Commit 2ac55d41 authored by Josef Bacik's avatar Josef Bacik Committed by Chris Mason

Btrfs: cache the extent state everywhere we possibly can V2

This patch just goes through and fixes everybody that does

lock_extent()
blah
unlock_extent()

to use

lock_extent_bits()
blah
unlock_extent_cached()

and pass around a extent_state so we only have to do the searches once per
function.  This gives me about a 3 mb/s boots on my random write test.  I have
not converted some things, like the relocation and ioctl's, since they aren't
heavily used and the relocation stuff is in the middle of being re-written.  I
also changed the clear_extent_bit() to only unset the cached state if we are
clearing EXTENT_LOCKED and related stuff, so we can do things like this

lock_extent_bits()
clear delalloc bits
unlock_extent_cached()

without losing our cached state.  I tested this thoroughly and turned on
LEAK_DEBUG to make sure we weren't leaking extent states, everything worked out
fine.
Signed-off-by: default avatarJosef Bacik <josef@redhat.com>
Signed-off-by: default avatarChris Mason <chris.mason@oracle.com>
parent 5a1a3df1
......@@ -2311,7 +2311,8 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
u32 min_type);
int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput);
int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end);
int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
struct extent_state **cached_state);
int btrfs_writepages(struct address_space *mapping,
struct writeback_control *wbc);
int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
......
......@@ -263,13 +263,15 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
static int verify_parent_transid(struct extent_io_tree *io_tree,
struct extent_buffer *eb, u64 parent_transid)
{
struct extent_state *cached_state = NULL;
int ret;
if (!parent_transid || btrfs_header_generation(eb) == parent_transid)
return 0;
lock_extent(io_tree, eb->start, eb->start + eb->len - 1, GFP_NOFS);
if (extent_buffer_uptodate(io_tree, eb) &&
lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
0, &cached_state, GFP_NOFS);
if (extent_buffer_uptodate(io_tree, eb, cached_state) &&
btrfs_header_generation(eb) == parent_transid) {
ret = 0;
goto out;
......@@ -282,10 +284,10 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
(unsigned long long)btrfs_header_generation(eb));
}
ret = 1;
clear_extent_buffer_uptodate(io_tree, eb);
clear_extent_buffer_uptodate(io_tree, eb, &cached_state);
out:
unlock_extent(io_tree, eb->start, eb->start + eb->len - 1,
GFP_NOFS);
unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
&cached_state, GFP_NOFS);
return ret;
}
......@@ -2497,7 +2499,8 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
int ret;
struct inode *btree_inode = buf->first_page->mapping->host;
ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf);
ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf,
NULL);
if (!ret)
return ret;
......
......@@ -6561,6 +6561,7 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root,
struct btrfs_key key;
struct inode *inode = NULL;
struct btrfs_file_extent_item *fi;
struct extent_state *cached_state = NULL;
u64 num_bytes;
u64 skip_objectid = 0;
u32 nritems;
......@@ -6589,12 +6590,14 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root,
}
num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
lock_extent(&BTRFS_I(inode)->io_tree, key.offset,
key.offset + num_bytes - 1, GFP_NOFS);
lock_extent_bits(&BTRFS_I(inode)->io_tree, key.offset,
key.offset + num_bytes - 1, 0, &cached_state,
GFP_NOFS);
btrfs_drop_extent_cache(inode, key.offset,
key.offset + num_bytes - 1, 1);
unlock_extent(&BTRFS_I(inode)->io_tree, key.offset,
key.offset + num_bytes - 1, GFP_NOFS);
unlock_extent_cached(&BTRFS_I(inode)->io_tree, key.offset,
key.offset + num_bytes - 1, &cached_state,
GFP_NOFS);
cond_resched();
}
iput(inode);
......
......@@ -513,7 +513,10 @@ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
u64 last_end;
int err;
int set = 0;
int clear = 0;
if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY))
clear = 1;
again:
if (!prealloc && (mask & __GFP_WAIT)) {
prealloc = alloc_extent_state(mask);
......@@ -524,13 +527,19 @@ again:
spin_lock(&tree->lock);
if (cached_state) {
cached = *cached_state;
if (clear) {
*cached_state = NULL;
cached_state = NULL;
}
if (cached && cached->tree && cached->start == start) {
if (clear)
atomic_dec(&cached->refs);
state = cached;
goto hit_next;
}
if (clear)
free_extent_state(cached);
}
/*
......@@ -946,11 +955,11 @@ int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
}
int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask)
struct extent_state **cached_state, gfp_t mask)
{
return set_extent_bit(tree, start, end,
EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE,
0, NULL, NULL, mask);
0, NULL, cached_state, mask);
}
int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
......@@ -984,10 +993,11 @@ int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
}
static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
u64 end, gfp_t mask)
u64 end, struct extent_state **cached_state,
gfp_t mask)
{
return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
NULL, mask);
cached_state, mask);
}
int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
......@@ -1727,7 +1737,7 @@ static void end_bio_extent_writepage(struct bio *bio, int err)
}
if (!uptodate) {
clear_extent_uptodate(tree, start, end, GFP_NOFS);
clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS);
ClearPageUptodate(page);
SetPageError(page);
}
......@@ -2710,6 +2720,7 @@ int extent_readpages(struct extent_io_tree *tree,
int extent_invalidatepage(struct extent_io_tree *tree,
struct page *page, unsigned long offset)
{
struct extent_state *cached_state = NULL;
u64 start = ((u64)page->index << PAGE_CACHE_SHIFT);
u64 end = start + PAGE_CACHE_SIZE - 1;
size_t blocksize = page->mapping->host->i_sb->s_blocksize;
......@@ -2718,12 +2729,12 @@ int extent_invalidatepage(struct extent_io_tree *tree,
if (start > end)
return 0;
lock_extent(tree, start, end, GFP_NOFS);
lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS);
wait_on_page_writeback(page);
clear_extent_bit(tree, start, end,
EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
EXTENT_DO_ACCOUNTING,
1, 1, NULL, GFP_NOFS);
1, 1, &cached_state, GFP_NOFS);
return 0;
}
......@@ -2926,16 +2937,17 @@ sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
get_extent_t *get_extent)
{
struct inode *inode = mapping->host;
struct extent_state *cached_state = NULL;
u64 start = iblock << inode->i_blkbits;
sector_t sector = 0;
size_t blksize = (1 << inode->i_blkbits);
struct extent_map *em;
lock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
GFP_NOFS);
lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
0, &cached_state, GFP_NOFS);
em = get_extent(inode, NULL, 0, start, blksize, 0);
unlock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
GFP_NOFS);
unlock_extent_cached(&BTRFS_I(inode)->io_tree, start,
start + blksize - 1, &cached_state, GFP_NOFS);
if (!em || IS_ERR(em))
return 0;
......@@ -2957,6 +2969,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
u32 flags = 0;
u64 disko = 0;
struct extent_map *em = NULL;
struct extent_state *cached_state = NULL;
int end = 0;
u64 em_start = 0, em_len = 0;
unsigned long emflags;
......@@ -2965,8 +2978,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
if (len == 0)
return -EINVAL;
lock_extent(&BTRFS_I(inode)->io_tree, start, start + len,
GFP_NOFS);
lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
&cached_state, GFP_NOFS);
em = get_extent(inode, NULL, 0, off, max - off, 0);
if (!em)
goto out;
......@@ -3029,8 +3042,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
out_free:
free_extent_map(em);
out:
unlock_extent(&BTRFS_I(inode)->io_tree, start, start + len,
GFP_NOFS);
unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len,
&cached_state, GFP_NOFS);
return ret;
}
......@@ -3270,7 +3283,8 @@ int set_extent_buffer_dirty(struct extent_io_tree *tree,
}
int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
struct extent_buffer *eb)
struct extent_buffer *eb,
struct extent_state **cached_state)
{
unsigned long i;
struct page *page;
......@@ -3280,7 +3294,7 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
GFP_NOFS);
cached_state, GFP_NOFS);
for (i = 0; i < num_pages; i++) {
page = extent_buffer_page(eb, i);
if (page)
......@@ -3340,7 +3354,8 @@ int extent_range_uptodate(struct extent_io_tree *tree,
}
int extent_buffer_uptodate(struct extent_io_tree *tree,
struct extent_buffer *eb)
struct extent_buffer *eb,
struct extent_state *cached_state)
{
int ret = 0;
unsigned long num_pages;
......@@ -3352,7 +3367,7 @@ int extent_buffer_uptodate(struct extent_io_tree *tree,
return 1;
ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
EXTENT_UPTODATE, 1, NULL);
EXTENT_UPTODATE, 1, cached_state);
if (ret)
return ret;
......
......@@ -163,6 +163,8 @@ int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
int bits, struct extent_state **cached, gfp_t mask);
int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
struct extent_state **cached, gfp_t mask);
int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask);
int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
......@@ -196,7 +198,7 @@ int clear_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
int clear_extent_ordered_metadata(struct extent_io_tree *tree, u64 start,
u64 end, gfp_t mask);
int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask);
struct extent_state **cached_state, gfp_t mask);
int set_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask);
int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
......@@ -281,9 +283,11 @@ int test_extent_buffer_dirty(struct extent_io_tree *tree,
int set_extent_buffer_uptodate(struct extent_io_tree *tree,
struct extent_buffer *eb);
int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
struct extent_buffer *eb);
struct extent_buffer *eb,
struct extent_state **cached_state);
int extent_buffer_uptodate(struct extent_io_tree *tree,
struct extent_buffer *eb);
struct extent_buffer *eb,
struct extent_state *cached_state);
int map_extent_buffer(struct extent_buffer *eb, unsigned long offset,
unsigned long min_len, char **token, char **map,
unsigned long *map_start,
......
......@@ -123,7 +123,8 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans,
root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
end_of_last_block = start_pos + num_bytes - 1;
err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block);
err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
NULL);
if (err)
return err;
......@@ -753,6 +754,7 @@ static noinline int prepare_pages(struct btrfs_root *root, struct file *file,
loff_t pos, unsigned long first_index,
unsigned long last_index, size_t write_bytes)
{
struct extent_state *cached_state = NULL;
int i;
unsigned long index = pos >> PAGE_CACHE_SHIFT;
struct inode *inode = fdentry(file)->d_inode;
......@@ -781,16 +783,18 @@ again:
}
if (start_pos < inode->i_size) {
struct btrfs_ordered_extent *ordered;
lock_extent(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1, GFP_NOFS);
lock_extent_bits(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1, 0, &cached_state,
GFP_NOFS);
ordered = btrfs_lookup_first_ordered_extent(inode,
last_pos - 1);
if (ordered &&
ordered->file_offset + ordered->len > start_pos &&
ordered->file_offset < last_pos) {
btrfs_put_ordered_extent(ordered);
unlock_extent(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1, GFP_NOFS);
unlock_extent_cached(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1,
&cached_state, GFP_NOFS);
for (i = 0; i < num_pages; i++) {
unlock_page(pages[i]);
page_cache_release(pages[i]);
......@@ -802,12 +806,13 @@ again:
if (ordered)
btrfs_put_ordered_extent(ordered);
clear_extent_bits(&BTRFS_I(inode)->io_tree, start_pos,
clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos,
last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
EXTENT_DO_ACCOUNTING,
EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
GFP_NOFS);
unlock_extent_cached(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1, &cached_state,
GFP_NOFS);
unlock_extent(&BTRFS_I(inode)->io_tree,
start_pos, last_pos - 1, GFP_NOFS);
}
for (i = 0; i < num_pages; i++) {
clear_page_dirty_for_io(pages[i]);
......
This diff is collapsed.
......@@ -673,7 +673,7 @@ again:
page_end, EXTENT_DIRTY | EXTENT_DELALLOC |
EXTENT_DO_ACCOUNTING, GFP_NOFS);
btrfs_set_extent_delalloc(inode, page_start, page_end);
btrfs_set_extent_delalloc(inode, page_start, page_end, NULL);
ClearPageChecked(page);
set_page_dirty(page);
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
......
......@@ -2659,7 +2659,7 @@ static int relocate_file_extent_cluster(struct inode *inode,
EXTENT_BOUNDARY, GFP_NOFS);
nr++;
}
btrfs_set_extent_delalloc(inode, page_start, page_end);
btrfs_set_extent_delalloc(inode, page_start, page_end, NULL);
set_page_dirty(page);
dirty_page++;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment