From bf21206266d1ca81c4b0fbfc13a39ed3ac9e81cc Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Sun, 28 Sep 2008 18:36:54 -0400 Subject: [PATCH] Add remove-old-balloc patch --- remove-old-balloc | 1955 +++++++++++++++++++++++++++++++++++++++++++++++++++++ series | 1 + 2 files changed, 1956 insertions(+) create mode 100644 remove-old-balloc diff --git a/remove-old-balloc b/remove-old-balloc new file mode 100644 index 00000000..ea4280df --- /dev/null +++ b/remove-old-balloc @@ -0,0 +1,1955 @@ +ext4: Remove old legacy block allocator + +Signed-off-by: "Theodore Ts'o" +diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c +index 02068fa..59566c0 100644 +--- a/fs/ext4/balloc.c ++++ b/fs/ext4/balloc.c +@@ -83,6 +83,7 @@ static int ext4_group_used_meta_blocks(struct super_block *sb, + } + return used_blocks; + } ++ + /* Initializes an uninitialized block bitmap if given, and returns the + * number of blocks free in the group. */ + unsigned ext4_init_block_bitmap(struct super_block *sb, struct buffer_head *bh, +@@ -345,303 +346,6 @@ ext4_read_block_bitmap(struct super_block *sb, ext4_group_t block_group) + */ + return bh; + } +-/* +- * The reservation window structure operations +- * -------------------------------------------- +- * Operations include: +- * dump, find, add, remove, is_empty, find_next_reservable_window, etc. +- * +- * We use a red-black tree to represent per-filesystem reservation +- * windows. +- * +- */ +- +-/** +- * __rsv_window_dump() -- Dump the filesystem block allocation reservation map +- * @rb_root: root of per-filesystem reservation rb tree +- * @verbose: verbose mode +- * @fn: function which wishes to dump the reservation map +- * +- * If verbose is turned on, it will print the whole block reservation +- * windows(start, end). Otherwise, it will only print out the "bad" windows, +- * those windows that overlap with their immediate neighbors. +- */ +-#if 1 +-static void __rsv_window_dump(struct rb_root *root, int verbose, +- const char *fn) +-{ +- struct rb_node *n; +- struct ext4_reserve_window_node *rsv, *prev; +- int bad; +- +-restart: +- n = rb_first(root); +- bad = 0; +- prev = NULL; +- +- printk(KERN_DEBUG "Block Allocation Reservation " +- "Windows Map (%s):\n", fn); +- while (n) { +- rsv = rb_entry(n, struct ext4_reserve_window_node, rsv_node); +- if (verbose) +- printk(KERN_DEBUG "reservation window 0x%p " +- "start: %llu, end: %llu\n", +- rsv, rsv->rsv_start, rsv->rsv_end); +- if (rsv->rsv_start && rsv->rsv_start >= rsv->rsv_end) { +- printk(KERN_DEBUG "Bad reservation %p (start >= end)\n", +- rsv); +- bad = 1; +- } +- if (prev && prev->rsv_end >= rsv->rsv_start) { +- printk(KERN_DEBUG "Bad reservation %p " +- "(prev->end >= start)\n", rsv); +- bad = 1; +- } +- if (bad) { +- if (!verbose) { +- printk(KERN_DEBUG "Restarting reservation " +- "walk in verbose mode\n"); +- verbose = 1; +- goto restart; +- } +- } +- n = rb_next(n); +- prev = rsv; +- } +- printk(KERN_DEBUG "Window map complete.\n"); +- BUG_ON(bad); +-} +-#define rsv_window_dump(root, verbose) \ +- __rsv_window_dump((root), (verbose), __func__) +-#else +-#define rsv_window_dump(root, verbose) do {} while (0) +-#endif +- +-/** +- * goal_in_my_reservation() +- * @rsv: inode's reservation window +- * @grp_goal: given goal block relative to the allocation block group +- * @group: the current allocation block group +- * @sb: filesystem super block +- * +- * Test if the given goal block (group relative) is within the file's +- * own block reservation window range. +- * +- * If the reservation window is outside the goal allocation group, return 0; +- * grp_goal (given goal block) could be -1, which means no specific +- * goal block. In this case, always return 1. +- * If the goal block is within the reservation window, return 1; +- * otherwise, return 0; +- */ +-static int +-goal_in_my_reservation(struct ext4_reserve_window *rsv, ext4_grpblk_t grp_goal, +- ext4_group_t group, struct super_block *sb) +-{ +- ext4_fsblk_t group_first_block, group_last_block; +- +- group_first_block = ext4_group_first_block_no(sb, group); +- group_last_block = group_first_block + (EXT4_BLOCKS_PER_GROUP(sb) - 1); +- +- if ((rsv->_rsv_start > group_last_block) || +- (rsv->_rsv_end < group_first_block)) +- return 0; +- if ((grp_goal >= 0) && ((grp_goal + group_first_block < rsv->_rsv_start) +- || (grp_goal + group_first_block > rsv->_rsv_end))) +- return 0; +- return 1; +-} +- +-/** +- * search_reserve_window() +- * @rb_root: root of reservation tree +- * @goal: target allocation block +- * +- * Find the reserved window which includes the goal, or the previous one +- * if the goal is not in any window. +- * Returns NULL if there are no windows or if all windows start after the goal. +- */ +-static struct ext4_reserve_window_node * +-search_reserve_window(struct rb_root *root, ext4_fsblk_t goal) +-{ +- struct rb_node *n = root->rb_node; +- struct ext4_reserve_window_node *rsv; +- +- if (!n) +- return NULL; +- +- do { +- rsv = rb_entry(n, struct ext4_reserve_window_node, rsv_node); +- +- if (goal < rsv->rsv_start) +- n = n->rb_left; +- else if (goal > rsv->rsv_end) +- n = n->rb_right; +- else +- return rsv; +- } while (n); +- /* +- * We've fallen off the end of the tree: the goal wasn't inside +- * any particular node. OK, the previous node must be to one +- * side of the interval containing the goal. If it's the RHS, +- * we need to back up one. +- */ +- if (rsv->rsv_start > goal) { +- n = rb_prev(&rsv->rsv_node); +- rsv = rb_entry(n, struct ext4_reserve_window_node, rsv_node); +- } +- return rsv; +-} +- +-/** +- * ext4_rsv_window_add() -- Insert a window to the block reservation rb tree. +- * @sb: super block +- * @rsv: reservation window to add +- * +- * Must be called with rsv_lock hold. +- */ +-void ext4_rsv_window_add(struct super_block *sb, +- struct ext4_reserve_window_node *rsv) +-{ +- struct rb_root *root = &EXT4_SB(sb)->s_rsv_window_root; +- struct rb_node *node = &rsv->rsv_node; +- ext4_fsblk_t start = rsv->rsv_start; +- +- struct rb_node **p = &root->rb_node; +- struct rb_node *parent = NULL; +- struct ext4_reserve_window_node *this; +- +- while (*p) +- { +- parent = *p; +- this = rb_entry(parent, struct ext4_reserve_window_node, rsv_node); +- +- if (start < this->rsv_start) +- p = &(*p)->rb_left; +- else if (start > this->rsv_end) +- p = &(*p)->rb_right; +- else { +- rsv_window_dump(root, 1); +- BUG(); +- } +- } +- +- rb_link_node(node, parent, p); +- rb_insert_color(node, root); +-} +- +-/** +- * ext4_rsv_window_remove() -- unlink a window from the reservation rb tree +- * @sb: super block +- * @rsv: reservation window to remove +- * +- * Mark the block reservation window as not allocated, and unlink it +- * from the filesystem reservation window rb tree. Must be called with +- * rsv_lock hold. +- */ +-static void rsv_window_remove(struct super_block *sb, +- struct ext4_reserve_window_node *rsv) +-{ +- rsv->rsv_start = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- rsv->rsv_end = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- rsv->rsv_alloc_hit = 0; +- rb_erase(&rsv->rsv_node, &EXT4_SB(sb)->s_rsv_window_root); +-} +- +-/* +- * rsv_is_empty() -- Check if the reservation window is allocated. +- * @rsv: given reservation window to check +- * +- * returns 1 if the end block is EXT4_RESERVE_WINDOW_NOT_ALLOCATED. +- */ +-static inline int rsv_is_empty(struct ext4_reserve_window *rsv) +-{ +- /* a valid reservation end block could not be 0 */ +- return rsv->_rsv_end == EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +-} +- +-/** +- * ext4_init_block_alloc_info() +- * @inode: file inode structure +- * +- * Allocate and initialize the reservation window structure, and +- * link the window to the ext4 inode structure at last +- * +- * The reservation window structure is only dynamically allocated +- * and linked to ext4 inode the first time the open file +- * needs a new block. So, before every ext4_new_block(s) call, for +- * regular files, we should check whether the reservation window +- * structure exists or not. In the latter case, this function is called. +- * Fail to do so will result in block reservation being turned off for that +- * open file. +- * +- * This function is called from ext4_get_blocks_handle(), also called +- * when setting the reservation window size through ioctl before the file +- * is open for write (needs block allocation). +- * +- * Needs down_write(i_data_sem) protection prior to call this function. +- */ +-void ext4_init_block_alloc_info(struct inode *inode) +-{ +- struct ext4_inode_info *ei = EXT4_I(inode); +- struct ext4_block_alloc_info *block_i = ei->i_block_alloc_info; +- struct super_block *sb = inode->i_sb; +- +- block_i = kmalloc(sizeof(*block_i), GFP_NOFS); +- if (block_i) { +- struct ext4_reserve_window_node *rsv = &block_i->rsv_window_node; +- +- rsv->rsv_start = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- rsv->rsv_end = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- +- /* +- * if filesystem is mounted with NORESERVATION, the goal +- * reservation window size is set to zero to indicate +- * block reservation is off +- */ +- if (!test_opt(sb, RESERVATION)) +- rsv->rsv_goal_size = 0; +- else +- rsv->rsv_goal_size = EXT4_DEFAULT_RESERVE_BLOCKS; +- rsv->rsv_alloc_hit = 0; +- block_i->last_alloc_logical_block = 0; +- block_i->last_alloc_physical_block = 0; +- } +- ei->i_block_alloc_info = block_i; +-} +- +-/** +- * ext4_discard_reservation() +- * @inode: inode +- * +- * Discard(free) block reservation window on last file close, or truncate +- * or at last iput(). +- * +- * It is being called in three cases: +- * ext4_release_file(): last writer close the file +- * ext4_clear_inode(): last iput(), when nobody link to this file. +- * ext4_truncate(): when the block indirect map is about to change. +- * +- */ +-void ext4_discard_reservation(struct inode *inode) +-{ +- struct ext4_inode_info *ei = EXT4_I(inode); +- struct ext4_block_alloc_info *block_i = ei->i_block_alloc_info; +- struct ext4_reserve_window_node *rsv; +- spinlock_t *rsv_lock = &EXT4_SB(inode->i_sb)->s_rsv_window_lock; +- +- ext4_mb_discard_inode_preallocations(inode); +- +- if (!block_i) +- return; +- +- rsv = &block_i->rsv_window_node; +- if (!rsv_is_empty(&rsv->rsv_window)) { +- spin_lock(rsv_lock); +- if (!rsv_is_empty(&rsv->rsv_window)) +- rsv_window_remove(inode->i_sb, rsv); +- spin_unlock(rsv_lock); +- } +-} + + /** + * ext4_free_blocks_sb() -- Free given blocks and update quota +@@ -650,6 +354,13 @@ void ext4_discard_reservation(struct inode *inode) + * @block: start physcial block to free + * @count: number of blocks to free + * @pdquot_freed_blocks: pointer to quota ++ * ++ * XXX This function is only used by the on-line resizing code, which ++ * should probably be fixed up to call the mballoc variant. There ++ * this needs to be cleaned up later; in fact, I'm not convinced this ++ * is 100% correct in the face of the mballoc code. The online resizing ++ * code needs to be fixed up to more tightly (and correctly) interlock ++ * with the mballoc code. + */ + void ext4_free_blocks_sb(handle_t *handle, struct super_block *sb, + ext4_fsblk_t block, unsigned long count, +@@ -861,747 +572,13 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode, + + sb = inode->i_sb; + +- if (!test_opt(sb, MBALLOC) || !EXT4_SB(sb)->s_group_info) +- ext4_free_blocks_sb(handle, sb, block, count, +- &dquot_freed_blocks); +- else +- ext4_mb_free_blocks(handle, inode, block, count, +- metadata, &dquot_freed_blocks); ++ ext4_mb_free_blocks(handle, inode, block, count, ++ metadata, &dquot_freed_blocks); + if (dquot_freed_blocks) + DQUOT_FREE_BLOCK(inode, dquot_freed_blocks); + return; + } + +-/** +- * ext4_test_allocatable() +- * @nr: given allocation block group +- * @bh: bufferhead contains the bitmap of the given block group +- * +- * For ext4 allocations, we must not reuse any blocks which are +- * allocated in the bitmap buffer's "last committed data" copy. This +- * prevents deletes from freeing up the page for reuse until we have +- * committed the delete transaction. +- * +- * If we didn't do this, then deleting something and reallocating it as +- * data would allow the old block to be overwritten before the +- * transaction committed (because we force data to disk before commit). +- * This would lead to corruption if we crashed between overwriting the +- * data and committing the delete. +- * +- * @@@ We may want to make this allocation behaviour conditional on +- * data-writes at some point, and disable it for metadata allocations or +- * sync-data inodes. +- */ +-static int ext4_test_allocatable(ext4_grpblk_t nr, struct buffer_head *bh) +-{ +- int ret; +- struct journal_head *jh = bh2jh(bh); +- +- if (ext4_test_bit(nr, bh->b_data)) +- return 0; +- +- jbd_lock_bh_state(bh); +- if (!jh->b_committed_data) +- ret = 1; +- else +- ret = !ext4_test_bit(nr, jh->b_committed_data); +- jbd_unlock_bh_state(bh); +- return ret; +-} +- +-/** +- * bitmap_search_next_usable_block() +- * @start: the starting block (group relative) of the search +- * @bh: bufferhead contains the block group bitmap +- * @maxblocks: the ending block (group relative) of the reservation +- * +- * The bitmap search --- search forward alternately through the actual +- * bitmap on disk and the last-committed copy in journal, until we find a +- * bit free in both bitmaps. +- */ +-static ext4_grpblk_t +-bitmap_search_next_usable_block(ext4_grpblk_t start, struct buffer_head *bh, +- ext4_grpblk_t maxblocks) +-{ +- ext4_grpblk_t next; +- struct journal_head *jh = bh2jh(bh); +- +- while (start < maxblocks) { +- next = ext4_find_next_zero_bit(bh->b_data, maxblocks, start); +- if (next >= maxblocks) +- return -1; +- if (ext4_test_allocatable(next, bh)) +- return next; +- jbd_lock_bh_state(bh); +- if (jh->b_committed_data) +- start = ext4_find_next_zero_bit(jh->b_committed_data, +- maxblocks, next); +- jbd_unlock_bh_state(bh); +- } +- return -1; +-} +- +-/** +- * find_next_usable_block() +- * @start: the starting block (group relative) to find next +- * allocatable block in bitmap. +- * @bh: bufferhead contains the block group bitmap +- * @maxblocks: the ending block (group relative) for the search +- * +- * Find an allocatable block in a bitmap. We honor both the bitmap and +- * its last-committed copy (if that exists), and perform the "most +- * appropriate allocation" algorithm of looking for a free block near +- * the initial goal; then for a free byte somewhere in the bitmap; then +- * for any free bit in the bitmap. +- */ +-static ext4_grpblk_t +-find_next_usable_block(ext4_grpblk_t start, struct buffer_head *bh, +- ext4_grpblk_t maxblocks) +-{ +- ext4_grpblk_t here, next; +- char *p, *r; +- +- if (start > 0) { +- /* +- * The goal was occupied; search forward for a free +- * block within the next XX blocks. +- * +- * end_goal is more or less random, but it has to be +- * less than EXT4_BLOCKS_PER_GROUP. Aligning up to the +- * next 64-bit boundary is simple.. +- */ +- ext4_grpblk_t end_goal = (start + 63) & ~63; +- if (end_goal > maxblocks) +- end_goal = maxblocks; +- here = ext4_find_next_zero_bit(bh->b_data, end_goal, start); +- if (here < end_goal && ext4_test_allocatable(here, bh)) +- return here; +- ext4_debug("Bit not found near goal\n"); +- } +- +- here = start; +- if (here < 0) +- here = 0; +- +- p = ((char *)bh->b_data) + (here >> 3); +- r = memscan(p, 0, ((maxblocks + 7) >> 3) - (here >> 3)); +- next = (r - ((char *)bh->b_data)) << 3; +- +- if (next < maxblocks && next >= start && ext4_test_allocatable(next, bh)) +- return next; +- +- /* +- * The bitmap search --- search forward alternately through the actual +- * bitmap and the last-committed copy until we find a bit free in +- * both +- */ +- here = bitmap_search_next_usable_block(here, bh, maxblocks); +- return here; +-} +- +-/** +- * claim_block() +- * @block: the free block (group relative) to allocate +- * @bh: the bufferhead containts the block group bitmap +- * +- * We think we can allocate this block in this bitmap. Try to set the bit. +- * If that succeeds then check that nobody has allocated and then freed the +- * block since we saw that is was not marked in b_committed_data. If it _was_ +- * allocated and freed then clear the bit in the bitmap again and return +- * zero (failure). +- */ +-static inline int +-claim_block(spinlock_t *lock, ext4_grpblk_t block, struct buffer_head *bh) +-{ +- struct journal_head *jh = bh2jh(bh); +- int ret; +- +- if (ext4_set_bit_atomic(lock, block, bh->b_data)) +- return 0; +- jbd_lock_bh_state(bh); +- if (jh->b_committed_data && ext4_test_bit(block, jh->b_committed_data)) { +- ext4_clear_bit_atomic(lock, block, bh->b_data); +- ret = 0; +- } else { +- ret = 1; +- } +- jbd_unlock_bh_state(bh); +- return ret; +-} +- +-/** +- * ext4_try_to_allocate() +- * @sb: superblock +- * @handle: handle to this transaction +- * @group: given allocation block group +- * @bitmap_bh: bufferhead holds the block bitmap +- * @grp_goal: given target block within the group +- * @count: target number of blocks to allocate +- * @my_rsv: reservation window +- * +- * Attempt to allocate blocks within a give range. Set the range of allocation +- * first, then find the first free bit(s) from the bitmap (within the range), +- * and at last, allocate the blocks by claiming the found free bit as allocated. +- * +- * To set the range of this allocation: +- * if there is a reservation window, only try to allocate block(s) from the +- * file's own reservation window; +- * Otherwise, the allocation range starts from the give goal block, ends at +- * the block group's last block. +- * +- * If we failed to allocate the desired block then we may end up crossing to a +- * new bitmap. In that case we must release write access to the old one via +- * ext4_journal_release_buffer(), else we'll run out of credits. +- */ +-static ext4_grpblk_t +-ext4_try_to_allocate(struct super_block *sb, handle_t *handle, +- ext4_group_t group, struct buffer_head *bitmap_bh, +- ext4_grpblk_t grp_goal, unsigned long *count, +- struct ext4_reserve_window *my_rsv) +-{ +- ext4_fsblk_t group_first_block; +- ext4_grpblk_t start, end; +- unsigned long num = 0; +- +- /* we do allocation within the reservation window if we have a window */ +- if (my_rsv) { +- group_first_block = ext4_group_first_block_no(sb, group); +- if (my_rsv->_rsv_start >= group_first_block) +- start = my_rsv->_rsv_start - group_first_block; +- else +- /* reservation window cross group boundary */ +- start = 0; +- end = my_rsv->_rsv_end - group_first_block + 1; +- if (end > EXT4_BLOCKS_PER_GROUP(sb)) +- /* reservation window crosses group boundary */ +- end = EXT4_BLOCKS_PER_GROUP(sb); +- if ((start <= grp_goal) && (grp_goal < end)) +- start = grp_goal; +- else +- grp_goal = -1; +- } else { +- if (grp_goal > 0) +- start = grp_goal; +- else +- start = 0; +- end = EXT4_BLOCKS_PER_GROUP(sb); +- } +- +- BUG_ON(start > EXT4_BLOCKS_PER_GROUP(sb)); +- +-repeat: +- if (grp_goal < 0 || !ext4_test_allocatable(grp_goal, bitmap_bh)) { +- grp_goal = find_next_usable_block(start, bitmap_bh, end); +- if (grp_goal < 0) +- goto fail_access; +- if (!my_rsv) { +- int i; +- +- for (i = 0; i < 7 && grp_goal > start && +- ext4_test_allocatable(grp_goal - 1, +- bitmap_bh); +- i++, grp_goal--) +- ; +- } +- } +- start = grp_goal; +- +- if (!claim_block(sb_bgl_lock(EXT4_SB(sb), group), +- grp_goal, bitmap_bh)) { +- /* +- * The block was allocated by another thread, or it was +- * allocated and then freed by another thread +- */ +- start++; +- grp_goal++; +- if (start >= end) +- goto fail_access; +- goto repeat; +- } +- num++; +- grp_goal++; +- while (num < *count && grp_goal < end +- && ext4_test_allocatable(grp_goal, bitmap_bh) +- && claim_block(sb_bgl_lock(EXT4_SB(sb), group), +- grp_goal, bitmap_bh)) { +- num++; +- grp_goal++; +- } +- *count = num; +- return grp_goal - num; +-fail_access: +- *count = num; +- return -1; +-} +- +-/** +- * find_next_reservable_window(): +- * find a reservable space within the given range. +- * It does not allocate the reservation window for now: +- * alloc_new_reservation() will do the work later. +- * +- * @search_head: the head of the searching list; +- * This is not necessarily the list head of the whole filesystem +- * +- * We have both head and start_block to assist the search +- * for the reservable space. The list starts from head, +- * but we will shift to the place where start_block is, +- * then start from there, when looking for a reservable space. +- * +- * @size: the target new reservation window size +- * +- * @group_first_block: the first block we consider to start +- * the real search from +- * +- * @last_block: +- * the maximum block number that our goal reservable space +- * could start from. This is normally the last block in this +- * group. The search will end when we found the start of next +- * possible reservable space is out of this boundary. +- * This could handle the cross boundary reservation window +- * request. +- * +- * basically we search from the given range, rather than the whole +- * reservation double linked list, (start_block, last_block) +- * to find a free region that is of my size and has not +- * been reserved. +- * +- */ +-static int find_next_reservable_window( +- struct ext4_reserve_window_node *search_head, +- struct ext4_reserve_window_node *my_rsv, +- struct super_block *sb, +- ext4_fsblk_t start_block, +- ext4_fsblk_t last_block) +-{ +- struct rb_node *next; +- struct ext4_reserve_window_node *rsv, *prev; +- ext4_fsblk_t cur; +- int size = my_rsv->rsv_goal_size; +- +- /* TODO: make the start of the reservation window byte-aligned */ +- /* cur = *start_block & ~7;*/ +- cur = start_block; +- rsv = search_head; +- if (!rsv) +- return -1; +- +- while (1) { +- if (cur <= rsv->rsv_end) +- cur = rsv->rsv_end + 1; +- +- /* TODO? +- * in the case we could not find a reservable space +- * that is what is expected, during the re-search, we could +- * remember what's the largest reservable space we could have +- * and return that one. +- * +- * For now it will fail if we could not find the reservable +- * space with expected-size (or more)... +- */ +- if (cur > last_block) +- return -1; /* fail */ +- +- prev = rsv; +- next = rb_next(&rsv->rsv_node); +- rsv = rb_entry(next, struct ext4_reserve_window_node, rsv_node); +- +- /* +- * Reached the last reservation, we can just append to the +- * previous one. +- */ +- if (!next) +- break; +- +- if (cur + size <= rsv->rsv_start) { +- /* +- * Found a reserveable space big enough. We could +- * have a reservation across the group boundary here +- */ +- break; +- } +- } +- /* +- * we come here either : +- * when we reach the end of the whole list, +- * and there is empty reservable space after last entry in the list. +- * append it to the end of the list. +- * +- * or we found one reservable space in the middle of the list, +- * return the reservation window that we could append to. +- * succeed. +- */ +- +- if ((prev != my_rsv) && (!rsv_is_empty(&my_rsv->rsv_window))) +- rsv_window_remove(sb, my_rsv); +- +- /* +- * Let's book the whole avaliable window for now. We will check the +- * disk bitmap later and then, if there are free blocks then we adjust +- * the window size if it's larger than requested. +- * Otherwise, we will remove this node from the tree next time +- * call find_next_reservable_window. +- */ +- my_rsv->rsv_start = cur; +- my_rsv->rsv_end = cur + size - 1; +- my_rsv->rsv_alloc_hit = 0; +- +- if (prev != my_rsv) +- ext4_rsv_window_add(sb, my_rsv); +- +- return 0; +-} +- +-/** +- * alloc_new_reservation()--allocate a new reservation window +- * +- * To make a new reservation, we search part of the filesystem +- * reservation list (the list that inside the group). We try to +- * allocate a new reservation window near the allocation goal, +- * or the beginning of the group, if there is no goal. +- * +- * We first find a reservable space after the goal, then from +- * there, we check the bitmap for the first free block after +- * it. If there is no free block until the end of group, then the +- * whole group is full, we failed. Otherwise, check if the free +- * block is inside the expected reservable space, if so, we +- * succeed. +- * If the first free block is outside the reservable space, then +- * start from the first free block, we search for next available +- * space, and go on. +- * +- * on succeed, a new reservation will be found and inserted into the list +- * It contains at least one free block, and it does not overlap with other +- * reservation windows. +- * +- * failed: we failed to find a reservation window in this group +- * +- * @rsv: the reservation +- * +- * @grp_goal: The goal (group-relative). It is where the search for a +- * free reservable space should start from. +- * if we have a grp_goal(grp_goal >0 ), then start from there, +- * no grp_goal(grp_goal = -1), we start from the first block +- * of the group. +- * +- * @sb: the super block +- * @group: the group we are trying to allocate in +- * @bitmap_bh: the block group block bitmap +- * +- */ +-static int alloc_new_reservation(struct ext4_reserve_window_node *my_rsv, +- ext4_grpblk_t grp_goal, struct super_block *sb, +- ext4_group_t group, struct buffer_head *bitmap_bh) +-{ +- struct ext4_reserve_window_node *search_head; +- ext4_fsblk_t group_first_block, group_end_block, start_block; +- ext4_grpblk_t first_free_block; +- struct rb_root *fs_rsv_root = &EXT4_SB(sb)->s_rsv_window_root; +- unsigned long size; +- int ret; +- spinlock_t *rsv_lock = &EXT4_SB(sb)->s_rsv_window_lock; +- +- group_first_block = ext4_group_first_block_no(sb, group); +- group_end_block = group_first_block + (EXT4_BLOCKS_PER_GROUP(sb) - 1); +- +- if (grp_goal < 0) +- start_block = group_first_block; +- else +- start_block = grp_goal + group_first_block; +- +- size = my_rsv->rsv_goal_size; +- +- if (!rsv_is_empty(&my_rsv->rsv_window)) { +- /* +- * if the old reservation is cross group boundary +- * and if the goal is inside the old reservation window, +- * we will come here when we just failed to allocate from +- * the first part of the window. We still have another part +- * that belongs to the next group. In this case, there is no +- * point to discard our window and try to allocate a new one +- * in this group(which will fail). we should +- * keep the reservation window, just simply move on. +- * +- * Maybe we could shift the start block of the reservation +- * window to the first block of next group. +- */ +- +- if ((my_rsv->rsv_start <= group_end_block) && +- (my_rsv->rsv_end > group_end_block) && +- (start_block >= my_rsv->rsv_start)) +- return -1; +- +- if ((my_rsv->rsv_alloc_hit > +- (my_rsv->rsv_end - my_rsv->rsv_start + 1) / 2)) { +- /* +- * if the previously allocation hit ratio is +- * greater than 1/2, then we double the size of +- * the reservation window the next time, +- * otherwise we keep the same size window +- */ +- size = size * 2; +- if (size > EXT4_MAX_RESERVE_BLOCKS) +- size = EXT4_MAX_RESERVE_BLOCKS; +- my_rsv->rsv_goal_size = size; +- } +- } +- +- spin_lock(rsv_lock); +- /* +- * shift the search start to the window near the goal block +- */ +- search_head = search_reserve_window(fs_rsv_root, start_block); +- +- /* +- * find_next_reservable_window() simply finds a reservable window +- * inside the given range(start_block, group_end_block). +- * +- * To make sure the reservation window has a free bit inside it, we +- * need to check the bitmap after we found a reservable window. +- */ +-retry: +- ret = find_next_reservable_window(search_head, my_rsv, sb, +- start_block, group_end_block); +- +- if (ret == -1) { +- if (!rsv_is_empty(&my_rsv->rsv_window)) +- rsv_window_remove(sb, my_rsv); +- spin_unlock(rsv_lock); +- return -1; +- } +- +- /* +- * On success, find_next_reservable_window() returns the +- * reservation window where there is a reservable space after it. +- * Before we reserve this reservable space, we need +- * to make sure there is at least a free block inside this region. +- * +- * searching the first free bit on the block bitmap and copy of +- * last committed bitmap alternatively, until we found a allocatable +- * block. Search start from the start block of the reservable space +- * we just found. +- */ +- spin_unlock(rsv_lock); +- first_free_block = bitmap_search_next_usable_block( +- my_rsv->rsv_start - group_first_block, +- bitmap_bh, group_end_block - group_first_block + 1); +- +- if (first_free_block < 0) { +- /* +- * no free block left on the bitmap, no point +- * to reserve the space. return failed. +- */ +- spin_lock(rsv_lock); +- if (!rsv_is_empty(&my_rsv->rsv_window)) +- rsv_window_remove(sb, my_rsv); +- spin_unlock(rsv_lock); +- return -1; /* failed */ +- } +- +- start_block = first_free_block + group_first_block; +- /* +- * check if the first free block is within the +- * free space we just reserved +- */ +- if (start_block >= my_rsv->rsv_start && start_block <= my_rsv->rsv_end) +- return 0; /* success */ +- /* +- * if the first free bit we found is out of the reservable space +- * continue search for next reservable space, +- * start from where the free block is, +- * we also shift the list head to where we stopped last time +- */ +- search_head = my_rsv; +- spin_lock(rsv_lock); +- goto retry; +-} +- +-/** +- * try_to_extend_reservation() +- * @my_rsv: given reservation window +- * @sb: super block +- * @size: the delta to extend +- * +- * Attempt to expand the reservation window large enough to have +- * required number of free blocks +- * +- * Since ext4_try_to_allocate() will always allocate blocks within +- * the reservation window range, if the window size is too small, +- * multiple blocks allocation has to stop at the end of the reservation +- * window. To make this more efficient, given the total number of +- * blocks needed and the current size of the window, we try to +- * expand the reservation window size if necessary on a best-effort +- * basis before ext4_new_blocks() tries to allocate blocks, +- */ +-static void try_to_extend_reservation(struct ext4_reserve_window_node *my_rsv, +- struct super_block *sb, int size) +-{ +- struct ext4_reserve_window_node *next_rsv; +- struct rb_node *next; +- spinlock_t *rsv_lock = &EXT4_SB(sb)->s_rsv_window_lock; +- +- if (!spin_trylock(rsv_lock)) +- return; +- +- next = rb_next(&my_rsv->rsv_node); +- +- if (!next) +- my_rsv->rsv_end += size; +- else { +- next_rsv = rb_entry(next, struct ext4_reserve_window_node, rsv_node); +- +- if ((next_rsv->rsv_start - my_rsv->rsv_end - 1) >= size) +- my_rsv->rsv_end += size; +- else +- my_rsv->rsv_end = next_rsv->rsv_start - 1; +- } +- spin_unlock(rsv_lock); +-} +- +-/** +- * ext4_try_to_allocate_with_rsv() +- * @sb: superblock +- * @handle: handle to this transaction +- * @group: given allocation block group +- * @bitmap_bh: bufferhead holds the block bitmap +- * @grp_goal: given target block within the group +- * @count: target number of blocks to allocate +- * @my_rsv: reservation window +- * @errp: pointer to store the error code +- * +- * This is the main function used to allocate a new block and its reservation +- * window. +- * +- * Each time when a new block allocation is need, first try to allocate from +- * its own reservation. If it does not have a reservation window, instead of +- * looking for a free bit on bitmap first, then look up the reservation list to +- * see if it is inside somebody else's reservation window, we try to allocate a +- * reservation window for it starting from the goal first. Then do the block +- * allocation within the reservation window. +- * +- * This will avoid keeping on searching the reservation list again and +- * again when somebody is looking for a free block (without +- * reservation), and there are lots of free blocks, but they are all +- * being reserved. +- * +- * We use a red-black tree for the per-filesystem reservation list. +- * +- */ +-static ext4_grpblk_t +-ext4_try_to_allocate_with_rsv(struct super_block *sb, handle_t *handle, +- ext4_group_t group, struct buffer_head *bitmap_bh, +- ext4_grpblk_t grp_goal, +- struct ext4_reserve_window_node *my_rsv, +- unsigned long *count, int *errp) +-{ +- ext4_fsblk_t group_first_block, group_last_block; +- ext4_grpblk_t ret = 0; +- int fatal; +- unsigned long num = *count; +- +- *errp = 0; +- +- /* +- * Make sure we use undo access for the bitmap, because it is critical +- * that we do the frozen_data COW on bitmap buffers in all cases even +- * if the buffer is in BJ_Forget state in the committing transaction. +- */ +- BUFFER_TRACE(bitmap_bh, "get undo access for new block"); +- fatal = ext4_journal_get_undo_access(handle, bitmap_bh); +- if (fatal) { +- *errp = fatal; +- return -1; +- } +- +- /* +- * we don't deal with reservation when +- * filesystem is mounted without reservation +- * or the file is not a regular file +- * or last attempt to allocate a block with reservation turned on failed +- */ +- if (my_rsv == NULL) { +- ret = ext4_try_to_allocate(sb, handle, group, bitmap_bh, +- grp_goal, count, NULL); +- goto out; +- } +- /* +- * grp_goal is a group relative block number (if there is a goal) +- * 0 <= grp_goal < EXT4_BLOCKS_PER_GROUP(sb) +- * first block is a filesystem wide block number +- * first block is the block number of the first block in this group +- */ +- group_first_block = ext4_group_first_block_no(sb, group); +- group_last_block = group_first_block + (EXT4_BLOCKS_PER_GROUP(sb) - 1); +- +- /* +- * Basically we will allocate a new block from inode's reservation +- * window. +- * +- * We need to allocate a new reservation window, if: +- * a) inode does not have a reservation window; or +- * b) last attempt to allocate a block from existing reservation +- * failed; or +- * c) we come here with a goal and with a reservation window +- * +- * We do not need to allocate a new reservation window if we come here +- * at the beginning with a goal and the goal is inside the window, or +- * we don't have a goal but already have a reservation window. +- * then we could go to allocate from the reservation window directly. +- */ +- while (1) { +- if (rsv_is_empty(&my_rsv->rsv_window) || (ret < 0) || +- !goal_in_my_reservation(&my_rsv->rsv_window, +- grp_goal, group, sb)) { +- if (my_rsv->rsv_goal_size < *count) +- my_rsv->rsv_goal_size = *count; +- ret = alloc_new_reservation(my_rsv, grp_goal, sb, +- group, bitmap_bh); +- if (ret < 0) +- break; /* failed */ +- +- if (!goal_in_my_reservation(&my_rsv->rsv_window, +- grp_goal, group, sb)) +- grp_goal = -1; +- } else if (grp_goal >= 0) { +- int curr = my_rsv->rsv_end - +- (grp_goal + group_first_block) + 1; +- +- if (curr < *count) +- try_to_extend_reservation(my_rsv, sb, +- *count - curr); +- } +- +- if ((my_rsv->rsv_start > group_last_block) || +- (my_rsv->rsv_end < group_first_block)) { +- rsv_window_dump(&EXT4_SB(sb)->s_rsv_window_root, 1); +- BUG(); +- } +- ret = ext4_try_to_allocate(sb, handle, group, bitmap_bh, +- grp_goal, &num, &my_rsv->rsv_window); +- if (ret >= 0) { +- my_rsv->rsv_alloc_hit += num; +- *count = num; +- break; /* succeed */ +- } +- num = *count; +- } +-out: +- if (ret >= 0) { +- BUFFER_TRACE(bitmap_bh, "journal_dirty_metadata for " +- "bitmap block"); +- fatal = ext4_journal_dirty_metadata(handle, bitmap_bh); +- if (fatal) { +- *errp = fatal; +- return -1; +- } +- return ret; +- } +- +- BUFFER_TRACE(bitmap_bh, "journal_release_buffer"); +- ext4_journal_release_buffer(handle, bitmap_bh); +- return ret; +-} +- + int ext4_claim_free_blocks(struct ext4_sb_info *sbi, + s64 nblocks) + { +@@ -1702,310 +679,6 @@ int ext4_should_retry_alloc(struct super_block *sb, int *retries) + return jbd2_journal_force_commit_nested(EXT4_SB(sb)->s_journal); + } + +-/** +- * ext4_old_new_blocks() -- core block bitmap based block allocation function +- * +- * @handle: handle to this transaction +- * @inode: file inode +- * @goal: given target block(filesystem wide) +- * @count: target number of blocks to allocate +- * @errp: error code +- * +- * ext4_old_new_blocks uses a goal block to assist allocation and look up +- * the block bitmap directly to do block allocation. It tries to +- * allocate block(s) from the block group contains the goal block first. If +- * that fails, it will try to allocate block(s) from other block groups +- * without any specific goal block. +- * +- * This function is called when -o nomballoc mount option is enabled +- * +- */ +-ext4_fsblk_t ext4_old_new_blocks(handle_t *handle, struct inode *inode, +- ext4_fsblk_t goal, unsigned long *count, int *errp) +-{ +- struct buffer_head *bitmap_bh = NULL; +- struct buffer_head *gdp_bh; +- ext4_group_t group_no; +- ext4_group_t goal_group; +- ext4_grpblk_t grp_target_blk; /* blockgroup relative goal block */ +- ext4_grpblk_t grp_alloc_blk; /* blockgroup-relative allocated block*/ +- ext4_fsblk_t ret_block; /* filesyetem-wide allocated block */ +- ext4_group_t bgi; /* blockgroup iteration index */ +- int fatal = 0, err; +- int performed_allocation = 0; +- ext4_grpblk_t free_blocks; /* number of free blocks in a group */ +- struct super_block *sb; +- struct ext4_group_desc *gdp; +- struct ext4_super_block *es; +- struct ext4_sb_info *sbi; +- struct ext4_reserve_window_node *my_rsv = NULL; +- struct ext4_block_alloc_info *block_i; +- unsigned short windowsz = 0; +- ext4_group_t ngroups; +- unsigned long num = *count; +- +- sb = inode->i_sb; +- if (!sb) { +- *errp = -ENODEV; +- printk(KERN_ERR "ext4_new_block: nonexistent superblock"); +- return 0; +- } +- +- sbi = EXT4_SB(sb); +- if (!EXT4_I(inode)->i_delalloc_reserved_flag) { +- /* +- * With delalloc we already reserved the blocks +- */ +- while (*count && ext4_claim_free_blocks(sbi, *count)) { +- /* let others to free the space */ +- yield(); +- *count = *count >> 1; +- } +- if (!*count) { +- *errp = -ENOSPC; +- return 0; /*return with ENOSPC error */ +- } +- num = *count; +- } +- /* +- * Check quota for allocation of this block. +- */ +- if (DQUOT_ALLOC_BLOCK(inode, num)) { +- *errp = -EDQUOT; +- return 0; +- } +- +- sbi = EXT4_SB(sb); +- es = EXT4_SB(sb)->s_es; +- ext4_debug("goal=%llu.\n", goal); +- /* +- * Allocate a block from reservation only when +- * filesystem is mounted with reservation(default,-o reservation), and +- * it's a regular file, and +- * the desired window size is greater than 0 (One could use ioctl +- * command EXT4_IOC_SETRSVSZ to set the window size to 0 to turn off +- * reservation on that particular file) +- */ +- block_i = EXT4_I(inode)->i_block_alloc_info; +- if (block_i && ((windowsz = block_i->rsv_window_node.rsv_goal_size) > 0)) +- my_rsv = &block_i->rsv_window_node; +- +- /* +- * First, test whether the goal block is free. +- */ +- if (goal < le32_to_cpu(es->s_first_data_block) || +- goal >= ext4_blocks_count(es)) +- goal = le32_to_cpu(es->s_first_data_block); +- ext4_get_group_no_and_offset(sb, goal, &group_no, &grp_target_blk); +- goal_group = group_no; +-retry_alloc: +- gdp = ext4_get_group_desc(sb, group_no, &gdp_bh); +- if (!gdp) +- goto io_error; +- +- free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); +- +- if (free_blocks > 0) { +- /* +- * try to allocate with group target block +- * in the goal group. If we have low free_blocks +- * count turn off reservation +- */ +- if (my_rsv && (free_blocks < windowsz) +- && (rsv_is_empty(&my_rsv->rsv_window))) +- my_rsv = NULL; +- +- bitmap_bh = ext4_read_block_bitmap(sb, group_no); +- if (!bitmap_bh) +- goto io_error; +- grp_alloc_blk = ext4_try_to_allocate_with_rsv(sb, handle, +- group_no, bitmap_bh, grp_target_blk, +- my_rsv, &num, &fatal); +- if (fatal) +- goto out; +- if (grp_alloc_blk >= 0) +- goto allocated; +- } +- +- ngroups = EXT4_SB(sb)->s_groups_count; +- smp_rmb(); +- +- /* +- * Now search the rest of the groups. We assume that +- * group_no and gdp correctly point to the last group visited. +- */ +- for (bgi = 0; bgi < ngroups; bgi++) { +- group_no++; +- if (group_no >= ngroups) +- group_no = 0; +- gdp = ext4_get_group_desc(sb, group_no, &gdp_bh); +- if (!gdp) +- goto io_error; +- free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); +- /* +- * skip this group if the number of +- * free blocks is less than half of the reservation +- * window size. +- */ +- if (my_rsv && (free_blocks <= (windowsz/2))) +- continue; +- +- brelse(bitmap_bh); +- bitmap_bh = ext4_read_block_bitmap(sb, group_no); +- if (!bitmap_bh) +- goto io_error; +- /* +- * try to allocate block(s) from this group, without a goal(-1). +- */ +- grp_alloc_blk = ext4_try_to_allocate_with_rsv(sb, handle, +- group_no, bitmap_bh, -1, my_rsv, +- &num, &fatal); +- if (fatal) +- goto out; +- if (grp_alloc_blk >= 0) +- goto allocated; +- } +- /* +- * We may end up a bogus ealier ENOSPC error due to +- * filesystem is "full" of reservations, but +- * there maybe indeed free blocks avaliable on disk +- * In this case, we just forget about the reservations +- * just do block allocation as without reservations. +- */ +- if (my_rsv) { +- my_rsv = NULL; +- windowsz = 0; +- group_no = goal_group; +- goto retry_alloc; +- } +- /* No space left on the device */ +- *errp = -ENOSPC; +- goto out; +- +-allocated: +- +- ext4_debug("using block group %lu(%d)\n", +- group_no, gdp->bg_free_blocks_count); +- +- BUFFER_TRACE(gdp_bh, "get_write_access"); +- fatal = ext4_journal_get_write_access(handle, gdp_bh); +- if (fatal) +- goto out; +- +- ret_block = grp_alloc_blk + ext4_group_first_block_no(sb, group_no); +- +- if (in_range(ext4_block_bitmap(sb, gdp), ret_block, num) || +- in_range(ext4_inode_bitmap(sb, gdp), ret_block, num) || +- in_range(ret_block, ext4_inode_table(sb, gdp), +- EXT4_SB(sb)->s_itb_per_group) || +- in_range(ret_block + num - 1, ext4_inode_table(sb, gdp), +- EXT4_SB(sb)->s_itb_per_group)) { +- ext4_error(sb, "ext4_new_block", +- "Allocating block in system zone - " +- "blocks from %llu, length %lu", +- ret_block, num); +- /* +- * claim_block marked the blocks we allocated +- * as in use. So we may want to selectively +- * mark some of the blocks as free +- */ +- goto retry_alloc; +- } +- +- performed_allocation = 1; +- +-#ifdef CONFIG_JBD2_DEBUG +- { +- struct buffer_head *debug_bh; +- +- /* Record bitmap buffer state in the newly allocated block */ +- debug_bh = sb_find_get_block(sb, ret_block); +- if (debug_bh) { +- BUFFER_TRACE(debug_bh, "state when allocated"); +- BUFFER_TRACE2(debug_bh, bitmap_bh, "bitmap state"); +- brelse(debug_bh); +- } +- } +- jbd_lock_bh_state(bitmap_bh); +- spin_lock(sb_bgl_lock(sbi, group_no)); +- if (buffer_jbd(bitmap_bh) && bh2jh(bitmap_bh)->b_committed_data) { +- int i; +- +- for (i = 0; i < num; i++) { +- if (ext4_test_bit(grp_alloc_blk+i, +- bh2jh(bitmap_bh)->b_committed_data)) { +- printk(KERN_ERR "%s: block was unexpectedly " +- "set in b_committed_data\n", __func__); +- } +- } +- } +- ext4_debug("found bit %d\n", grp_alloc_blk); +- spin_unlock(sb_bgl_lock(sbi, group_no)); +- jbd_unlock_bh_state(bitmap_bh); +-#endif +- +- if (ret_block + num - 1 >= ext4_blocks_count(es)) { +- ext4_error(sb, "ext4_new_block", +- "block(%llu) >= blocks count(%llu) - " +- "block_group = %lu, es == %p ", ret_block, +- ext4_blocks_count(es), group_no, es); +- goto out; +- } +- +- /* +- * It is up to the caller to add the new buffer to a journal +- * list of some description. We don't know in advance whether +- * the caller wants to use it as metadata or data. +- */ +- spin_lock(sb_bgl_lock(sbi, group_no)); +- if (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) +- gdp->bg_flags &= cpu_to_le16(~EXT4_BG_BLOCK_UNINIT); +- le16_add_cpu(&gdp->bg_free_blocks_count, -num); +- gdp->bg_checksum = ext4_group_desc_csum(sbi, group_no, gdp); +- spin_unlock(sb_bgl_lock(sbi, group_no)); +- percpu_counter_sub(&sbi->s_freeblocks_counter, num); +- /* +- * Now reduce the dirty block count also. Should not go negative +- */ +- percpu_counter_sub(&sbi->s_dirtyblocks_counter, num); +- if (sbi->s_log_groups_per_flex) { +- ext4_group_t flex_group = ext4_flex_group(sbi, group_no); +- spin_lock(sb_bgl_lock(sbi, flex_group)); +- sbi->s_flex_groups[flex_group].free_blocks -= num; +- spin_unlock(sb_bgl_lock(sbi, flex_group)); +- } +- +- BUFFER_TRACE(gdp_bh, "journal_dirty_metadata for group descriptor"); +- err = ext4_journal_dirty_metadata(handle, gdp_bh); +- if (!fatal) +- fatal = err; +- +- sb->s_dirt = 1; +- if (fatal) +- goto out; +- +- *errp = 0; +- brelse(bitmap_bh); +- DQUOT_FREE_BLOCK(inode, *count-num); +- *count = num; +- return ret_block; +- +-io_error: +- *errp = -EIO; +-out: +- if (fatal) { +- *errp = fatal; +- ext4_std_error(sb, fatal); +- } +- /* +- * Undo the block allocation +- */ +- if (!performed_allocation) +- DQUOT_FREE_BLOCK(inode, *count); +- brelse(bitmap_bh); +- return 0; +-} +- + #define EXT4_META_BLOCK 0x1 + + static ext4_fsblk_t do_blk_alloc(handle_t *handle, struct inode *inode, +@@ -2015,10 +688,6 @@ static ext4_fsblk_t do_blk_alloc(handle_t *handle, struct inode *inode, + struct ext4_allocation_request ar; + ext4_fsblk_t ret; + +- if (!test_opt(inode->i_sb, MBALLOC)) { +- return ext4_old_new_blocks(handle, inode, goal, count, errp); +- } +- + memset(&ar, 0, sizeof(ar)); + /* Fill with neighbour allocated blocks */ + +@@ -2239,3 +908,4 @@ unsigned long ext4_bg_num_gdb(struct super_block *sb, ext4_group_t group) + return ext4_bg_num_gdb_meta(sb,group); + + } ++ +diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h +index 922d187..c50c04c 100644 +--- a/fs/ext4/ext4.h ++++ b/fs/ext4/ext4.h +@@ -539,7 +539,6 @@ do { \ + #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ + #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ + #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */ +-#define EXT4_MOUNT_MBALLOC 0x4000000 /* Buddy allocation support */ + #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ + /* Compatibility, for having both ext2_fs.h and ext4_fs.h included at once */ + #ifndef _LINUX_EXT2_FS_H +@@ -1002,8 +1001,6 @@ extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, + extern ext4_fsblk_t ext4_new_blocks(handle_t *handle, struct inode *inode, + ext4_lblk_t iblock, ext4_fsblk_t goal, + unsigned long *count, int *errp); +-extern ext4_fsblk_t ext4_old_new_blocks(handle_t *handle, struct inode *inode, +- ext4_fsblk_t goal, unsigned long *count, int *errp); + extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); + extern ext4_fsblk_t ext4_has_free_blocks(struct ext4_sb_info *sbi, + s64 nblocks); +@@ -1018,8 +1015,6 @@ extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, + ext4_group_t block_group, + struct buffer_head ** bh); + extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); +-extern void ext4_init_block_alloc_info(struct inode *); +-extern void ext4_rsv_window_add(struct super_block *sb, struct ext4_reserve_window_node *rsv); + + /* dir.c */ + extern int ext4_check_dir_entry(const char *, struct inode *, +@@ -1054,7 +1049,7 @@ extern int ext4_mb_release(struct super_block *); + extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, + struct ext4_allocation_request *, int *); + extern int ext4_mb_reserve_blocks(struct super_block *, int); +-extern void ext4_mb_discard_inode_preallocations(struct inode *); ++extern void ext4_discard_preallocations(struct inode *); + extern int __init init_ext4_mballoc(void); + extern void exit_ext4_mballoc(void); + extern void ext4_mb_free_blocks(handle_t *, struct inode *, +@@ -1084,7 +1079,6 @@ extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, + struct kstat *stat); + extern void ext4_delete_inode(struct inode *); + extern int ext4_sync_inode(handle_t *, struct inode *); +-extern void ext4_discard_reservation(struct inode *); + extern void ext4_dirty_inode(struct inode *); + extern int ext4_change_inode_journal_flag(struct inode *, int); + extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); +diff --git a/fs/ext4/ext4_i.h b/fs/ext4/ext4_i.h +index ef7409f..2875eec 100644 +--- a/fs/ext4/ext4_i.h ++++ b/fs/ext4/ext4_i.h +@@ -33,38 +33,6 @@ typedef __u32 ext4_lblk_t; + /* data type for block group number */ + typedef unsigned long ext4_group_t; + +-struct ext4_reserve_window { +- ext4_fsblk_t _rsv_start; /* First byte reserved */ +- ext4_fsblk_t _rsv_end; /* Last byte reserved or 0 */ +-}; +- +-struct ext4_reserve_window_node { +- struct rb_node rsv_node; +- __u32 rsv_goal_size; +- __u32 rsv_alloc_hit; +- struct ext4_reserve_window rsv_window; +-}; +- +-struct ext4_block_alloc_info { +- /* information about reservation window */ +- struct ext4_reserve_window_node rsv_window_node; +- /* +- * was i_next_alloc_block in ext4_inode_info +- * is the logical (file-relative) number of the +- * most-recently-allocated block in this file. +- * We use this for detecting linearly ascending allocation requests. +- */ +- ext4_lblk_t last_alloc_logical_block; +- /* +- * Was i_next_alloc_goal in ext4_inode_info +- * is the *physical* companion to i_next_alloc_block. +- * it the physical block number of the block which was most-recentl +- * allocated to this file. This give us the goal (target) for the next +- * allocation when we detect linearly ascending requests. +- */ +- ext4_fsblk_t last_alloc_physical_block; +-}; +- + #define rsv_start rsv_window._rsv_start + #define rsv_end rsv_window._rsv_end + +@@ -97,9 +65,6 @@ struct ext4_inode_info { + ext4_group_t i_block_group; + __u32 i_state; /* Dynamic state flags for ext4 */ + +- /* block reservation info */ +- struct ext4_block_alloc_info *i_block_alloc_info; +- + ext4_lblk_t i_dir_start_lookup; + #ifdef CONFIG_EXT4DEV_FS_XATTR + /* +diff --git a/fs/ext4/ext4_sb.h b/fs/ext4/ext4_sb.h +index 94e0757..6a0b40d 100644 +--- a/fs/ext4/ext4_sb.h ++++ b/fs/ext4/ext4_sb.h +@@ -67,7 +67,6 @@ struct ext4_sb_info { + /* root of the per fs reservation window tree */ + spinlock_t s_rsv_window_lock; + struct rb_root s_rsv_window_root; +- struct ext4_reserve_window_node s_rsv_window_head; + + /* Journaling */ + struct inode *s_journal_inode; +diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c +index e8758df..c8f81f2 100644 +--- a/fs/ext4/extents.c ++++ b/fs/ext4/extents.c +@@ -2697,11 +2697,8 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, + goto out2; + } + /* +- * Okay, we need to do block allocation. Lazily initialize the block +- * allocation info here if necessary. ++ * Okay, we need to do block allocation. + */ +- if (S_ISREG(inode->i_mode) && (!EXT4_I(inode)->i_block_alloc_info)) +- ext4_init_block_alloc_info(inode); + + /* find neighbour allocated blocks */ + ar.lleft = iblock; +@@ -2761,7 +2758,7 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, + /* free data blocks we just allocated */ + /* not a good idea to call discard here directly, + * but otherwise we'd need to call it every free() */ +- ext4_mb_discard_inode_preallocations(inode); ++ ext4_discard_preallocations(inode); + ext4_free_blocks(handle, inode, ext_pblock(&newex), + ext4_ext_get_actual_len(&newex), 0); + goto out2; +@@ -2825,7 +2822,7 @@ void ext4_ext_truncate(struct inode *inode) + down_write(&EXT4_I(inode)->i_data_sem); + ext4_ext_invalidate_cache(inode); + +- ext4_discard_reservation(inode); ++ ext4_discard_preallocations(inode); + + /* + * TODO: optimization is possible here. +diff --git a/fs/ext4/file.c b/fs/ext4/file.c +index 11b289f..62796b7 100644 +--- a/fs/ext4/file.c ++++ b/fs/ext4/file.c +@@ -38,7 +38,7 @@ static int ext4_release_file(struct inode *inode, struct file *filp) + (atomic_read(&inode->i_writecount) == 1)) + { + down_write(&EXT4_I(inode)->i_data_sem); +- ext4_discard_reservation(inode); ++ ext4_discard_preallocations(inode); + up_write(&EXT4_I(inode)->i_data_sem); + } + if (is_dx(inode) && filp->private_data) +diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c +index 5e66a2f..1343bf1 100644 +--- a/fs/ext4/ialloc.c ++++ b/fs/ext4/ialloc.c +@@ -817,7 +817,6 @@ got: + ei->i_flags &= ~EXT4_DIRSYNC_FL; + ei->i_file_acl = 0; + ei->i_dtime = 0; +- ei->i_block_alloc_info = NULL; + ei->i_block_group = group; + + ext4_set_inode_flags(inode); +diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c +index 5bd700f..193b09f 100644 +--- a/fs/ext4/inode.c ++++ b/fs/ext4/inode.c +@@ -486,18 +486,9 @@ static ext4_fsblk_t ext4_find_near(struct inode *inode, Indirect *ind) + static ext4_fsblk_t ext4_find_goal(struct inode *inode, ext4_lblk_t block, + Indirect *partial) + { +- struct ext4_block_alloc_info *block_i; +- +- block_i = EXT4_I(inode)->i_block_alloc_info; +- + /* +- * try the heuristic for sequential allocation, +- * failing that at least try to get decent locality. ++ * XXX need to get goal block from mballoc's data structures + */ +- if (block_i && (block == block_i->last_alloc_logical_block + 1) +- && (block_i->last_alloc_physical_block != 0)) { +- return block_i->last_alloc_physical_block + 1; +- } + + return ext4_find_near(inode, partial); + } +@@ -757,10 +748,8 @@ static int ext4_splice_branch(handle_t *handle, struct inode *inode, + { + int i; + int err = 0; +- struct ext4_block_alloc_info *block_i; + ext4_fsblk_t current_block; + +- block_i = EXT4_I(inode)->i_block_alloc_info; + /* + * If we're splicing into a [td]indirect block (as opposed to the + * inode) then we need to get write access to the [td]indirect block +@@ -786,17 +775,6 @@ static int ext4_splice_branch(handle_t *handle, struct inode *inode, + *(where->p + i) = cpu_to_le32(current_block++); + } + +- /* +- * update the most recently allocated logical & physical block +- * in i_block_alloc_info, to assist find the proper goal block for next +- * allocation +- */ +- if (block_i) { +- block_i->last_alloc_logical_block = block + blks - 1; +- block_i->last_alloc_physical_block = +- le32_to_cpu(where[num].key) + blks - 1; +- } +- + /* We are done with atomic stuff, now do the rest of housekeeping */ + + inode->i_ctime = ext4_current_time(inode); +@@ -914,12 +892,8 @@ int ext4_get_blocks_handle(handle_t *handle, struct inode *inode, + goto cleanup; + + /* +- * Okay, we need to do block allocation. Lazily initialize the block +- * allocation info here if necessary ++ * Okay, we need to do block allocation. + */ +- if (S_ISREG(inode->i_mode) && (!ei->i_block_alloc_info)) +- ext4_init_block_alloc_info(inode); +- + goal = ext4_find_goal(inode, iblock, partial); + + /* the number of blocks need to allocate for [d,t]indirect blocks */ +@@ -3738,7 +3712,7 @@ void ext4_truncate(struct inode *inode) + */ + down_write(&ei->i_data_sem); + +- ext4_discard_reservation(inode); ++ ext4_discard_preallocations(inode); + + /* + * The orphan list entry will now protect us from any crash which +@@ -4071,7 +4045,6 @@ struct inode *ext4_iget(struct super_block *sb, unsigned long ino) + ei->i_acl = EXT4_ACL_NOT_CACHED; + ei->i_default_acl = EXT4_ACL_NOT_CACHED; + #endif +- ei->i_block_alloc_info = NULL; + + ret = __ext4_get_inode_loc(inode, &iloc, 0); + if (ret < 0) +diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c +index 3e14060..ea27eaa 100644 +--- a/fs/ext4/ioctl.c ++++ b/fs/ext4/ioctl.c +@@ -23,7 +23,6 @@ long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) + struct inode *inode = filp->f_dentry->d_inode; + struct ext4_inode_info *ei = EXT4_I(inode); + unsigned int flags; +- unsigned short rsv_window_size; + + ext4_debug("cmd = %u, arg = %lu\n", cmd, arg); + +@@ -190,49 +189,6 @@ setversion_out: + return ret; + } + #endif +- case EXT4_IOC_GETRSVSZ: +- if (test_opt(inode->i_sb, RESERVATION) +- && S_ISREG(inode->i_mode) +- && ei->i_block_alloc_info) { +- rsv_window_size = ei->i_block_alloc_info->rsv_window_node.rsv_goal_size; +- return put_user(rsv_window_size, (int __user *)arg); +- } +- return -ENOTTY; +- case EXT4_IOC_SETRSVSZ: { +- int err; +- +- if (!test_opt(inode->i_sb, RESERVATION) || !S_ISREG(inode->i_mode)) +- return -ENOTTY; +- +- if (!is_owner_or_cap(inode)) +- return -EACCES; +- +- if (get_user(rsv_window_size, (int __user *)arg)) +- return -EFAULT; +- +- err = mnt_want_write(filp->f_path.mnt); +- if (err) +- return err; +- +- if (rsv_window_size > EXT4_MAX_RESERVE_BLOCKS) +- rsv_window_size = EXT4_MAX_RESERVE_BLOCKS; +- +- /* +- * need to allocate reservation structure for this inode +- * before set the window size +- */ +- down_write(&ei->i_data_sem); +- if (!ei->i_block_alloc_info) +- ext4_init_block_alloc_info(inode); +- +- if (ei->i_block_alloc_info){ +- struct ext4_reserve_window_node *rsv = &ei->i_block_alloc_info->rsv_window_node; +- rsv->rsv_goal_size = rsv_window_size; +- } +- up_write(&ei->i_data_sem); +- mnt_drop_write(filp->f_path.mnt); +- return 0; +- } + case EXT4_IOC_GROUP_EXTEND: { + ext4_fsblk_t n_blocks_count; + struct super_block *sb = inode->i_sb; +diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c +index 3373f89..c84b68f 100644 +--- a/fs/ext4/mballoc.c ++++ b/fs/ext4/mballoc.c +@@ -534,9 +534,6 @@ static int __mb_check_buddy(struct ext4_buddy *e4b, char *file, + void *buddy; + void *buddy2; + +- if (!test_opt(sb, MBALLOC)) +- return 0; +- + { + static int mb_check_counter; + if (mb_check_counter++ % 100 != 0) +@@ -2487,19 +2484,14 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery) + 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; + } + 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); + return -ENOMEM; + } +@@ -2522,7 +2514,6 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery) + /* 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; +@@ -2544,7 +2535,6 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery) + + 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; +@@ -2590,9 +2580,6 @@ int ext4_mb_release(struct super_block *sb) + 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, +@@ -3799,7 +3786,7 @@ out: + * + * 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; +@@ -3811,7 +3798,7 @@ void ext4_mb_discard_inode_preallocations(struct inode *inode) + 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; + } +@@ -4293,11 +4280,6 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle, + 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; +- } + if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag) { + /* + * With delalloc we already reserved the blocks +diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c +index b60afbc..b6ec184 100644 +--- a/fs/ext4/resize.c ++++ b/fs/ext4/resize.c +@@ -870,11 +870,10 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input) + * We can allocate memory for mb_alloc based on the new group + * descriptor + */ +- if (test_opt(sb, MBALLOC)) { +- err = ext4_mb_add_more_groupinfo(sb, input->group, gdp); +- if (err) +- goto exit_journal; +- } ++ err = ext4_mb_add_more_groupinfo(sb, input->group, gdp); ++ if (err) ++ goto exit_journal; ++ + /* + * Make the new blocks and inodes valid next. We do this before + * increasing the group count so that once the group is enabled, +@@ -1086,8 +1085,15 @@ int ext4_group_extend(struct super_block *sb, struct ext4_super_block *es, + /* + * Mark mballoc pages as not up to date so that they will be updated + * next time they are loaded by ext4_mb_load_buddy. ++ * ++ * XXX Bad, Bad, BAD!!! We should not be overloading the ++ * Uptodate flag, particularly on thte bitmap bh, as way of ++ * hinting to ext4_mb_load_buddy() that it needs to be ++ * overloaded. A user could take a LVM snapshot, then do an ++ * on-line fsck, and clear the uptodate flag, and this would ++ * not be a bug in userspace, but a bug in the kernel. FIXME!!! + */ +- if (test_opt(sb, MBALLOC)) { ++ { + struct ext4_sb_info *sbi = EXT4_SB(sb); + struct inode *inode = sbi->s_buddy_cache; + int blocks_per_page; +diff --git a/fs/ext4/super.c b/fs/ext4/super.c +index 9094095..515af05 100644 +--- a/fs/ext4/super.c ++++ b/fs/ext4/super.c +@@ -573,7 +573,6 @@ static struct inode *ext4_alloc_inode(struct super_block *sb) + ei->i_acl = EXT4_ACL_NOT_CACHED; + ei->i_default_acl = EXT4_ACL_NOT_CACHED; + #endif +- ei->i_block_alloc_info = NULL; + ei->vfs_inode.i_version = 1; + ei->vfs_inode.i_data.writeback_index = 0; + memset(&ei->i_cached_extent, 0, sizeof(struct ext4_ext_cache)); +@@ -632,7 +631,6 @@ static void destroy_inodecache(void) + + static void ext4_clear_inode(struct inode *inode) + { +- struct ext4_block_alloc_info *rsv = EXT4_I(inode)->i_block_alloc_info; + #ifdef CONFIG_EXT4DEV_FS_POSIX_ACL + if (EXT4_I(inode)->i_acl && + EXT4_I(inode)->i_acl != EXT4_ACL_NOT_CACHED) { +@@ -645,10 +643,7 @@ static void ext4_clear_inode(struct inode *inode) + EXT4_I(inode)->i_default_acl = EXT4_ACL_NOT_CACHED; + } + #endif +- ext4_discard_reservation(inode); +- EXT4_I(inode)->i_block_alloc_info = NULL; +- if (unlikely(rsv)) +- kfree(rsv); ++ ext4_discard_preallocations(inode); + jbd2_journal_release_jbd_inode(EXT4_SB(inode->i_sb)->s_journal, + &EXT4_I(inode)->jinode); + } +@@ -759,8 +754,6 @@ static int ext4_show_options(struct seq_file *seq, struct vfsmount *vfs) + seq_puts(seq, ",nobh"); + if (!test_opt(sb, EXTENTS)) + seq_puts(seq, ",noextents"); +- if (!test_opt(sb, MBALLOC)) +- seq_puts(seq, ",nomballoc"); + if (test_opt(sb, I_VERSION)) + seq_puts(seq, ",i_version"); + if (!test_opt(sb, DELALLOC)) +@@ -1372,12 +1365,6 @@ set_qf_format: + case Opt_nodelalloc: + clear_opt(sbi->s_mount_opt, DELALLOC); + break; +- case Opt_mballoc: +- set_opt(sbi->s_mount_opt, MBALLOC); +- break; +- case Opt_nomballoc: +- clear_opt(sbi->s_mount_opt, MBALLOC); +- break; + case Opt_stripe: + if (match_int(&args[0], &option)) + return 0; +@@ -2039,11 +2026,6 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) + ext4_warning(sb, __func__, + "extents feature not enabled on this filesystem, " + "use tune2fs.\n"); +- /* +- * turn on mballoc code by default in ext4 filesystem +- * Use -o nomballoc to turn it off +- */ +- set_opt(sbi->s_mount_opt, MBALLOC); + + /* + * enable delayed allocation by default +@@ -2300,19 +2282,6 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) + goto failed_mount3; + } + +- /* per fileystem reservation list head & lock */ +- spin_lock_init(&sbi->s_rsv_window_lock); +- sbi->s_rsv_window_root = RB_ROOT; +- /* Add a single, static dummy reservation to the start of the +- * reservation window list --- it gives us a placeholder for +- * append-at-start-of-list which makes the allocation logic +- * _much_ simpler. */ +- sbi->s_rsv_window_head.rsv_start = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- sbi->s_rsv_window_head.rsv_end = EXT4_RESERVE_WINDOW_NOT_ALLOCATED; +- sbi->s_rsv_window_head.rsv_alloc_hit = 0; +- sbi->s_rsv_window_head.rsv_goal_size = 0; +- ext4_rsv_window_add(sb, &sbi->s_rsv_window_head); +- + sbi->s_stripe = ext4_get_stripe_size(sbi); + + /* +@@ -2509,7 +2478,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) + printk(KERN_INFO "EXT4-fs: delayed allocation enabled\n"); + + ext4_ext_init(sb); +- ext4_mb_init(sb, needs_recovery); ++ err = ext4_mb_init(sb, needs_recovery); ++ if (err) { ++ printk(KERN_ERR "EXT4-fs: failed to initalize mballoc (%d)\n", ++ err); ++ goto failed_mount4; ++ } + + lock_kernel(); + return 0; diff --git a/series b/series index 284615c1..8be8f53f 100644 --- a/series +++ b/series @@ -47,6 +47,7 @@ ext4-fix-namei-stack-usage make-proc-generic centralize-proc-functions inode-readahead +remove-old-balloc ########################################## # unstable patches -- 2.11.4.GIT