X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=fs%2Fext3%2Fballoc.c;h=fd75cf8f7181e29ff1b0c78b48afa05e3d72539e;hb=refs%2Fheads%2Fvserver;hp=893b08bd5c1798a2375382dbc787222bcc1f9126;hpb=5273a3df6485dc2ad6aa7ddd441b9a21970f003b;p=linux-2.6.git diff --git a/fs/ext3/balloc.c b/fs/ext3/balloc.c index 893b08bd5..fd75cf8f7 100644 --- a/fs/ext3/balloc.c +++ b/fs/ext3/balloc.c @@ -11,14 +11,16 @@ * David S. Miller (davem@caip.rutgers.edu), 1995 */ -#include #include +#include #include #include #include #include #include #include +#include +#include /* * balloc.c contains the blocks allocation and deallocation routines @@ -38,42 +40,54 @@ #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) +/** + * ext3_get_group_desc() -- load group descriptor from disk + * @sb: super block + * @block_group: given block group + * @bh: pointer to the buffer head to store the block + * group descriptor + */ struct ext3_group_desc * ext3_get_group_desc(struct super_block * sb, unsigned int block_group, struct buffer_head ** bh) { unsigned long group_desc; - unsigned long desc; - struct ext3_group_desc * gdp; + unsigned long offset; + struct ext3_group_desc * desc; + struct ext3_sb_info *sbi = EXT3_SB(sb); - if (block_group >= EXT3_SB(sb)->s_groups_count) { + if (block_group >= sbi->s_groups_count) { ext3_error (sb, "ext3_get_group_desc", "block_group >= groups_count - " "block_group = %d, groups_count = %lu", - block_group, EXT3_SB(sb)->s_groups_count); + block_group, sbi->s_groups_count); return NULL; } + smp_rmb(); - group_desc = block_group / EXT3_DESC_PER_BLOCK(sb); - desc = block_group % EXT3_DESC_PER_BLOCK(sb); - if (!EXT3_SB(sb)->s_group_desc[group_desc]) { + group_desc = block_group >> EXT3_DESC_PER_BLOCK_BITS(sb); + offset = block_group & (EXT3_DESC_PER_BLOCK(sb) - 1); + if (!sbi->s_group_desc[group_desc]) { ext3_error (sb, "ext3_get_group_desc", "Group descriptor not loaded - " "block_group = %d, group_desc = %lu, desc = %lu", - block_group, group_desc, desc); + block_group, group_desc, offset); return NULL; } - gdp = (struct ext3_group_desc *) - EXT3_SB(sb)->s_group_desc[group_desc]->b_data; + desc = (struct ext3_group_desc *) sbi->s_group_desc[group_desc]->b_data; if (bh) - *bh = EXT3_SB(sb)->s_group_desc[group_desc]; - return gdp + desc; + *bh = sbi->s_group_desc[group_desc]; + return desc + offset; } -/* - * Read the bitmap for a given block_group, reading into the specified +/** + * read_block_bitmap() + * @sb: super block + * @block_group: given block group + * + * Read the bitmap for a given block_group, reading into the specified * slot in the superblock's bitmap cache. * * Return buffer_head on success or NULL in case of failure. @@ -91,46 +105,343 @@ read_block_bitmap(struct super_block *sb, unsigned int block_group) if (!bh) ext3_error (sb, "read_block_bitmap", "Cannot read block bitmap - " - "block_group = %d, block_bitmap = %lu", - block_group, (unsigned long) desc->bg_block_bitmap); + "block_group = %d, block_bitmap = %u", + block_group, le32_to_cpu(desc->bg_block_bitmap)); error_out: 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 ext3_reserve_window_node *rsv, *prev; + int bad; + +restart: + n = rb_first(root); + bad = 0; + prev = NULL; + + printk("Block Allocation Reservation Windows Map (%s):\n", fn); + while (n) { + rsv = rb_entry(n, struct ext3_reserve_window_node, rsv_node); + if (verbose) + printk("reservation window 0x%p " + "start: %lu, end: %lu\n", + rsv, rsv->rsv_start, rsv->rsv_end); + if (rsv->rsv_start && rsv->rsv_start >= rsv->rsv_end) { + printk("Bad reservation %p (start >= end)\n", + rsv); + bad = 1; + } + if (prev && prev->rsv_end >= rsv->rsv_start) { + printk("Bad reservation %p (prev->end >= start)\n", + rsv); + bad = 1; + } + if (bad) { + if (!verbose) { + printk("Restarting reservation walk in verbose mode\n"); + verbose = 1; + goto restart; + } + } + n = rb_next(n); + prev = rsv; + } + printk("Window map complete.\n"); + if (bad) + BUG(); +} +#define rsv_window_dump(root, verbose) \ + __rsv_window_dump((root), (verbose), __FUNCTION__) +#else +#define rsv_window_dump(root, verbose) do {} while (0) +#endif -/* Free given blocks, update quota and i_blocks field */ -void ext3_free_blocks (handle_t *handle, struct inode * inode, - unsigned long block, unsigned long count) +/** + * 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 ext3_reserve_window *rsv, ext3_grpblk_t grp_goal, + unsigned int group, struct super_block * sb) +{ + ext3_fsblk_t group_first_block, group_last_block; + + group_first_block = ext3_group_first_block_no(sb, group); + group_last_block = group_first_block + (EXT3_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 ext3_reserve_window_node * +search_reserve_window(struct rb_root *root, ext3_fsblk_t goal) +{ + struct rb_node *n = root->rb_node; + struct ext3_reserve_window_node *rsv; + + if (!n) + return NULL; + + do { + rsv = rb_entry(n, struct ext3_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 ext3_reserve_window_node, rsv_node); + } + return rsv; +} + +/** + * ext3_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 ext3_rsv_window_add(struct super_block *sb, + struct ext3_reserve_window_node *rsv) +{ + struct rb_root *root = &EXT3_SB(sb)->s_rsv_window_root; + struct rb_node *node = &rsv->rsv_node; + ext3_fsblk_t start = rsv->rsv_start; + + struct rb_node ** p = &root->rb_node; + struct rb_node * parent = NULL; + struct ext3_reserve_window_node *this; + + while (*p) + { + parent = *p; + this = rb_entry(parent, struct ext3_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); +} + +/** + * ext3_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 ext3_reserve_window_node *rsv) +{ + rsv->rsv_start = EXT3_RESERVE_WINDOW_NOT_ALLOCATED; + rsv->rsv_end = EXT3_RESERVE_WINDOW_NOT_ALLOCATED; + rsv->rsv_alloc_hit = 0; + rb_erase(&rsv->rsv_node, &EXT3_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 EXT3_RESERVE_WINDOW_NOT_ALLOCATED. + */ +static inline int rsv_is_empty(struct ext3_reserve_window *rsv) +{ + /* a valid reservation end block could not be 0 */ + return rsv->_rsv_end == EXT3_RESERVE_WINDOW_NOT_ALLOCATED; +} + +/** + * ext3_init_block_alloc_info() + * @inode: file inode structure + * + * Allocate and initialize the reservation window structure, and + * link the window to the ext3 inode structure at last + * + * The reservation window structure is only dynamically allocated + * and linked to ext3 inode the first time the open file + * needs a new block. So, before every ext3_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 ext3_get_blocks_handle(), also called + * when setting the reservation window size through ioctl before the file + * is open for write (needs block allocation). + * + * Needs truncate_mutex protection prior to call this function. + */ +void ext3_init_block_alloc_info(struct inode *inode) +{ + struct ext3_inode_info *ei = EXT3_I(inode); + struct ext3_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 ext3_reserve_window_node *rsv = &block_i->rsv_window_node; + + rsv->rsv_start = EXT3_RESERVE_WINDOW_NOT_ALLOCATED; + rsv->rsv_end = EXT3_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 = EXT3_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; +} + +/** + * ext3_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: + * ext3_release_file(): last writer close the file + * ext3_clear_inode(): last iput(), when nobody link to this file. + * ext3_truncate(): when the block indirect map is about to change. + * + */ +void ext3_discard_reservation(struct inode *inode) +{ + struct ext3_inode_info *ei = EXT3_I(inode); + struct ext3_block_alloc_info *block_i = ei->i_block_alloc_info; + struct ext3_reserve_window_node *rsv; + spinlock_t *rsv_lock = &EXT3_SB(inode->i_sb)->s_rsv_window_lock; + + 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); + } +} + +/** + * ext3_free_blocks_sb() -- Free given blocks and update quota + * @handle: handle to this transaction + * @sb: super block + * @block: start physcial block to free + * @count: number of blocks to free + * @pdquot_freed_blocks: pointer to quota + */ +void ext3_free_blocks_sb(handle_t *handle, struct super_block *sb, + ext3_fsblk_t block, unsigned long count, + unsigned long *pdquot_freed_blocks) { struct buffer_head *bitmap_bh = NULL; struct buffer_head *gd_bh; unsigned long block_group; - unsigned long bit; + ext3_grpblk_t bit; unsigned long i; unsigned long overflow; - struct super_block * sb; - struct ext3_group_desc * gdp; + struct ext3_group_desc * desc; struct ext3_super_block * es; struct ext3_sb_info *sbi; int err = 0, ret; - int dquot_freed_blocks = 0; + ext3_grpblk_t group_freed; - sb = inode->i_sb; - if (!sb) { - printk ("ext3_free_blocks: nonexistent device"); - return; - } + *pdquot_freed_blocks = 0; sbi = EXT3_SB(sb); - es = EXT3_SB(sb)->s_es; + es = sbi->s_es; if (block < le32_to_cpu(es->s_first_data_block) || block + count < block || block + count > le32_to_cpu(es->s_blocks_count)) { ext3_error (sb, "ext3_free_blocks", "Freeing blocks not in datazone - " - "block = %lu, count = %lu", block, count); + "block = "E3FSBLK", count = %lu", block, count); goto error_return; } - ext3_debug ("freeing block %lu\n", block); + ext3_debug ("freeing block(s) %lu-%lu\n", block, block + count - 1); do_more: overflow = 0; @@ -150,19 +461,19 @@ do_more: bitmap_bh = read_block_bitmap(sb, block_group); if (!bitmap_bh) goto error_return; - gdp = ext3_get_group_desc (sb, block_group, &gd_bh); - if (!gdp) + desc = ext3_get_group_desc (sb, block_group, &gd_bh); + if (!desc) goto error_return; - if (in_range (le32_to_cpu(gdp->bg_block_bitmap), block, count) || - in_range (le32_to_cpu(gdp->bg_inode_bitmap), block, count) || - in_range (block, le32_to_cpu(gdp->bg_inode_table), - EXT3_SB(sb)->s_itb_per_group) || - in_range (block + count - 1, le32_to_cpu(gdp->bg_inode_table), - EXT3_SB(sb)->s_itb_per_group)) + if (in_range (le32_to_cpu(desc->bg_block_bitmap), block, count) || + in_range (le32_to_cpu(desc->bg_inode_bitmap), block, count) || + in_range (block, le32_to_cpu(desc->bg_inode_table), + sbi->s_itb_per_group) || + in_range (block + count - 1, le32_to_cpu(desc->bg_inode_table), + sbi->s_itb_per_group)) ext3_error (sb, "ext3_free_blocks", "Freeing blocks in system zones - " - "Block = %lu, count = %lu", + "Block = "E3FSBLK", count = %lu", block, count); /* @@ -171,7 +482,7 @@ do_more: */ /* @@@ check errors */ BUFFER_TRACE(bitmap_bh, "getting undo access"); - err = ext3_journal_get_undo_access(handle, bitmap_bh, NULL); + err = ext3_journal_get_undo_access(handle, bitmap_bh); if (err) goto error_return; @@ -187,7 +498,7 @@ do_more: jbd_lock_bh_state(bitmap_bh); - for (i = 0; i < count; i++) { + for (i = 0, group_freed = 0; i < count; i++) { /* * An HJ special. This is expensive... */ @@ -207,10 +518,15 @@ do_more: } jbd_lock_bh_state(bitmap_bh); #endif + if (need_resched()) { + jbd_unlock_bh_state(bitmap_bh); + cond_resched(); + jbd_lock_bh_state(bitmap_bh); + } /* @@@ This prevents newly-allocated data from being * freed and then reallocated within the same - * transaction. - * + * transaction. + * * Ideally we would want to allow that to happen, but to * do so requires making journal_forget() capable of * revoking the queued write of a data block, which @@ -223,7 +539,7 @@ do_more: * safe not to set the allocation bit in the committed * bitmap, because we know that there is no outstanding * activity on the buffer any more and so it is safe to - * reallocate it. + * reallocate it. */ BUFFER_TRACE(bitmap_bh, "set in b_committed_data"); J_ASSERT_BH(bitmap_bh, @@ -241,19 +557,20 @@ do_more: bit + i, bitmap_bh->b_data)) { jbd_unlock_bh_state(bitmap_bh); ext3_error(sb, __FUNCTION__, - "bit already cleared for block %lu", block + i); + "bit already cleared for block "E3FSBLK, + block + i); jbd_lock_bh_state(bitmap_bh); BUFFER_TRACE(bitmap_bh, "bit already cleared"); } else { - dquot_freed_blocks++; + group_freed++; } } jbd_unlock_bh_state(bitmap_bh); spin_lock(sb_bgl_lock(sbi, block_group)); - gdp->bg_free_blocks_count = - cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count) + - dquot_freed_blocks); + desc->bg_free_blocks_count = + cpu_to_le16(le16_to_cpu(desc->bg_free_blocks_count) + + group_freed); spin_unlock(sb_bgl_lock(sbi, block_group)); percpu_counter_mod(&sbi->s_freeblocks_counter, count); @@ -265,6 +582,7 @@ do_more: BUFFER_TRACE(gd_bh, "dirtied group descriptor block"); ret = ext3_journal_dirty_metadata(handle, gd_bh); if (!err) err = ret; + *pdquot_freed_blocks += group_freed; if (overflow && !err) { block += count; @@ -275,12 +593,40 @@ do_more: error_return: brelse(bitmap_bh); ext3_std_error(sb, err); - if (dquot_freed_blocks) + return; +} + +/** + * ext3_free_blocks() -- Free given blocks and update quota + * @handle: handle for this transaction + * @inode: inode + * @block: start physical block to free + * @count: number of blocks to count + */ +void ext3_free_blocks(handle_t *handle, struct inode *inode, + ext3_fsblk_t block, unsigned long count) +{ + struct super_block * sb; + unsigned long dquot_freed_blocks; + + sb = inode->i_sb; + if (!sb) { + printk ("ext3_free_blocks: nonexistent device"); + return; + } + ext3_free_blocks_sb(handle, sb, block, count, &dquot_freed_blocks); + if (dquot_freed_blocks) { + DLIMIT_FREE_BLOCK(inode, dquot_freed_blocks); DQUOT_FREE_BLOCK(inode, dquot_freed_blocks); + } return; } -/* +/** + * ext3_test_allocatable() + * @nr: given allocation block group + * @bh: bufferhead contains the bitmap of the given block group + * * For ext3 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 @@ -290,13 +636,13 @@ error_return: * 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. + * 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 inline int ext3_test_allocatable(int nr, struct buffer_head *bh) +static int ext3_test_allocatable(ext3_grpblk_t nr, struct buffer_head *bh) { int ret; struct journal_head *jh = bh2jh(bh); @@ -313,30 +659,70 @@ static inline int ext3_test_allocatable(int nr, struct buffer_head *bh) return ret; } -/* - * Find an allocatable block in a bitmap. We honour both the bitmap and +/** + * 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 ext3_grpblk_t +bitmap_search_next_usable_block(ext3_grpblk_t start, struct buffer_head *bh, + ext3_grpblk_t maxblocks) +{ + ext3_grpblk_t next; + struct journal_head *jh = bh2jh(bh); + + while (start < maxblocks) { + next = ext3_find_next_zero_bit(bh->b_data, maxblocks, start); + if (next >= maxblocks) + return -1; + if (ext3_test_allocatable(next, bh)) + return next; + jbd_lock_bh_state(bh); + if (jh->b_committed_data) + start = ext3_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 int -find_next_usable_block(int start, struct buffer_head *bh, int maxblocks) +static ext3_grpblk_t +find_next_usable_block(ext3_grpblk_t start, struct buffer_head *bh, + ext3_grpblk_t maxblocks) { - int here, next; + ext3_grpblk_t here, next; char *p, *r; - struct journal_head *jh = bh2jh(bh); if (start > 0) { /* - * The goal was occupied; search forward for a free + * 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 EXT3_BLOCKS_PER_GROUP. Aligning up to the * next 64-bit boundary is simple.. */ - int end_goal = (start + 63) & ~63; + ext3_grpblk_t end_goal = (start + 63) & ~63; + if (end_goal > maxblocks) + end_goal = maxblocks; here = ext3_find_next_zero_bit(bh->b_data, end_goal, start); if (here < end_goal && ext3_test_allocatable(here, bh)) return here; @@ -348,10 +734,10 @@ find_next_usable_block(int start, struct buffer_head *bh, int maxblocks) here = 0; p = ((char *)bh->b_data) + (here >> 3); - r = memscan(p, 0, (maxblocks - here + 7) >> 3); + r = memscan(p, 0, ((maxblocks + 7) >> 3) - (here >> 3)); next = (r - ((char *)bh->b_data)) << 3; - if (next < maxblocks && ext3_test_allocatable(next, bh)) + if (next < maxblocks && next >= start && ext3_test_allocatable(next, bh)) return next; /* @@ -359,22 +745,15 @@ find_next_usable_block(int start, struct buffer_head *bh, int maxblocks) * bitmap and the last-committed copy until we find a bit free in * both */ - while (here < maxblocks) { - next = ext3_find_next_zero_bit(bh->b_data, maxblocks, here); - if (next >= maxblocks) - return -1; - if (ext3_test_allocatable(next, bh)) - return next; - jbd_lock_bh_state(bh); - if (jh->b_committed_data) - here = ext3_find_next_zero_bit(jh->b_committed_data, - maxblocks, next); - jbd_unlock_bh_state(bh); - } - return -1; + 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_ @@ -382,7 +761,7 @@ find_next_usable_block(int start, struct buffer_head *bh, int maxblocks) * zero (failure). */ static inline int -claim_block(spinlock_t *lock, int block, struct buffer_head *bh) +claim_block(spinlock_t *lock, ext3_grpblk_t block, struct buffer_head *bh) { struct journal_head *jh = bh2jh(bh); int ret; @@ -400,18 +779,474 @@ claim_block(spinlock_t *lock, int block, struct buffer_head *bh) return ret; } -/* +/** + * ext3_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 * ext3_journal_release_buffer(), else we'll run out of credits. */ -static int +static ext3_grpblk_t ext3_try_to_allocate(struct super_block *sb, handle_t *handle, int group, - struct buffer_head *bitmap_bh, int goal, int *errp) + struct buffer_head *bitmap_bh, ext3_grpblk_t grp_goal, + unsigned long *count, struct ext3_reserve_window *my_rsv) { - int i; + ext3_fsblk_t group_first_block; + ext3_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 = ext3_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 > EXT3_BLOCKS_PER_GROUP(sb)) + /* reservation window crosses group boundary */ + end = EXT3_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 = EXT3_BLOCKS_PER_GROUP(sb); + } + + BUG_ON(start > EXT3_BLOCKS_PER_GROUP(sb)); + +repeat: + if (grp_goal < 0 || !ext3_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 && + ext3_test_allocatable(grp_goal - 1, + bitmap_bh); + i++, grp_goal--) + ; + } + } + start = grp_goal; + + if (!claim_block(sb_bgl_lock(EXT3_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 + && ext3_test_allocatable(grp_goal, bitmap_bh) + && claim_block(sb_bgl_lock(EXT3_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 ext3_reserve_window_node *search_head, + struct ext3_reserve_window_node *my_rsv, + struct super_block * sb, + ext3_fsblk_t start_block, + ext3_fsblk_t last_block) +{ + struct rb_node *next; + struct ext3_reserve_window_node *rsv, *prev; + ext3_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 ext3_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) + ext3_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 ext3_reserve_window_node *my_rsv, + ext3_grpblk_t grp_goal, struct super_block *sb, + unsigned int group, struct buffer_head *bitmap_bh) +{ + struct ext3_reserve_window_node *search_head; + ext3_fsblk_t group_first_block, group_end_block, start_block; + ext3_grpblk_t first_free_block; + struct rb_root *fs_rsv_root = &EXT3_SB(sb)->s_rsv_window_root; + unsigned long size; + int ret; + spinlock_t *rsv_lock = &EXT3_SB(sb)->s_rsv_window_lock; + + group_first_block = ext3_group_first_block_no(sb, group); + group_end_block = group_first_block + (EXT3_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 > EXT3_MAX_RESERVE_BLOCKS) + size = EXT3_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 ext3_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 ext3_new_blocks() tries to allocate blocks, + */ +static void try_to_extend_reservation(struct ext3_reserve_window_node *my_rsv, + struct super_block *sb, int size) +{ + struct ext3_reserve_window_node *next_rsv; + struct rb_node *next; + spinlock_t *rsv_lock = &EXT3_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 ext3_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); +} + +/** + * ext3_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 ext3_grpblk_t +ext3_try_to_allocate_with_rsv(struct super_block *sb, handle_t *handle, + unsigned int group, struct buffer_head *bitmap_bh, + ext3_grpblk_t grp_goal, + struct ext3_reserve_window_node * my_rsv, + unsigned long *count, int *errp) +{ + ext3_fsblk_t group_first_block, group_last_block; + ext3_grpblk_t ret = 0; int fatal; - int credits = 0; + unsigned long num = *count; *errp = 0; @@ -421,78 +1256,199 @@ ext3_try_to_allocate(struct super_block *sb, handle_t *handle, int group, * if the buffer is in BJ_Forget state in the committing transaction. */ BUFFER_TRACE(bitmap_bh, "get undo access for new block"); - fatal = ext3_journal_get_undo_access(handle, bitmap_bh, &credits); + fatal = ext3_journal_get_undo_access(handle, bitmap_bh); if (fatal) { *errp = fatal; - goto fail; + return -1; } -repeat: - if (goal < 0 || !ext3_test_allocatable(goal, bitmap_bh)) { - goal = find_next_usable_block(goal, bitmap_bh, - EXT3_BLOCKS_PER_GROUP(sb)); - if (goal < 0) - goto fail_access; - - for (i = 0; i < 7 && goal > 0 && - ext3_test_allocatable(goal - 1, bitmap_bh); - i++, goal--); + /* + * 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 = ext3_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 < EXT3_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 = ext3_group_first_block_no(sb, group); + group_last_block = group_first_block + (EXT3_BLOCKS_PER_GROUP(sb) - 1); - if (!claim_block(sb_bgl_lock(EXT3_SB(sb), group), goal, bitmap_bh)) { - /* - * The block was allocated by another thread, or it was - * allocated and then freed by another thread - */ - goal++; - if (goal >= EXT3_BLOCKS_PER_GROUP(sb)) - goto fail_access; - goto repeat; - } + /* + * 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); + } - BUFFER_TRACE(bitmap_bh, "journal_dirty_metadata for bitmap block"); - fatal = ext3_journal_dirty_metadata(handle, bitmap_bh); - if (fatal) { - *errp = fatal; - goto fail; + if ((my_rsv->rsv_start > group_last_block) || + (my_rsv->rsv_end < group_first_block)) { + rsv_window_dump(&EXT3_SB(sb)->s_rsv_window_root, 1); + BUG(); + } + ret = ext3_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 = ext3_journal_dirty_metadata(handle, bitmap_bh); + if (fatal) { + *errp = fatal; + return -1; + } + return ret; } - return goal; -fail_access: BUFFER_TRACE(bitmap_bh, "journal_release_buffer"); - ext3_journal_release_buffer(handle, bitmap_bh, credits); -fail: - return -1; + ext3_journal_release_buffer(handle, bitmap_bh); + return ret; } -/* - * ext3_new_block uses a goal block to assist allocation. If the goal is - * free, or there is a free block within 32 blocks of the goal, that block - * is allocated. Otherwise a forward search is made for a free block; within - * each block group the search first looks for an entire free byte in the block - * bitmap, and then for any free bit if that fails. - * This function also updates quota and i_blocks field. +/** + * ext3_has_free_blocks() + * @sbi: in-core super block structure. + * + * Check if filesystem has at least 1 free block available for allocation. + */ +static int ext3_has_free_blocks(struct super_block *sb) +{ + struct ext3_sb_info *sbi = EXT3_SB(sb); + unsigned long long free_blocks, root_blocks; + int cond; + + free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter); + root_blocks = le32_to_cpu(sbi->s_es->s_r_blocks_count); + + vxdprintk(VXD_CBIT(dlim, 3), + "ext3_has_free_blocks(%p): free=%llu, root=%llu", + sb, free_blocks, root_blocks); + + DLIMIT_ADJUST_BLOCK(sb, dx_current_tag(), &free_blocks, &root_blocks); + + cond = (free_blocks < root_blocks + 1 && + !capable(CAP_SYS_RESOURCE) && + sbi->s_resuid != current->fsuid && + (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))); + + vxdprintk(VXD_CBIT(dlim, 3), + "ext3_has_free_blocks(%p): %llu<%llu+1, %c, %u!=%u r=%d", + sb, free_blocks, root_blocks, + !capable(CAP_SYS_RESOURCE)?'1':'0', + sbi->s_resuid, current->fsuid, cond?0:1); + + return (cond ? 0 : 1); +} + +/** + * ext3_should_retry_alloc() + * @sb: super block + * @retries number of attemps has been made + * + * ext3_should_retry_alloc() is called when ENOSPC is returned, and if + * it is profitable to retry the operation, this function will wait + * for the current or commiting transaction to complete, and then + * return TRUE. + * + * if the total number of retries exceed three times, return FALSE. + */ +int ext3_should_retry_alloc(struct super_block *sb, int *retries) +{ + if (!ext3_has_free_blocks(sb) || (*retries)++ > 3) + return 0; + + jbd_debug(1, "%s: retrying operation after ENOSPC\n", sb->s_id); + + return journal_force_commit_nested(EXT3_SB(sb)->s_journal); +} + +/** + * ext3_new_blocks() -- core block(s) 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 + * + * ext3_new_blocks uses a goal block to assist 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. + * */ -int -ext3_new_block(handle_t *handle, struct inode *inode, unsigned long goal, - u32 *prealloc_count, u32 *prealloc_block, int *errp) -{ - struct buffer_head *bitmap_bh = NULL; /* bh */ - struct buffer_head *gdp_bh; /* bh2 */ - int group_no; /* i */ - int ret_block; /* j */ - int bgi; /* blockgroup iteration index */ - int target_block; /* tmp */ +ext3_fsblk_t ext3_new_blocks(handle_t *handle, struct inode *inode, + ext3_fsblk_t goal, unsigned long *count, int *errp) +{ + struct buffer_head *bitmap_bh = NULL; + struct buffer_head *gdp_bh; + int group_no; + int goal_group; + ext3_grpblk_t grp_target_blk; /* blockgroup relative goal block */ + ext3_grpblk_t grp_alloc_blk; /* blockgroup-relative allocated block*/ + ext3_fsblk_t ret_block; /* filesyetem-wide allocated block */ + int bgi; /* blockgroup iteration index */ int fatal = 0, err; int performed_allocation = 0; - int free_blocks, root_blocks; + ext3_grpblk_t free_blocks; /* number of free blocks in a group */ struct super_block *sb; struct ext3_group_desc *gdp; struct ext3_super_block *es; struct ext3_sb_info *sbi; + struct ext3_reserve_window_node *my_rsv = NULL; + struct ext3_block_alloc_info *block_i; + unsigned short windowsz = 0; #ifdef EXT3FS_DEBUG static int goal_hits, goal_attempts; #endif + unsigned long ngroups; + unsigned long num = *count; + *errp = -ENOSPC; sb = inode->i_sb; if (!sb) { @@ -503,20 +1459,29 @@ ext3_new_block(handle_t *handle, struct inode *inode, unsigned long goal, /* * Check quota for allocation of this block. */ - if (DQUOT_ALLOC_BLOCK(inode, 1)) { + if (DQUOT_ALLOC_BLOCK(inode, num)) { *errp = -EDQUOT; return 0; } + if (DLIMIT_ALLOC_BLOCK(inode, 1)) + goto out_dlimit; sbi = EXT3_SB(sb); es = EXT3_SB(sb)->s_es; ext3_debug("goal=%lu.\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 EXT3_IOC_SETRSVSZ to set the window size to 0 to turn off + * reservation on that particular file) + */ + block_i = EXT3_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; - free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter); - root_blocks = le32_to_cpu(es->s_r_blocks_count); - if (free_blocks < root_blocks + 1 && !capable(CAP_SYS_RESOURCE) && - sbi->s_resuid != current->fsuid && - (sbi->s_resgid == 0 || !in_group_p (sbi->s_resgid))) { + if (!ext3_has_free_blocks(sb)) { *errp = -ENOSPC; goto out; } @@ -529,54 +1494,87 @@ ext3_new_block(handle_t *handle, struct inode *inode, unsigned long goal, goal = le32_to_cpu(es->s_first_data_block); group_no = (goal - le32_to_cpu(es->s_first_data_block)) / EXT3_BLOCKS_PER_GROUP(sb); + goal_group = group_no; +retry_alloc: gdp = ext3_get_group_desc(sb, group_no, &gdp_bh); if (!gdp) goto io_error; free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); + /* + * if there is not enough free blocks to make a new resevation + * turn off reservation for this allocation + */ + if (my_rsv && (free_blocks < windowsz) + && (rsv_is_empty(&my_rsv->rsv_window))) + my_rsv = NULL; + if (free_blocks > 0) { - ret_block = ((goal - le32_to_cpu(es->s_first_data_block)) % + grp_target_blk = ((goal - le32_to_cpu(es->s_first_data_block)) % EXT3_BLOCKS_PER_GROUP(sb)); bitmap_bh = read_block_bitmap(sb, group_no); if (!bitmap_bh) goto io_error; - ret_block = ext3_try_to_allocate(sb, handle, group_no, - bitmap_bh, ret_block, &fatal); + grp_alloc_blk = ext3_try_to_allocate_with_rsv(sb, handle, + group_no, bitmap_bh, grp_target_blk, + my_rsv, &num, &fatal); if (fatal) goto out; - if (ret_block >= 0) + if (grp_alloc_blk >= 0) goto allocated; } + ngroups = EXT3_SB(sb)->s_groups_count; + smp_rmb(); + /* - * Now search the rest of the groups. We assume that + * Now search the rest of the groups. We assume that * i and gdp correctly point to the last group visited. */ - for (bgi = 0; bgi < EXT3_SB(sb)->s_groups_count; bgi++) { + for (bgi = 0; bgi < ngroups; bgi++) { group_no++; - if (group_no >= EXT3_SB(sb)->s_groups_count) + if (group_no >= ngroups) group_no = 0; gdp = ext3_get_group_desc(sb, group_no, &gdp_bh); - if (!gdp) { - *errp = -EIO; - goto out; - } + if (!gdp) + goto io_error; free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); - if (free_blocks <= 0) + /* + * skip this group if the number of + * free blocks is less than half of the reservation + * window size. + */ + if (free_blocks <= (windowsz/2)) continue; brelse(bitmap_bh); bitmap_bh = read_block_bitmap(sb, group_no); if (!bitmap_bh) goto io_error; - ret_block = ext3_try_to_allocate(sb, handle, group_no, - bitmap_bh, -1, &fatal); + /* + * try to allocate block(s) from this group, without a goal(-1). + */ + grp_alloc_blk = ext3_try_to_allocate_with_rsv(sb, handle, + group_no, bitmap_bh, -1, my_rsv, + &num, &fatal); if (fatal) goto out; - if (ret_block >= 0) + 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; @@ -591,16 +1589,18 @@ allocated: if (fatal) goto out; - target_block = ret_block + group_no * EXT3_BLOCKS_PER_GROUP(sb) - + le32_to_cpu(es->s_first_data_block); + ret_block = grp_alloc_blk + ext3_group_first_block_no(sb, group_no); - if (target_block == le32_to_cpu(gdp->bg_block_bitmap) || - target_block == le32_to_cpu(gdp->bg_inode_bitmap) || - in_range(target_block, le32_to_cpu(gdp->bg_inode_table), + if (in_range(le32_to_cpu(gdp->bg_block_bitmap), ret_block, num) || + in_range(le32_to_cpu(gdp->bg_inode_bitmap), ret_block, num) || + in_range(ret_block, le32_to_cpu(gdp->bg_inode_table), + EXT3_SB(sb)->s_itb_per_group) || + in_range(ret_block + num - 1, le32_to_cpu(gdp->bg_inode_table), EXT3_SB(sb)->s_itb_per_group)) ext3_error(sb, "ext3_new_block", "Allocating block in system zone - " - "block = %u", target_block); + "blocks from "E3FSBLK", length %lu", + ret_block, num); performed_allocation = 1; @@ -609,7 +1609,7 @@ allocated: struct buffer_head *debug_bh; /* Record bitmap buffer state in the newly allocated block */ - debug_bh = sb_find_get_block(sb, target_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"); @@ -619,23 +1619,24 @@ allocated: 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) { - if (ext3_test_bit(ret_block, - bh2jh(bitmap_bh)->b_committed_data)) { - printk("%s: block was unexpectedly set in " - "b_committed_data\n", __FUNCTION__); + int i; + + for (i = 0; i < num; i++) { + if (ext3_test_bit(grp_alloc_blk+i, + bh2jh(bitmap_bh)->b_committed_data)) { + printk("%s: block was unexpectedly set in " + "b_committed_data\n", __FUNCTION__); + } } } - ext3_debug("found bit %d\n", ret_block); + ext3_debug("found bit %d\n", grp_alloc_blk); spin_unlock(sb_bgl_lock(sbi, group_no)); jbd_unlock_bh_state(bitmap_bh); #endif - /* ret_block was blockgroup-relative. Now it becomes fs-relative */ - ret_block = target_block; - - if (ret_block >= le32_to_cpu(es->s_blocks_count)) { + if (ret_block + num - 1 >= le32_to_cpu(es->s_blocks_count)) { ext3_error(sb, "ext3_new_block", - "block(%d) >= blocks count(%d) - " + "block("E3FSBLK") >= blocks count(%d) - " "block_group = %d, es == %p ", ret_block, le32_to_cpu(es->s_blocks_count), group_no, es); goto out; @@ -646,14 +1647,14 @@ allocated: * list of some description. We don't know in advance whether * the caller wants to use it as metadata or data. */ - ext3_debug("allocating block %d. Goal hits %d of %d.\n", + ext3_debug("allocating block %lu. Goal hits %d of %d.\n", ret_block, goal_hits, goal_attempts); spin_lock(sb_bgl_lock(sbi, group_no)); gdp->bg_free_blocks_count = - cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count) - 1); + cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count)-num); spin_unlock(sb_bgl_lock(sbi, group_no)); - percpu_counter_mod(&sbi->s_freeblocks_counter, -1); + percpu_counter_mod(&sbi->s_freeblocks_counter, -num); BUFFER_TRACE(gdp_bh, "journal_dirty_metadata for group descriptor"); err = ext3_journal_dirty_metadata(handle, gdp_bh); @@ -666,11 +1667,16 @@ allocated: *errp = 0; brelse(bitmap_bh); + DQUOT_FREE_BLOCK(inode, *count-num); + *count = num; return ret_block; io_error: *errp = -EIO; out: + if (!performed_allocation) + DLIMIT_FREE_BLOCK(inode, 1); +out_dlimit: if (fatal) { *errp = fatal; ext3_std_error(sb, fatal); @@ -679,27 +1685,44 @@ out: * Undo the block allocation */ if (!performed_allocation) - DQUOT_FREE_BLOCK(inode, 1); + DQUOT_FREE_BLOCK(inode, *count); brelse(bitmap_bh); return 0; } -unsigned long ext3_count_free_blocks(struct super_block *sb) +ext3_fsblk_t ext3_new_block(handle_t *handle, struct inode *inode, + ext3_fsblk_t goal, int *errp) +{ + unsigned long count = 1; + + return ext3_new_blocks(handle, inode, goal, &count, errp); +} + +/** + * ext3_count_free_blocks() -- count filesystem free blocks + * @sb: superblock + * + * Adds up the number of free blocks from each block group. + */ +ext3_fsblk_t ext3_count_free_blocks(struct super_block *sb) { - unsigned long desc_count; + ext3_fsblk_t desc_count; struct ext3_group_desc *gdp; int i; + unsigned long ngroups = EXT3_SB(sb)->s_groups_count; #ifdef EXT3FS_DEBUG struct ext3_super_block *es; - unsigned long bitmap_count, x; + ext3_fsblk_t bitmap_count; + unsigned long x; struct buffer_head *bitmap_bh = NULL; - lock_super(sb); es = EXT3_SB(sb)->s_es; desc_count = 0; bitmap_count = 0; gdp = NULL; - for (i = 0; i < EXT3_SB(sb)->s_groups_count; i++) { + + smp_rmb(); + for (i = 0; i < ngroups; i++) { gdp = ext3_get_group_desc(sb, i, NULL); if (!gdp) continue; @@ -715,13 +1738,15 @@ unsigned long ext3_count_free_blocks(struct super_block *sb) bitmap_count += x; } brelse(bitmap_bh); - printk("ext3_count_free_blocks: stored = %u, computed = %lu, %lu\n", - le32_to_cpu(es->s_free_blocks_count), desc_count, bitmap_count); - unlock_super(sb); + printk("ext3_count_free_blocks: stored = "E3FSBLK + ", computed = "E3FSBLK", "E3FSBLK"\n", + le32_to_cpu(es->s_free_blocks_count), + desc_count, bitmap_count); return bitmap_count; #else desc_count = 0; - for (i = 0; i < EXT3_SB(sb)->s_groups_count; i++) { + smp_rmb(); + for (i = 0; i < ngroups; i++) { gdp = ext3_get_group_desc(sb, i, NULL); if (!gdp) continue; @@ -732,9 +1757,8 @@ unsigned long ext3_count_free_blocks(struct super_block *sb) #endif } -static inline int block_in_use(unsigned long block, - struct super_block * sb, - unsigned char * map) +static inline int +block_in_use(ext3_fsblk_t block, struct super_block *sb, unsigned char *map) { return ext3_test_bit ((block - le32_to_cpu(EXT3_SB(sb)->s_es->s_first_data_block)) % @@ -743,21 +1767,21 @@ static inline int block_in_use(unsigned long block, static inline int test_root(int a, int b) { - if (a == 0) - return 1; - while (1) { - if (a == 1) - return 1; - if (a % b) - return 0; - a = a / b; - } + int num = b; + + while (a > num) + num *= b; + return num == a; } -int ext3_group_sparse(int group) +static int ext3_group_sparse(int group) { - return (test_root(group, 3) || test_root(group, 5) || - test_root(group, 7)); + if (group <= 1) + return 1; + if (!(group & 1)) + return 0; + return (test_root(group, 7) || test_root(group, 5) || + test_root(group, 3)); } /** @@ -770,12 +1794,33 @@ int ext3_group_sparse(int group) */ int ext3_bg_has_super(struct super_block *sb, int group) { - if (EXT3_HAS_RO_COMPAT_FEATURE(sb,EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER)&& - !ext3_group_sparse(group)) + if (EXT3_HAS_RO_COMPAT_FEATURE(sb, + EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER) && + !ext3_group_sparse(group)) return 0; return 1; } +static unsigned long ext3_bg_num_gdb_meta(struct super_block *sb, int group) +{ + unsigned long metagroup = group / EXT3_DESC_PER_BLOCK(sb); + unsigned long first = metagroup * EXT3_DESC_PER_BLOCK(sb); + unsigned long last = first + EXT3_DESC_PER_BLOCK(sb) - 1; + + if (group == first || group == first + 1 || group == last) + return 1; + return 0; +} + +static unsigned long ext3_bg_num_gdb_nometa(struct super_block *sb, int group) +{ + if (EXT3_HAS_RO_COMPAT_FEATURE(sb, + EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER) && + !ext3_group_sparse(group)) + return 0; + return EXT3_SB(sb)->s_gdb_count; +} + /** * ext3_bg_num_gdb - number of blocks used by the group table in group * @sb: superblock for filesystem @@ -787,82 +1832,14 @@ int ext3_bg_has_super(struct super_block *sb, int group) */ unsigned long ext3_bg_num_gdb(struct super_block *sb, int group) { - if (EXT3_HAS_RO_COMPAT_FEATURE(sb,EXT3_FEATURE_RO_COMPAT_SPARSE_SUPER)&& - !ext3_group_sparse(group)) - return 0; - return EXT3_SB(sb)->s_gdb_count; -} + unsigned long first_meta_bg = + le32_to_cpu(EXT3_SB(sb)->s_es->s_first_meta_bg); + unsigned long metagroup = group / EXT3_DESC_PER_BLOCK(sb); -#ifdef CONFIG_EXT3_CHECK -/* Called at mount-time, super-block is locked */ -void ext3_check_blocks_bitmap (struct super_block * sb) -{ - struct ext3_super_block *es; - unsigned long desc_count, bitmap_count, x, j; - unsigned long desc_blocks; - struct buffer_head *bitmap_bh = NULL; - struct ext3_group_desc *gdp; - int i; + if (!EXT3_HAS_INCOMPAT_FEATURE(sb,EXT3_FEATURE_INCOMPAT_META_BG) || + metagroup < first_meta_bg) + return ext3_bg_num_gdb_nometa(sb,group); - es = EXT3_SB(sb)->s_es; - desc_count = 0; - bitmap_count = 0; - gdp = NULL; - for (i = 0; i < EXT3_SB(sb)->s_groups_count; i++) { - gdp = ext3_get_group_desc (sb, i, NULL); - if (!gdp) - continue; - desc_count += le16_to_cpu(gdp->bg_free_blocks_count); - brelse(bitmap_bh); - bitmap_bh = read_block_bitmap(sb, i); - if (bitmap_bh == NULL) - continue; + return ext3_bg_num_gdb_meta(sb,group); - if (ext3_bg_has_super(sb, i) && - !ext3_test_bit(0, bitmap_bh->b_data)) - ext3_error(sb, __FUNCTION__, - "Superblock in group %d is marked free", i); - - desc_blocks = ext3_bg_num_gdb(sb, i); - for (j = 0; j < desc_blocks; j++) - if (!ext3_test_bit(j + 1, bitmap_bh->b_data)) - ext3_error(sb, __FUNCTION__, - "Descriptor block #%ld in group " - "%d is marked free", j, i); - - if (!block_in_use (le32_to_cpu(gdp->bg_block_bitmap), - sb, bitmap_bh->b_data)) - ext3_error (sb, "ext3_check_blocks_bitmap", - "Block bitmap for group %d is marked free", - i); - - if (!block_in_use (le32_to_cpu(gdp->bg_inode_bitmap), - sb, bitmap_bh->b_data)) - ext3_error (sb, "ext3_check_blocks_bitmap", - "Inode bitmap for group %d is marked free", - i); - - for (j = 0; j < EXT3_SB(sb)->s_itb_per_group; j++) - if (!block_in_use (le32_to_cpu(gdp->bg_inode_table) + j, - sb, bitmap_bh->b_data)) - ext3_error (sb, "ext3_check_blocks_bitmap", - "Block #%d of the inode table in " - "group %d is marked free", j, i); - - x = ext3_count_free(bitmap_bh, sb->s_blocksize); - if (le16_to_cpu(gdp->bg_free_blocks_count) != x) - ext3_error (sb, "ext3_check_blocks_bitmap", - "Wrong free blocks count for group %d, " - "stored = %d, counted = %lu", i, - le16_to_cpu(gdp->bg_free_blocks_count), x); - bitmap_count += x; - } - brelse(bitmap_bh); - if (le32_to_cpu(es->s_free_blocks_count) != bitmap_count) - ext3_error (sb, "ext3_check_blocks_bitmap", - "Wrong free blocks count in super block, " - "stored = %lu, counted = %lu", - (unsigned long)le32_to_cpu(es->s_free_blocks_count), - bitmap_count); } -#endif