->sdev;
struct btrfs_header *h;
struct btrfs_root *root = sdev->dev->dev_root;
struct btrfs_fs_info *fs_info = root->fs_info;
- u8 csum[BTRFS_CSUM_SIZE];
+ u8 calculated_csum[BTRFS_CSUM_SIZE];
+ u8 on_disk_csum[BTRFS_CSUM_SIZE];
+ struct page *page;
+ void *mapped_buffer;
+ u64 mapped_size;
+ void *p;
u32 crc = ~(u32)0;
int fail = 0;
int crc_fail = 0;
- mapped_buffer = kmap_atomic(page, KM_USER0);
+ u64 len;
+ int index;
+
+ BUG_ON(sblock->page_count < 1);
+ page = sblock->pagev[0].page;
++ mapped_buffer = kmap_atomic(page);
+ h = (struct btrfs_header *)mapped_buffer;
+ memcpy(on_disk_csum, h->csum, sdev->csum_size);
/*
* we don't use the getter functions here, as we
* a) don't have an extent buffer and
* b) the page is already kmapped
*/
- h = (struct btrfs_header *)buffer;
- if (logical != le64_to_cpu(h->bytenr))
+ if (sblock->pagev[0].logical != le64_to_cpu(h->bytenr))
++fail;
- if (spag->generation != le64_to_cpu(h->generation))
+ if (sblock->pagev[0].generation != le64_to_cpu(h->generation))
++fail;
if (memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE))
BTRFS_UUID_SIZE))
++fail;
- crc = btrfs_csum_data(root, buffer + BTRFS_CSUM_SIZE, crc,
- PAGE_SIZE - BTRFS_CSUM_SIZE);
- btrfs_csum_final(crc, csum);
- if (memcmp(csum, h->csum, sdev->csum_size))
+ BUG_ON(sdev->nodesize != sdev->leafsize);
+ len = sdev->nodesize - BTRFS_CSUM_SIZE;
+ mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
+ p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
+ index = 0;
+ for (;;) {
+ u64 l = min_t(u64, len, mapped_size);
+
+ crc = btrfs_csum_data(root, p, crc, l);
- kunmap_atomic(mapped_buffer, KM_USER0);
++ kunmap_atomic(mapped_buffer);
+ len -= l;
+ if (len == 0)
+ break;
+ index++;
+ BUG_ON(index >= sblock->page_count);
+ BUG_ON(!sblock->pagev[index].page);
+ page = sblock->pagev[index].page;
- mapped_buffer = kmap_atomic(page, KM_USER0);
++ mapped_buffer = kmap_atomic(page);
+ mapped_size = PAGE_SIZE;
+ p = mapped_buffer;
+ }
+
+ btrfs_csum_final(crc, calculated_csum);
+ if (memcmp(calculated_csum, on_disk_csum, sdev->csum_size))
++crc_fail;
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.tree_extents_scrubbed;
- sdev->stat.tree_bytes_scrubbed += PAGE_SIZE;
- if (crc_fail)
- ++sdev->stat.csum_errors;
- if (fail)
- ++sdev->stat.verify_errors;
- spin_unlock(&sdev->stat_lock);
+ if (crc_fail || fail) {
+ spin_lock(&sdev->stat_lock);
+ if (crc_fail)
+ ++sdev->stat.csum_errors;
+ if (fail)
+ ++sdev->stat.verify_errors;
+ spin_unlock(&sdev->stat_lock);
+ }
return fail || crc_fail;
}
- static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer)
+ static int scrub_checksum_super(struct scrub_block *sblock)
{
struct btrfs_super_block *s;
- u64 logical;
- struct scrub_dev *sdev = sbio->sdev;
+ struct scrub_dev *sdev = sblock->sdev;
struct btrfs_root *root = sdev->dev->dev_root;
struct btrfs_fs_info *fs_info = root->fs_info;
- u8 csum[BTRFS_CSUM_SIZE];
+ u8 calculated_csum[BTRFS_CSUM_SIZE];
+ u8 on_disk_csum[BTRFS_CSUM_SIZE];
+ struct page *page;
+ void *mapped_buffer;
+ u64 mapped_size;
+ void *p;
u32 crc = ~(u32)0;
int fail = 0;
+ u64 len;
+ int index;
- s = (struct btrfs_super_block *)buffer;
- logical = sbio->logical;
+ BUG_ON(sblock->page_count < 1);
+ page = sblock->pagev[0].page;
- mapped_buffer = kmap_atomic(page, KM_USER0);
++ mapped_buffer = kmap_atomic(page);
+ s = (struct btrfs_super_block *)mapped_buffer;
+ memcpy(on_disk_csum, s->csum, sdev->csum_size);
- if (logical != le64_to_cpu(s->bytenr))
+ if (sblock->pagev[0].logical != le64_to_cpu(s->bytenr))
++fail;
- if (sbio->spag[0].generation != le64_to_cpu(s->generation))
+ if (sblock->pagev[0].generation != le64_to_cpu(s->generation))
++fail;
if (memcmp(s->fsid, fs_info->fsid, BTRFS_UUID_SIZE))
++fail;
- crc = btrfs_csum_data(root, buffer + BTRFS_CSUM_SIZE, crc,
- PAGE_SIZE - BTRFS_CSUM_SIZE);
- btrfs_csum_final(crc, csum);
- if (memcmp(csum, s->csum, sbio->sdev->csum_size))
+ len = BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE;
+ mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
+ p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
+ index = 0;
+ for (;;) {
+ u64 l = min_t(u64, len, mapped_size);
+
+ crc = btrfs_csum_data(root, p, crc, l);
- kunmap_atomic(mapped_buffer, KM_USER0);
++ kunmap_atomic(mapped_buffer);
+ len -= l;
+ if (len == 0)
+ break;
+ index++;
+ BUG_ON(index >= sblock->page_count);
+ BUG_ON(!sblock->pagev[index].page);
+ page = sblock->pagev[index].page;
- mapped_buffer = kmap_atomic(page, KM_USER0);
++ mapped_buffer = kmap_atomic(page);
+ mapped_size = PAGE_SIZE;
+ p = mapped_buffer;
+ }
+
+ btrfs_csum_final(crc, calculated_csum);
+ if (memcmp(calculated_csum, on_disk_csum, sdev->csum_size))
++fail;
if (fail) {
return fail;
}
- static int scrub_submit(struct scrub_dev *sdev)
+ static void scrub_block_get(struct scrub_block *sblock)
+ {
+ atomic_inc(&sblock->ref_count);
+ }
+
+ static void scrub_block_put(struct scrub_block *sblock)
+ {
+ if (atomic_dec_and_test(&sblock->ref_count)) {
+ int i;
+
+ for (i = 0; i < sblock->page_count; i++)
+ if (sblock->pagev[i].page)
+ __free_page(sblock->pagev[i].page);
+ kfree(sblock);
+ }
+ }
+
+ static void scrub_submit(struct scrub_dev *sdev)
{
struct scrub_bio *sbio;
if (sdev->curr == -1)
- return 0;
+ return;
sbio = sdev->bios[sdev->curr];
- sbio->err = 0;
sdev->curr = -1;
atomic_inc(&sdev->in_flight);
btrfsic_submit_bio(READ, sbio->bio);
-
- return 0;
}
- static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
- u64 physical, u64 flags, u64 gen, int mirror_num,
- u8 *csum, int force)
+ static int scrub_add_page_to_bio(struct scrub_dev *sdev,
+ struct scrub_page *spage)
{
+ struct scrub_block *sblock = spage->sblock;
struct scrub_bio *sbio;
- struct page *page;
int ret;
again:
if (sdev->curr != -1) {
sdev->first_free = sdev->bios[sdev->curr]->next_free;
sdev->bios[sdev->curr]->next_free = -1;
- sdev->bios[sdev->curr]->count = 0;
+ sdev->bios[sdev->curr]->page_count = 0;
spin_unlock(&sdev->list_lock);
} else {
spin_unlock(&sdev->list_lock);
}
}
sbio = sdev->bios[sdev->curr];
- if (sbio->count == 0) {
+ if (sbio->page_count == 0) {
struct bio *bio;
- sbio->physical = physical;
- sbio->logical = logical;
- bio = bio_alloc(GFP_NOFS, SCRUB_PAGES_PER_BIO);
- if (!bio)
- return -ENOMEM;
+ sbio->physical = spage->physical;
+ sbio->logical = spage->logical;
+ bio = sbio->bio;
+ if (!bio) {
+ bio = bio_alloc(GFP_NOFS, sdev->pages_per_bio);
+ if (!bio)
+ return -ENOMEM;
+ sbio->bio = bio;
+ }
bio->bi_private = sbio;
bio->bi_end_io = scrub_bio_end_io;
bio->bi_bdev = sdev->dev->bdev;
- bio->bi_sector = sbio->physical >> 9;
+ bio->bi_sector = spage->physical >> 9;
sbio->err = 0;
- sbio->bio = bio;
- } else if (sbio->physical + sbio->count * PAGE_SIZE != physical ||
- sbio->logical + sbio->count * PAGE_SIZE != logical) {
- ret = scrub_submit(sdev);
- if (ret)
- return ret;
+ } else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
+ spage->physical ||
+ sbio->logical + sbio->page_count * PAGE_SIZE !=
+ spage->logical) {
+ scrub_submit(sdev);
goto again;
}
- sbio->spag[sbio->count].flags = flags;
- sbio->spag[sbio->count].generation = gen;
- sbio->spag[sbio->count].have_csum = 0;
- sbio->spag[sbio->count].mirror_num = mirror_num;
-
- page = alloc_page(GFP_NOFS);
- if (!page)
- return -ENOMEM;
- ret = bio_add_page(sbio->bio, page, PAGE_SIZE, 0);
- if (!ret) {
- __free_page(page);
- ret = scrub_submit(sdev);
- if (ret)
- return ret;
+ sbio->pagev[sbio->page_count] = spage;
+ ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
+ if (ret != PAGE_SIZE) {
+ if (sbio->page_count < 1) {
+ bio_put(sbio->bio);
+ sbio->bio = NULL;
+ return -EIO;
+ }
+ scrub_submit(sdev);
goto again;
}
- if (csum) {
- sbio->spag[sbio->count].have_csum = 1;
- memcpy(sbio->spag[sbio->count].csum, csum, sdev->csum_size);
+ scrub_block_get(sblock); /* one for the added page */
+ atomic_inc(&sblock->outstanding_pages);
+ sbio->page_count++;
+ if (sbio->page_count == sdev->pages_per_bio)
+ scrub_submit(sdev);
+
+ return 0;
+ }
+
+ static int scrub_pages(struct scrub_dev *sdev, u64 logical, u64 len,
+ u64 physical, u64 flags, u64 gen, int mirror_num,
+ u8 *csum, int force)
+ {
+ struct scrub_block *sblock;
+ int index;
+
+ sblock = kzalloc(sizeof(*sblock), GFP_NOFS);
+ if (!sblock) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.malloc_errors++;
+ spin_unlock(&sdev->stat_lock);
+ return -ENOMEM;
}
- ++sbio->count;
- if (sbio->count == SCRUB_PAGES_PER_BIO || force) {
+
+ /* one ref inside this function, plus one for each page later on */
+ atomic_set(&sblock->ref_count, 1);
+ sblock->sdev = sdev;
+ sblock->no_io_error_seen = 1;
+
+ for (index = 0; len > 0; index++) {
+ struct scrub_page *spage = sblock->pagev + index;
+ u64 l = min_t(u64, len, PAGE_SIZE);
+
+ BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
+ spage->page = alloc_page(GFP_NOFS);
+ if (!spage->page) {
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.malloc_errors++;
+ spin_unlock(&sdev->stat_lock);
+ while (index > 0) {
+ index--;
+ __free_page(sblock->pagev[index].page);
+ }
+ kfree(sblock);
+ return -ENOMEM;
+ }
+ spage->sblock = sblock;
+ spage->bdev = sdev->dev->bdev;
+ spage->flags = flags;
+ spage->generation = gen;
+ spage->logical = logical;
+ spage->physical = physical;
+ spage->mirror_num = mirror_num;
+ if (csum) {
+ spage->have_csum = 1;
+ memcpy(spage->csum, csum, sdev->csum_size);
+ } else {
+ spage->have_csum = 0;
+ }
+ sblock->page_count++;
+ len -= l;
+ logical += l;
+ physical += l;
+ }
+
+ BUG_ON(sblock->page_count == 0);
+ for (index = 0; index < sblock->page_count; index++) {
+ struct scrub_page *spage = sblock->pagev + index;
int ret;
- ret = scrub_submit(sdev);
- if (ret)
+ ret = scrub_add_page_to_bio(sdev, spage);
+ if (ret) {
+ scrub_block_put(sblock);
return ret;
+ }
}
+ if (force)
+ scrub_submit(sdev);
+
+ /* last one frees, either here or in bio completion for last page */
+ scrub_block_put(sblock);
return 0;
}
+ static void scrub_bio_end_io(struct bio *bio, int err)
+ {
+ struct scrub_bio *sbio = bio->bi_private;
+ struct scrub_dev *sdev = sbio->sdev;
+ struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
+
+ sbio->err = err;
+ sbio->bio = bio;
+
+ btrfs_queue_worker(&fs_info->scrub_workers, &sbio->work);
+ }
+
+ static void scrub_bio_end_io_worker(struct btrfs_work *work)
+ {
+ struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
+ struct scrub_dev *sdev = sbio->sdev;
+ int i;
+
+ BUG_ON(sbio->page_count > SCRUB_PAGES_PER_BIO);
+ if (sbio->err) {
+ for (i = 0; i < sbio->page_count; i++) {
+ struct scrub_page *spage = sbio->pagev[i];
+
+ spage->io_error = 1;
+ spage->sblock->no_io_error_seen = 0;
+ }
+ }
+
+ /* now complete the scrub_block items that have all pages completed */
+ for (i = 0; i < sbio->page_count; i++) {
+ struct scrub_page *spage = sbio->pagev[i];
+ struct scrub_block *sblock = spage->sblock;
+
+ if (atomic_dec_and_test(&sblock->outstanding_pages))
+ scrub_block_complete(sblock);
+ scrub_block_put(sblock);
+ }
+
+ if (sbio->err) {
+ /* what is this good for??? */
+ sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
+ sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
+ sbio->bio->bi_phys_segments = 0;
+ sbio->bio->bi_idx = 0;
+
+ for (i = 0; i < sbio->page_count; i++) {
+ struct bio_vec *bi;
+ bi = &sbio->bio->bi_io_vec[i];
+ bi->bv_offset = 0;
+ bi->bv_len = PAGE_SIZE;
+ }
+ }
+
+ bio_put(sbio->bio);
+ sbio->bio = NULL;
+ spin_lock(&sdev->list_lock);
+ sbio->next_free = sdev->first_free;
+ sdev->first_free = sbio->index;
+ spin_unlock(&sdev->list_lock);
+ atomic_dec(&sdev->in_flight);
+ wake_up(&sdev->list_wait);
+ }
+
+ static void scrub_block_complete(struct scrub_block *sblock)
+ {
+ if (!sblock->no_io_error_seen)
+ scrub_handle_errored_block(sblock);
+ else
+ scrub_checksum(sblock);
+ }
+
static int scrub_find_csum(struct scrub_dev *sdev, u64 logical, u64 len,
u8 *csum)
{
int ret = 0;
unsigned long i;
unsigned long num_sectors;
- u32 sectorsize = sdev->dev->dev_root->sectorsize;
while (!list_empty(&sdev->csum_list)) {
sum = list_first_entry(&sdev->csum_list,
if (!sum)
return 0;
- num_sectors = sum->len / sectorsize;
+ num_sectors = sum->len / sdev->sectorsize;
for (i = 0; i < num_sectors; ++i) {
if (sum->sums[i].bytenr == logical) {
memcpy(csum, &sum->sums[i].sum, sdev->csum_size);
{
int ret;
u8 csum[BTRFS_CSUM_SIZE];
+ u32 blocksize;
+
+ if (flags & BTRFS_EXTENT_FLAG_DATA) {
+ blocksize = sdev->sectorsize;
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.data_extents_scrubbed++;
+ sdev->stat.data_bytes_scrubbed += len;
+ spin_unlock(&sdev->stat_lock);
+ } else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+ BUG_ON(sdev->nodesize != sdev->leafsize);
+ blocksize = sdev->nodesize;
+ spin_lock(&sdev->stat_lock);
+ sdev->stat.tree_extents_scrubbed++;
+ sdev->stat.tree_bytes_scrubbed += len;
+ spin_unlock(&sdev->stat_lock);
+ } else {
+ blocksize = sdev->sectorsize;
+ BUG_ON(1);
+ }
while (len) {
- u64 l = min_t(u64, len, PAGE_SIZE);
+ u64 l = min_t(u64, len, blocksize);
int have_csum = 0;
if (flags & BTRFS_EXTENT_FLAG_DATA) {
if (have_csum == 0)
++sdev->stat.no_csum;
}
- ret = scrub_page(sdev, logical, l, physical, flags, gen,
- mirror_num, have_csum ? csum : NULL, 0);
+ ret = scrub_pages(sdev, logical, l, physical, flags, gen,
+ mirror_num, have_csum ? csum : NULL, 0);
if (ret)
return ret;
len -= l;
if (!path)
return -ENOMEM;
+ /*
+ * work on commit root. The related disk blocks are static as
+ * long as COW is applied. This means, it is save to rewrite
+ * them to repair disk errors without any race conditions
+ */
path->search_commit_root = 1;
path->skip_locking = 1;
struct btrfs_device *device = sdev->dev;
struct btrfs_root *root = device->dev_root;
+ if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
+ return -EIO;
+
gen = root->fs_info->last_trans_committed;
for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
bytenr = btrfs_sb_offset(i);
- if (bytenr + BTRFS_SUPER_INFO_SIZE >= device->total_bytes)
+ if (bytenr + BTRFS_SUPER_INFO_SIZE > device->total_bytes)
break;
- ret = scrub_page(sdev, bytenr, PAGE_SIZE, bytenr,
- BTRFS_EXTENT_FLAG_SUPER, gen, i, NULL, 1);
+ ret = scrub_pages(sdev, bytenr, BTRFS_SUPER_INFO_SIZE, bytenr,
+ BTRFS_EXTENT_FLAG_SUPER, gen, i, NULL, 1);
if (ret)
return ret;
}
/*
* check some assumptions
*/
- if (root->sectorsize != PAGE_SIZE ||
- root->sectorsize != root->leafsize ||
- root->sectorsize != root->nodesize) {
- printk(KERN_ERR "btrfs_scrub: size assumptions fail\n");
+ if (root->nodesize != root->leafsize) {
+ printk(KERN_ERR
+ "btrfs_scrub: size assumption nodesize == leafsize (%d == %d) fails\n",
+ root->nodesize, root->leafsize);
+ return -EINVAL;
+ }
+
+ if (root->nodesize > BTRFS_STRIPE_LEN) {
+ /*
+ * in this case scrub is unable to calculate the checksum
+ * the way scrub is implemented. Do not handle this
+ * situation at all because it won't ever happen.
+ */
+ printk(KERN_ERR
+ "btrfs_scrub: size assumption nodesize <= BTRFS_STRIPE_LEN (%d <= %d) fails\n",
+ root->nodesize, BTRFS_STRIPE_LEN);
+ return -EINVAL;
+ }
+
+ if (root->sectorsize != PAGE_SIZE) {
+ /* not supported for data w/o checksums */
+ printk(KERN_ERR
+ "btrfs_scrub: size assumption sectorsize != PAGE_SIZE (%d != %lld) fails\n",
+ root->sectorsize, (unsigned long long)PAGE_SIZE);
return -EINVAL;
}
return ret;
}
- int btrfs_scrub_pause(struct btrfs_root *root)
+ void btrfs_scrub_pause(struct btrfs_root *root)
{
struct btrfs_fs_info *fs_info = root->fs_info;
mutex_lock(&fs_info->scrub_lock);
}
mutex_unlock(&fs_info->scrub_lock);
-
- return 0;
}
- int btrfs_scrub_continue(struct btrfs_root *root)
+ void btrfs_scrub_continue(struct btrfs_root *root)
{
struct btrfs_fs_info *fs_info = root->fs_info;
atomic_dec(&fs_info->scrub_pause_req);
wake_up(&fs_info->scrub_pause_wait);
- return 0;
}
- int btrfs_scrub_pause_super(struct btrfs_root *root)
+ void btrfs_scrub_pause_super(struct btrfs_root *root)
{
down_write(&root->fs_info->scrub_super_lock);
- return 0;
}
- int btrfs_scrub_continue_super(struct btrfs_root *root)
+ void btrfs_scrub_continue_super(struct btrfs_root *root)
{
up_write(&root->fs_info->scrub_super_lock);
- return 0;
}
- int btrfs_scrub_cancel(struct btrfs_root *root)
+ int __btrfs_scrub_cancel(struct btrfs_fs_info *fs_info)
{
- struct btrfs_fs_info *fs_info = root->fs_info;
mutex_lock(&fs_info->scrub_lock);
if (!atomic_read(&fs_info->scrubs_running)) {
return 0;
}
+ int btrfs_scrub_cancel(struct btrfs_root *root)
+ {
+ return __btrfs_scrub_cancel(root->fs_info);
+ }
+
int btrfs_scrub_cancel_dev(struct btrfs_root *root, struct btrfs_device *dev)
{
struct btrfs_fs_info *fs_info = root->fs_info;
return 0;
}
+
int btrfs_scrub_cancel_devid(struct btrfs_root *root, u64 devid)
{
struct btrfs_fs_info *fs_info = root->fs_info;
case -EROFS:
errstr = "Readonly filesystem";
break;
+ case -EEXIST:
+ errstr = "Object already exists";
+ break;
default:
if (nbuf) {
if (snprintf(nbuf, 16, "error %d", -errno) >= 0)
if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
sb->s_flags |= MS_RDONLY;
printk(KERN_INFO "btrfs is forced readonly\n");
+ __btrfs_scrub_cancel(fs_info);
+ // WARN_ON(1);
}
}
* invokes the approciate error response.
*/
void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
- unsigned int line, int errno)
+ unsigned int line, int errno, const char *fmt, ...)
{
struct super_block *sb = fs_info->sb;
char nbuf[16];
const char *errstr;
+ va_list args;
+ va_start(args, fmt);
/*
* Special case: if the error is EROFS, and we're already
* under MS_RDONLY, then it is safe here.
*/
if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
+ return;
+
+ errstr = btrfs_decode_error(fs_info, errno, nbuf);
+ if (fmt) {
+ struct va_format vaf = {
+ .fmt = fmt,
+ .va = &args,
+ };
+
+ printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s (%pV)\n",
+ sb->s_id, function, line, errstr, &vaf);
+ } else {
+ printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s\n",
+ sb->s_id, function, line, errstr);
+ }
+
+ /* Don't go through full error handling during mount */
+ if (sb->s_flags & MS_BORN) {
+ save_error_info(fs_info);
+ btrfs_handle_error(fs_info);
+ }
+ va_end(args);
+ }
+
+ const char *logtypes[] = {
+ "emergency",
+ "alert",
+ "critical",
+ "error",
+ "warning",
+ "notice",
+ "info",
+ "debug",
+ };
+
+ void btrfs_printk(struct btrfs_fs_info *fs_info, const char *fmt, ...)
+ {
+ struct super_block *sb = fs_info->sb;
+ char lvl[4];
+ struct va_format vaf;
+ va_list args;
+ const char *type = logtypes[4];
+
+ va_start(args, fmt);
+
+ if (fmt[0] == '<' && isdigit(fmt[1]) && fmt[2] == '>') {
+ strncpy(lvl, fmt, 3);
+ fmt += 3;
+ type = logtypes[fmt[1] - '0'];
+ } else
+ *lvl = '\0';
+
+ vaf.fmt = fmt;
+ vaf.va = &args;
+ printk("%sBTRFS %s (device %s): %pV", lvl, type, sb->s_id, &vaf);
+ }
+
+ /*
+ * We only mark the transaction aborted and then set the file system read-only.
+ * This will prevent new transactions from starting or trying to join this
+ * one.
+ *
+ * This means that error recovery at the call site is limited to freeing
+ * any local memory allocations and passing the error code up without
+ * further cleanup. The transaction should complete as it normally would
+ * in the call path but will return -EIO.
+ *
+ * We'll complete the cleanup in btrfs_end_transaction and
+ * btrfs_commit_transaction.
+ */
+ void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, const char *function,
+ unsigned int line, int errno)
+ {
+ WARN_ONCE(1, KERN_DEBUG "btrfs: Transaction aborted");
+ trans->aborted = errno;
+ /* Nothing used. The other threads that have joined this
+ * transaction may be able to continue. */
+ if (!trans->blocks_used) {
+ btrfs_printk(root->fs_info, "Aborting unused transaction.\n");
return;
+ }
+ trans->transaction->aborted = errno;
+ __btrfs_std_error(root->fs_info, function, line, errno, NULL);
+ }
+ /*
+ * __btrfs_panic decodes unexpected, fatal errors from the caller,
+ * issues an alert, and either panics or BUGs, depending on mount options.
+ */
+ void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
+ unsigned int line, int errno, const char *fmt, ...)
+ {
+ char nbuf[16];
+ char *s_id = "<unknown>";
+ const char *errstr;
+ struct va_format vaf = { .fmt = fmt };
+ va_list args;
- errstr = btrfs_decode_error(fs_info, errno, nbuf);
- printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s\n",
- sb->s_id, function, line, errstr);
- save_error_info(fs_info);
+ if (fs_info)
+ s_id = fs_info->sb->s_id;
- btrfs_handle_error(fs_info);
+ va_start(args, fmt);
+ vaf.va = &args;
+
+ errstr = btrfs_decode_error(fs_info, errno, nbuf);
+ if (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)
+ panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
+ s_id, function, line, &vaf, errstr);
+
+ printk(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
+ s_id, function, line, &vaf, errstr);
+ va_end(args);
+ /* Caller calls BUG() */
}
static void btrfs_put_super(struct super_block *sb)
Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
Opt_check_integrity, Opt_check_integrity_including_extent_data,
- Opt_check_integrity_print_mask,
+ Opt_check_integrity_print_mask, Opt_fatal_errors,
Opt_err,
};
{Opt_check_integrity, "check_int"},
{Opt_check_integrity_including_extent_data, "check_int_data"},
{Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
+ {Opt_fatal_errors, "fatal_errors=%s"},
{Opt_err, NULL},
};
/*
* Regular mount options parser. Everything that is needed only when
* reading in a new superblock is parsed here.
+ * XXX JDM: This needs to be cleaned up for remount.
*/
int btrfs_parse_options(struct btrfs_root *root, char *options)
{
ret = -EINVAL;
goto out;
#endif
+ case Opt_fatal_errors:
+ if (strcmp(args[0].from, "panic") == 0)
+ btrfs_set_opt(info->mount_opt,
+ PANIC_ON_FATAL_ERROR);
+ else if (strcmp(args[0].from, "bug") == 0)
+ btrfs_clear_opt(info->mount_opt,
+ PANIC_ON_FATAL_ERROR);
+ else {
+ ret = -EINVAL;
+ goto out;
+ }
+ break;
case Opt_err:
printk(KERN_INFO "btrfs: unrecognized mount option "
"'%s'\n", p);
void *data, int silent)
{
struct inode *inode;
- struct dentry *root_dentry;
struct btrfs_fs_info *fs_info = btrfs_sb(sb);
struct btrfs_key key;
int err;
goto fail_close;
}
- root_dentry = d_alloc_root(inode);
- if (!root_dentry) {
- iput(inode);
+ sb->s_root = d_make_root(inode);
+ if (!sb->s_root) {
err = -ENOMEM;
goto fail_close;
}
- sb->s_root = root_dentry;
-
save_mount_options(sb, data);
cleancache_init_fs(sb);
sb->s_flags |= MS_ACTIVE;
seq_puts(seq, ",inode_cache");
if (btrfs_test_opt(root, SKIP_BALANCE))
seq_puts(seq, ",skip_balance");
+ if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
+ seq_puts(seq, ",fatal_errors=panic");
return 0;
}
{
struct btrfs_fs_info *fs_info = btrfs_sb(sb);
struct btrfs_root *root = fs_info->tree_root;
+ unsigned old_flags = sb->s_flags;
+ unsigned long old_opts = fs_info->mount_opt;
+ unsigned long old_compress_type = fs_info->compress_type;
+ u64 old_max_inline = fs_info->max_inline;
+ u64 old_alloc_start = fs_info->alloc_start;
+ int old_thread_pool_size = fs_info->thread_pool_size;
+ unsigned int old_metadata_ratio = fs_info->metadata_ratio;
int ret;
ret = btrfs_parse_options(root, data);
- if (ret)
- return -EINVAL;
+ if (ret) {
+ ret = -EINVAL;
+ goto restore;
+ }
if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
return 0;
if (*flags & MS_RDONLY) {
sb->s_flags |= MS_RDONLY;
- ret = btrfs_commit_super(root);
- WARN_ON(ret);
+ ret = btrfs_commit_super(root);
+ if (ret)
+ goto restore;
} else {
if (fs_info->fs_devices->rw_devices == 0)
- return -EACCES;
+ ret = -EACCES;
+ goto restore;
if (btrfs_super_log_root(fs_info->super_copy) != 0)
- return -EINVAL;
+ ret = -EINVAL;
+ goto restore;
ret = btrfs_cleanup_fs_roots(fs_info);
- WARN_ON(ret);
+ if (ret)
+ goto restore;
/* recover relocation */
ret = btrfs_recover_relocation(root);
- WARN_ON(ret);
+ if (ret)
+ goto restore;
sb->s_flags &= ~MS_RDONLY;
}
return 0;
+
+ restore:
+ /* We've hit an error - don't reset MS_RDONLY */
+ if (sb->s_flags & MS_RDONLY)
+ old_flags |= MS_RDONLY;
+ sb->s_flags = old_flags;
+ fs_info->mount_opt = old_opts;
+ fs_info->compress_type = old_compress_type;
+ fs_info->max_inline = old_max_inline;
+ fs_info->alloc_start = old_alloc_start;
+ fs_info->thread_pool_size = old_thread_pool_size;
+ fs_info->metadata_ratio = old_metadata_ratio;
+ return ret;
}
/* Used to sort the devices by max_avail(descending sort) */
if (err)
return err;
- err = btrfs_init_compress();
- if (err)
- goto free_sysfs;
+ btrfs_init_compress();
err = btrfs_init_cachep();
if (err)
if (err)
goto unregister_ioctl;
+ btrfs_init_lockdep();
+
printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
return 0;
btrfs_destroy_cachep();
free_compress:
btrfs_exit_compress();
- free_sysfs:
btrfs_exit_sysfs();
return err;
}