* The allocation request involve request for multiple number of blocks
* near to the goal(block) value specified.
*
- * During initialization phase of the allocator we decide to use the group
- * preallocation or inode preallocation depending on the size file. The
- * size of the file could be the resulting file size we would have after
- * allocation or the current file size which ever is larger. If the size is
- * less that sbi->s_mb_stream_request we select the group
- * preallocation. The default value of s_mb_stream_request is 16
- * blocks. This can also be tuned via
- * /proc/fs/ext4/<partition>/stream_req. The value is represented in terms
- * of number of blocks.
+ * During initialization phase of the allocator we decide to use the
+ * group preallocation or inode preallocation depending on the size of
+ * the file. The size of the file could be the resulting file size we
+ * would have after allocation, or the current file size, which ever
+ * is larger. If the size is less than sbi->s_mb_stream_request we
+ * select to use the group preallocation. The default value of
+ * s_mb_stream_request is 16 blocks. This can also be tuned via
+ * /sys/fs/ext4/<partition>/mb_stream_req. The value is represented in
+ * terms of number of blocks.
*
* The main motivation for having small file use group preallocation is to
- * ensure that we have small file closer in the disk.
+ * ensure that we have small files closer together on the disk.
*
- * First stage the allocator looks at the inode prealloc list
- * ext4_inode_info->i_prealloc_list contain list of prealloc spaces for
- * this particular inode. The inode prealloc space is represented as:
+ * First stage the allocator looks at the inode prealloc list,
+ * ext4_inode_info->i_prealloc_list, which contains list of prealloc
+ * spaces for this particular inode. The inode prealloc space is
+ * represented as:
*
* pa_lstart -> the logical start block for this prealloc space
* pa_pstart -> the physical start block for this prealloc space
* inode as:
*
* { page }
- * [ group 0 buddy][ group 0 bitmap] [group 1][ group 1]...
+ * [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]...
*
*
* one block each for bitmap and buddy information. So for each group we
* list. In case of inode preallocation we follow a list of heuristics
* based on file size. This can be found in ext4_mb_normalize_request. If
* we are doing a group prealloc we try to normalize the request to
- * sbi->s_mb_group_prealloc. Default value of s_mb_group_prealloc is set to
+ * sbi->s_mb_group_prealloc. Default value of s_mb_group_prealloc is
* 512 blocks. This can be tuned via
- * /proc/fs/ext4/<partition/group_prealloc. The value is represented in
+ * /sys/fs/ext4/<partition/mb_group_prealloc. The value is represented in
* terms of number of blocks. If we have mounted the file system with -O
* stripe=<value> option the group prealloc request is normalized to the
* stripe value (sbi->s_stripe)
*
- * The regular allocator(using the buddy cache) support few tunables.
+ * The regular allocator(using the buddy cache) supports few tunables.
*
- * /proc/fs/ext4/<partition>/min_to_scan
- * /proc/fs/ext4/<partition>/max_to_scan
- * /proc/fs/ext4/<partition>/order2_req
+ * /sys/fs/ext4/<partition>/mb_min_to_scan
+ * /sys/fs/ext4/<partition>/mb_max_to_scan
+ * /sys/fs/ext4/<partition>/mb_order2_req
*
- * The regular allocator use buddy scan only if the request len is power of
+ * The regular allocator uses buddy scan only if the request len is power of
* 2 blocks and the order of allocation is >= sbi->s_mb_order2_reqs. The
* value of s_mb_order2_reqs can be tuned via
- * /proc/fs/ext4/<partition>/order2_req. If the request len is equal to
+ * /sys/fs/ext4/<partition>/mb_order2_req. If the request len is equal to
* stripe size (sbi->s_stripe), we try to search for contigous block in
- * stripe size. This should result in better allocation on RAID setup. If
- * not we search in the specific group using bitmap for best extents. The
- * tunable min_to_scan and max_to_scan controll the behaviour here.
+ * stripe size. This should result in better allocation on RAID setups. If
+ * not, we search in the specific group using bitmap for best extents. The
+ * tunable min_to_scan and max_to_scan control the behaviour here.
* min_to_scan indicate how long the mballoc __must__ look for a best
- * extent and max_to_scanindicate how long the mballoc __can__ look for a
+ * extent and max_to_scan indicates how long the mballoc __can__ look for a
* best extent in the found extents. Searching for the blocks starts with
* the group specified as the goal value in allocation context via
* ac_g_ex. Each group is first checked based on the criteria whether it
* object
*
*/
+static struct kmem_cache *ext4_pspace_cachep;
+static struct kmem_cache *ext4_ac_cachep;
+static struct kmem_cache *ext4_free_ext_cachep;
+static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap,
+ ext4_group_t group);
+static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap,
+ ext4_group_t group);
+static void release_blocks_on_commit(journal_t *journal, transaction_t *txn);
+
+
static inline void *mb_correct_addr_and_bit(int *bit, void *addr)
{
blocknr += first + i;
blocknr +=
le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
-
- ext4_error(sb, __func__, "double-free of inode"
- " %lu's block %llu(bit %u in group %lu)\n",
+ ext4_grp_locked_error(sb, e4b->bd_group,
+ __func__, "double-free of inode"
+ " %lu's block %llu(bit %u in group %u)",
inode ? inode->i_ino : 0, blocknr,
first + i, e4b->bd_group);
}
b2 = (unsigned char *) bitmap;
for (i = 0; i < e4b->bd_sb->s_blocksize; i++) {
if (b1[i] != b2[i]) {
- printk("corruption in group %lu at byte %u(%u):"
- " %x in copy != %x on disk/prealloc\n",
- e4b->bd_group, i, i * 8, b1[i], b2[i]);
+ printk(KERN_ERR "corruption in group %u "
+ "at byte %u(%u): %x in copy != %x "
+ "on disk/prealloc\n",
+ e4b->bd_group, i, i * 8, b1[i], b2[i]);
BUG();
}
}
void *buddy;
void *buddy2;
- if (!test_opt(sb, MBALLOC))
- return 0;
-
{
static int mb_check_counter;
if (mb_check_counter++ % 100 != 0)
grp->bb_fragments = fragments;
if (free != grp->bb_free) {
- ext4_error(sb, __func__,
- "EXT4-fs: group %lu: %u blocks in bitmap, %u in gd\n",
+ ext4_grp_locked_error(sb, group, __func__,
+ "EXT4-fs: group %u: %u blocks in bitmap, %u in gd",
group, free, grp->bb_free);
/*
* If we intent to continue, we consider group descritor
* stored in the inode as
*
* { page }
- * [ group 0 buddy][ group 0 bitmap] [group 1][ group 1]...
+ * [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]...
*
*
* one block each for bitmap and buddy information.
if (bh[i] == NULL)
goto out;
- if (bh_uptodate_or_lock(bh[i]))
+ if (bitmap_uptodate(bh[i]))
continue;
+ lock_buffer(bh[i]);
+ if (bitmap_uptodate(bh[i])) {
+ unlock_buffer(bh[i]);
+ continue;
+ }
+ spin_lock(sb_bgl_lock(EXT4_SB(sb), first_group + i));
if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) {
ext4_init_block_bitmap(sb, bh[i],
first_group + i, desc);
+ set_bitmap_uptodate(bh[i]);
set_buffer_uptodate(bh[i]);
+ spin_unlock(sb_bgl_lock(EXT4_SB(sb), first_group + i));
+ unlock_buffer(bh[i]);
+ continue;
+ }
+ spin_unlock(sb_bgl_lock(EXT4_SB(sb), first_group + i));
+ if (buffer_uptodate(bh[i])) {
+ /*
+ * if not uninit if bh is uptodate,
+ * bitmap is also uptodate
+ */
+ set_bitmap_uptodate(bh[i]);
unlock_buffer(bh[i]);
continue;
}
get_bh(bh[i]);
+ /*
+ * submit the buffer_head for read. We can
+ * safely mark the bitmap as uptodate now.
+ * We do it here so the bitmap uptodate bit
+ * get set with buffer lock held.
+ */
+ set_bitmap_uptodate(bh[i]);
bh[i]->b_end_io = end_buffer_read_sync;
submit_bh(READ, bh[i]);
- mb_debug("read bitmap for group %lu\n", first_group + i);
+ mb_debug("read bitmap for group %u\n", first_group + i);
}
/* wait for I/O completion */
err = 0;
first_block = page->index * blocks_per_page;
+ /* init the page */
+ memset(page_address(page), 0xff, PAGE_CACHE_SIZE);
for (i = 0; i < blocks_per_page; i++) {
int group;
struct ext4_group_info *grinfo;
BUG_ON(incore == NULL);
mb_debug("put buddy for group %u in page %lu/%x\n",
group, page->index, i * blocksize);
- memset(data, 0xff, blocksize);
grinfo = ext4_get_group_info(sb, group);
grinfo->bb_fragments = 0;
memset(grinfo->bb_counters, 0,
/*
* incore got set to the group block bitmap below
*/
+ ext4_lock_group(sb, group);
ext4_mb_generate_buddy(sb, data, incore, group);
+ ext4_unlock_group(sb, group);
incore = NULL;
} else {
/* this is block of bitmap */
/* mark all preallocated blks used in in-core bitmap */
ext4_mb_generate_from_pa(sb, data, group);
+ ext4_mb_generate_from_freelist(sb, data, group);
ext4_unlock_group(sb, group);
/* set incore so that the buddy information can be
ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
struct ext4_buddy *e4b)
{
- struct ext4_sb_info *sbi = EXT4_SB(sb);
- struct inode *inode = sbi->s_buddy_cache;
int blocks_per_page;
int block;
int pnum;
int poff;
struct page *page;
int ret;
+ struct ext4_group_info *grp;
+ struct ext4_sb_info *sbi = EXT4_SB(sb);
+ struct inode *inode = sbi->s_buddy_cache;
- mb_debug("load group %lu\n", group);
+ mb_debug("load group %u\n", group);
blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
+ grp = ext4_get_group_info(sb, group);
e4b->bd_blkbits = sb->s_blocksize_bits;
e4b->bd_info = ext4_get_group_info(sb, group);
e4b->bd_group = group;
e4b->bd_buddy_page = NULL;
e4b->bd_bitmap_page = NULL;
+ e4b->alloc_semp = &grp->alloc_sem;
+
+ /* Take the read lock on the group alloc
+ * sem. This would make sure a parallel
+ * ext4_mb_init_group happening on other
+ * groups mapped by the page is blocked
+ * till we are done with allocation
+ */
+ down_read(e4b->alloc_semp);
/*
* the buddy cache inode stores the block bitmap
page = find_get_page(inode->i_mapping, pnum);
if (page == NULL || !PageUptodate(page)) {
if (page)
+ /*
+ * drop the page reference and try
+ * to get the page with lock. If we
+ * are not uptodate that implies
+ * somebody just created the page but
+ * is yet to initialize the same. So
+ * wait for it to initialize.
+ */
page_cache_release(page);
page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
if (page) {
page_cache_release(e4b->bd_buddy_page);
e4b->bd_buddy = NULL;
e4b->bd_bitmap = NULL;
+
+ /* Done with the buddy cache */
+ up_read(e4b->alloc_semp);
return ret;
}
page_cache_release(e4b->bd_bitmap_page);
if (e4b->bd_buddy_page)
page_cache_release(e4b->bd_buddy_page);
+ /* Done with the buddy cache */
+ if (e4b->alloc_semp)
+ up_read(e4b->alloc_semp);
}
cur += 32;
continue;
}
- mb_clear_bit_atomic(lock, cur, bm);
+ if (lock)
+ mb_clear_bit_atomic(lock, cur, bm);
+ else
+ mb_clear_bit(cur, bm);
cur++;
}
}
cur += 32;
continue;
}
- mb_set_bit_atomic(lock, cur, bm);
+ if (lock)
+ mb_set_bit_atomic(lock, cur, bm);
+ else
+ mb_set_bit(cur, bm);
cur++;
}
}
blocknr += block;
blocknr +=
le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
- ext4_unlock_group(sb, e4b->bd_group);
- ext4_error(sb, __func__, "double-free of inode"
- " %lu's block %llu(bit %u in group %lu)\n",
+ ext4_grp_locked_error(sb, e4b->bd_group,
+ __func__, "double-free of inode"
+ " %lu's block %llu(bit %u in group %u)",
inode ? inode->i_ino : 0, blocknr, block,
e4b->bd_group);
- ext4_lock_group(sb, e4b->bd_group);
}
mb_clear_bit(block, EXT4_MB_BITMAP(e4b));
e4b->bd_info->bb_counters[order]++;
ac->ac_tail = ret & 0xffff;
ac->ac_buddy = ret >> 16;
- /* XXXXXXX: SUCH A HORRIBLE **CK */
- /*FIXME!! Why ? */
+ /*
+ * take the page reference. We want the page to be pinned
+ * so that we don't get a ext4_mb_init_cache_call for this
+ * group until we update the bitmap. That would mean we
+ * double allocate blocks. The reference is dropped
+ * in ext4_mb_release_context
+ */
ac->ac_bitmap_page = e4b->bd_bitmap_page;
get_page(ac->ac_bitmap_page);
ac->ac_buddy_page = e4b->bd_buddy_page;
get_page(ac->ac_buddy_page);
-
+ /* on allocation we use ac to track the held semaphore */
+ ac->alloc_semp = e4b->alloc_semp;
+ e4b->alloc_semp = NULL;
/* store last allocated for subsequent stream allocation */
if ((ac->ac_flags & EXT4_MB_HINT_DATA)) {
spin_lock(&sbi->s_md_lock);
struct ext4_free_extent ex;
int max;
+ if (ac->ac_status == AC_STATUS_FOUND)
+ return;
/*
* We don't want to scan for a whole year
*/
struct ext4_free_extent *gex = &ac->ac_g_ex;
BUG_ON(ex->fe_len <= 0);
- BUG_ON(ex->fe_len >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
+ BUG_ON(ex->fe_len > EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
BUG_ON(ex->fe_start >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
BUG_ON(ac->ac_status != AC_STATUS_CONTINUE);
* free blocks even though group info says we
* we have free blocks
*/
- ext4_error(sb, __func__, "%d free blocks as per "
- "group info. But bitmap says 0\n",
+ ext4_grp_locked_error(sb, e4b->bd_group,
+ __func__, "%d free blocks as per "
+ "group info. But bitmap says 0",
free);
break;
}
mb_find_extent(e4b, 0, i, ac->ac_g_ex.fe_len, &ex);
BUG_ON(ex.fe_len <= 0);
if (free < ex.fe_len) {
- ext4_error(sb, __func__, "%d free blocks as per "
- "group info. But got %d blocks\n",
+ ext4_grp_locked_error(sb, e4b->bd_group,
+ __func__, "%d free blocks as per "
+ "group info. But got %d blocks",
free, ex.fe_len);
/*
* The number of free blocks differs. This mostly
{
unsigned free, fragments;
unsigned i, bits;
+ int flex_size = ext4_flex_bg_size(EXT4_SB(ac->ac_sb));
struct ext4_group_desc *desc;
struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))
return 0;
+ /* Avoid using the first bg of a flexgroup for data files */
+ if ((ac->ac_flags & EXT4_MB_HINT_DATA) &&
+ (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) &&
+ ((group % flex_size) == 0))
+ return 0;
+
bits = ac->ac_sb->s_blocksize_bits + 1;
for (i = ac->ac_2order; i <= bits; i++)
if (grp->bb_counters[i] > 0)
return 0;
}
+/*
+ * lock the group_info alloc_sem of all the groups
+ * belonging to the same buddy cache page. This
+ * make sure other parallel operation on the buddy
+ * cache doesn't happen whild holding the buddy cache
+ * lock
+ */
+int ext4_mb_get_buddy_cache_lock(struct super_block *sb, ext4_group_t group)
+{
+ int i;
+ int block, pnum;
+ int blocks_per_page;
+ int groups_per_page;
+ ext4_group_t first_group;
+ struct ext4_group_info *grp;
+
+ blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
+ /*
+ * the buddy cache inode stores the block bitmap
+ * and buddy information in consecutive blocks.
+ * So for each group we need two blocks.
+ */
+ block = group * 2;
+ pnum = block / blocks_per_page;
+ first_group = pnum * blocks_per_page / 2;
+
+ groups_per_page = blocks_per_page >> 1;
+ if (groups_per_page == 0)
+ groups_per_page = 1;
+ /* read all groups the page covers into the cache */
+ for (i = 0; i < groups_per_page; i++) {
+
+ if ((first_group + i) >= EXT4_SB(sb)->s_groups_count)
+ break;
+ grp = ext4_get_group_info(sb, first_group + i);
+ /* take all groups write allocation
+ * semaphore. This make sure there is
+ * no block allocation going on in any
+ * of that groups
+ */
+ down_write_nested(&grp->alloc_sem, i);
+ }
+ return i;
+}
+
+void ext4_mb_put_buddy_cache_lock(struct super_block *sb,
+ ext4_group_t group, int locked_group)
+{
+ int i;
+ int block, pnum;
+ int blocks_per_page;
+ ext4_group_t first_group;
+ struct ext4_group_info *grp;
+
+ blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
+ /*
+ * the buddy cache inode stores the block bitmap
+ * and buddy information in consecutive blocks.
+ * So for each group we need two blocks.
+ */
+ block = group * 2;
+ pnum = block / blocks_per_page;
+ first_group = pnum * blocks_per_page / 2;
+ /* release locks on all the groups */
+ for (i = 0; i < locked_group; i++) {
+
+ grp = ext4_get_group_info(sb, first_group + i);
+ /* take all groups write allocation
+ * semaphore. This make sure there is
+ * no block allocation going on in any
+ * of that groups
+ */
+ up_write(&grp->alloc_sem);
+ }
+
+}
+
+static int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
+{
+
+ int ret;
+ void *bitmap;
+ int blocks_per_page;
+ int block, pnum, poff;
+ int num_grp_locked = 0;
+ struct ext4_group_info *this_grp;
+ struct ext4_sb_info *sbi = EXT4_SB(sb);
+ struct inode *inode = sbi->s_buddy_cache;
+ struct page *page = NULL, *bitmap_page = NULL;
+
+ mb_debug("init group %lu\n", group);
+ blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
+ this_grp = ext4_get_group_info(sb, group);
+ /*
+ * This ensures we don't add group
+ * to this buddy cache via resize
+ */
+ num_grp_locked = ext4_mb_get_buddy_cache_lock(sb, group);
+ if (!EXT4_MB_GRP_NEED_INIT(this_grp)) {
+ /*
+ * somebody initialized the group
+ * return without doing anything
+ */
+ ret = 0;
+ goto err;
+ }
+ /*
+ * the buddy cache inode stores the block bitmap
+ * and buddy information in consecutive blocks.
+ * So for each group we need two blocks.
+ */
+ block = group * 2;
+ pnum = block / blocks_per_page;
+ poff = block % blocks_per_page;
+ page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
+ if (page) {
+ BUG_ON(page->mapping != inode->i_mapping);
+ ret = ext4_mb_init_cache(page, NULL);
+ if (ret) {
+ unlock_page(page);
+ goto err;
+ }
+ unlock_page(page);
+ }
+ if (page == NULL || !PageUptodate(page)) {
+ ret = -EIO;
+ goto err;
+ }
+ mark_page_accessed(page);
+ bitmap_page = page;
+ bitmap = page_address(page) + (poff * sb->s_blocksize);
+
+ /* init buddy cache */
+ block++;
+ pnum = block / blocks_per_page;
+ poff = block % blocks_per_page;
+ page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
+ if (page == bitmap_page) {
+ /*
+ * If both the bitmap and buddy are in
+ * the same page we don't need to force
+ * init the buddy
+ */
+ unlock_page(page);
+ } else if (page) {
+ BUG_ON(page->mapping != inode->i_mapping);
+ ret = ext4_mb_init_cache(page, bitmap);
+ if (ret) {
+ unlock_page(page);
+ goto err;
+ }
+ unlock_page(page);
+ }
+ if (page == NULL || !PageUptodate(page)) {
+ ret = -EIO;
+ goto err;
+ }
+ mark_page_accessed(page);
+err:
+ ext4_mb_put_buddy_cache_lock(sb, group, num_grp_locked);
+ if (bitmap_page)
+ page_cache_release(bitmap_page);
+ if (page)
+ page_cache_release(page);
+ return ret;
+}
+
static noinline_for_stack int
ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
{
/*
* We search using buddy data only if the order of the request
* is greater than equal to the sbi_s_mb_order2_reqs
- * You can tune it via /proc/fs/ext4/<partition>/order2_req
+ * You can tune it via /sys/fs/ext4/<partition>/mb_order2_req
*/
if (i >= sbi->s_mb_order2_reqs) {
/*
group = 0;
/* quick check to skip empty groups */
- grp = ext4_get_group_info(ac->ac_sb, group);
+ grp = ext4_get_group_info(sb, group);
if (grp->bb_free == 0)
continue;
* we need full data about the group
* to make a good selection
*/
- err = ext4_mb_load_buddy(sb, group, &e4b);
+ err = ext4_mb_init_group(sb, group);
if (err)
goto out;
- ext4_mb_release_desc(&e4b);
}
/*
if (hs->op == EXT4_MB_HISTORY_ALLOC) {
fmt = "%-5u %-8u %-23s %-23s %-23s %-5u %-5u %-2u "
"%-5u %-5s %-5u %-6u\n";
- sprintf(buf2, "%lu/%d/%u@%u", hs->result.fe_group,
+ sprintf(buf2, "%u/%d/%u@%u", hs->result.fe_group,
hs->result.fe_start, hs->result.fe_len,
hs->result.fe_logical);
- sprintf(buf, "%lu/%d/%u@%u", hs->orig.fe_group,
+ sprintf(buf, "%u/%d/%u@%u", hs->orig.fe_group,
hs->orig.fe_start, hs->orig.fe_len,
hs->orig.fe_logical);
- sprintf(buf3, "%lu/%d/%u@%u", hs->goal.fe_group,
+ sprintf(buf3, "%u/%d/%u@%u", hs->goal.fe_group,
hs->goal.fe_start, hs->goal.fe_len,
hs->goal.fe_logical);
seq_printf(seq, fmt, hs->pid, hs->ino, buf, buf3, buf2,
hs->buddy ? 1 << hs->buddy : 0);
} else if (hs->op == EXT4_MB_HISTORY_PREALLOC) {
fmt = "%-5u %-8u %-23s %-23s %-23s\n";
- sprintf(buf2, "%lu/%d/%u@%u", hs->result.fe_group,
+ sprintf(buf2, "%u/%d/%u@%u", hs->result.fe_group,
hs->result.fe_start, hs->result.fe_len,
hs->result.fe_logical);
- sprintf(buf, "%lu/%d/%u@%u", hs->orig.fe_group,
+ sprintf(buf, "%u/%d/%u@%u", hs->orig.fe_group,
hs->orig.fe_start, hs->orig.fe_len,
hs->orig.fe_logical);
seq_printf(seq, fmt, hs->pid, hs->ino, buf, "", buf2);
} else if (hs->op == EXT4_MB_HISTORY_DISCARD) {
- sprintf(buf2, "%lu/%d/%u", hs->result.fe_group,
+ sprintf(buf2, "%u/%d/%u", hs->result.fe_group,
hs->result.fe_start, hs->result.fe_len);
seq_printf(seq, "%-5u %-8u %-23s discard\n",
hs->pid, hs->ino, buf2);
} else if (hs->op == EXT4_MB_HISTORY_FREE) {
- sprintf(buf2, "%lu/%d/%u", hs->result.fe_group,
+ sprintf(buf2, "%u/%d/%u", hs->result.fe_group,
hs->result.fe_start, hs->result.fe_len);
seq_printf(seq, "%-5u %-8u %-23s free\n",
hs->pid, hs->ino, buf2);
return NULL;
group = *pos + 1;
- return (void *) group;
+ return (void *) ((unsigned long) group);
}
static void *ext4_mb_seq_groups_next(struct seq_file *seq, void *v, loff_t *pos)
if (*pos < 0 || *pos >= sbi->s_groups_count)
return NULL;
group = *pos + 1;
- return (void *) group;;
+ return (void *) ((unsigned long) group);
}
static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v)
{
struct super_block *sb = seq->private;
- long group = (long) v;
+ ext4_group_t group = (ext4_group_t) ((unsigned long) v);
int i;
int err;
struct ext4_buddy e4b;
sizeof(struct ext4_group_info);
err = ext4_mb_load_buddy(sb, group, &e4b);
if (err) {
- seq_printf(seq, "#%-5lu: I/O error\n", group);
+ seq_printf(seq, "#%-5u: I/O error\n", group);
return 0;
}
ext4_lock_group(sb, group);
ext4_unlock_group(sb, group);
ext4_mb_release_desc(&e4b);
- seq_printf(seq, "#%-5lu: %-5u %-5u %-5u [", group, sg.info.bb_free,
+ seq_printf(seq, "#%-5u: %-5u %-5u %-5u [", group, sg.info.bb_free,
sg.info.bb_fragments, sg.info.bb_first_free);
for (i = 0; i <= 13; i++)
seq_printf(seq, " %-5u", i <= sb->s_blocksize_bits + 1 ?
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
- remove_proc_entry("mb_groups", sbi->s_mb_proc);
- remove_proc_entry("mb_history", sbi->s_mb_proc);
-
+ if (sbi->s_proc != NULL) {
+ remove_proc_entry("mb_groups", sbi->s_proc);
+ remove_proc_entry("mb_history", sbi->s_proc);
+ }
kfree(sbi->s_mb_history);
}
struct ext4_sb_info *sbi = EXT4_SB(sb);
int i;
- if (sbi->s_mb_proc != NULL) {
- proc_create_data("mb_history", S_IRUGO, sbi->s_mb_proc,
+ if (sbi->s_proc != NULL) {
+ proc_create_data("mb_history", S_IRUGO, sbi->s_proc,
&ext4_mb_seq_history_fops, sb);
- proc_create_data("mb_groups", S_IRUGO, sbi->s_mb_proc,
+ proc_create_data("mb_groups", S_IRUGO, sbi->s_proc,
&ext4_mb_seq_groups_fops, sb);
}
ext4_free_blocks_after_init(sb, group, desc);
} else {
meta_group_info[i]->bb_free =
- le16_to_cpu(desc->bg_free_blocks_count);
+ ext4_free_blks_count(sb, desc);
}
INIT_LIST_HEAD(&meta_group_info[i]->bb_prealloc_list);
+ init_rwsem(&meta_group_info[i]->alloc_sem);
+ meta_group_info[i]->bb_free_root.rb_node = NULL;;
#ifdef DOUBLE_CHECK
{
} /* ext4_mb_add_groupinfo */
/*
- * Add a group to the existing groups.
- * This function is used for online resize
- */
-int ext4_mb_add_more_groupinfo(struct super_block *sb, ext4_group_t group,
- struct ext4_group_desc *desc)
-{
- struct ext4_sb_info *sbi = EXT4_SB(sb);
- struct inode *inode = sbi->s_buddy_cache;
- int blocks_per_page;
- int block;
- int pnum;
- struct page *page;
- int err;
-
- /* Add group based on group descriptor*/
- err = ext4_mb_add_groupinfo(sb, group, desc);
- if (err)
- return err;
-
- /*
- * Cache pages containing dynamic mb_alloc datas (buddy and bitmap
- * datas) are set not up to date so that they will be re-initilaized
- * during the next call to ext4_mb_load_buddy
- */
-
- /* Set buddy page as not up to date */
- blocks_per_page = PAGE_CACHE_SIZE / sb->s_blocksize;
- block = group * 2;
- pnum = block / blocks_per_page;
- page = find_get_page(inode->i_mapping, pnum);
- if (page != NULL) {
- ClearPageUptodate(page);
- page_cache_release(page);
- }
-
- /* Set bitmap page as not up to date */
- block++;
- pnum = block / blocks_per_page;
- page = find_get_page(inode->i_mapping, pnum);
- if (page != NULL) {
- ClearPageUptodate(page);
- page_cache_release(page);
- }
-
- return 0;
-}
-
-/*
* Update an existing group.
* This function is used for online resize
*/
desc = ext4_get_group_desc(sb, i, NULL);
if (desc == NULL) {
printk(KERN_ERR
- "EXT4-fs: can't read descriptor %lu\n", i);
+ "EXT4-fs: can't read descriptor %u\n", i);
goto err_freebuddy;
}
if (ext4_mb_add_groupinfo(sb, i, desc) != 0)
int ext4_mb_init(struct super_block *sb, int needs_recovery)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
- unsigned i;
+ unsigned i, j;
unsigned offset;
unsigned max;
int ret;
- if (!test_opt(sb, MBALLOC))
- return 0;
-
i = (sb->s_blocksize_bits + 2) * sizeof(unsigned short);
sbi->s_mb_offsets = kmalloc(i, GFP_KERNEL);
if (sbi->s_mb_offsets == NULL) {
- clear_opt(sbi->s_mount_opt, MBALLOC);
return -ENOMEM;
}
+
+ i = (sb->s_blocksize_bits + 2) * sizeof(unsigned int);
sbi->s_mb_maxs = kmalloc(i, GFP_KERNEL);
if (sbi->s_mb_maxs == NULL) {
- clear_opt(sbi->s_mount_opt, MBALLOC);
- kfree(sbi->s_mb_maxs);
+ kfree(sbi->s_mb_offsets);
return -ENOMEM;
}
/* init file for buddy data */
ret = ext4_mb_init_backend(sb);
if (ret != 0) {
- clear_opt(sbi->s_mount_opt, MBALLOC);
kfree(sbi->s_mb_offsets);
kfree(sbi->s_mb_maxs);
return ret;
}
spin_lock_init(&sbi->s_md_lock);
- INIT_LIST_HEAD(&sbi->s_active_transaction);
- INIT_LIST_HEAD(&sbi->s_closed_transaction);
- INIT_LIST_HEAD(&sbi->s_committed_transaction);
spin_lock_init(&sbi->s_bal_lock);
sbi->s_mb_max_to_scan = MB_DEFAULT_MAX_TO_SCAN;
sbi->s_mb_history_filter = EXT4_MB_HISTORY_DEFAULT;
sbi->s_mb_group_prealloc = MB_DEFAULT_GROUP_PREALLOC;
- i = sizeof(struct ext4_locality_group) * NR_CPUS;
- sbi->s_locality_groups = kmalloc(i, GFP_KERNEL);
+ sbi->s_locality_groups = alloc_percpu(struct ext4_locality_group);
if (sbi->s_locality_groups == NULL) {
- clear_opt(sbi->s_mount_opt, MBALLOC);
kfree(sbi->s_mb_offsets);
kfree(sbi->s_mb_maxs);
return -ENOMEM;
}
- for (i = 0; i < NR_CPUS; i++) {
+ for_each_possible_cpu(i) {
struct ext4_locality_group *lg;
- lg = &sbi->s_locality_groups[i];
+ lg = per_cpu_ptr(sbi->s_locality_groups, i);
mutex_init(&lg->lg_mutex);
- INIT_LIST_HEAD(&lg->lg_prealloc_list);
+ for (j = 0; j < PREALLOC_TB_SIZE; j++)
+ INIT_LIST_HEAD(&lg->lg_prealloc_list[j]);
spin_lock_init(&lg->lg_prealloc_lock);
}
- ext4_mb_init_per_dev_proc(sb);
ext4_mb_history_init(sb);
- printk("EXT4-fs: mballoc enabled\n");
+ if (sbi->s_journal)
+ sbi->s_journal->j_commit_callback = release_blocks_on_commit;
+
+ printk(KERN_INFO "EXT4-fs: mballoc enabled\n");
return 0;
}
pa = list_entry(cur, struct ext4_prealloc_space, pa_group_list);
list_del(&pa->pa_group_list);
count++;
- kfree(pa);
+ kmem_cache_free(ext4_pspace_cachep, pa);
}
if (count)
mb_debug("mballoc: %u PAs left\n", count);
struct ext4_group_info *grinfo;
struct ext4_sb_info *sbi = EXT4_SB(sb);
- if (!test_opt(sb, MBALLOC))
- return 0;
-
- /* release freed, non-committed blocks */
- spin_lock(&sbi->s_md_lock);
- list_splice_init(&sbi->s_closed_transaction,
- &sbi->s_committed_transaction);
- list_splice_init(&sbi->s_active_transaction,
- &sbi->s_committed_transaction);
- spin_unlock(&sbi->s_md_lock);
- ext4_mb_free_committed_blocks(sb);
-
if (sbi->s_group_info) {
for (i = 0; i < sbi->s_groups_count; i++) {
grinfo = ext4_get_group_info(sb, i);
atomic_read(&sbi->s_mb_discarded));
}
- kfree(sbi->s_locality_groups);
-
+ free_percpu(sbi->s_locality_groups);
ext4_mb_history_release(sb);
- ext4_mb_destroy_per_dev_proc(sb);
return 0;
}
-static noinline_for_stack void
-ext4_mb_free_committed_blocks(struct super_block *sb)
+/*
+ * This function is called by the jbd2 layer once the commit has finished,
+ * so we know we can free the blocks that were released with that commit.
+ */
+static void release_blocks_on_commit(journal_t *journal, transaction_t *txn)
{
- struct ext4_sb_info *sbi = EXT4_SB(sb);
- int err;
- int i;
- int count = 0;
- int count2 = 0;
- struct ext4_free_metadata *md;
+ struct super_block *sb = journal->j_private;
struct ext4_buddy e4b;
+ struct ext4_group_info *db;
+ int err, count = 0, count2 = 0;
+ struct ext4_free_data *entry;
+ ext4_fsblk_t discard_block;
+ struct list_head *l, *ltmp;
- if (list_empty(&sbi->s_committed_transaction))
- return;
-
- /* there is committed blocks to be freed yet */
- do {
- /* get next array of blocks */
- md = NULL;
- spin_lock(&sbi->s_md_lock);
- if (!list_empty(&sbi->s_committed_transaction)) {
- md = list_entry(sbi->s_committed_transaction.next,
- struct ext4_free_metadata, list);
- list_del(&md->list);
- }
- spin_unlock(&sbi->s_md_lock);
-
- if (md == NULL)
- break;
+ list_for_each_safe(l, ltmp, &txn->t_private_list) {
+ entry = list_entry(l, struct ext4_free_data, list);
- mb_debug("gonna free %u blocks in group %lu (0x%p):",
- md->num, md->group, md);
+ mb_debug("gonna free %u blocks in group %u (0x%p):",
+ entry->count, entry->group, entry);
- err = ext4_mb_load_buddy(sb, md->group, &e4b);
+ err = ext4_mb_load_buddy(sb, entry->group, &e4b);
/* we expect to find existing buddy because it's pinned */
BUG_ON(err != 0);
+ db = e4b.bd_info;
/* there are blocks to put in buddy to make them really free */
- count += md->num;
+ count += entry->count;
count2++;
- ext4_lock_group(sb, md->group);
- for (i = 0; i < md->num; i++) {
- mb_debug(" %u", md->blocks[i]);
- mb_free_blocks(NULL, &e4b, md->blocks[i], 1);
+ ext4_lock_group(sb, entry->group);
+ /* Take it out of per group rb tree */
+ rb_erase(&entry->node, &(db->bb_free_root));
+ mb_free_blocks(NULL, &e4b, entry->start_blk, entry->count);
+
+ if (!db->bb_free_root.rb_node) {
+ /* No more items in the per group rb tree
+ * balance refcounts from ext4_mb_free_metadata()
+ */
+ page_cache_release(e4b.bd_buddy_page);
+ page_cache_release(e4b.bd_bitmap_page);
}
- mb_debug("\n");
- ext4_unlock_group(sb, md->group);
-
- /* balance refcounts from ext4_mb_free_metadata() */
- page_cache_release(e4b.bd_buddy_page);
- page_cache_release(e4b.bd_bitmap_page);
-
- kfree(md);
+ ext4_unlock_group(sb, entry->group);
+ discard_block = (ext4_fsblk_t) entry->group * EXT4_BLOCKS_PER_GROUP(sb)
+ + entry->start_blk
+ + le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
+ trace_mark(ext4_discard_blocks, "dev %s blk %llu count %u",
+ sb->s_id, (unsigned long long) discard_block,
+ entry->count);
+ sb_issue_discard(sb, discard_block, entry->count);
+
+ kmem_cache_free(ext4_free_ext_cachep, entry);
ext4_mb_release_desc(&e4b);
-
- } while (md);
-
- mb_debug("freed %u blocks in %u structures\n", count, count2);
-}
-
-#define EXT4_MB_STATS_NAME "stats"
-#define EXT4_MB_MAX_TO_SCAN_NAME "max_to_scan"
-#define EXT4_MB_MIN_TO_SCAN_NAME "min_to_scan"
-#define EXT4_MB_ORDER2_REQ "order2_req"
-#define EXT4_MB_STREAM_REQ "stream_req"
-#define EXT4_MB_GROUP_PREALLOC "group_prealloc"
-
-
-
-#define MB_PROC_FOPS(name) \
-static int ext4_mb_##name##_proc_show(struct seq_file *m, void *v) \
-{ \
- struct ext4_sb_info *sbi = m->private; \
- \
- seq_printf(m, "%ld\n", sbi->s_mb_##name); \
- return 0; \
-} \
- \
-static int ext4_mb_##name##_proc_open(struct inode *inode, struct file *file)\
-{ \
- return single_open(file, ext4_mb_##name##_proc_show, PDE(inode)->data);\
-} \
- \
-static ssize_t ext4_mb_##name##_proc_write(struct file *file, \
- const char __user *buf, size_t cnt, loff_t *ppos) \
-{ \
- struct ext4_sb_info *sbi = PDE(file->f_path.dentry->d_inode)->data;\
- char str[32]; \
- long value; \
- if (cnt >= sizeof(str)) \
- return -EINVAL; \
- if (copy_from_user(str, buf, cnt)) \
- return -EFAULT; \
- value = simple_strtol(str, NULL, 0); \
- if (value <= 0) \
- return -ERANGE; \
- sbi->s_mb_##name = value; \
- return cnt; \
-} \
- \
-static const struct file_operations ext4_mb_##name##_proc_fops = { \
- .owner = THIS_MODULE, \
- .open = ext4_mb_##name##_proc_open, \
- .read = seq_read, \
- .llseek = seq_lseek, \
- .release = single_release, \
- .write = ext4_mb_##name##_proc_write, \
-};
-
-MB_PROC_FOPS(stats);
-MB_PROC_FOPS(max_to_scan);
-MB_PROC_FOPS(min_to_scan);
-MB_PROC_FOPS(order2_reqs);
-MB_PROC_FOPS(stream_request);
-MB_PROC_FOPS(group_prealloc);
-
-#define MB_PROC_HANDLER(name, var) \
-do { \
- proc = proc_create_data(name, mode, sbi->s_mb_proc, \
- &ext4_mb_##var##_proc_fops, sbi); \
- if (proc == NULL) { \
- printk(KERN_ERR "EXT4-fs: can't to create %s\n", name); \
- goto err_out; \
- } \
-} while (0)
-
-static int ext4_mb_init_per_dev_proc(struct super_block *sb)
-{
- mode_t mode = S_IFREG | S_IRUGO | S_IWUSR;
- struct ext4_sb_info *sbi = EXT4_SB(sb);
- struct proc_dir_entry *proc;
- char devname[64];
-
- if (proc_root_ext4 == NULL) {
- sbi->s_mb_proc = NULL;
- return -EINVAL;
}
- bdevname(sb->s_bdev, devname);
- sbi->s_mb_proc = proc_mkdir(devname, proc_root_ext4);
-
- MB_PROC_HANDLER(EXT4_MB_STATS_NAME, stats);
- MB_PROC_HANDLER(EXT4_MB_MAX_TO_SCAN_NAME, max_to_scan);
- MB_PROC_HANDLER(EXT4_MB_MIN_TO_SCAN_NAME, min_to_scan);
- MB_PROC_HANDLER(EXT4_MB_ORDER2_REQ, order2_reqs);
- MB_PROC_HANDLER(EXT4_MB_STREAM_REQ, stream_request);
- MB_PROC_HANDLER(EXT4_MB_GROUP_PREALLOC, group_prealloc);
-
- return 0;
-
-err_out:
- printk(KERN_ERR "EXT4-fs: Unable to create %s\n", devname);
- remove_proc_entry(EXT4_MB_GROUP_PREALLOC, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_STREAM_REQ, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_ORDER2_REQ, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_MIN_TO_SCAN_NAME, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_MAX_TO_SCAN_NAME, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_STATS_NAME, sbi->s_mb_proc);
- remove_proc_entry(devname, proc_root_ext4);
- sbi->s_mb_proc = NULL;
-
- return -ENOMEM;
-}
-
-static int ext4_mb_destroy_per_dev_proc(struct super_block *sb)
-{
- struct ext4_sb_info *sbi = EXT4_SB(sb);
- char devname[64];
- if (sbi->s_mb_proc == NULL)
- return -EINVAL;
-
- bdevname(sb->s_bdev, devname);
- remove_proc_entry(EXT4_MB_GROUP_PREALLOC, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_STREAM_REQ, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_ORDER2_REQ, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_MIN_TO_SCAN_NAME, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_MAX_TO_SCAN_NAME, sbi->s_mb_proc);
- remove_proc_entry(EXT4_MB_STATS_NAME, sbi->s_mb_proc);
- remove_proc_entry(devname, proc_root_ext4);
-
- return 0;
+ mb_debug("freed %u blocks in %u structures\n", count, count2);
}
int __init init_ext4_mballoc(void)
kmem_cache_destroy(ext4_pspace_cachep);
return -ENOMEM;
}
-#ifdef CONFIG_PROC_FS
- proc_root_ext4 = proc_mkdir("fs/ext4", NULL);
- if (proc_root_ext4 == NULL)
- printk(KERN_ERR "EXT4-fs: Unable to create fs/ext4\n");
-#endif
+
+ ext4_free_ext_cachep =
+ kmem_cache_create("ext4_free_block_extents",
+ sizeof(struct ext4_free_data),
+ 0, SLAB_RECLAIM_ACCOUNT, NULL);
+ if (ext4_free_ext_cachep == NULL) {
+ kmem_cache_destroy(ext4_pspace_cachep);
+ kmem_cache_destroy(ext4_ac_cachep);
+ return -ENOMEM;
+ }
return 0;
}
/* XXX: synchronize_rcu(); */
kmem_cache_destroy(ext4_pspace_cachep);
kmem_cache_destroy(ext4_ac_cachep);
-#ifdef CONFIG_PROC_FS
- remove_proc_entry("fs/ext4", NULL);
-#endif
+ kmem_cache_destroy(ext4_free_ext_cachep);
}
*/
static noinline_for_stack int
ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
- handle_t *handle)
+ handle_t *handle, unsigned int reserv_blks)
{
struct buffer_head *bitmap_bh = NULL;
struct ext4_super_block *es;
if (!gdp)
goto out_err;
- ext4_debug("using block group %lu(%d)\n", ac->ac_b_ex.fe_group,
- gdp->bg_free_blocks_count);
+ ext4_debug("using block group %u(%d)\n", ac->ac_b_ex.fe_group,
+ ext4_free_blks_count(sb, gdp));
err = ext4_journal_get_write_access(handle, gdp_bh);
if (err)
in_range(block + len - 1, ext4_inode_table(sb, gdp),
EXT4_SB(sb)->s_itb_per_group)) {
ext4_error(sb, __func__,
- "Allocating block in system zone - block = %llu",
- block);
+ "Allocating block %llu in system zone of %d group\n",
+ block, ac->ac_b_ex.fe_group);
/* File system mounted not to panic on error
* Fix the bitmap and repeat the block allocation
* We leak some of the blocks here.
mb_set_bits(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group),
bitmap_bh->b_data, ac->ac_b_ex.fe_start,
ac->ac_b_ex.fe_len);
- err = ext4_journal_dirty_metadata(handle, bitmap_bh);
+ err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh);
if (!err)
err = -EAGAIN;
goto out_err;
}
}
#endif
- mb_set_bits(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group), bitmap_bh->b_data,
- ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len);
-
spin_lock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
+ mb_set_bits(NULL, bitmap_bh->b_data,
+ ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len);
if (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) {
gdp->bg_flags &= cpu_to_le16(~EXT4_BG_BLOCK_UNINIT);
- gdp->bg_free_blocks_count =
- cpu_to_le16(ext4_free_blocks_after_init(sb,
- ac->ac_b_ex.fe_group,
- gdp));
+ ext4_free_blks_set(sb, gdp,
+ ext4_free_blocks_after_init(sb,
+ ac->ac_b_ex.fe_group, gdp));
}
- le16_add_cpu(&gdp->bg_free_blocks_count, -ac->ac_b_ex.fe_len);
+ len = ext4_free_blks_count(sb, gdp) - ac->ac_b_ex.fe_len;
+ ext4_free_blks_set(sb, gdp, len);
gdp->bg_checksum = ext4_group_desc_csum(sbi, ac->ac_b_ex.fe_group, gdp);
spin_unlock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
percpu_counter_sub(&sbi->s_freeblocks_counter, ac->ac_b_ex.fe_len);
+ /*
+ * Now reduce the dirty block count also. Should not go negative
+ */
+ if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED))
+ /* release all the reserved blocks if non delalloc */
+ percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks);
+ else {
+ percpu_counter_sub(&sbi->s_dirtyblocks_counter,
+ ac->ac_b_ex.fe_len);
+ /* convert reserved quota blocks to real quota blocks */
+ vfs_dq_claim_block(ac->ac_inode, ac->ac_b_ex.fe_len);
+ }
if (sbi->s_log_groups_per_flex) {
ext4_group_t flex_group = ext4_flex_group(sbi,
ac->ac_b_ex.fe_group);
- spin_lock(sb_bgl_lock(sbi, flex_group));
- sbi->s_flex_groups[flex_group].free_blocks -= ac->ac_b_ex.fe_len;
- spin_unlock(sb_bgl_lock(sbi, flex_group));
+ atomic_sub(ac->ac_b_ex.fe_len,
+ &sbi->s_flex_groups[flex_group].free_blocks);
}
- err = ext4_journal_dirty_metadata(handle, bitmap_bh);
+ err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh);
if (err)
goto out_err;
- err = ext4_journal_dirty_metadata(handle, gdp_bh);
+ err = ext4_handle_dirty_metadata(handle, NULL, gdp_bh);
out_err:
sb->s_dirt = 1;
* here we normalize request for locality group
* Group request are normalized to s_strip size if we set the same via mount
* option. If not we set it to s_mb_group_prealloc which can be configured via
- * /proc/fs/ext4/<partition>/group_prealloc
+ * /sys/fs/ext4/<partition>/mb_group_prealloc
*
* XXX: should we try to preallocate more than the group has now?
*/
/* check we don't cross already preallocated blocks */
rcu_read_lock();
list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) {
- unsigned long pa_end;
+ ext4_lblk_t pa_end;
if (pa->pa_deleted)
continue;
/* XXX: extra loop to check we really don't overlap preallocations */
rcu_read_lock();
list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) {
- unsigned long pa_end;
+ ext4_lblk_t pa_end;
spin_lock(&pa->pa_lock);
if (pa->pa_deleted == 0) {
pa_end = pa->pa_lstart + pa->pa_len;
}
BUG_ON(start + size <= ac->ac_o_ex.fe_logical &&
start > ac->ac_o_ex.fe_logical);
- BUG_ON(size <= 0 || size >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
+ BUG_ON(size <= 0 || size > EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
/* now prepare goal request */
struct ext4_prealloc_space *pa)
{
unsigned int len = ac->ac_o_ex.fe_len;
+
ext4_get_group_no_and_offset(ac->ac_sb, pa->pa_pstart,
&ac->ac_b_ex.fe_group,
&ac->ac_b_ex.fe_start);
}
/*
+ * Return the prealloc space that have minimal distance
+ * from the goal block. @cpa is the prealloc
+ * space that is having currently known minimal distance
+ * from the goal block.
+ */
+static struct ext4_prealloc_space *
+ext4_mb_check_group_pa(ext4_fsblk_t goal_block,
+ struct ext4_prealloc_space *pa,
+ struct ext4_prealloc_space *cpa)
+{
+ ext4_fsblk_t cur_distance, new_distance;
+
+ if (cpa == NULL) {
+ atomic_inc(&pa->pa_count);
+ return pa;
+ }
+ cur_distance = abs(goal_block - cpa->pa_pstart);
+ new_distance = abs(goal_block - pa->pa_pstart);
+
+ if (cur_distance < new_distance)
+ return cpa;
+
+ /* drop the previous reference */
+ atomic_dec(&cpa->pa_count);
+ atomic_inc(&pa->pa_count);
+ return pa;
+}
+
+/*
* search goal blocks in preallocated space
*/
static noinline_for_stack int
ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
{
+ int order, i;
struct ext4_inode_info *ei = EXT4_I(ac->ac_inode);
struct ext4_locality_group *lg;
- struct ext4_prealloc_space *pa;
+ struct ext4_prealloc_space *pa, *cpa = NULL;
+ ext4_fsblk_t goal_block;
/* only data can be preallocated */
if (!(ac->ac_flags & EXT4_MB_HINT_DATA))
lg = ac->ac_lg;
if (lg == NULL)
return 0;
+ order = fls(ac->ac_o_ex.fe_len) - 1;
+ if (order > PREALLOC_TB_SIZE - 1)
+ /* The max size of hash table is PREALLOC_TB_SIZE */
+ order = PREALLOC_TB_SIZE - 1;
+
+ goal_block = ac->ac_g_ex.fe_group * EXT4_BLOCKS_PER_GROUP(ac->ac_sb) +
+ ac->ac_g_ex.fe_start +
+ le32_to_cpu(EXT4_SB(ac->ac_sb)->s_es->s_first_data_block);
+ /*
+ * search for the prealloc space that is having
+ * minimal distance from the goal block.
+ */
+ for (i = order; i < PREALLOC_TB_SIZE; i++) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(pa, &lg->lg_prealloc_list[i],
+ pa_inode_list) {
+ spin_lock(&pa->pa_lock);
+ if (pa->pa_deleted == 0 &&
+ pa->pa_free >= ac->ac_o_ex.fe_len) {
- rcu_read_lock();
- list_for_each_entry_rcu(pa, &lg->lg_prealloc_list, pa_inode_list) {
- spin_lock(&pa->pa_lock);
- if (pa->pa_deleted == 0 && pa->pa_free >= ac->ac_o_ex.fe_len) {
- atomic_inc(&pa->pa_count);
- ext4_mb_use_group_pa(ac, pa);
+ cpa = ext4_mb_check_group_pa(goal_block,
+ pa, cpa);
+ }
spin_unlock(&pa->pa_lock);
- ac->ac_criteria = 20;
- rcu_read_unlock();
- return 1;
}
- spin_unlock(&pa->pa_lock);
+ rcu_read_unlock();
+ }
+ if (cpa) {
+ ext4_mb_use_group_pa(ac, cpa);
+ ac->ac_criteria = 20;
+ return 1;
}
- rcu_read_unlock();
-
return 0;
}
/*
+ * the function goes through all block freed in the group
+ * but not yet committed and marks them used in in-core bitmap.
+ * buddy must be generated from this bitmap
+ * Need to be called with ext4 group lock (ext4_lock_group)
+ */
+static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap,
+ ext4_group_t group)
+{
+ struct rb_node *n;
+ struct ext4_group_info *grp;
+ struct ext4_free_data *entry;
+
+ grp = ext4_get_group_info(sb, group);
+ n = rb_first(&(grp->bb_free_root));
+
+ while (n) {
+ entry = rb_entry(n, struct ext4_free_data, node);
+ mb_set_bits(sb_bgl_lock(EXT4_SB(sb), group),
+ bitmap, entry->start_blk,
+ entry->count);
+ n = rb_next(n);
+ }
+ return;
+}
+
+/*
* the function goes through all preallocation in this group and marks them
* used in in-core bitmap. buddy must be generated from this bitmap
* Need to be called with ext4 group lock (ext4_lock_group)
preallocated += len;
count++;
}
- mb_debug("prellocated %u for group %lu\n", preallocated, group);
+ mb_debug("prellocated %u for group %u\n", preallocated, group);
}
static void ext4_mb_pa_callback(struct rcu_head *head)
static void ext4_mb_put_pa(struct ext4_allocation_context *ac,
struct super_block *sb, struct ext4_prealloc_space *pa)
{
- unsigned long grp;
+ ext4_group_t grp;
+ ext4_fsblk_t grp_blk;
if (!atomic_dec_and_test(&pa->pa_count) || pa->pa_free != 0)
return;
pa->pa_deleted = 1;
spin_unlock(&pa->pa_lock);
- /* -1 is to protect from crossing allocation group */
- ext4_get_group_no_and_offset(sb, pa->pa_pstart - 1, &grp, NULL);
+ grp_blk = pa->pa_pstart;
+ /*
+ * If doing group-based preallocation, pa_pstart may be in the
+ * next group when pa is used up
+ */
+ if (pa->pa_type == MB_GROUP_PA)
+ grp_blk--;
+
+ ext4_get_group_no_and_offset(sb, grp_blk, &grp, NULL);
/*
* possible race:
pa->pa_free = pa->pa_len;
atomic_set(&pa->pa_count, 1);
spin_lock_init(&pa->pa_lock);
+ INIT_LIST_HEAD(&pa->pa_inode_list);
+ INIT_LIST_HEAD(&pa->pa_group_list);
pa->pa_deleted = 0;
- pa->pa_linear = 0;
+ pa->pa_type = MB_INODE_PA;
mb_debug("new inode pa %p: %llu/%u for %u\n", pa,
pa->pa_pstart, pa->pa_len, pa->pa_lstart);
+ trace_mark(ext4_mb_new_inode_pa,
+ "dev %s ino %lu pstart %llu len %u lstart %u",
+ sb->s_id, ac->ac_inode->i_ino,
+ pa->pa_pstart, pa->pa_len, pa->pa_lstart);
ext4_mb_use_inode_pa(ac, pa);
atomic_add(pa->pa_free, &EXT4_SB(sb)->s_mb_preallocated);
pa->pa_free = pa->pa_len;
atomic_set(&pa->pa_count, 1);
spin_lock_init(&pa->pa_lock);
+ INIT_LIST_HEAD(&pa->pa_inode_list);
+ INIT_LIST_HEAD(&pa->pa_group_list);
pa->pa_deleted = 0;
- pa->pa_linear = 1;
+ pa->pa_type = MB_GROUP_PA;
mb_debug("new group pa %p: %llu/%u for %u\n", pa,
- pa->pa_pstart, pa->pa_len, pa->pa_lstart);
+ pa->pa_pstart, pa->pa_len, pa->pa_lstart);
+ trace_mark(ext4_mb_new_group_pa, "dev %s pstart %llu len %u lstart %u",
+ sb->s_id, pa->pa_pstart, pa->pa_len, pa->pa_lstart);
ext4_mb_use_group_pa(ac, pa);
atomic_add(pa->pa_free, &EXT4_SB(sb)->s_mb_preallocated);
list_add(&pa->pa_group_list, &grp->bb_prealloc_list);
ext4_unlock_group(sb, ac->ac_b_ex.fe_group);
- spin_lock(pa->pa_obj_lock);
- list_add_tail_rcu(&pa->pa_inode_list, &lg->lg_prealloc_list);
- spin_unlock(pa->pa_obj_lock);
-
+ /*
+ * We will later add the new pa to the right bucket
+ * after updating the pa_free in ext4_mb_release_context
+ */
return 0;
}
{
struct super_block *sb = e4b->bd_sb;
struct ext4_sb_info *sbi = EXT4_SB(sb);
- unsigned long end;
- unsigned long next;
+ unsigned int end;
+ unsigned int next;
ext4_group_t group;
ext4_grpblk_t bit;
+ unsigned long long grp_blk_start;
sector_t start;
int err = 0;
int free = 0;
BUG_ON(pa->pa_deleted == 0);
ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
+ grp_blk_start = pa->pa_pstart - bit;
BUG_ON(group != e4b->bd_group && pa->pa_len != 0);
end = bit + pa->pa_len;
ext4_mb_store_history(ac);
}
+ trace_mark(ext4_mb_release_inode_pa,
+ "dev %s ino %lu block %llu count %u",
+ sb->s_id, pa->pa_inode->i_ino, grp_blk_start + bit,
+ next - bit);
mb_free_blocks(pa->pa_inode, e4b, bit, next - bit);
bit = next + 1;
}
pa, (unsigned long) pa->pa_lstart,
(unsigned long) pa->pa_pstart,
(unsigned long) pa->pa_len);
- ext4_error(sb, __func__, "free %u, pa_free %u\n",
- free, pa->pa_free);
+ ext4_grp_locked_error(sb, group,
+ __func__, "free %u, pa_free %u",
+ free, pa->pa_free);
/*
* pa is already deleted so we use the value obtained
* from the bitmap and continue.
if (ac)
ac->ac_op = EXT4_MB_HISTORY_DISCARD;
+ trace_mark(ext4_mb_release_group_pa, "dev %s pstart %llu len %d",
+ sb->s_id, pa->pa_pstart, pa->pa_len);
BUG_ON(pa->pa_deleted == 0);
ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
BUG_ON(group != e4b->bd_group && pa->pa_len != 0);
int busy = 0;
int free = 0;
- mb_debug("discard preallocation for group %lu\n", group);
+ mb_debug("discard preallocation for group %u\n", group);
if (list_empty(&grp->bb_prealloc_list))
return 0;
bitmap_bh = ext4_read_block_bitmap(sb, group);
if (bitmap_bh == NULL) {
- /* error handling here */
- ext4_mb_release_desc(&e4b);
- BUG_ON(bitmap_bh == NULL);
+ ext4_error(sb, __func__, "Error in reading block "
+ "bitmap for %u", group);
+ return 0;
}
err = ext4_mb_load_buddy(sb, group, &e4b);
- BUG_ON(err != 0); /* error handling here */
+ if (err) {
+ ext4_error(sb, __func__, "Error in loading buddy "
+ "information for %u", group);
+ put_bh(bitmap_bh);
+ return 0;
+ }
if (needed == 0)
needed = EXT4_BLOCKS_PER_GROUP(sb) + 1;
- grp = ext4_get_group_info(sb, group);
INIT_LIST_HEAD(&list);
-
ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
repeat:
ext4_lock_group(sb, group);
list_del_rcu(&pa->pa_inode_list);
spin_unlock(pa->pa_obj_lock);
- if (pa->pa_linear)
+ if (pa->pa_type == MB_GROUP_PA)
ext4_mb_release_group_pa(&e4b, pa, ac);
else
ext4_mb_release_inode_pa(&e4b, bitmap_bh, pa, ac);
*
* FIXME!! Make sure it is valid at all the call sites
*/
-void ext4_mb_discard_inode_preallocations(struct inode *inode)
+void ext4_discard_preallocations(struct inode *inode)
{
struct ext4_inode_info *ei = EXT4_I(inode);
struct super_block *sb = inode->i_sb;
struct ext4_buddy e4b;
int err;
- if (!test_opt(sb, MBALLOC) || !S_ISREG(inode->i_mode)) {
+ if (!S_ISREG(inode->i_mode)) {
/*BUG_ON(!list_empty(&ei->i_prealloc_list));*/
return;
}
mb_debug("discard preallocation for inode %lu\n", inode->i_ino);
+ trace_mark(ext4_discard_preallocations, "dev %s ino %lu", sb->s_id,
+ inode->i_ino);
INIT_LIST_HEAD(&list);
spin_unlock(&ei->i_prealloc_lock);
list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) {
- BUG_ON(pa->pa_linear != 0);
+ BUG_ON(pa->pa_type != MB_INODE_PA);
ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, NULL);
err = ext4_mb_load_buddy(sb, group, &e4b);
- BUG_ON(err != 0); /* error handling here */
+ if (err) {
+ ext4_error(sb, __func__, "Error in loading buddy "
+ "information for %u", group);
+ continue;
+ }
bitmap_bh = ext4_read_block_bitmap(sb, group);
if (bitmap_bh == NULL) {
- /* error handling here */
+ ext4_error(sb, __func__, "Error in reading block "
+ "bitmap for %u", group);
ext4_mb_release_desc(&e4b);
- BUG_ON(bitmap_bh == NULL);
+ continue;
}
ext4_lock_group(sb, group);
* file is determined by the current size or the resulting size after
* allocation which ever is larger
*
- * One can tune this size via /proc/fs/ext4/<partition>/stream_req
+ * One can tune this size via /sys/fs/ext4/<partition>/mb_stream_req
*/
static void ext4_mb_group_or_file(struct ext4_allocation_context *ac)
{
* per cpu locality group is to reduce the contention between block
* request from multiple CPUs.
*/
- ac->ac_lg = &sbi->s_locality_groups[get_cpu()];
- put_cpu();
+ ac->ac_lg = per_cpu_ptr(sbi->s_locality_groups, raw_smp_processor_id());
/* we're going to use group allocation */
ac->ac_flags |= EXT4_MB_HINT_GROUP_ALLOC;
struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_super_block *es = sbi->s_es;
ext4_group_t group;
- unsigned long len;
- unsigned long goal;
+ unsigned int len;
+ ext4_fsblk_t goal;
ext4_grpblk_t block;
/* we can't allocate > group size */
ac->ac_pa = NULL;
ac->ac_bitmap_page = NULL;
ac->ac_buddy_page = NULL;
+ ac->alloc_semp = NULL;
ac->ac_lg = NULL;
/* we have to define context: we'll we work with a file or
}
+static noinline_for_stack void
+ext4_mb_discard_lg_preallocations(struct super_block *sb,
+ struct ext4_locality_group *lg,
+ int order, int total_entries)
+{
+ ext4_group_t group = 0;
+ struct ext4_buddy e4b;
+ struct list_head discard_list;
+ struct ext4_prealloc_space *pa, *tmp;
+ struct ext4_allocation_context *ac;
+
+ mb_debug("discard locality group preallocation\n");
+
+ INIT_LIST_HEAD(&discard_list);
+ ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
+
+ spin_lock(&lg->lg_prealloc_lock);
+ list_for_each_entry_rcu(pa, &lg->lg_prealloc_list[order],
+ pa_inode_list) {
+ spin_lock(&pa->pa_lock);
+ if (atomic_read(&pa->pa_count)) {
+ /*
+ * This is the pa that we just used
+ * for block allocation. So don't
+ * free that
+ */
+ spin_unlock(&pa->pa_lock);
+ continue;
+ }
+ if (pa->pa_deleted) {
+ spin_unlock(&pa->pa_lock);
+ continue;
+ }
+ /* only lg prealloc space */
+ BUG_ON(pa->pa_type != MB_GROUP_PA);
+
+ /* seems this one can be freed ... */
+ pa->pa_deleted = 1;
+ spin_unlock(&pa->pa_lock);
+
+ list_del_rcu(&pa->pa_inode_list);
+ list_add(&pa->u.pa_tmp_list, &discard_list);
+
+ total_entries--;
+ if (total_entries <= 5) {
+ /*
+ * we want to keep only 5 entries
+ * allowing it to grow to 8. This
+ * mak sure we don't call discard
+ * soon for this list.
+ */
+ break;
+ }
+ }
+ spin_unlock(&lg->lg_prealloc_lock);
+
+ list_for_each_entry_safe(pa, tmp, &discard_list, u.pa_tmp_list) {
+
+ ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, NULL);
+ if (ext4_mb_load_buddy(sb, group, &e4b)) {
+ ext4_error(sb, __func__, "Error in loading buddy "
+ "information for %u", group);
+ continue;
+ }
+ ext4_lock_group(sb, group);
+ list_del(&pa->pa_group_list);
+ ext4_mb_release_group_pa(&e4b, pa, ac);
+ ext4_unlock_group(sb, group);
+
+ ext4_mb_release_desc(&e4b);
+ list_del(&pa->u.pa_tmp_list);
+ call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback);
+ }
+ if (ac)
+ kmem_cache_free(ext4_ac_cachep, ac);
+}
+
+/*
+ * We have incremented pa_count. So it cannot be freed at this
+ * point. Also we hold lg_mutex. So no parallel allocation is
+ * possible from this lg. That means pa_free cannot be updated.
+ *
+ * A parallel ext4_mb_discard_group_preallocations is possible.
+ * which can cause the lg_prealloc_list to be updated.
+ */
+
+static void ext4_mb_add_n_trim(struct ext4_allocation_context *ac)
+{
+ int order, added = 0, lg_prealloc_count = 1;
+ struct super_block *sb = ac->ac_sb;
+ struct ext4_locality_group *lg = ac->ac_lg;
+ struct ext4_prealloc_space *tmp_pa, *pa = ac->ac_pa;
+
+ order = fls(pa->pa_free) - 1;
+ if (order > PREALLOC_TB_SIZE - 1)
+ /* The max size of hash table is PREALLOC_TB_SIZE */
+ order = PREALLOC_TB_SIZE - 1;
+ /* Add the prealloc space to lg */
+ rcu_read_lock();
+ list_for_each_entry_rcu(tmp_pa, &lg->lg_prealloc_list[order],
+ pa_inode_list) {
+ spin_lock(&tmp_pa->pa_lock);
+ if (tmp_pa->pa_deleted) {
+ spin_unlock(&tmp_pa->pa_lock);
+ continue;
+ }
+ if (!added && pa->pa_free < tmp_pa->pa_free) {
+ /* Add to the tail of the previous entry */
+ list_add_tail_rcu(&pa->pa_inode_list,
+ &tmp_pa->pa_inode_list);
+ added = 1;
+ /*
+ * we want to count the total
+ * number of entries in the list
+ */
+ }
+ spin_unlock(&tmp_pa->pa_lock);
+ lg_prealloc_count++;
+ }
+ if (!added)
+ list_add_tail_rcu(&pa->pa_inode_list,
+ &lg->lg_prealloc_list[order]);
+ rcu_read_unlock();
+
+ /* Now trim the list to be not more than 8 elements */
+ if (lg_prealloc_count > 8) {
+ ext4_mb_discard_lg_preallocations(sb, lg,
+ order, lg_prealloc_count);
+ return;
+ }
+ return ;
+}
+
/*
* release all resource we used in allocation
*/
static int ext4_mb_release_context(struct ext4_allocation_context *ac)
{
- if (ac->ac_pa) {
- if (ac->ac_pa->pa_linear) {
+ struct ext4_prealloc_space *pa = ac->ac_pa;
+ if (pa) {
+ if (pa->pa_type == MB_GROUP_PA) {
/* see comment in ext4_mb_use_group_pa() */
- spin_lock(&ac->ac_pa->pa_lock);
- ac->ac_pa->pa_pstart += ac->ac_b_ex.fe_len;
- ac->ac_pa->pa_lstart += ac->ac_b_ex.fe_len;
- ac->ac_pa->pa_free -= ac->ac_b_ex.fe_len;
- ac->ac_pa->pa_len -= ac->ac_b_ex.fe_len;
- spin_unlock(&ac->ac_pa->pa_lock);
+ spin_lock(&pa->pa_lock);
+ pa->pa_pstart += ac->ac_b_ex.fe_len;
+ pa->pa_lstart += ac->ac_b_ex.fe_len;
+ pa->pa_free -= ac->ac_b_ex.fe_len;
+ pa->pa_len -= ac->ac_b_ex.fe_len;
+ spin_unlock(&pa->pa_lock);
}
- ext4_mb_put_pa(ac, ac->ac_sb, ac->ac_pa);
+ }
+ if (ac->alloc_semp)
+ up_read(ac->alloc_semp);
+ if (pa) {
+ /*
+ * We want to add the pa to the right bucket.
+ * Remove it from the list and while adding
+ * make sure the list to which we are adding
+ * doesn't grow big. We need to release
+ * alloc_semp before calling ext4_mb_add_n_trim()
+ */
+ if ((pa->pa_type == MB_GROUP_PA) && likely(pa->pa_free)) {
+ spin_lock(pa->pa_obj_lock);
+ list_del_rcu(&pa->pa_inode_list);
+ spin_unlock(pa->pa_obj_lock);
+ ext4_mb_add_n_trim(ac);
+ }
+ ext4_mb_put_pa(ac, ac->ac_sb, pa);
}
if (ac->ac_bitmap_page)
page_cache_release(ac->ac_bitmap_page);
int ret;
int freed = 0;
+ trace_mark(ext4_mb_discard_preallocations, "dev %s needed %d",
+ sb->s_id, needed);
for (i = 0; i < EXT4_SB(sb)->s_groups_count && needed > 0; i++) {
ret = ext4_mb_discard_group_preallocations(sb, i, needed);
freed += ret;
ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
struct ext4_allocation_request *ar, int *errp)
{
+ int freed;
struct ext4_allocation_context *ac = NULL;
struct ext4_sb_info *sbi;
struct super_block *sb;
ext4_fsblk_t block = 0;
- int freed;
- int inquota;
+ unsigned int inquota = 0;
+ unsigned int reserv_blks = 0;
sb = ar->inode->i_sb;
sbi = EXT4_SB(sb);
- if (!test_opt(sb, MBALLOC)) {
- block = ext4_old_new_blocks(handle, ar->inode, ar->goal,
- &(ar->len), errp);
- return block;
- }
- ar->len = ext4_has_free_blocks(sbi, ar->len);
+ trace_mark(ext4_request_blocks, "dev %s flags %u len %u ino %lu "
+ "lblk %llu goal %llu lleft %llu lright %llu "
+ "pleft %llu pright %llu ",
+ sb->s_id, ar->flags, ar->len,
+ ar->inode ? ar->inode->i_ino : 0,
+ (unsigned long long) ar->logical,
+ (unsigned long long) ar->goal,
+ (unsigned long long) ar->lleft,
+ (unsigned long long) ar->lright,
+ (unsigned long long) ar->pleft,
+ (unsigned long long) ar->pright);
- if (ar->len == 0) {
- *errp = -ENOSPC;
- return 0;
- }
-
- while (ar->len && DQUOT_ALLOC_BLOCK(ar->inode, ar->len)) {
- ar->flags |= EXT4_MB_HINT_NOPREALLOC;
- ar->len--;
- }
- if (ar->len == 0) {
- *errp = -EDQUOT;
- return 0;
+ /*
+ * For delayed allocation, we could skip the ENOSPC and
+ * EDQUOT check, as blocks and quotas have been already
+ * reserved when data being copied into pagecache.
+ */
+ if (EXT4_I(ar->inode)->i_delalloc_reserved_flag)
+ ar->flags |= EXT4_MB_DELALLOC_RESERVED;
+ else {
+ /* Without delayed allocation we need to verify
+ * there is enough free blocks to do block allocation
+ * and verify allocation doesn't exceed the quota limits.
+ */
+ while (ar->len && ext4_claim_free_blocks(sbi, ar->len)) {
+ /* let others to free the space */
+ yield();
+ ar->len = ar->len >> 1;
+ }
+ if (!ar->len) {
+ *errp = -ENOSPC;
+ return 0;
+ }
+ reserv_blks = ar->len;
+ while (ar->len && vfs_dq_alloc_block(ar->inode, ar->len)) {
+ ar->flags |= EXT4_MB_HINT_NOPREALLOC;
+ ar->len--;
+ }
+ inquota = ar->len;
+ if (ar->len == 0) {
+ *errp = -EDQUOT;
+ goto out3;
+ }
}
- inquota = ar->len;
ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
if (!ac) {
goto out1;
}
- ext4_mb_poll_new_transaction(sb, handle);
-
*errp = ext4_mb_initialize_context(ac, ar);
if (*errp) {
ar->len = 0;
ac->ac_o_ex.fe_len < ac->ac_b_ex.fe_len)
ext4_mb_new_preallocation(ac);
}
-
if (likely(ac->ac_status == AC_STATUS_FOUND)) {
- *errp = ext4_mb_mark_diskspace_used(ac, handle);
+ *errp = ext4_mb_mark_diskspace_used(ac, handle, reserv_blks);
if (*errp == -EAGAIN) {
+ /*
+ * drop the reference that we took
+ * in ext4_mb_use_best_found
+ */
+ ext4_mb_release_context(ac);
ac->ac_b_ex.fe_group = 0;
ac->ac_b_ex.fe_start = 0;
ac->ac_b_ex.fe_len = 0;
out2:
kmem_cache_free(ext4_ac_cachep, ac);
out1:
- if (ar->len < inquota)
- DQUOT_FREE_BLOCK(ar->inode, inquota - ar->len);
+ if (inquota && ar->len < inquota)
+ vfs_dq_free_block(ar->inode, inquota - ar->len);
+out3:
+ if (!ar->len) {
+ if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag)
+ /* release all the reserved blocks if non delalloc */
+ percpu_counter_sub(&sbi->s_dirtyblocks_counter,
+ reserv_blks);
+ }
+
+ trace_mark(ext4_allocate_blocks,
+ "dev %s block %llu flags %u len %u ino %lu "
+ "logical %llu goal %llu lleft %llu lright %llu "
+ "pleft %llu pright %llu ",
+ sb->s_id, (unsigned long long) block,
+ ar->flags, ar->len, ar->inode ? ar->inode->i_ino : 0,
+ (unsigned long long) ar->logical,
+ (unsigned long long) ar->goal,
+ (unsigned long long) ar->lleft,
+ (unsigned long long) ar->lright,
+ (unsigned long long) ar->pleft,
+ (unsigned long long) ar->pright);
return block;
}
-static void ext4_mb_poll_new_transaction(struct super_block *sb,
- handle_t *handle)
-{
- struct ext4_sb_info *sbi = EXT4_SB(sb);
-
- if (sbi->s_last_transaction == handle->h_transaction->t_tid)
- return;
-
- /* new transaction! time to close last one and free blocks for
- * committed transaction. we know that only transaction can be
- * active, so previos transaction can be being logged and we
- * know that transaction before previous is known to be already
- * logged. this means that now we may free blocks freed in all
- * transactions before previous one. hope I'm clear enough ... */
- spin_lock(&sbi->s_md_lock);
- if (sbi->s_last_transaction != handle->h_transaction->t_tid) {
- mb_debug("new transaction %lu, old %lu\n",
- (unsigned long) handle->h_transaction->t_tid,
- (unsigned long) sbi->s_last_transaction);
- list_splice_init(&sbi->s_closed_transaction,
- &sbi->s_committed_transaction);
- list_splice_init(&sbi->s_active_transaction,
- &sbi->s_closed_transaction);
- sbi->s_last_transaction = handle->h_transaction->t_tid;
- }
- spin_unlock(&sbi->s_md_lock);
-
- ext4_mb_free_committed_blocks(sb);
+/*
+ * We can merge two free data extents only if the physical blocks
+ * are contiguous, AND the extents were freed by the same transaction,
+ * AND the blocks are associated with the same group.
+ */
+static int can_merge(struct ext4_free_data *entry1,
+ struct ext4_free_data *entry2)
+{
+ if ((entry1->t_tid == entry2->t_tid) &&
+ (entry1->group == entry2->group) &&
+ ((entry1->start_blk + entry1->count) == entry2->start_blk))
+ return 1;
+ return 0;
}
static noinline_for_stack int
ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
- ext4_group_t group, ext4_grpblk_t block, int count)
+ struct ext4_free_data *new_entry)
{
+ ext4_grpblk_t block;
+ struct ext4_free_data *entry;
struct ext4_group_info *db = e4b->bd_info;
struct super_block *sb = e4b->bd_sb;
struct ext4_sb_info *sbi = EXT4_SB(sb);
- struct ext4_free_metadata *md;
- int i;
+ struct rb_node **n = &db->bb_free_root.rb_node, *node;
+ struct rb_node *parent = NULL, *new_node;
+ BUG_ON(!ext4_handle_valid(handle));
BUG_ON(e4b->bd_bitmap_page == NULL);
BUG_ON(e4b->bd_buddy_page == NULL);
- ext4_lock_group(sb, group);
- for (i = 0; i < count; i++) {
- md = db->bb_md_cur;
- if (md && db->bb_tid != handle->h_transaction->t_tid) {
- db->bb_md_cur = NULL;
- md = NULL;
+ new_node = &new_entry->node;
+ block = new_entry->start_blk;
+
+ if (!*n) {
+ /* first free block exent. We need to
+ protect buddy cache from being freed,
+ * otherwise we'll refresh it from
+ * on-disk bitmap and lose not-yet-available
+ * blocks */
+ page_cache_get(e4b->bd_buddy_page);
+ page_cache_get(e4b->bd_bitmap_page);
+ }
+ while (*n) {
+ parent = *n;
+ entry = rb_entry(parent, struct ext4_free_data, node);
+ if (block < entry->start_blk)
+ n = &(*n)->rb_left;
+ else if (block >= (entry->start_blk + entry->count))
+ n = &(*n)->rb_right;
+ else {
+ ext4_grp_locked_error(sb, e4b->bd_group, __func__,
+ "Double free of blocks %d (%d %d)",
+ block, entry->start_blk, entry->count);
+ return 0;
}
+ }
- if (md == NULL) {
- ext4_unlock_group(sb, group);
- md = kmalloc(sizeof(*md), GFP_NOFS);
- if (md == NULL)
- return -ENOMEM;
- md->num = 0;
- md->group = group;
-
- ext4_lock_group(sb, group);
- if (db->bb_md_cur == NULL) {
- spin_lock(&sbi->s_md_lock);
- list_add(&md->list, &sbi->s_active_transaction);
- spin_unlock(&sbi->s_md_lock);
- /* protect buddy cache from being freed,
- * otherwise we'll refresh it from
- * on-disk bitmap and lose not-yet-available
- * blocks */
- page_cache_get(e4b->bd_buddy_page);
- page_cache_get(e4b->bd_bitmap_page);
- db->bb_md_cur = md;
- db->bb_tid = handle->h_transaction->t_tid;
- mb_debug("new md 0x%p for group %lu\n",
- md, md->group);
- } else {
- kfree(md);
- md = db->bb_md_cur;
- }
+ rb_link_node(new_node, parent, n);
+ rb_insert_color(new_node, &db->bb_free_root);
+
+ /* Now try to see the extent can be merged to left and right */
+ node = rb_prev(new_node);
+ if (node) {
+ entry = rb_entry(node, struct ext4_free_data, node);
+ if (can_merge(entry, new_entry)) {
+ new_entry->start_blk = entry->start_blk;
+ new_entry->count += entry->count;
+ rb_erase(node, &(db->bb_free_root));
+ spin_lock(&sbi->s_md_lock);
+ list_del(&entry->list);
+ spin_unlock(&sbi->s_md_lock);
+ kmem_cache_free(ext4_free_ext_cachep, entry);
}
+ }
- BUG_ON(md->num >= EXT4_BB_MAX_BLOCKS);
- md->blocks[md->num] = block + i;
- md->num++;
- if (md->num == EXT4_BB_MAX_BLOCKS) {
- /* no more space, put full container on a sb's list */
- db->bb_md_cur = NULL;
+ node = rb_next(new_node);
+ if (node) {
+ entry = rb_entry(node, struct ext4_free_data, node);
+ if (can_merge(new_entry, entry)) {
+ new_entry->count += entry->count;
+ rb_erase(node, &(db->bb_free_root));
+ spin_lock(&sbi->s_md_lock);
+ list_del(&entry->list);
+ spin_unlock(&sbi->s_md_lock);
+ kmem_cache_free(ext4_free_ext_cachep, entry);
}
}
- ext4_unlock_group(sb, group);
+ /* Add the extent to transaction's private list */
+ spin_lock(&sbi->s_md_lock);
+ list_add(&new_entry->list, &handle->h_transaction->t_private_list);
+ spin_unlock(&sbi->s_md_lock);
return 0;
}
struct ext4_allocation_context *ac = NULL;
struct ext4_group_desc *gdp;
struct ext4_super_block *es;
- unsigned long overflow;
+ unsigned int overflow;
ext4_grpblk_t bit;
struct buffer_head *gd_bh;
ext4_group_t block_group;
*freed = 0;
- ext4_mb_poll_new_transaction(sb, handle);
-
sbi = EXT4_SB(sb);
es = EXT4_SB(sb)->s_es;
if (block < le32_to_cpu(es->s_first_data_block) ||
}
ext4_debug("freeing block %lu\n", block);
+ trace_mark(ext4_free_blocks,
+ "dev %s block %llu count %lu metadata %d ino %lu",
+ sb->s_id, (unsigned long long) block, count, metadata,
+ inode ? inode->i_ino : 0);
ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
if (ac) {
count -= overflow;
}
bitmap_bh = ext4_read_block_bitmap(sb, block_group);
- if (!bitmap_bh)
+ if (!bitmap_bh) {
+ err = -EIO;
goto error_return;
+ }
gdp = ext4_get_group_desc(sb, block_group, &gd_bh);
- if (!gdp)
+ if (!gdp) {
+ err = -EIO;
goto error_return;
+ }
if (in_range(ext4_block_bitmap(sb, gdp), block, count) ||
in_range(ext4_inode_bitmap(sb, gdp), block, count) ||
err = ext4_journal_get_write_access(handle, gd_bh);
if (err)
goto error_return;
-
- err = ext4_mb_load_buddy(sb, block_group, &e4b);
- if (err)
- goto error_return;
-
#ifdef AGGRESSIVE_CHECK
{
int i;
BUG_ON(!mb_test_bit(bit + i, bitmap_bh->b_data));
}
#endif
- mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data,
- bit, count);
-
- /* We dirtied the bitmap block */
- BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
- err = ext4_journal_dirty_metadata(handle, bitmap_bh);
-
if (ac) {
ac->ac_b_ex.fe_group = block_group;
ac->ac_b_ex.fe_start = bit;
ext4_mb_store_history(ac);
}
- if (metadata) {
- /* blocks being freed are metadata. these blocks shouldn't
- * be used until this transaction is committed */
- ext4_mb_free_metadata(handle, &e4b, block_group, bit, count);
+ err = ext4_mb_load_buddy(sb, block_group, &e4b);
+ if (err)
+ goto error_return;
+ if (metadata && ext4_handle_valid(handle)) {
+ struct ext4_free_data *new_entry;
+ /*
+ * blocks being freed are metadata. these blocks shouldn't
+ * be used until this transaction is committed
+ */
+ new_entry = kmem_cache_alloc(ext4_free_ext_cachep, GFP_NOFS);
+ new_entry->start_blk = bit;
+ new_entry->group = block_group;
+ new_entry->count = count;
+ new_entry->t_tid = handle->h_transaction->t_tid;
+ ext4_lock_group(sb, block_group);
+ mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data,
+ bit, count);
+ ext4_mb_free_metadata(handle, &e4b, new_entry);
+ ext4_unlock_group(sb, block_group);
} else {
ext4_lock_group(sb, block_group);
+ /* need to update group_info->bb_free and bitmap
+ * with group lock held. generate_buddy look at
+ * them with group lock_held
+ */
+ mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data,
+ bit, count);
mb_free_blocks(inode, &e4b, bit, count);
ext4_mb_return_to_preallocation(inode, &e4b, block, count);
ext4_unlock_group(sb, block_group);
}
spin_lock(sb_bgl_lock(sbi, block_group));
- le16_add_cpu(&gdp->bg_free_blocks_count, count);
+ ret = ext4_free_blks_count(sb, gdp) + count;
+ ext4_free_blks_set(sb, gdp, ret);
gdp->bg_checksum = ext4_group_desc_csum(sbi, block_group, gdp);
spin_unlock(sb_bgl_lock(sbi, block_group));
percpu_counter_add(&sbi->s_freeblocks_counter, count);
if (sbi->s_log_groups_per_flex) {
ext4_group_t flex_group = ext4_flex_group(sbi, block_group);
- spin_lock(sb_bgl_lock(sbi, flex_group));
- sbi->s_flex_groups[flex_group].free_blocks += count;
- spin_unlock(sb_bgl_lock(sbi, flex_group));
+ atomic_add(count, &sbi->s_flex_groups[flex_group].free_blocks);
}
ext4_mb_release_desc(&e4b);
*freed += count;
+ /* We dirtied the bitmap block */
+ BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
+ err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh);
+
/* And the group descriptor block */
BUFFER_TRACE(gd_bh, "dirtied group descriptor block");
- ret = ext4_journal_dirty_metadata(handle, gd_bh);
+ ret = ext4_handle_dirty_metadata(handle, NULL, gd_bh);
if (!err)
err = ret;