diff options
Diffstat (limited to 'fs/ext4/mballoc.c')
| -rw-r--r-- | fs/ext4/mballoc.c | 4197 |
1 files changed, 2920 insertions, 1277 deletions
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index a9083113a8c0..56d50fd3310b 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -16,7 +16,9 @@ #include <linux/slab.h> #include <linux/nospec.h> #include <linux/backing-dev.h> +#include <linux/freezer.h> #include <trace/events/ext4.h> +#include <kunit/static_stub.h> /* * MUSTDO: @@ -96,14 +98,14 @@ * block bitmap and buddy information. The information are stored in the * inode as: * - * { page } + * { folio } * [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]... * * * one block each for bitmap and buddy information. So for each group we - * take up 2 blocks. A page can contain blocks_per_page (PAGE_SIZE / - * blocksize) blocks. So it can have information regarding groups_per_page - * which is blocks_per_page/2 + * take up 2 blocks. A folio can contain blocks_per_folio (folio_size / + * blocksize) blocks. So it can have information regarding groups_per_folio + * which is blocks_per_folio/2 * * The buddy cache inode is not stored on disk. The inode is thrown * away when the filesystem is unmounted. @@ -124,14 +126,73 @@ * /sys/fs/ext4/<partition>/mb_group_prealloc. The value is represented in * terms of number of blocks. If we have mounted the file system with -O * stripe=<value> option the group prealloc request is normalized to the - * the smallest multiple of the stripe value (sbi->s_stripe) which is + * smallest multiple of the stripe value (sbi->s_stripe) which is * greater than the default mb_group_prealloc. * + * If "mb_optimize_scan" mount option is set, we maintain in memory group info + * structures in two data structures: + * + * 1) Array of largest free order xarrays (sbi->s_mb_largest_free_orders) + * + * Locking: Writers use xa_lock, readers use rcu_read_lock. + * + * This is an array of xarrays where the index in the array represents the + * largest free order in the buddy bitmap of the participating group infos of + * that xarray. So, there are exactly MB_NUM_ORDERS(sb) (which means total + * number of buddy bitmap orders possible) number of xarrays. Group-infos are + * placed in appropriate xarrays. + * + * 2) Average fragment size xarrays (sbi->s_mb_avg_fragment_size) + * + * Locking: Writers use xa_lock, readers use rcu_read_lock. + * + * This is an array of xarrays where in the i-th xarray there are groups with + * average fragment size >= 2^i and < 2^(i+1). The average fragment size + * is computed as ext4_group_info->bb_free / ext4_group_info->bb_fragments. + * Note that we don't bother with a special xarray for completely empty + * groups so we only have MB_NUM_ORDERS(sb) xarrays. Group-infos are placed + * in appropriate xarrays. + * + * In xarray, the index is the block group number, the value is the block group + * information, and a non-empty value indicates the block group is present in + * the current xarray. + * + * When "mb_optimize_scan" mount option is set, mballoc consults the above data + * structures to decide the order in which groups are to be traversed for + * fulfilling an allocation request. + * + * At CR_POWER2_ALIGNED , we look for groups which have the largest_free_order + * >= the order of the request. We directly look at the largest free order list + * in the data structure (1) above where largest_free_order = order of the + * request. If that list is empty, we look at remaining list in the increasing + * order of largest_free_order. This allows us to perform CR_POWER2_ALIGNED + * lookup in O(1) time. + * + * At CR_GOAL_LEN_FAST, we only consider groups where + * average fragment size > request size. So, we lookup a group which has average + * fragment size just above or equal to request size using our average fragment + * size group lists (data structure 2) in O(1) time. + * + * At CR_BEST_AVAIL_LEN, we aim to optimize allocations which can't be satisfied + * in CR_GOAL_LEN_FAST. The fact that we couldn't find a group in + * CR_GOAL_LEN_FAST suggests that there is no BG that has avg + * fragment size > goal length. So before falling to the slower + * CR_GOAL_LEN_SLOW, in CR_BEST_AVAIL_LEN we proactively trim goal length and + * then use the same fragment lists as CR_GOAL_LEN_FAST to find a BG with a big + * enough average fragment size. This increases the chances of finding a + * suitable block group in O(1) time and results in faster allocation at the + * cost of reduced size of allocation. + * + * If "mb_optimize_scan" mount option is not set, mballoc traverses groups in + * linear order which requires O(N) search time for each CR_POWER2_ALIGNED and + * CR_GOAL_LEN_FAST phase. + * * The regular allocator (using the buddy cache) supports a few tunables. * * /sys/fs/ext4/<partition>/mb_min_to_scan * /sys/fs/ext4/<partition>/mb_max_to_scan * /sys/fs/ext4/<partition>/mb_order2_req + * /sys/fs/ext4/<partition>/mb_max_linear_groups * * The regular allocator uses buddy scan only if the request len is power of * 2 blocks and the order of allocation is >= sbi->s_mb_order2_reqs. The @@ -149,6 +210,16 @@ * can be used for allocation. ext4_mb_good_group explains how the groups are * checked. * + * When "mb_optimize_scan" is turned on, as mentioned above, the groups may not + * get traversed linearly. That may result in subsequent allocations being not + * close to each other. And so, the underlying device may get filled up in a + * non-linear fashion. While that may not matter on non-rotational devices, for + * rotational devices that may result in higher seek times. "mb_max_linear_groups" + * tells mballoc how many groups mballoc should search linearly before + * performing consulting above data structures for more efficient lookups. For + * non rotational devices, this value defaults to 0 and for rotational devices + * this is set to MB_DEFAULT_LINEAR_LIMIT. + * * Both the prealloc space are getting populated as above. So for the first * request we will hit the buddy cache which will result in this prealloc * space getting filled. The prealloc space is then later used for the @@ -299,6 +370,8 @@ * - bitlock on a group (group) * - object (inode/locality) (object) * - per-pa lock (pa) + * - cr_power2_aligned lists lock (cr_power2_aligned) + * - cr_goal_len_fast lists lock (cr_goal_len_fast) * * Paths: * - new pa @@ -328,6 +401,9 @@ * group * object * + * - allocation path (ext4_mb_regular_allocator) + * group + * cr_power2_aligned/cr_goal_len_fast */ static struct kmem_cache *ext4_pspace_cachep; static struct kmem_cache *ext4_ac_cachep; @@ -347,10 +423,15 @@ static const char * const ext4_groupinfo_slab_names[NR_GRPINFO_CACHES] = { static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap, ext4_group_t group); -static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap, - ext4_group_t group); static void ext4_mb_new_preallocation(struct ext4_allocation_context *ac); +static int ext4_mb_scan_group(struct ext4_allocation_context *ac, + ext4_group_t group); + +static int ext4_try_to_trim_range(struct super_block *sb, + struct ext4_buddy *e4b, ext4_grpblk_t start, + ext4_grpblk_t max, ext4_grpblk_t minblocks); + /* * The algorithm using this percpu seq counter goes below: * 1. We sample the percpu discard_pa_seq counter before trying for block @@ -488,14 +569,14 @@ static void mb_free_blocks_double(struct inode *inode, struct ext4_buddy *e4b, blocknr = ext4_group_first_block_no(sb, e4b->bd_group); blocknr += EXT4_C2B(EXT4_SB(sb), first + i); + ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, + EXT4_GROUP_INFO_BBITMAP_CORRUPT); ext4_grp_locked_error(sb, e4b->bd_group, inode ? inode->i_ino : 0, blocknr, "freeing block already freed " "(bit %u)", first + i); - ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, - EXT4_GROUP_INFO_BBITMAP_CORRUPT); } mb_clear_bit(first + i, e4b->bd_info->bb_bitmap); } @@ -601,7 +682,25 @@ do { \ } \ } while (0) -static int __mb_check_buddy(struct ext4_buddy *e4b, char *file, +/* + * Perform buddy integrity check with the following steps: + * + * 1. Top-down validation (from highest order down to order 1, excluding order-0 bitmap): + * For each pair of adjacent orders, if a higher-order bit is set (indicating a free block), + * at most one of the two corresponding lower-order bits may be clear (free). + * + * 2. Order-0 (bitmap) validation, performed on bit pairs: + * - If either bit in a pair is set (1, allocated), then all corresponding higher-order bits + * must not be free (0). + * - If both bits in a pair are clear (0, free), then exactly one of the corresponding + * higher-order bits must be free (0). + * + * 3. Preallocation (pa) list validation: + * For each preallocated block (pa) in the group: + * - Verify that pa_pstart falls within the bounds of this block group. + * - Ensure the corresponding bit(s) in the order-0 bitmap are marked as allocated (1). + */ +static void __mb_check_buddy(struct ext4_buddy *e4b, char *file, const char *function, int line) { struct super_block *sb = e4b->bd_sb; @@ -619,11 +718,8 @@ static int __mb_check_buddy(struct ext4_buddy *e4b, char *file, void *buddy; void *buddy2; - { - static int mb_check_counter; - if (mb_check_counter++ % 100 != 0) - return 0; - } + if (e4b->bd_info->bb_check_counter++ % 10) + return; while (order > 1) { buddy = mb_find_buddy(e4b, order, &max); @@ -637,26 +733,14 @@ static int __mb_check_buddy(struct ext4_buddy *e4b, char *file, for (i = 0; i < max; i++) { if (mb_test_bit(i, buddy)) { - /* only single bit in buddy2 may be 1 */ + /* only single bit in buddy2 may be 0 */ if (!mb_test_bit(i << 1, buddy2)) { MB_CHECK_ASSERT( mb_test_bit((i<<1)+1, buddy2)); - } else if (!mb_test_bit((i << 1) + 1, buddy2)) { - MB_CHECK_ASSERT( - mb_test_bit(i << 1, buddy2)); } continue; } - /* both bits in buddy2 must be 1 */ - MB_CHECK_ASSERT(mb_test_bit(i << 1, buddy2)); - MB_CHECK_ASSERT(mb_test_bit((i << 1) + 1, buddy2)); - - for (j = 0; j < (1 << order); j++) { - k = (i * (1 << order)) + j; - MB_CHECK_ASSERT( - !mb_test_bit(k, e4b->bd_bitmap)); - } count++; } MB_CHECK_ASSERT(e4b->bd_info->bb_counters[order] == count); @@ -672,31 +756,40 @@ static int __mb_check_buddy(struct ext4_buddy *e4b, char *file, fragments++; fstart = i; } - continue; + } else { + fstart = -1; } - fstart = -1; - /* check used bits only */ - for (j = 0; j < e4b->bd_blkbits + 1; j++) { - buddy2 = mb_find_buddy(e4b, j, &max2); - k = i >> j; - MB_CHECK_ASSERT(k < max2); - MB_CHECK_ASSERT(mb_test_bit(k, buddy2)); + if (!(i & 1)) { + int in_use, zero_bit_count = 0; + + in_use = mb_test_bit(i, buddy) || mb_test_bit(i + 1, buddy); + for (j = 1; j < e4b->bd_blkbits + 2; j++) { + buddy2 = mb_find_buddy(e4b, j, &max2); + k = i >> j; + MB_CHECK_ASSERT(k < max2); + if (!mb_test_bit(k, buddy2)) + zero_bit_count++; + } + MB_CHECK_ASSERT(zero_bit_count == !in_use); } } MB_CHECK_ASSERT(!EXT4_MB_GRP_NEED_INIT(e4b->bd_info)); MB_CHECK_ASSERT(e4b->bd_info->bb_fragments == fragments); grp = ext4_get_group_info(sb, e4b->bd_group); + if (!grp) + return; list_for_each(cur, &grp->bb_prealloc_list) { ext4_group_t groupnr; struct ext4_prealloc_space *pa; pa = list_entry(cur, struct ext4_prealloc_space, pa_group_list); + if (!pa->pa_len) + continue; ext4_get_group_no_and_offset(sb, pa->pa_pstart, &groupnr, &k); MB_CHECK_ASSERT(groupnr == e4b->bd_group); for (i = 0; i < pa->pa_len; i++) MB_CHECK_ASSERT(mb_test_bit(k + i, buddy)); } - return 0; } #undef MB_CHECK_ASSERT #define mb_check_buddy(e4b) __mb_check_buddy(e4b, \ @@ -747,6 +840,392 @@ static void ext4_mb_mark_free_simple(struct super_block *sb, } } +static int mb_avg_fragment_size_order(struct super_block *sb, ext4_grpblk_t len) +{ + int order; + + /* + * We don't bother with a special lists groups with only 1 block free + * extents and for completely empty groups. + */ + order = fls(len) - 2; + if (order < 0) + return 0; + if (order == MB_NUM_ORDERS(sb)) + order--; + if (WARN_ON_ONCE(order > MB_NUM_ORDERS(sb))) + order = MB_NUM_ORDERS(sb) - 1; + return order; +} + +/* Move group to appropriate avg_fragment_size list */ +static void +mb_update_avg_fragment_size(struct super_block *sb, struct ext4_group_info *grp) +{ + struct ext4_sb_info *sbi = EXT4_SB(sb); + int new, old; + + if (!test_opt2(sb, MB_OPTIMIZE_SCAN)) + return; + + old = grp->bb_avg_fragment_size_order; + new = grp->bb_fragments == 0 ? -1 : + mb_avg_fragment_size_order(sb, grp->bb_free / grp->bb_fragments); + if (new == old) + return; + + if (old >= 0) + xa_erase(&sbi->s_mb_avg_fragment_size[old], grp->bb_group); + + grp->bb_avg_fragment_size_order = new; + if (new >= 0) { + /* + * Cannot use __GFP_NOFAIL because we hold the group lock. + * Although allocation for insertion may fails, it's not fatal + * as we have linear traversal to fall back on. + */ + int err = xa_insert(&sbi->s_mb_avg_fragment_size[new], + grp->bb_group, grp, GFP_ATOMIC); + if (err) + mb_debug(sb, "insert group: %u to s_mb_avg_fragment_size[%d] failed, err %d", + grp->bb_group, new, err); + } +} + +static int ext4_mb_scan_groups_xa_range(struct ext4_allocation_context *ac, + struct xarray *xa, + ext4_group_t start, ext4_group_t end) +{ + struct super_block *sb = ac->ac_sb; + struct ext4_sb_info *sbi = EXT4_SB(sb); + enum criteria cr = ac->ac_criteria; + ext4_group_t ngroups = ext4_get_groups_count(sb); + unsigned long group = start; + struct ext4_group_info *grp; + + if (WARN_ON_ONCE(end > ngroups || start >= end)) + return 0; + + xa_for_each_range(xa, group, grp, start, end - 1) { + int err; + + if (sbi->s_mb_stats) + atomic64_inc(&sbi->s_bal_cX_groups_considered[cr]); + + err = ext4_mb_scan_group(ac, grp->bb_group); + if (err || ac->ac_status != AC_STATUS_CONTINUE) + return err; + + cond_resched(); + } + + return 0; +} + +/* + * Find a suitable group of given order from the largest free orders xarray. + */ +static inline int +ext4_mb_scan_groups_largest_free_order_range(struct ext4_allocation_context *ac, + int order, ext4_group_t start, + ext4_group_t end) +{ + struct xarray *xa = &EXT4_SB(ac->ac_sb)->s_mb_largest_free_orders[order]; + + if (xa_empty(xa)) + return 0; + + return ext4_mb_scan_groups_xa_range(ac, xa, start, end); +} + +/* + * Choose next group by traversing largest_free_order lists. Updates *new_cr if + * cr level needs an update. + */ +static int ext4_mb_scan_groups_p2_aligned(struct ext4_allocation_context *ac, + ext4_group_t group) +{ + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + int i; + int ret = 0; + ext4_group_t start, end; + + start = group; + end = ext4_get_groups_count(ac->ac_sb); +wrap_around: + for (i = ac->ac_2order; i < MB_NUM_ORDERS(ac->ac_sb); i++) { + ret = ext4_mb_scan_groups_largest_free_order_range(ac, i, + start, end); + if (ret || ac->ac_status != AC_STATUS_CONTINUE) + return ret; + } + if (start) { + end = start; + start = 0; + goto wrap_around; + } + + if (sbi->s_mb_stats) + atomic64_inc(&sbi->s_bal_cX_failed[ac->ac_criteria]); + + /* Increment cr and search again if no group is found */ + ac->ac_criteria = CR_GOAL_LEN_FAST; + return ret; +} + +/* + * Find a suitable group of given order from the average fragments xarray. + */ +static int +ext4_mb_scan_groups_avg_frag_order_range(struct ext4_allocation_context *ac, + int order, ext4_group_t start, + ext4_group_t end) +{ + struct xarray *xa = &EXT4_SB(ac->ac_sb)->s_mb_avg_fragment_size[order]; + + if (xa_empty(xa)) + return 0; + + return ext4_mb_scan_groups_xa_range(ac, xa, start, end); +} + +/* + * Choose next group by traversing average fragment size list of suitable + * order. Updates *new_cr if cr level needs an update. + */ +static int ext4_mb_scan_groups_goal_fast(struct ext4_allocation_context *ac, + ext4_group_t group) +{ + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + int i, ret = 0; + ext4_group_t start, end; + + start = group; + end = ext4_get_groups_count(ac->ac_sb); +wrap_around: + i = mb_avg_fragment_size_order(ac->ac_sb, ac->ac_g_ex.fe_len); + for (; i < MB_NUM_ORDERS(ac->ac_sb); i++) { + ret = ext4_mb_scan_groups_avg_frag_order_range(ac, i, + start, end); + if (ret || ac->ac_status != AC_STATUS_CONTINUE) + return ret; + } + if (start) { + end = start; + start = 0; + goto wrap_around; + } + + if (sbi->s_mb_stats) + atomic64_inc(&sbi->s_bal_cX_failed[ac->ac_criteria]); + /* + * CR_BEST_AVAIL_LEN works based on the concept that we have + * a larger normalized goal len request which can be trimmed to + * a smaller goal len such that it can still satisfy original + * request len. However, allocation request for non-regular + * files never gets normalized. + * See function ext4_mb_normalize_request() (EXT4_MB_HINT_DATA). + */ + if (ac->ac_flags & EXT4_MB_HINT_DATA) + ac->ac_criteria = CR_BEST_AVAIL_LEN; + else + ac->ac_criteria = CR_GOAL_LEN_SLOW; + + return ret; +} + +/* + * We couldn't find a group in CR_GOAL_LEN_FAST so try to find the highest free fragment + * order we have and proactively trim the goal request length to that order to + * find a suitable group faster. + * + * This optimizes allocation speed at the cost of slightly reduced + * preallocations. However, we make sure that we don't trim the request too + * much and fall to CR_GOAL_LEN_SLOW in that case. + */ +static int ext4_mb_scan_groups_best_avail(struct ext4_allocation_context *ac, + ext4_group_t group) +{ + int ret = 0; + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + int i, order, min_order; + unsigned long num_stripe_clusters = 0; + ext4_group_t start, end; + + /* + * mb_avg_fragment_size_order() returns order in a way that makes + * retrieving back the length using (1 << order) inaccurate. Hence, use + * fls() instead since we need to know the actual length while modifying + * goal length. + */ + order = fls(ac->ac_g_ex.fe_len) - 1; + if (WARN_ON_ONCE(order - 1 > MB_NUM_ORDERS(ac->ac_sb))) + order = MB_NUM_ORDERS(ac->ac_sb); + min_order = order - sbi->s_mb_best_avail_max_trim_order; + if (min_order < 0) + min_order = 0; + + if (sbi->s_stripe > 0) { + /* + * We are assuming that stripe size is always a multiple of + * cluster ratio otherwise __ext4_fill_super exists early. + */ + num_stripe_clusters = EXT4_NUM_B2C(sbi, sbi->s_stripe); + if (1 << min_order < num_stripe_clusters) + /* + * We consider 1 order less because later we round + * up the goal len to num_stripe_clusters + */ + min_order = fls(num_stripe_clusters) - 1; + } + + if (1 << min_order < ac->ac_o_ex.fe_len) + min_order = fls(ac->ac_o_ex.fe_len); + + start = group; + end = ext4_get_groups_count(ac->ac_sb); +wrap_around: + for (i = order; i >= min_order; i--) { + int frag_order; + /* + * Scale down goal len to make sure we find something + * in the free fragments list. Basically, reduce + * preallocations. + */ + ac->ac_g_ex.fe_len = 1 << i; + + if (num_stripe_clusters > 0) { + /* + * Try to round up the adjusted goal length to + * stripe size (in cluster units) multiple for + * efficiency. + */ + ac->ac_g_ex.fe_len = roundup(ac->ac_g_ex.fe_len, + num_stripe_clusters); + } + + frag_order = mb_avg_fragment_size_order(ac->ac_sb, + ac->ac_g_ex.fe_len); + + ret = ext4_mb_scan_groups_avg_frag_order_range(ac, frag_order, + start, end); + if (ret || ac->ac_status != AC_STATUS_CONTINUE) + return ret; + } + if (start) { + end = start; + start = 0; + goto wrap_around; + } + + /* Reset goal length to original goal length before falling into CR_GOAL_LEN_SLOW */ + ac->ac_g_ex.fe_len = ac->ac_orig_goal_len; + if (sbi->s_mb_stats) + atomic64_inc(&sbi->s_bal_cX_failed[ac->ac_criteria]); + ac->ac_criteria = CR_GOAL_LEN_SLOW; + + return ret; +} + +static inline int should_optimize_scan(struct ext4_allocation_context *ac) +{ + if (unlikely(!test_opt2(ac->ac_sb, MB_OPTIMIZE_SCAN))) + return 0; + if (ac->ac_criteria >= CR_GOAL_LEN_SLOW) + return 0; + if (!ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)) + return 0; + return 1; +} + +/* + * next linear group for allocation. + */ +static void next_linear_group(ext4_group_t *group, ext4_group_t ngroups) +{ + /* + * Artificially restricted ngroups for non-extent + * files makes group > ngroups possible on first loop. + */ + *group = *group + 1 >= ngroups ? 0 : *group + 1; +} + +static int ext4_mb_scan_groups_linear(struct ext4_allocation_context *ac, + ext4_group_t ngroups, ext4_group_t *start, ext4_group_t count) +{ + int ret, i; + enum criteria cr = ac->ac_criteria; + struct super_block *sb = ac->ac_sb; + struct ext4_sb_info *sbi = EXT4_SB(sb); + ext4_group_t group = *start; + + for (i = 0; i < count; i++, next_linear_group(&group, ngroups)) { + ret = ext4_mb_scan_group(ac, group); + if (ret || ac->ac_status != AC_STATUS_CONTINUE) + return ret; + cond_resched(); + } + + *start = group; + if (count == ngroups) + ac->ac_criteria++; + + /* Processed all groups and haven't found blocks */ + if (sbi->s_mb_stats && i == ngroups) + atomic64_inc(&sbi->s_bal_cX_failed[cr]); + + return 0; +} + +static int ext4_mb_scan_groups(struct ext4_allocation_context *ac) +{ + int ret = 0; + ext4_group_t start; + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + ext4_group_t ngroups = ext4_get_groups_count(ac->ac_sb); + + /* non-extent files are limited to low blocks/groups */ + if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS))) + ngroups = sbi->s_blockfile_groups; + + /* searching for the right group start from the goal value specified */ + start = ac->ac_g_ex.fe_group; + ac->ac_prefetch_grp = start; + ac->ac_prefetch_nr = 0; + + if (!should_optimize_scan(ac)) + return ext4_mb_scan_groups_linear(ac, ngroups, &start, ngroups); + + /* + * Optimized scanning can return non adjacent groups which can cause + * seek overhead for rotational disks. So try few linear groups before + * trying optimized scan. + */ + if (sbi->s_mb_max_linear_groups) + ret = ext4_mb_scan_groups_linear(ac, ngroups, &start, + sbi->s_mb_max_linear_groups); + if (ret || ac->ac_status != AC_STATUS_CONTINUE) + return ret; + + switch (ac->ac_criteria) { + case CR_POWER2_ALIGNED: + return ext4_mb_scan_groups_p2_aligned(ac, start); + case CR_GOAL_LEN_FAST: + return ext4_mb_scan_groups_goal_fast(ac, start); + case CR_BEST_AVAIL_LEN: + return ext4_mb_scan_groups_best_avail(ac, start); + default: + /* + * TODO: For CR_GOAL_LEN_SLOW, we can arrange groups in an + * rb tree sorted by bb_free. But until that happens, we should + * never come here. + */ + WARN_ON(1); + } + + return 0; +} + /* * Cache the order of the largest free extent we have available in this block * group. @@ -754,25 +1233,44 @@ static void ext4_mb_mark_free_simple(struct super_block *sb, static void mb_set_largest_free_order(struct super_block *sb, struct ext4_group_info *grp) { - int i; - int bits; - - grp->bb_largest_free_order = -1; /* uninit */ + struct ext4_sb_info *sbi = EXT4_SB(sb); + int new, old = grp->bb_largest_free_order; - bits = sb->s_blocksize_bits + 1; - for (i = bits; i >= 0; i--) { - if (grp->bb_counters[i] > 0) { - grp->bb_largest_free_order = i; + for (new = MB_NUM_ORDERS(sb) - 1; new >= 0; new--) + if (grp->bb_counters[new] > 0) break; - } + + /* No need to move between order lists? */ + if (new == old) + return; + + if (old >= 0) { + struct xarray *xa = &sbi->s_mb_largest_free_orders[old]; + + if (!xa_empty(xa) && xa_load(xa, grp->bb_group)) + xa_erase(xa, grp->bb_group); + } + + grp->bb_largest_free_order = new; + if (test_opt2(sb, MB_OPTIMIZE_SCAN) && new >= 0 && grp->bb_free) { + /* + * Cannot use __GFP_NOFAIL because we hold the group lock. + * Although allocation for insertion may fails, it's not fatal + * as we have linear traversal to fall back on. + */ + int err = xa_insert(&sbi->s_mb_largest_free_orders[new], + grp->bb_group, grp, GFP_ATOMIC); + if (err) + mb_debug(sb, "insert group: %u to s_mb_largest_free_orders[%d] failed, err %d", + grp->bb_group, new, err); } } static noinline_for_stack void ext4_mb_generate_buddy(struct super_block *sb, - void *buddy, void *bitmap, ext4_group_t group) + void *buddy, void *bitmap, ext4_group_t group, + struct ext4_group_info *grp) { - struct ext4_group_info *grp = ext4_get_group_info(sb, group); struct ext4_sb_info *sbi = EXT4_SB(sb); ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb); ext4_grpblk_t i = 0; @@ -815,14 +1313,13 @@ void ext4_mb_generate_buddy(struct super_block *sb, EXT4_GROUP_INFO_BBITMAP_CORRUPT); } mb_set_largest_free_order(sb, grp); + mb_update_avg_fragment_size(sb, grp); clear_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &(grp->bb_state)); period = get_cycles() - period; - spin_lock(&sbi->s_bal_lock); - sbi->s_mb_buddies_generated++; - sbi->s_mb_generation_time += period; - spin_unlock(&sbi->s_bal_lock); + atomic_inc(&sbi->s_mb_buddies_generated); + atomic64_add(period, &sbi->s_mb_generation_time); } static void mb_regenerate_buddy(struct ext4_buddy *e4b) @@ -831,16 +1328,16 @@ static void mb_regenerate_buddy(struct ext4_buddy *e4b) int order = 1; void *buddy; - while ((buddy = mb_find_buddy(e4b, order++, &count))) { - ext4_set_bits(buddy, 0, count); - } + while ((buddy = mb_find_buddy(e4b, order++, &count))) + mb_set_bits(buddy, 0, count); + e4b->bd_info->bb_fragments = 0; memset(e4b->bd_info->bb_counters, 0, sizeof(*e4b->bd_info->bb_counters) * (e4b->bd_sb->s_blocksize_bits + 2)); ext4_mb_generate_buddy(e4b->bd_sb, e4b->bd_buddy, - e4b->bd_bitmap, e4b->bd_group); + e4b->bd_bitmap, e4b->bd_group, e4b->bd_info); } /* The buddy information is attached the buddy cache inode @@ -849,26 +1346,25 @@ static void mb_regenerate_buddy(struct ext4_buddy *e4b) * block bitmap and buddy information. The information are * stored in the inode as * - * { page } + * { folio } * [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]... * * * one block each for bitmap and buddy information. - * So for each group we take up 2 blocks. A page can - * contain blocks_per_page (PAGE_SIZE / blocksize) blocks. - * So it can have information regarding groups_per_page which - * is blocks_per_page/2 + * So for each group we take up 2 blocks. A folio can + * contain blocks_per_folio (folio_size / blocksize) blocks. + * So it can have information regarding groups_per_folio which + * is blocks_per_folio/2 * * Locking note: This routine takes the block group lock of all groups - * for this page; do not hold this lock when calling this routine! + * for this folio; do not hold this lock when calling this routine! */ - -static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) +static int ext4_mb_init_cache(struct folio *folio, char *incore, gfp_t gfp) { ext4_group_t ngroups; - int blocksize; - int blocks_per_page; - int groups_per_page; + unsigned int blocksize; + int blocks_per_folio; + int groups_per_folio; int err = 0; int i; ext4_group_t first_group, group; @@ -881,48 +1377,46 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) char *bitmap; struct ext4_group_info *grinfo; - inode = page->mapping->host; + inode = folio->mapping->host; sb = inode->i_sb; ngroups = ext4_get_groups_count(sb); blocksize = i_blocksize(inode); - blocks_per_page = PAGE_SIZE / blocksize; + blocks_per_folio = folio_size(folio) / blocksize; + WARN_ON_ONCE(!blocks_per_folio); + groups_per_folio = DIV_ROUND_UP(blocks_per_folio, 2); - mb_debug(sb, "init page %lu\n", page->index); - - groups_per_page = blocks_per_page >> 1; - if (groups_per_page == 0) - groups_per_page = 1; + mb_debug(sb, "init folio %lu\n", folio->index); /* allocate buffer_heads to read bitmaps */ - if (groups_per_page > 1) { - i = sizeof(struct buffer_head *) * groups_per_page; + if (groups_per_folio > 1) { + i = sizeof(struct buffer_head *) * groups_per_folio; bh = kzalloc(i, gfp); - if (bh == NULL) { - err = -ENOMEM; - goto out; - } + if (bh == NULL) + return -ENOMEM; } else bh = &bhs; - first_group = page->index * blocks_per_page / 2; - - /* read all groups the page covers into the cache */ - for (i = 0, group = first_group; i < groups_per_page; i++, group++) { + /* read all groups the folio covers into the cache */ + first_group = EXT4_PG_TO_LBLK(inode, folio->index) / 2; + for (i = 0, group = first_group; i < groups_per_folio; i++, group++) { if (group >= ngroups) break; grinfo = ext4_get_group_info(sb, group); + if (!grinfo) + continue; /* - * If page is uptodate then we came here after online resize + * If folio is uptodate then we came here after online resize * which added some new uninitialized group info structs, so - * we must skip all initialized uptodate buddies on the page, + * we must skip all initialized uptodate buddies on the folio, * which may be currently in use by an allocating task. */ - if (PageUptodate(page) && !EXT4_MB_GRP_NEED_INIT(grinfo)) { + if (folio_test_uptodate(folio) && + !EXT4_MB_GRP_NEED_INIT(grinfo)) { bh[i] = NULL; continue; } - bh[i] = ext4_read_block_bitmap_nowait(sb, group); + bh[i] = ext4_read_block_bitmap_nowait(sb, group, false); if (IS_ERR(bh[i])) { err = PTR_ERR(bh[i]); bh[i] = NULL; @@ -932,7 +1426,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) } /* wait for I/O completion */ - for (i = 0, group = first_group; i < groups_per_page; i++, group++) { + for (i = 0, group = first_group; i < groups_per_folio; i++, group++) { int err2; if (!bh[i]) @@ -942,8 +1436,8 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) err = err2; } - first_block = page->index * blocks_per_page; - for (i = 0; i < blocks_per_page; i++) { + first_block = EXT4_PG_TO_LBLK(inode, folio->index); + for (i = 0; i < blocks_per_folio; i++) { group = (first_block + i) >> 1; if (group >= ngroups) break; @@ -963,38 +1457,42 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) * above * */ - data = page_address(page) + (i * blocksize); + data = folio_address(folio) + (i * blocksize); bitmap = bh[group - first_group]->b_data; /* * We place the buddy block and bitmap block * close together */ + grinfo = ext4_get_group_info(sb, group); + if (!grinfo) { + err = -EFSCORRUPTED; + goto out; + } if ((first_block + i) & 1) { /* this is block of buddy */ BUG_ON(incore == NULL); - mb_debug(sb, "put buddy for group %u in page %lu/%x\n", - group, page->index, i * blocksize); + mb_debug(sb, "put buddy for group %u in folio %lu/%x\n", + group, folio->index, i * blocksize); trace_ext4_mb_buddy_bitmap_load(sb, group); - grinfo = ext4_get_group_info(sb, group); grinfo->bb_fragments = 0; memset(grinfo->bb_counters, 0, sizeof(*grinfo->bb_counters) * - (sb->s_blocksize_bits+2)); + (MB_NUM_ORDERS(sb))); /* * incore got set to the group block bitmap below */ ext4_lock_group(sb, group); /* init the buddy */ memset(data, 0xff, blocksize); - ext4_mb_generate_buddy(sb, data, incore, group); + ext4_mb_generate_buddy(sb, data, incore, group, grinfo); ext4_unlock_group(sb, group); incore = NULL; } else { /* this is block of bitmap */ BUG_ON(incore != NULL); - mb_debug(sb, "put bitmap for group %u in page %lu/%x\n", - group, page->index, i * blocksize); + mb_debug(sb, "put bitmap for group %u in folio %lu/%x\n", + group, folio->index, i * blocksize); trace_ext4_mb_bitmap_load(sb, group); /* see comments in ext4_mb_put_pa() */ @@ -1003,7 +1501,7 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) /* mark all preallocated blks used in in-core bitmap */ ext4_mb_generate_from_pa(sb, data, group); - ext4_mb_generate_from_freelist(sb, data, group); + WARN_ON_ONCE(!RB_EMPTY_ROOT(&grinfo->bb_free_root)); ext4_unlock_group(sb, group); /* set incore so that the buddy information can be @@ -1012,11 +1510,11 @@ static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp) incore = data; } } - SetPageUptodate(page); + folio_mark_uptodate(folio); out: if (bh) { - for (i = 0; i < groups_per_page; i++) + for (i = 0; i < groups_per_folio; i++) brelse(bh[i]); if (bh != &bhs) kfree(bh); @@ -1025,68 +1523,71 @@ out: } /* - * Lock the buddy and bitmap pages. This make sure other parallel init_group - * on the same buddy page doesn't happen whild holding the buddy page lock. - * Return locked buddy and bitmap pages on e4b struct. If buddy and bitmap - * are on the same page e4b->bd_buddy_page is NULL and return value is 0. + * Lock the buddy and bitmap folios. This makes sure other parallel init_group + * on the same buddy folio doesn't happen while holding the buddy folio lock. + * Return locked buddy and bitmap folios on e4b struct. If buddy and bitmap + * are on the same folio e4b->bd_buddy_folio is NULL and return value is 0. */ -static int ext4_mb_get_buddy_page_lock(struct super_block *sb, +static int ext4_mb_get_buddy_folio_lock(struct super_block *sb, ext4_group_t group, struct ext4_buddy *e4b, gfp_t gfp) { struct inode *inode = EXT4_SB(sb)->s_buddy_cache; - int block, pnum, poff; - int blocks_per_page; - struct page *page; + int block, pnum; + struct folio *folio; - e4b->bd_buddy_page = NULL; - e4b->bd_bitmap_page = NULL; + e4b->bd_buddy_folio = NULL; + e4b->bd_bitmap_folio = NULL; - blocks_per_page = PAGE_SIZE / sb->s_blocksize; /* * the buddy cache inode stores the block bitmap * and buddy information in consecutive blocks. * So for each group we need two blocks. */ block = group * 2; - pnum = block / blocks_per_page; - poff = block % blocks_per_page; - page = find_or_create_page(inode->i_mapping, pnum, gfp); - if (!page) - return -ENOMEM; - BUG_ON(page->mapping != inode->i_mapping); - e4b->bd_bitmap_page = page; - e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize); + pnum = EXT4_LBLK_TO_PG(inode, block); + folio = __filemap_get_folio(inode->i_mapping, pnum, + FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp); + if (IS_ERR(folio)) + return PTR_ERR(folio); + BUG_ON(folio->mapping != inode->i_mapping); + WARN_ON_ONCE(folio_size(folio) < sb->s_blocksize); + e4b->bd_bitmap_folio = folio; + e4b->bd_bitmap = folio_address(folio) + + offset_in_folio(folio, EXT4_LBLK_TO_B(inode, block)); - if (blocks_per_page >= 2) { - /* buddy and bitmap are on the same page */ + block++; + pnum = EXT4_LBLK_TO_PG(inode, block); + if (folio_contains(folio, pnum)) { + /* buddy and bitmap are on the same folio */ return 0; } - block++; - pnum = block / blocks_per_page; - page = find_or_create_page(inode->i_mapping, pnum, gfp); - if (!page) - return -ENOMEM; - BUG_ON(page->mapping != inode->i_mapping); - e4b->bd_buddy_page = page; + /* we need another folio for the buddy */ + folio = __filemap_get_folio(inode->i_mapping, pnum, + FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp); + if (IS_ERR(folio)) + return PTR_ERR(folio); + BUG_ON(folio->mapping != inode->i_mapping); + WARN_ON_ONCE(folio_size(folio) < sb->s_blocksize); + e4b->bd_buddy_folio = folio; return 0; } -static void ext4_mb_put_buddy_page_lock(struct ext4_buddy *e4b) +static void ext4_mb_put_buddy_folio_lock(struct ext4_buddy *e4b) { - if (e4b->bd_bitmap_page) { - unlock_page(e4b->bd_bitmap_page); - put_page(e4b->bd_bitmap_page); + if (e4b->bd_bitmap_folio) { + folio_unlock(e4b->bd_bitmap_folio); + folio_put(e4b->bd_bitmap_folio); } - if (e4b->bd_buddy_page) { - unlock_page(e4b->bd_buddy_page); - put_page(e4b->bd_buddy_page); + if (e4b->bd_buddy_folio) { + folio_unlock(e4b->bd_buddy_folio); + folio_put(e4b->bd_buddy_folio); } } /* * Locking note: This routine calls ext4_mb_init_cache(), which takes the - * block group lock of all groups for this page; do not hold the BG lock when + * block group lock of all groups for this folio; do not hold the BG lock when * calling this routine! */ static noinline_for_stack @@ -1095,22 +1596,25 @@ int ext4_mb_init_group(struct super_block *sb, ext4_group_t group, gfp_t gfp) struct ext4_group_info *this_grp; struct ext4_buddy e4b; - struct page *page; + struct folio *folio; int ret = 0; might_sleep(); mb_debug(sb, "init group %u\n", group); this_grp = ext4_get_group_info(sb, group); + if (!this_grp) + return -EFSCORRUPTED; + /* * This ensures that we don't reinit the buddy cache - * page which map to the group from which we are already + * folio which map to the group from which we are already * allocating. If we are looking at the buddy cache we would * have taken a reference using ext4_mb_load_buddy and that - * would have pinned buddy page to page cache. - * The call to ext4_mb_get_buddy_page_lock will mark the - * page accessed. + * would have pinned buddy folio to page cache. + * The call to ext4_mb_get_buddy_folio_lock will mark the + * folio accessed. */ - ret = ext4_mb_get_buddy_page_lock(sb, group, &e4b, gfp); + ret = ext4_mb_get_buddy_folio_lock(sb, group, &e4b, gfp); if (ret || !EXT4_MB_GRP_NEED_INIT(this_grp)) { /* * somebody initialized the group @@ -1119,52 +1623,50 @@ int ext4_mb_init_group(struct super_block *sb, ext4_group_t group, gfp_t gfp) goto err; } - page = e4b.bd_bitmap_page; - ret = ext4_mb_init_cache(page, NULL, gfp); + folio = e4b.bd_bitmap_folio; + ret = ext4_mb_init_cache(folio, NULL, gfp); if (ret) goto err; - if (!PageUptodate(page)) { + if (!folio_test_uptodate(folio)) { ret = -EIO; goto err; } - if (e4b.bd_buddy_page == NULL) { + if (e4b.bd_buddy_folio == NULL) { /* * If both the bitmap and buddy are in - * the same page we don't need to force + * the same folio we don't need to force * init the buddy */ ret = 0; goto err; } /* init buddy cache */ - page = e4b.bd_buddy_page; - ret = ext4_mb_init_cache(page, e4b.bd_bitmap, gfp); + folio = e4b.bd_buddy_folio; + ret = ext4_mb_init_cache(folio, e4b.bd_bitmap, gfp); if (ret) goto err; - if (!PageUptodate(page)) { + if (!folio_test_uptodate(folio)) { ret = -EIO; goto err; } err: - ext4_mb_put_buddy_page_lock(&e4b); + ext4_mb_put_buddy_folio_lock(&e4b); return ret; } /* * Locking note: This routine calls ext4_mb_init_cache(), which takes the - * block group lock of all groups for this page; do not hold the BG lock when + * block group lock of all groups for this folio; do not hold the BG lock when * calling this routine! */ static noinline_for_stack int ext4_mb_load_buddy_gfp(struct super_block *sb, ext4_group_t group, struct ext4_buddy *e4b, gfp_t gfp) { - int blocks_per_page; int block; int pnum; - int poff; - struct page *page; + struct folio *folio; int ret; struct ext4_group_info *grp; struct ext4_sb_info *sbi = EXT4_SB(sb); @@ -1173,15 +1675,16 @@ ext4_mb_load_buddy_gfp(struct super_block *sb, ext4_group_t group, might_sleep(); mb_debug(sb, "load group %u\n", group); - blocks_per_page = PAGE_SIZE / sb->s_blocksize; grp = ext4_get_group_info(sb, group); + if (!grp) + return -EFSCORRUPTED; e4b->bd_blkbits = sb->s_blocksize_bits; e4b->bd_info = grp; e4b->bd_sb = sb; e4b->bd_group = group; - e4b->bd_buddy_page = NULL; - e4b->bd_bitmap_page = NULL; + e4b->bd_buddy_folio = NULL; + e4b->bd_bitmap_folio = NULL; if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) { /* @@ -1199,98 +1702,115 @@ ext4_mb_load_buddy_gfp(struct super_block *sb, ext4_group_t group, * So for each group we need two blocks. */ block = group * 2; - pnum = block / blocks_per_page; - poff = block % blocks_per_page; - - /* we could use find_or_create_page(), but it locks page - * what we'd like to avoid in fast path ... */ - page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED); - if (page == NULL || !PageUptodate(page)) { - if (page) + pnum = EXT4_LBLK_TO_PG(inode, block); + + /* Avoid locking the folio in the fast path ... */ + folio = __filemap_get_folio(inode->i_mapping, pnum, FGP_ACCESSED, 0); + if (IS_ERR(folio) || !folio_test_uptodate(folio)) { + if (!IS_ERR(folio)) /* - * drop the page reference and try - * to get the page with lock. If we + * drop the folio reference and try + * to get the folio with lock. If we * are not uptodate that implies - * somebody just created the page but - * is yet to initialize the same. So + * somebody just created the folio but + * is yet to initialize it. So * wait for it to initialize. */ - put_page(page); - page = find_or_create_page(inode->i_mapping, pnum, gfp); - if (page) { - BUG_ON(page->mapping != inode->i_mapping); - if (!PageUptodate(page)) { - ret = ext4_mb_init_cache(page, NULL, gfp); + folio_put(folio); + folio = __filemap_get_folio(inode->i_mapping, pnum, + FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp); + if (!IS_ERR(folio)) { + if (WARN_RATELIMIT(folio->mapping != inode->i_mapping, + "ext4: bitmap's mapping != inode->i_mapping\n")) { + /* should never happen */ + folio_unlock(folio); + ret = -EINVAL; + goto err; + } + if (!folio_test_uptodate(folio)) { + ret = ext4_mb_init_cache(folio, NULL, gfp); if (ret) { - unlock_page(page); + folio_unlock(folio); goto err; } - mb_cmp_bitmaps(e4b, page_address(page) + - (poff * sb->s_blocksize)); + mb_cmp_bitmaps(e4b, folio_address(folio) + + offset_in_folio(folio, + EXT4_LBLK_TO_B(inode, block))); } - unlock_page(page); + folio_unlock(folio); } } - if (page == NULL) { - ret = -ENOMEM; + if (IS_ERR(folio)) { + ret = PTR_ERR(folio); goto err; } - if (!PageUptodate(page)) { + if (!folio_test_uptodate(folio)) { ret = -EIO; goto err; } - /* Pages marked accessed already */ - e4b->bd_bitmap_page = page; - e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize); + /* Folios marked accessed already */ + e4b->bd_bitmap_folio = folio; + e4b->bd_bitmap = folio_address(folio) + + offset_in_folio(folio, EXT4_LBLK_TO_B(inode, block)); block++; - pnum = block / blocks_per_page; - poff = block % blocks_per_page; - - page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED); - if (page == NULL || !PageUptodate(page)) { - if (page) - put_page(page); - page = find_or_create_page(inode->i_mapping, pnum, gfp); - if (page) { - BUG_ON(page->mapping != inode->i_mapping); - if (!PageUptodate(page)) { - ret = ext4_mb_init_cache(page, e4b->bd_bitmap, + pnum = EXT4_LBLK_TO_PG(inode, block); + /* buddy and bitmap are on the same folio? */ + if (folio_contains(folio, pnum)) { + folio_get(folio); + goto update_buddy; + } + + /* we need another folio for the buddy */ + folio = __filemap_get_folio(inode->i_mapping, pnum, FGP_ACCESSED, 0); + if (IS_ERR(folio) || !folio_test_uptodate(folio)) { + if (!IS_ERR(folio)) + folio_put(folio); + folio = __filemap_get_folio(inode->i_mapping, pnum, + FGP_LOCK | FGP_ACCESSED | FGP_CREAT, gfp); + if (!IS_ERR(folio)) { + if (WARN_RATELIMIT(folio->mapping != inode->i_mapping, + "ext4: buddy bitmap's mapping != inode->i_mapping\n")) { + /* should never happen */ + folio_unlock(folio); + ret = -EINVAL; + goto err; + } + if (!folio_test_uptodate(folio)) { + ret = ext4_mb_init_cache(folio, e4b->bd_bitmap, gfp); if (ret) { - unlock_page(page); + folio_unlock(folio); goto err; } } - unlock_page(page); + folio_unlock(folio); } } - if (page == NULL) { - ret = -ENOMEM; + if (IS_ERR(folio)) { + ret = PTR_ERR(folio); goto err; } - if (!PageUptodate(page)) { + if (!folio_test_uptodate(folio)) { ret = -EIO; goto err; } - /* Pages marked accessed already */ - e4b->bd_buddy_page = page; - e4b->bd_buddy = page_address(page) + (poff * sb->s_blocksize); - - BUG_ON(e4b->bd_bitmap_page == NULL); - BUG_ON(e4b->bd_buddy_page == NULL); +update_buddy: + /* Folios marked accessed already */ + e4b->bd_buddy_folio = folio; + e4b->bd_buddy = folio_address(folio) + + offset_in_folio(folio, EXT4_LBLK_TO_B(inode, block)); return 0; err: - if (page) - put_page(page); - if (e4b->bd_bitmap_page) - put_page(e4b->bd_bitmap_page); - if (e4b->bd_buddy_page) - put_page(e4b->bd_buddy_page); + if (!IS_ERR_OR_NULL(folio)) + folio_put(folio); + if (e4b->bd_bitmap_folio) + folio_put(e4b->bd_bitmap_folio); + e4b->bd_buddy = NULL; e4b->bd_bitmap = NULL; return ret; @@ -1304,31 +1824,27 @@ static int ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group, static void ext4_mb_unload_buddy(struct ext4_buddy *e4b) { - if (e4b->bd_bitmap_page) - put_page(e4b->bd_bitmap_page); - if (e4b->bd_buddy_page) - put_page(e4b->bd_buddy_page); + if (e4b->bd_bitmap_folio) + folio_put(e4b->bd_bitmap_folio); + if (e4b->bd_buddy_folio) + folio_put(e4b->bd_buddy_folio); } static int mb_find_order_for_block(struct ext4_buddy *e4b, int block) { - int order = 1; - int bb_incr = 1 << (e4b->bd_blkbits - 1); + int order = 1, max; void *bb; BUG_ON(e4b->bd_bitmap == e4b->bd_buddy); BUG_ON(block >= (1 << (e4b->bd_blkbits + 3))); - bb = e4b->bd_buddy; while (order <= e4b->bd_blkbits + 1) { - block = block >> 1; - if (!mb_test_bit(block, bb)) { + bb = mb_find_buddy(e4b, order, &max); + if (!mb_test_bit(block >> order, bb)) { /* this block is part of buddy of order 'order' */ return order; } - bb += bb_incr; - bb_incr >>= 1; order++; } return 0; @@ -1379,7 +1895,7 @@ static int mb_test_and_clear_bits(void *bm, int cur, int len) return zero_bit; } -void ext4_set_bits(void *bm, int cur, int len) +void mb_set_bits(void *bm, int cur, int len) { __u32 *addr; @@ -1397,9 +1913,6 @@ void ext4_set_bits(void *bm, int cur, int len) } } -/* - * _________________________________________________________________ */ - static inline int mb_buddy_adjust_border(int* bit, void* bitmap, int side) { if (mb_test_bit(*bit + side, bitmap)) { @@ -1460,7 +1973,8 @@ static void mb_buddy_mark_free(struct ext4_buddy *e4b, int first, int last) break; order++; - if (first == last || !(buddy2 = mb_find_buddy(e4b, order, &max))) { + buddy2 = mb_find_buddy(e4b, order, &max); + if (!buddy2) { mb_clear_bits(buddy, first, last - first + 1); e4b->bd_info->bb_counters[order - 1] += last - first + 1; break; @@ -1491,11 +2005,6 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b, mb_check_buddy(e4b); mb_free_blocks_double(inode, e4b, first, count); - this_cpu_inc(discard_pa_seq); - e4b->bd_info->bb_free += count; - if (first < e4b->bd_info->bb_first_free) - e4b->bd_info->bb_first_free = first; - /* access memory sequentially: check left neighbour, * clear range and then check right neighbour */ @@ -1509,20 +2018,31 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b, struct ext4_sb_info *sbi = EXT4_SB(sb); ext4_fsblk_t blocknr; + /* + * Fastcommit replay can free already freed blocks which + * corrupts allocation info. Regenerate it. + */ + if (sbi->s_mount_state & EXT4_FC_REPLAY) { + mb_regenerate_buddy(e4b); + goto check; + } + blocknr = ext4_group_first_block_no(sb, e4b->bd_group); blocknr += EXT4_C2B(sbi, block); - ext4_grp_locked_error(sb, e4b->bd_group, - inode ? inode->i_ino : 0, - blocknr, - "freeing already freed block " - "(bit %u); block bitmap corrupt.", - block); ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, EXT4_GROUP_INFO_BBITMAP_CORRUPT); - mb_regenerate_buddy(e4b); - goto done; + ext4_grp_locked_error(sb, e4b->bd_group, + inode ? inode->i_ino : 0, blocknr, + "freeing already freed block (bit %u); block bitmap corrupt.", + block); + return; } + this_cpu_inc(discard_pa_seq); + e4b->bd_info->bb_free += count; + if (first < e4b->bd_info->bb_first_free) + e4b->bd_info->bb_first_free = first; + /* let's maintain fragments counter */ if (left_is_free && right_is_free) e4b->bd_info->bb_fragments--; @@ -1547,16 +2067,16 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b, if (first <= last) mb_buddy_mark_free(e4b, first >> 1, last >> 1); -done: mb_set_largest_free_order(sb, e4b->bd_info); + mb_update_avg_fragment_size(sb, e4b->bd_info); +check: mb_check_buddy(e4b); } static int mb_find_extent(struct ext4_buddy *e4b, int block, int needed, struct ext4_free_extent *ex) { - int next = block; - int max, order; + int max, order, next; void *buddy; assert_spin_locked(ext4_group_lock_ptr(e4b->bd_sb, e4b->bd_group)); @@ -1574,16 +2094,12 @@ static int mb_find_extent(struct ext4_buddy *e4b, int block, /* find actual order */ order = mb_find_order_for_block(e4b, block); - block = block >> order; - ex->fe_len = 1 << order; - ex->fe_start = block << order; + ex->fe_len = (1 << order) - (block & ((1 << order) - 1)); + ex->fe_start = block; ex->fe_group = e4b->bd_group; - /* calc difference from given start */ - next = next - ex->fe_start; - ex->fe_len -= next; - ex->fe_start += next; + block = block >> order; while (needed > ex->fe_len && mb_find_buddy(e4b, order, &max)) { @@ -1604,10 +2120,11 @@ static int mb_find_extent(struct ext4_buddy *e4b, int block, if (ex->fe_start + ex->fe_len > EXT4_CLUSTERS_PER_GROUP(e4b->bd_sb)) { /* Should never happen! (but apparently sometimes does?!?) */ WARN_ON(1); - ext4_error(e4b->bd_sb, "corruption or bug in mb_find_extent " - "block=%d, order=%d needed=%d ex=%u/%d/%d@%u", - block, order, needed, ex->fe_group, ex->fe_start, - ex->fe_len, ex->fe_logical); + ext4_grp_locked_error(e4b->bd_sb, e4b->bd_group, 0, 0, + "corruption or bug in mb_find_extent " + "block=%d, order=%d needed=%d ex=%u/%d/%d@%u", + block, order, needed, ex->fe_group, ex->fe_start, + ex->fe_len, ex->fe_logical); ex->fe_len = 0; ex->fe_start = 0; ex->fe_group = 0; @@ -1620,12 +2137,12 @@ static int mb_mark_used(struct ext4_buddy *e4b, struct ext4_free_extent *ex) int ord; int mlen = 0; int max = 0; - int cur; int start = ex->fe_start; int len = ex->fe_len; unsigned ret = 0; int len0 = len; void *buddy; + int ord_start, ord_end; BUG_ON(start + len > (e4b->bd_sb->s_blocksize << 3)); BUG_ON(e4b->bd_group != ex->fe_group); @@ -1669,23 +2186,34 @@ static int mb_mark_used(struct ext4_buddy *e4b, struct ext4_free_extent *ex) if (ret == 0) ret = len | (ord << 16); - /* we have to split large buddy */ BUG_ON(ord <= 0); buddy = mb_find_buddy(e4b, ord, &max); mb_set_bit(start >> ord, buddy); e4b->bd_info->bb_counters[ord]--; - ord--; - cur = (start >> ord) & ~1U; - buddy = mb_find_buddy(e4b, ord, &max); - mb_clear_bit(cur, buddy); - mb_clear_bit(cur + 1, buddy); - e4b->bd_info->bb_counters[ord]++; - e4b->bd_info->bb_counters[ord]++; + ord_start = (start >> ord) << ord; + ord_end = ord_start + (1 << ord); + /* first chunk */ + if (start > ord_start) + ext4_mb_mark_free_simple(e4b->bd_sb, e4b->bd_buddy, + ord_start, start - ord_start, + e4b->bd_info); + + /* last chunk */ + if (start + len < ord_end) { + ext4_mb_mark_free_simple(e4b->bd_sb, e4b->bd_buddy, + start + len, + ord_end - (start + len), + e4b->bd_info); + break; + } + len = start + len - ord_end; + start = ord_end; } mb_set_largest_free_order(e4b->bd_sb, e4b->bd_info); - ext4_set_bits(e4b->bd_bitmap, ex->fe_start, len0); + mb_update_avg_fragment_size(e4b->bd_sb, e4b->bd_info); + mb_set_bits(e4b->bd_bitmap, ex->fe_start, len0); mb_check_buddy(e4b); return ret; @@ -1716,23 +2244,23 @@ static void ext4_mb_use_best_found(struct ext4_allocation_context *ac, ac->ac_buddy = ret >> 16; /* - * take the page reference. We want the page to be pinned + * take the folio reference. We want the folio to be pinned * so that we don't get a ext4_mb_init_cache_call for this * group until we update the bitmap. That would mean we * double allocate blocks. The reference is dropped * in ext4_mb_release_context */ - ac->ac_bitmap_page = e4b->bd_bitmap_page; - get_page(ac->ac_bitmap_page); - ac->ac_buddy_page = e4b->bd_buddy_page; - get_page(ac->ac_buddy_page); + ac->ac_bitmap_folio = e4b->bd_bitmap_folio; + folio_get(ac->ac_bitmap_folio); + ac->ac_buddy_folio = e4b->bd_buddy_folio; + folio_get(ac->ac_buddy_folio); /* store last allocated for subsequent stream allocation */ if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) { - spin_lock(&sbi->s_md_lock); - sbi->s_mb_last_group = ac->ac_f_ex.fe_group; - sbi->s_mb_last_start = ac->ac_f_ex.fe_start; - spin_unlock(&sbi->s_md_lock); + int hash = ac->ac_inode->i_ino % sbi->s_mb_nr_global_goals; + + WRITE_ONCE(sbi->s_mb_last_groups[hash], ac->ac_f_ex.fe_group); } + /* * As we've just preallocated more space than * user requested originally, we store allocated @@ -1743,10 +2271,6 @@ static void ext4_mb_use_best_found(struct ext4_allocation_context *ac, } -/* - * regular allocator, for general purposes allocation - */ - static void ext4_mb_check_limits(struct ext4_allocation_context *ac, struct ext4_buddy *e4b, int finish_group) @@ -1754,8 +2278,6 @@ static void ext4_mb_check_limits(struct ext4_allocation_context *ac, struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); struct ext4_free_extent *bex = &ac->ac_b_ex; struct ext4_free_extent *gex = &ac->ac_g_ex; - struct ext4_free_extent ex; - int max; if (ac->ac_status == AC_STATUS_FOUND) return; @@ -1774,17 +2296,8 @@ static void ext4_mb_check_limits(struct ext4_allocation_context *ac, if (bex->fe_len < gex->fe_len) return; - if ((finish_group || ac->ac_found > sbi->s_mb_min_to_scan) - && bex->fe_group == e4b->bd_group) { - /* recheck chunk's availability - we don't know - * when it was found (within this lock-unlock - * period or not) */ - max = mb_find_extent(e4b, bex->fe_start, gex->fe_len, &ex); - if (max >= gex->fe_len) { - ext4_mb_use_best_found(ac, e4b); - return; - } - } + if (finish_group || ac->ac_found > sbi->s_mb_min_to_scan) + ext4_mb_use_best_found(ac, e4b); } /* @@ -1795,6 +2308,20 @@ static void ext4_mb_check_limits(struct ext4_allocation_context *ac, * in the context. Later, the best found extent will be used, if * mballoc can't find good enough extent. * + * The algorithm used is roughly as follows: + * + * * If free extent found is exactly as big as goal, then + * stop the scan and use it immediately + * + * * If free extent found is smaller than goal, then keep retrying + * upto a max of sbi->s_mb_max_to_scan times (default 200). After + * that stop scanning and use whatever we have. + * + * * If free extent found is bigger than goal, then keep retrying + * upto a max of sbi->s_mb_min_to_scan times (default 10) before + * stopping the scan and using the extent. + * + * * FIXME: real allocation policy is to be designed yet! */ static void ext4_mb_measure_extent(struct ext4_allocation_context *ac, @@ -1810,6 +2337,7 @@ static void ext4_mb_measure_extent(struct ext4_allocation_context *ac, BUG_ON(ac->ac_status != AC_STATUS_CONTINUE); ac->ac_found++; + ac->ac_cX_found[ac->ac_criteria]++; /* * The special case - take what you catch first @@ -1857,7 +2385,7 @@ static void ext4_mb_measure_extent(struct ext4_allocation_context *ac, } static noinline_for_stack -int ext4_mb_try_best_found(struct ext4_allocation_context *ac, +void ext4_mb_try_best_found(struct ext4_allocation_context *ac, struct ext4_buddy *e4b) { struct ext4_free_extent ex = ac->ac_b_ex; @@ -1868,9 +2396,12 @@ int ext4_mb_try_best_found(struct ext4_allocation_context *ac, BUG_ON(ex.fe_len <= 0); err = ext4_mb_load_buddy(ac->ac_sb, group, e4b); if (err) - return err; + return; ext4_lock_group(ac->ac_sb, group); + if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) + goto out; + max = mb_find_extent(e4b, ex.fe_start, ex.fe_len, &ex); if (max > 0) { @@ -1878,10 +2409,9 @@ int ext4_mb_try_best_found(struct ext4_allocation_context *ac, ext4_mb_use_best_found(ac, e4b); } +out: ext4_unlock_group(ac->ac_sb, group); ext4_mb_unload_buddy(e4b); - - return 0; } static noinline_for_stack @@ -1895,7 +2425,9 @@ int ext4_mb_find_by_goal(struct ext4_allocation_context *ac, struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group); struct ext4_free_extent ex; - if (!(ac->ac_flags & EXT4_MB_HINT_TRY_GOAL)) + if (!grp) + return -EFSCORRUPTED; + if (!(ac->ac_flags & (EXT4_MB_HINT_TRY_GOAL | EXT4_MB_HINT_GOAL_ONLY))) return 0; if (grp->bb_free == 0) return 0; @@ -1904,21 +2436,19 @@ int ext4_mb_find_by_goal(struct ext4_allocation_context *ac, if (err) return err; - if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) { - ext4_mb_unload_buddy(e4b); - return 0; - } - ext4_lock_group(ac->ac_sb, group); + if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) + goto out; + max = mb_find_extent(e4b, ac->ac_g_ex.fe_start, ac->ac_g_ex.fe_len, &ex); ex.fe_logical = 0xDEADFA11; /* debug value */ - if (max >= ac->ac_g_ex.fe_len && ac->ac_g_ex.fe_len == sbi->s_stripe) { + if (max >= ac->ac_g_ex.fe_len && + ac->ac_g_ex.fe_len == EXT4_NUM_B2C(sbi, sbi->s_stripe)) { ext4_fsblk_t start; - start = ext4_group_first_block_no(ac->ac_sb, e4b->bd_group) + - ex.fe_start; + start = ext4_grp_offs_to_block(ac->ac_sb, &ex); /* use do_div to get remainder (would be 64-bit modulo) */ if (do_div(start, sbi->s_stripe) == 0) { ac->ac_found++; @@ -1942,6 +2472,7 @@ int ext4_mb_find_by_goal(struct ext4_allocation_context *ac, ac->ac_b_ex = ex; ext4_mb_use_best_found(ac, e4b); } +out: ext4_unlock_group(ac->ac_sb, group); ext4_mb_unload_buddy(e4b); @@ -1964,24 +2495,27 @@ void ext4_mb_simple_scan_group(struct ext4_allocation_context *ac, int max; BUG_ON(ac->ac_2order <= 0); - for (i = ac->ac_2order; i <= sb->s_blocksize_bits + 1; i++) { + for (i = ac->ac_2order; i < MB_NUM_ORDERS(sb); i++) { if (grp->bb_counters[i] == 0) continue; buddy = mb_find_buddy(e4b, i, &max); - BUG_ON(buddy == NULL); + if (WARN_RATELIMIT(buddy == NULL, + "ext4: mb_simple_scan_group: mb_find_buddy failed, (%d)\n", i)) + continue; k = mb_find_next_zero_bit(buddy, max, 0); if (k >= max) { + ext4_mark_group_bitmap_corrupted(ac->ac_sb, + e4b->bd_group, + EXT4_GROUP_INFO_BBITMAP_CORRUPT); ext4_grp_locked_error(ac->ac_sb, e4b->bd_group, 0, 0, "%d free clusters of order %d. But found 0", grp->bb_counters[i], i); - ext4_mark_group_bitmap_corrupted(ac->ac_sb, - e4b->bd_group, - EXT4_GROUP_INFO_BBITMAP_CORRUPT); break; } ac->ac_found++; + ac->ac_cX_found[ac->ac_criteria]++; ac->ac_b_ex.fe_len = 1 << i; ac->ac_b_ex.fe_start = k << i; @@ -2010,7 +2544,7 @@ void ext4_mb_complex_scan_group(struct ext4_allocation_context *ac, struct super_block *sb = ac->ac_sb; void *bitmap = e4b->bd_bitmap; struct ext4_free_extent ex; - int i; + int i, j, freelen; int free; free = e4b->bd_info->bb_free; @@ -2026,27 +2560,45 @@ void ext4_mb_complex_scan_group(struct ext4_allocation_context *ac, /* * IF we have corrupt bitmap, we won't find any * free blocks even though group info says we - * we have free blocks + * have free blocks */ + ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, + EXT4_GROUP_INFO_BBITMAP_CORRUPT); ext4_grp_locked_error(sb, e4b->bd_group, 0, 0, "%d free clusters as per " "group info. But bitmap says 0", free); - ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, - EXT4_GROUP_INFO_BBITMAP_CORRUPT); break; } + if (!ext4_mb_cr_expensive(ac->ac_criteria)) { + /* + * In CR_GOAL_LEN_FAST and CR_BEST_AVAIL_LEN, we are + * sure that this group will have a large enough + * continuous free extent, so skip over the smaller free + * extents + */ + j = mb_find_next_bit(bitmap, + EXT4_CLUSTERS_PER_GROUP(sb), i); + freelen = j - i; + + if (freelen < ac->ac_g_ex.fe_len) { + i = j; + free -= freelen; + continue; + } + } + mb_find_extent(e4b, i, ac->ac_g_ex.fe_len, &ex); if (WARN_ON(ex.fe_len <= 0)) break; if (free < ex.fe_len) { + ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, + EXT4_GROUP_INFO_BBITMAP_CORRUPT); ext4_grp_locked_error(sb, e4b->bd_group, 0, 0, "%d free clusters as per " "group info. But got %d blocks", free, ex.fe_len); - ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group, - EXT4_GROUP_INFO_BBITMAP_CORRUPT); /* * The number of free blocks differs. This mostly * indicate that the bitmap is corrupt. So exit @@ -2078,7 +2630,7 @@ void ext4_mb_scan_aligned(struct ext4_allocation_context *ac, struct ext4_free_extent ex; ext4_fsblk_t first_group_block; ext4_fsblk_t a; - ext4_grpblk_t i; + ext4_grpblk_t i, stripe; int max; BUG_ON(sbi->s_stripe == 0); @@ -2090,42 +2642,67 @@ void ext4_mb_scan_aligned(struct ext4_allocation_context *ac, do_div(a, sbi->s_stripe); i = (a * sbi->s_stripe) - first_group_block; + stripe = EXT4_NUM_B2C(sbi, sbi->s_stripe); + i = EXT4_B2C(sbi, i); while (i < EXT4_CLUSTERS_PER_GROUP(sb)) { if (!mb_test_bit(i, bitmap)) { - max = mb_find_extent(e4b, i, sbi->s_stripe, &ex); - if (max >= sbi->s_stripe) { + max = mb_find_extent(e4b, i, stripe, &ex); + if (max >= stripe) { ac->ac_found++; + ac->ac_cX_found[ac->ac_criteria]++; ex.fe_logical = 0xDEADF00D; /* debug value */ ac->ac_b_ex = ex; ext4_mb_use_best_found(ac, e4b); break; } } - i += sbi->s_stripe; + i += stripe; } } +static void __ext4_mb_scan_group(struct ext4_allocation_context *ac) +{ + bool is_stripe_aligned; + struct ext4_sb_info *sbi; + enum criteria cr = ac->ac_criteria; + + ac->ac_groups_scanned++; + if (cr == CR_POWER2_ALIGNED) + return ext4_mb_simple_scan_group(ac, ac->ac_e4b); + + sbi = EXT4_SB(ac->ac_sb); + is_stripe_aligned = false; + if ((sbi->s_stripe >= sbi->s_cluster_ratio) && + !(ac->ac_g_ex.fe_len % EXT4_NUM_B2C(sbi, sbi->s_stripe))) + is_stripe_aligned = true; + + if ((cr == CR_GOAL_LEN_FAST || cr == CR_BEST_AVAIL_LEN) && + is_stripe_aligned) + ext4_mb_scan_aligned(ac, ac->ac_e4b); + + if (ac->ac_status == AC_STATUS_CONTINUE) + ext4_mb_complex_scan_group(ac, ac->ac_e4b); +} + /* * This is also called BEFORE we load the buddy bitmap. * Returns either 1 or 0 indicating that the group is either suitable * for the allocation or not. */ static bool ext4_mb_good_group(struct ext4_allocation_context *ac, - ext4_group_t group, int cr) + ext4_group_t group, enum criteria cr) { ext4_grpblk_t free, fragments; int flex_size = ext4_flex_bg_size(EXT4_SB(ac->ac_sb)); struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group); - BUG_ON(cr < 0 || cr >= 4); + BUG_ON(cr < CR_POWER2_ALIGNED || cr >= EXT4_MB_NUM_CRS); - free = grp->bb_free; - if (free == 0) - return false; - if (cr <= 2 && free < ac->ac_g_ex.fe_len) + if (unlikely(!grp || EXT4_MB_GRP_BBITMAP_CORRUPT(grp))) return false; - if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp))) + free = grp->bb_free; + if (free == 0) return false; fragments = grp->bb_fragments; @@ -2133,7 +2710,7 @@ static bool ext4_mb_good_group(struct ext4_allocation_context *ac, return false; switch (cr) { - case 0: + case CR_POWER2_ALIGNED: BUG_ON(ac->ac_2order == 0); /* Avoid using the first bg of a flexgroup for data files */ @@ -2142,23 +2719,26 @@ static bool ext4_mb_good_group(struct ext4_allocation_context *ac, ((group % flex_size) == 0)) return false; - if ((ac->ac_2order > ac->ac_sb->s_blocksize_bits+1) || - (free / fragments) >= ac->ac_g_ex.fe_len) + if (free < ac->ac_g_ex.fe_len) + return false; + + if (ac->ac_2order >= MB_NUM_ORDERS(ac->ac_sb)) return true; if (grp->bb_largest_free_order < ac->ac_2order) return false; return true; - case 1: + case CR_GOAL_LEN_FAST: + case CR_BEST_AVAIL_LEN: if ((free / fragments) >= ac->ac_g_ex.fe_len) return true; break; - case 2: + case CR_GOAL_LEN_SLOW: if (free >= ac->ac_g_ex.fe_len) return true; break; - case 3: + case CR_ANY_FREE: return true; default: BUG(); @@ -2171,61 +2751,245 @@ static bool ext4_mb_good_group(struct ext4_allocation_context *ac, * This could return negative error code if something goes wrong * during ext4_mb_init_group(). This should not be called with * ext4_lock_group() held. + * + * Note: because we are conditionally operating with the group lock in + * the EXT4_MB_STRICT_CHECK case, we need to fake out sparse in this + * function using __acquire and __release. This means we need to be + * super careful before messing with the error path handling via "goto + * out"! */ static int ext4_mb_good_group_nolock(struct ext4_allocation_context *ac, - ext4_group_t group, int cr) + ext4_group_t group, enum criteria cr) { struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group); struct super_block *sb = ac->ac_sb; + struct ext4_sb_info *sbi = EXT4_SB(sb); bool should_lock = ac->ac_flags & EXT4_MB_STRICT_CHECK; ext4_grpblk_t free; int ret = 0; - if (should_lock) + if (!grp) + return -EFSCORRUPTED; + if (sbi->s_mb_stats) + atomic64_inc(&sbi->s_bal_cX_groups_considered[ac->ac_criteria]); + if (should_lock) { ext4_lock_group(sb, group); + __release(ext4_group_lock_ptr(sb, group)); + } free = grp->bb_free; if (free == 0) goto out; - if (cr <= 2 && free < ac->ac_g_ex.fe_len) + /* + * In all criterias except CR_ANY_FREE we try to avoid groups that + * can't possibly satisfy the full goal request due to insufficient + * free blocks. + */ + if (cr < CR_ANY_FREE && free < ac->ac_g_ex.fe_len) goto out; if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp))) goto out; - if (should_lock) + if (should_lock) { + __acquire(ext4_group_lock_ptr(sb, group)); ext4_unlock_group(sb, group); + } /* We only do this if the grp has never been initialized */ if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) { - ret = ext4_mb_init_group(ac->ac_sb, group, GFP_NOFS); + struct ext4_group_desc *gdp = + ext4_get_group_desc(sb, group, NULL); + int ret; + + /* + * CR_POWER2_ALIGNED/CR_GOAL_LEN_FAST is a very optimistic + * search to find large good chunks almost for free. If buddy + * data is not ready, then this optimization makes no sense. But + * we never skip the first block group in a flex_bg, since this + * gets used for metadata block allocation, and we want to make + * sure we locate metadata blocks in the first block group in + * the flex_bg if possible. + */ + if (!ext4_mb_cr_expensive(cr) && + (!sbi->s_log_groups_per_flex || + ((group & ((1 << sbi->s_log_groups_per_flex) - 1)) != 0)) && + !(ext4_has_group_desc_csum(sb) && + (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)))) + return 0; + ret = ext4_mb_init_group(sb, group, GFP_NOFS); if (ret) return ret; } - if (should_lock) + if (should_lock) { ext4_lock_group(sb, group); + __release(ext4_group_lock_ptr(sb, group)); + } ret = ext4_mb_good_group(ac, group, cr); out: - if (should_lock) + if (should_lock) { + __acquire(ext4_group_lock_ptr(sb, group)); ext4_unlock_group(sb, group); + } + return ret; +} + +/* + * Start prefetching @nr block bitmaps starting at @group. + * Return the next group which needs to be prefetched. + */ +ext4_group_t ext4_mb_prefetch(struct super_block *sb, ext4_group_t group, + unsigned int nr, int *cnt) +{ + ext4_group_t ngroups = ext4_get_groups_count(sb); + struct buffer_head *bh; + struct blk_plug plug; + + blk_start_plug(&plug); + while (nr-- > 0) { + struct ext4_group_desc *gdp = ext4_get_group_desc(sb, group, + NULL); + struct ext4_group_info *grp = ext4_get_group_info(sb, group); + + /* + * Prefetch block groups with free blocks; but don't + * bother if it is marked uninitialized on disk, since + * it won't require I/O to read. Also only try to + * prefetch once, so we avoid getblk() call, which can + * be expensive. + */ + if (gdp && grp && !EXT4_MB_GRP_TEST_AND_SET_READ(grp) && + EXT4_MB_GRP_NEED_INIT(grp) && + ext4_free_group_clusters(sb, gdp) > 0 ) { + bh = ext4_read_block_bitmap_nowait(sb, group, true); + if (bh && !IS_ERR(bh)) { + if (!buffer_uptodate(bh) && cnt) + (*cnt)++; + brelse(bh); + } + } + if (++group >= ngroups) + group = 0; + } + blk_finish_plug(&plug); + return group; +} + +/* + * Batch reads of the block allocation bitmaps to get + * multiple READs in flight; limit prefetching at inexpensive + * CR, otherwise mballoc can spend a lot of time loading + * imperfect groups + */ +static void ext4_mb_might_prefetch(struct ext4_allocation_context *ac, + ext4_group_t group) +{ + struct ext4_sb_info *sbi; + + if (ac->ac_prefetch_grp != group) + return; + + sbi = EXT4_SB(ac->ac_sb); + if (ext4_mb_cr_expensive(ac->ac_criteria) || + ac->ac_prefetch_ios < sbi->s_mb_prefetch_limit) { + unsigned int nr = sbi->s_mb_prefetch; + + if (ext4_has_feature_flex_bg(ac->ac_sb)) { + nr = 1 << sbi->s_log_groups_per_flex; + nr -= group & (nr - 1); + nr = umin(nr, sbi->s_mb_prefetch); + } + + ac->ac_prefetch_nr = nr; + ac->ac_prefetch_grp = ext4_mb_prefetch(ac->ac_sb, group, nr, + &ac->ac_prefetch_ios); + } +} + +/* + * Prefetching reads the block bitmap into the buffer cache; but we + * need to make sure that the buddy bitmap in the page cache has been + * initialized. Note that ext4_mb_init_group() will block if the I/O + * is not yet completed, or indeed if it was not initiated by + * ext4_mb_prefetch did not start the I/O. + * + * TODO: We should actually kick off the buddy bitmap setup in a work + * queue when the buffer I/O is completed, so that we don't block + * waiting for the block allocation bitmap read to finish when + * ext4_mb_prefetch_fini is called from ext4_mb_regular_allocator(). + */ +void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, + unsigned int nr) +{ + struct ext4_group_desc *gdp; + struct ext4_group_info *grp; + + while (nr-- > 0) { + if (!group) + group = ext4_get_groups_count(sb); + group--; + gdp = ext4_get_group_desc(sb, group, NULL); + grp = ext4_get_group_info(sb, group); + + if (grp && gdp && EXT4_MB_GRP_NEED_INIT(grp) && + ext4_free_group_clusters(sb, gdp) > 0) { + if (ext4_mb_init_group(sb, group, GFP_NOFS)) + break; + } + } +} + +static int ext4_mb_scan_group(struct ext4_allocation_context *ac, + ext4_group_t group) +{ + int ret; + struct super_block *sb = ac->ac_sb; + enum criteria cr = ac->ac_criteria; + + ext4_mb_might_prefetch(ac, group); + + /* prevent unnecessary buddy loading. */ + if (cr < CR_ANY_FREE && spin_is_locked(ext4_group_lock_ptr(sb, group))) + return 0; + + /* This now checks without needing the buddy folio */ + ret = ext4_mb_good_group_nolock(ac, group, cr); + if (ret <= 0) { + if (!ac->ac_first_err) + ac->ac_first_err = ret; + return 0; + } + + ret = ext4_mb_load_buddy(sb, group, ac->ac_e4b); + if (ret) + return ret; + + /* skip busy group */ + if (cr >= CR_ANY_FREE) + ext4_lock_group(sb, group); + else if (!ext4_try_lock_group(sb, group)) + goto out_unload; + + /* We need to check again after locking the block group. */ + if (unlikely(!ext4_mb_good_group(ac, group, cr))) + goto out_unlock; + + __ext4_mb_scan_group(ac); + +out_unlock: + ext4_unlock_group(sb, group); +out_unload: + ext4_mb_unload_buddy(ac->ac_e4b); return ret; } static noinline_for_stack int ext4_mb_regular_allocator(struct ext4_allocation_context *ac) { - ext4_group_t ngroups, group, i; - int cr = -1; - int err = 0, first_err = 0; - struct ext4_sb_info *sbi; - struct super_block *sb; + ext4_group_t i; + int err = 0; + struct super_block *sb = ac->ac_sb; + struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_buddy e4b; - sb = ac->ac_sb; - sbi = EXT4_SB(sb); - ngroups = ext4_get_groups_count(sb); - /* non-extent files are limited to low blocks/groups */ - if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS))) - ngroups = sbi->s_blockfile_groups; - BUG_ON(ac->ac_status == AC_STATUS_FOUND); /* first, try the goal */ @@ -2237,9 +3001,9 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac) goto out; /* - * ac->ac2_order is set only if the fe_len is a power of 2 - * if ac2_order is set we also set criteria to 0 so that we - * try exact allocation using buddy. + * ac->ac_2order is set only if the fe_len is a power of 2 + * if ac->ac_2order is set we also set criteria to CR_POWER2_ALIGNED + * so that we try exact allocation using buddy. */ i = fls(ac->ac_g_ex.fe_len); ac->ac_2order = 0; @@ -2250,89 +3014,41 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac) * We also support searching for power-of-two requests only for * requests upto maximum buddy size we have constructed. */ - if (i >= sbi->s_mb_order2_reqs && i <= sb->s_blocksize_bits + 2) { - /* - * This should tell if fe_len is exactly power of 2 - */ - if ((ac->ac_g_ex.fe_len & (~(1 << (i - 1)))) == 0) + if (i >= sbi->s_mb_order2_reqs && i <= MB_NUM_ORDERS(sb)) { + if (is_power_of_2(ac->ac_g_ex.fe_len)) ac->ac_2order = array_index_nospec(i - 1, - sb->s_blocksize_bits + 2); + MB_NUM_ORDERS(sb)); } /* if stream allocation is enabled, use global goal */ if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) { - /* TBD: may be hot point */ - spin_lock(&sbi->s_md_lock); - ac->ac_g_ex.fe_group = sbi->s_mb_last_group; - ac->ac_g_ex.fe_start = sbi->s_mb_last_start; - spin_unlock(&sbi->s_md_lock); + int hash = ac->ac_inode->i_ino % sbi->s_mb_nr_global_goals; + + ac->ac_g_ex.fe_group = READ_ONCE(sbi->s_mb_last_groups[hash]); + ac->ac_g_ex.fe_start = -1; + ac->ac_flags &= ~EXT4_MB_HINT_TRY_GOAL; } - /* Let's just scan groups to find more-less suitable blocks */ - cr = ac->ac_2order ? 0 : 1; /* - * cr == 0 try to get exact allocation, - * cr == 3 try to get anything + * Let's just scan groups to find more-less suitable blocks We + * start with CR_GOAL_LEN_FAST, unless it is power of 2 + * aligned, in which case let's do that faster approach first. */ -repeat: - for (; cr < 4 && ac->ac_status == AC_STATUS_CONTINUE; cr++) { - ac->ac_criteria = cr; - /* - * searching for the right group start - * from the goal value specified - */ - group = ac->ac_g_ex.fe_group; - - for (i = 0; i < ngroups; group++, i++) { - int ret = 0; - cond_resched(); - /* - * Artificially restricted ngroups for non-extent - * files makes group > ngroups possible on first loop. - */ - if (group >= ngroups) - group = 0; - - /* This now checks without needing the buddy page */ - ret = ext4_mb_good_group_nolock(ac, group, cr); - if (ret <= 0) { - if (!first_err) - first_err = ret; - continue; - } - - err = ext4_mb_load_buddy(sb, group, &e4b); - if (err) - goto out; - - ext4_lock_group(sb, group); - - /* - * We need to check again after locking the - * block group - */ - ret = ext4_mb_good_group(ac, group, cr); - if (ret == 0) { - ext4_unlock_group(sb, group); - ext4_mb_unload_buddy(&e4b); - continue; - } + ac->ac_criteria = CR_GOAL_LEN_FAST; + if (ac->ac_2order) + ac->ac_criteria = CR_POWER2_ALIGNED; - ac->ac_groups_scanned++; - if (cr == 0) - ext4_mb_simple_scan_group(ac, &e4b); - else if (cr == 1 && sbi->s_stripe && - !(ac->ac_g_ex.fe_len % sbi->s_stripe)) - ext4_mb_scan_aligned(ac, &e4b); - else - ext4_mb_complex_scan_group(ac, &e4b); - - ext4_unlock_group(sb, group); - ext4_mb_unload_buddy(&e4b); + ac->ac_e4b = &e4b; + ac->ac_prefetch_ios = 0; + ac->ac_first_err = 0; +repeat: + while (ac->ac_criteria < EXT4_MB_NUM_CRS) { + err = ext4_mb_scan_groups(ac); + if (err) + goto out; - if (ac->ac_status != AC_STATUS_CONTINUE) - break; - } + if (ac->ac_status != AC_STATUS_CONTINUE) + break; } if (ac->ac_b_ex.fe_len > 0 && ac->ac_status != AC_STATUS_FOUND && @@ -2341,38 +3057,53 @@ repeat: * We've been searching too long. Let's try to allocate * the best chunk we've found so far */ - ext4_mb_try_best_found(ac, &e4b); if (ac->ac_status != AC_STATUS_FOUND) { + int lost; + /* * Someone more lucky has already allocated it. * The only thing we can do is just take first * found block(s) - printk(KERN_DEBUG "EXT4-fs: someone won our chunk\n"); */ + lost = atomic_inc_return(&sbi->s_mb_lost_chunks); + mb_debug(sb, "lost chunk, group: %u, start: %d, len: %d, lost: %d\n", + ac->ac_b_ex.fe_group, ac->ac_b_ex.fe_start, + ac->ac_b_ex.fe_len, lost); + ac->ac_b_ex.fe_group = 0; ac->ac_b_ex.fe_start = 0; ac->ac_b_ex.fe_len = 0; ac->ac_status = AC_STATUS_CONTINUE; ac->ac_flags |= EXT4_MB_HINT_FIRST; - cr = 3; - atomic_inc(&sbi->s_mb_lost_chunks); + ac->ac_criteria = CR_ANY_FREE; goto repeat; } } + + if (sbi->s_mb_stats && ac->ac_status == AC_STATUS_FOUND) { + atomic64_inc(&sbi->s_bal_cX_hits[ac->ac_criteria]); + if (ac->ac_flags & EXT4_MB_STREAM_ALLOC && + ac->ac_b_ex.fe_group == ac->ac_g_ex.fe_group) + atomic_inc(&sbi->s_bal_stream_goals); + } out: - if (!err && ac->ac_status != AC_STATUS_FOUND && first_err) - err = first_err; + if (!err && ac->ac_status != AC_STATUS_FOUND && ac->ac_first_err) + err = ac->ac_first_err; mb_debug(sb, "Best len %d, origin len %d, ac_status %u, ac_flags 0x%x, cr %d ret %d\n", ac->ac_b_ex.fe_len, ac->ac_o_ex.fe_len, ac->ac_status, - ac->ac_flags, cr, err); + ac->ac_flags, ac->ac_criteria, err); + + if (ac->ac_prefetch_nr) + ext4_mb_prefetch_fini(sb, ac->ac_prefetch_grp, ac->ac_prefetch_nr); + return err; } static void *ext4_mb_seq_groups_start(struct seq_file *seq, loff_t *pos) { - struct super_block *sb = PDE_DATA(file_inode(seq->file)); + struct super_block *sb = pde_data(file_inode(seq->file)); ext4_group_t group; if (*pos < 0 || *pos >= ext4_get_groups_count(sb)) @@ -2383,7 +3114,7 @@ static void *ext4_mb_seq_groups_start(struct seq_file *seq, loff_t *pos) static void *ext4_mb_seq_groups_next(struct seq_file *seq, void *v, loff_t *pos) { - struct super_block *sb = PDE_DATA(file_inode(seq->file)); + struct super_block *sb = pde_data(file_inode(seq->file)); ext4_group_t group; ++*pos; @@ -2395,19 +3126,17 @@ static void *ext4_mb_seq_groups_next(struct seq_file *seq, void *v, loff_t *pos) static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v) { - struct super_block *sb = PDE_DATA(file_inode(seq->file)); + struct super_block *sb = pde_data(file_inode(seq->file)); ext4_group_t group = (ext4_group_t) ((unsigned long) v); - int i; - int err, buddy_loaded = 0; + int i, err; + char nbuf[16]; struct ext4_buddy e4b; struct ext4_group_info *grinfo; unsigned char blocksize_bits = min_t(unsigned char, sb->s_blocksize_bits, EXT4_MAX_BLOCK_LOG_SIZE); - struct sg { - struct ext4_group_info info; - ext4_grpblk_t counters[EXT4_MAX_BLOCK_LOG_SIZE + 2]; - } sg; + DEFINE_RAW_FLEX(struct ext4_group_info, sg, bb_counters, + EXT4_MAX_BLOCK_LOG_SIZE + 2); group--; if (group == 0) @@ -2415,32 +3144,36 @@ static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v) " 2^0 2^1 2^2 2^3 2^4 2^5 2^6 " " 2^7 2^8 2^9 2^10 2^11 2^12 2^13 ]\n"); - i = (blocksize_bits + 2) * sizeof(sg.info.bb_counters[0]) + + i = (blocksize_bits + 2) * sizeof(sg->bb_counters[0]) + sizeof(struct ext4_group_info); grinfo = ext4_get_group_info(sb, group); + if (!grinfo) + return 0; /* Load the group info in memory only if not already loaded. */ if (unlikely(EXT4_MB_GRP_NEED_INIT(grinfo))) { err = ext4_mb_load_buddy(sb, group, &e4b); if (err) { - seq_printf(seq, "#%-5u: I/O error\n", group); + seq_printf(seq, "#%-5u: %s\n", group, ext4_decode_error(NULL, err, nbuf)); return 0; } - buddy_loaded = 1; - } - - memcpy(&sg, ext4_get_group_info(sb, group), i); - - if (buddy_loaded) ext4_mb_unload_buddy(&e4b); + } - seq_printf(seq, "#%-5u: %-5u %-5u %-5u [", group, sg.info.bb_free, - sg.info.bb_fragments, sg.info.bb_first_free); + /* + * We care only about free space counters in the group info and + * these are safe to access even after the buddy has been unloaded + */ + memcpy(sg, grinfo, i); + seq_printf(seq, "#%-5u: %-5u %-5u %-5u [", group, sg->bb_free, + sg->bb_fragments, sg->bb_first_free); for (i = 0; i <= 13; i++) seq_printf(seq, " %-5u", i <= blocksize_bits + 1 ? - sg.info.bb_counters[i] : 0); - seq_printf(seq, " ]\n"); - + sg->bb_counters[i] : 0); + seq_puts(seq, " ]"); + if (EXT4_MB_GRP_BBITMAP_CORRUPT(sg)) + seq_puts(seq, " Block bitmap corrupted!"); + seq_putc(seq, '\n'); return 0; } @@ -2455,6 +3188,180 @@ const struct seq_operations ext4_mb_seq_groups_ops = { .show = ext4_mb_seq_groups_show, }; +int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset) +{ + struct super_block *sb = seq->private; + struct ext4_sb_info *sbi = EXT4_SB(sb); + + seq_puts(seq, "mballoc:\n"); + if (!sbi->s_mb_stats) { + seq_puts(seq, "\tmb stats collection turned off.\n"); + seq_puts( + seq, + "\tTo enable, please write \"1\" to sysfs file mb_stats.\n"); + return 0; + } + seq_printf(seq, "\treqs: %u\n", atomic_read(&sbi->s_bal_reqs)); + seq_printf(seq, "\tsuccess: %u\n", atomic_read(&sbi->s_bal_success)); + + seq_printf(seq, "\tgroups_scanned: %u\n", + atomic_read(&sbi->s_bal_groups_scanned)); + + /* CR_POWER2_ALIGNED stats */ + seq_puts(seq, "\tcr_p2_aligned_stats:\n"); + seq_printf(seq, "\t\thits: %llu\n", + atomic64_read(&sbi->s_bal_cX_hits[CR_POWER2_ALIGNED])); + seq_printf( + seq, "\t\tgroups_considered: %llu\n", + atomic64_read( + &sbi->s_bal_cX_groups_considered[CR_POWER2_ALIGNED])); + seq_printf(seq, "\t\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_cX_ex_scanned[CR_POWER2_ALIGNED])); + seq_printf(seq, "\t\tuseless_loops: %llu\n", + atomic64_read(&sbi->s_bal_cX_failed[CR_POWER2_ALIGNED])); + + /* CR_GOAL_LEN_FAST stats */ + seq_puts(seq, "\tcr_goal_fast_stats:\n"); + seq_printf(seq, "\t\thits: %llu\n", + atomic64_read(&sbi->s_bal_cX_hits[CR_GOAL_LEN_FAST])); + seq_printf(seq, "\t\tgroups_considered: %llu\n", + atomic64_read( + &sbi->s_bal_cX_groups_considered[CR_GOAL_LEN_FAST])); + seq_printf(seq, "\t\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_cX_ex_scanned[CR_GOAL_LEN_FAST])); + seq_printf(seq, "\t\tuseless_loops: %llu\n", + atomic64_read(&sbi->s_bal_cX_failed[CR_GOAL_LEN_FAST])); + + /* CR_BEST_AVAIL_LEN stats */ + seq_puts(seq, "\tcr_best_avail_stats:\n"); + seq_printf(seq, "\t\thits: %llu\n", + atomic64_read(&sbi->s_bal_cX_hits[CR_BEST_AVAIL_LEN])); + seq_printf( + seq, "\t\tgroups_considered: %llu\n", + atomic64_read( + &sbi->s_bal_cX_groups_considered[CR_BEST_AVAIL_LEN])); + seq_printf(seq, "\t\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_cX_ex_scanned[CR_BEST_AVAIL_LEN])); + seq_printf(seq, "\t\tuseless_loops: %llu\n", + atomic64_read(&sbi->s_bal_cX_failed[CR_BEST_AVAIL_LEN])); + + /* CR_GOAL_LEN_SLOW stats */ + seq_puts(seq, "\tcr_goal_slow_stats:\n"); + seq_printf(seq, "\t\thits: %llu\n", + atomic64_read(&sbi->s_bal_cX_hits[CR_GOAL_LEN_SLOW])); + seq_printf(seq, "\t\tgroups_considered: %llu\n", + atomic64_read( + &sbi->s_bal_cX_groups_considered[CR_GOAL_LEN_SLOW])); + seq_printf(seq, "\t\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_cX_ex_scanned[CR_GOAL_LEN_SLOW])); + seq_printf(seq, "\t\tuseless_loops: %llu\n", + atomic64_read(&sbi->s_bal_cX_failed[CR_GOAL_LEN_SLOW])); + + /* CR_ANY_FREE stats */ + seq_puts(seq, "\tcr_any_free_stats:\n"); + seq_printf(seq, "\t\thits: %llu\n", + atomic64_read(&sbi->s_bal_cX_hits[CR_ANY_FREE])); + seq_printf( + seq, "\t\tgroups_considered: %llu\n", + atomic64_read(&sbi->s_bal_cX_groups_considered[CR_ANY_FREE])); + seq_printf(seq, "\t\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_cX_ex_scanned[CR_ANY_FREE])); + seq_printf(seq, "\t\tuseless_loops: %llu\n", + atomic64_read(&sbi->s_bal_cX_failed[CR_ANY_FREE])); + + /* Aggregates */ + seq_printf(seq, "\textents_scanned: %u\n", + atomic_read(&sbi->s_bal_ex_scanned)); + seq_printf(seq, "\t\tgoal_hits: %u\n", atomic_read(&sbi->s_bal_goals)); + seq_printf(seq, "\t\tstream_goal_hits: %u\n", + atomic_read(&sbi->s_bal_stream_goals)); + seq_printf(seq, "\t\tlen_goal_hits: %u\n", + atomic_read(&sbi->s_bal_len_goals)); + seq_printf(seq, "\t\t2^n_hits: %u\n", atomic_read(&sbi->s_bal_2orders)); + seq_printf(seq, "\t\tbreaks: %u\n", atomic_read(&sbi->s_bal_breaks)); + seq_printf(seq, "\t\tlost: %u\n", atomic_read(&sbi->s_mb_lost_chunks)); + seq_printf(seq, "\tbuddies_generated: %u/%u\n", + atomic_read(&sbi->s_mb_buddies_generated), + ext4_get_groups_count(sb)); + seq_printf(seq, "\tbuddies_time_used: %llu\n", + atomic64_read(&sbi->s_mb_generation_time)); + seq_printf(seq, "\tpreallocated: %u\n", + atomic_read(&sbi->s_mb_preallocated)); + seq_printf(seq, "\tdiscarded: %u\n", atomic_read(&sbi->s_mb_discarded)); + return 0; +} + +static void *ext4_mb_seq_structs_summary_start(struct seq_file *seq, loff_t *pos) +{ + struct super_block *sb = pde_data(file_inode(seq->file)); + unsigned long position; + + if (*pos < 0 || *pos >= 2*MB_NUM_ORDERS(sb)) + return NULL; + position = *pos + 1; + return (void *) ((unsigned long) position); +} + +static void *ext4_mb_seq_structs_summary_next(struct seq_file *seq, void *v, loff_t *pos) +{ + struct super_block *sb = pde_data(file_inode(seq->file)); + unsigned long position; + + ++*pos; + if (*pos < 0 || *pos >= 2*MB_NUM_ORDERS(sb)) + return NULL; + position = *pos + 1; + return (void *) ((unsigned long) position); +} + +static int ext4_mb_seq_structs_summary_show(struct seq_file *seq, void *v) +{ + struct super_block *sb = pde_data(file_inode(seq->file)); + struct ext4_sb_info *sbi = EXT4_SB(sb); + unsigned long position = ((unsigned long) v); + struct ext4_group_info *grp; + unsigned int count; + unsigned long idx; + + position--; + if (position >= MB_NUM_ORDERS(sb)) { + position -= MB_NUM_ORDERS(sb); + if (position == 0) + seq_puts(seq, "avg_fragment_size_lists:\n"); + + count = 0; + xa_for_each(&sbi->s_mb_avg_fragment_size[position], idx, grp) + count++; + seq_printf(seq, "\tlist_order_%u_groups: %u\n", + (unsigned int)position, count); + return 0; + } + + if (position == 0) { + seq_printf(seq, "optimize_scan: %d\n", + test_opt2(sb, MB_OPTIMIZE_SCAN) ? 1 : 0); + seq_puts(seq, "max_free_order_lists:\n"); + } + count = 0; + xa_for_each(&sbi->s_mb_largest_free_orders[position], idx, grp) + count++; + seq_printf(seq, "\tlist_order_%u_groups: %u\n", + (unsigned int)position, count); + + return 0; +} + +static void ext4_mb_seq_structs_summary_stop(struct seq_file *seq, void *v) +{ +} + +const struct seq_operations ext4_mb_seq_structs_summary_ops = { + .start = ext4_mb_seq_structs_summary_start, + .next = ext4_mb_seq_structs_summary_next, + .stop = ext4_mb_seq_structs_summary_stop, + .show = ext4_mb_seq_structs_summary_show, +}; + static struct kmem_cache *get_groupinfo_cache(int blocksize_bits) { int cache_index = blocksize_bits - EXT4_MIN_BLOCK_LOG_SIZE; @@ -2495,7 +3402,7 @@ int ext4_mb_alloc_groupinfo(struct super_block *sb, ext4_group_t ngroups) sbi->s_group_info_size = size / sizeof(*sbi->s_group_info); if (old_groupinfo) ext4_kvfree_array_rcu(old_groupinfo); - ext4_debug("allocated s_groupinfo array for %d meta_bg's\n", + ext4_debug("allocated s_groupinfo array for %d meta_bg's\n", sbi->s_group_info_size); return 0; } @@ -2523,7 +3430,7 @@ int ext4_mb_add_groupinfo(struct super_block *sb, ext4_group_t group, if (meta_group_info == NULL) { ext4_msg(sb, KERN_ERR, "can't allocate mem " "for a buddy group"); - goto exit_meta_group_info; + return -ENOMEM; } rcu_read_lock(); rcu_dereference(sbi->s_group_info)[idx] = meta_group_info; @@ -2558,6 +3465,8 @@ int ext4_mb_add_groupinfo(struct super_block *sb, ext4_group_t group, init_rwsem(&meta_group_info[i]->alloc_sem); meta_group_info[i]->bb_free_root = RB_ROOT; meta_group_info[i]->bb_largest_free_order = -1; /* uninit */ + meta_group_info[i]->bb_avg_fragment_size_order = -1; /* uninit */ + meta_group_info[i]->bb_group = group; mb_group_bb_bitmap_alloc(sb, meta_group_info[i], group); return 0; @@ -2573,7 +3482,6 @@ exit_group_info: group_info[idx] = NULL; rcu_read_unlock(); } -exit_meta_group_info: return -ENOMEM; } /* ext4_mb_add_groupinfo */ @@ -2602,6 +3510,8 @@ static int ext4_mb_init_backend(struct super_block *sb) * this will avoid confusion if it ever shows up during debugging. */ sbi->s_buddy_cache->i_ino = EXT4_BAD_INO; EXT4_I(sbi->s_buddy_cache)->i_disksize = 0; + ext4_set_inode_mapping_order(sbi->s_buddy_cache); + for (i = 0; i < ngroups; i++) { cond_resched(); desc = ext4_get_group_desc(sb, i, NULL); @@ -2613,12 +3523,45 @@ static int ext4_mb_init_backend(struct super_block *sb) goto err_freebuddy; } + if (ext4_has_feature_flex_bg(sb)) { + /* a single flex group is supposed to be read by a single IO. + * 2 ^ s_log_groups_per_flex != UINT_MAX as s_mb_prefetch is + * unsigned integer, so the maximum shift is 32. + */ + if (sbi->s_es->s_log_groups_per_flex >= 32) { + ext4_msg(sb, KERN_ERR, "too many log groups per flexible block group"); + goto err_freebuddy; + } + sbi->s_mb_prefetch = min_t(uint, 1 << sbi->s_es->s_log_groups_per_flex, + BLK_MAX_SEGMENT_SIZE >> (sb->s_blocksize_bits - 9)); + sbi->s_mb_prefetch *= 8; /* 8 prefetch IOs in flight at most */ + } else { + sbi->s_mb_prefetch = 32; + } + if (sbi->s_mb_prefetch > ext4_get_groups_count(sb)) + sbi->s_mb_prefetch = ext4_get_groups_count(sb); + /* + * now many real IOs to prefetch within a single allocation at + * CR_POWER2_ALIGNED. Given CR_POWER2_ALIGNED is an CPU-related + * optimization we shouldn't try to load too many groups, at some point + * we should start to use what we've got in memory. + * with an average random access time 5ms, it'd take a second to get + * 200 groups (* N with flex_bg), so let's make this limit 4 + */ + sbi->s_mb_prefetch_limit = sbi->s_mb_prefetch * 4; + if (sbi->s_mb_prefetch_limit > ext4_get_groups_count(sb)) + sbi->s_mb_prefetch_limit = ext4_get_groups_count(sb); + return 0; err_freebuddy: cachep = get_groupinfo_cache(sb->s_blocksize_bits); - while (i-- > 0) - kmem_cache_free(cachep, ext4_get_group_info(sb, i)); + while (i-- > 0) { + struct ext4_group_info *grp = ext4_get_group_info(sb, i); + + if (grp) + kmem_cache_free(cachep, grp); + } i = sbi->s_group_info_size; rcu_read_lock(); group_info = rcu_dereference(sbi->s_group_info); @@ -2682,6 +3625,80 @@ static int ext4_groupinfo_create_slab(size_t size) return 0; } +static void ext4_discard_work(struct work_struct *work) +{ + struct ext4_sb_info *sbi = container_of(work, + struct ext4_sb_info, s_discard_work); + struct super_block *sb = sbi->s_sb; + struct ext4_free_data *fd, *nfd; + struct ext4_buddy e4b; + LIST_HEAD(discard_list); + ext4_group_t grp, load_grp; + int err = 0; + + spin_lock(&sbi->s_md_lock); + list_splice_init(&sbi->s_discard_list, &discard_list); + spin_unlock(&sbi->s_md_lock); + + load_grp = UINT_MAX; + list_for_each_entry_safe(fd, nfd, &discard_list, efd_list) { + /* + * If filesystem is umounting or no memory or suffering + * from no space, give up the discard + */ + if ((sb->s_flags & SB_ACTIVE) && !err && + !atomic_read(&sbi->s_retry_alloc_pending)) { + grp = fd->efd_group; + if (grp != load_grp) { + if (load_grp != UINT_MAX) + ext4_mb_unload_buddy(&e4b); + + err = ext4_mb_load_buddy(sb, grp, &e4b); + if (err) { + kmem_cache_free(ext4_free_data_cachep, fd); + load_grp = UINT_MAX; + continue; + } else { + load_grp = grp; + } + } + + ext4_lock_group(sb, grp); + ext4_try_to_trim_range(sb, &e4b, fd->efd_start_cluster, + fd->efd_start_cluster + fd->efd_count - 1, 1); + ext4_unlock_group(sb, grp); + } + kmem_cache_free(ext4_free_data_cachep, fd); + } + + if (load_grp != UINT_MAX) + ext4_mb_unload_buddy(&e4b); +} + +static inline void ext4_mb_avg_fragment_size_destroy(struct ext4_sb_info *sbi) +{ + if (!sbi->s_mb_avg_fragment_size) + return; + + for (int i = 0; i < MB_NUM_ORDERS(sbi->s_sb); i++) + xa_destroy(&sbi->s_mb_avg_fragment_size[i]); + + kfree(sbi->s_mb_avg_fragment_size); + sbi->s_mb_avg_fragment_size = NULL; +} + +static inline void ext4_mb_largest_free_orders_destroy(struct ext4_sb_info *sbi) +{ + if (!sbi->s_mb_largest_free_orders) + return; + + for (int i = 0; i < MB_NUM_ORDERS(sbi->s_sb); i++) + xa_destroy(&sbi->s_mb_largest_free_orders[i]); + + kfree(sbi->s_mb_largest_free_orders); + sbi->s_mb_largest_free_orders = NULL; +} + int ext4_mb_init(struct super_block *sb) { struct ext4_sb_info *sbi = EXT4_SB(sb); @@ -2690,7 +3707,7 @@ int ext4_mb_init(struct super_block *sb) unsigned max; int ret; - i = (sb->s_blocksize_bits + 2) * sizeof(*sbi->s_mb_offsets); + i = MB_NUM_ORDERS(sb) * sizeof(*sbi->s_mb_offsets); sbi->s_mb_offsets = kmalloc(i, GFP_KERNEL); if (sbi->s_mb_offsets == NULL) { @@ -2698,7 +3715,7 @@ int ext4_mb_init(struct super_block *sb) goto out; } - i = (sb->s_blocksize_bits + 2) * sizeof(*sbi->s_mb_maxs); + i = MB_NUM_ORDERS(sb) * sizeof(*sbi->s_mb_maxs); sbi->s_mb_maxs = kmalloc(i, GFP_KERNEL); if (sbi->s_mb_maxs == NULL) { ret = -ENOMEM; @@ -2724,18 +3741,43 @@ int ext4_mb_init(struct super_block *sb) offset_incr = offset_incr >> 1; max = max >> 1; i++; - } while (i <= sb->s_blocksize_bits + 1); + } while (i < MB_NUM_ORDERS(sb)); + + sbi->s_mb_avg_fragment_size = + kmalloc_array(MB_NUM_ORDERS(sb), sizeof(struct xarray), + GFP_KERNEL); + if (!sbi->s_mb_avg_fragment_size) { + ret = -ENOMEM; + goto out; + } + for (i = 0; i < MB_NUM_ORDERS(sb); i++) + xa_init(&sbi->s_mb_avg_fragment_size[i]); + + sbi->s_mb_largest_free_orders = + kmalloc_array(MB_NUM_ORDERS(sb), sizeof(struct xarray), + GFP_KERNEL); + if (!sbi->s_mb_largest_free_orders) { + ret = -ENOMEM; + goto out; + } + for (i = 0; i < MB_NUM_ORDERS(sb); i++) + xa_init(&sbi->s_mb_largest_free_orders[i]); spin_lock_init(&sbi->s_md_lock); - spin_lock_init(&sbi->s_bal_lock); - sbi->s_mb_free_pending = 0; - INIT_LIST_HEAD(&sbi->s_freed_data_list); + atomic_set(&sbi->s_mb_free_pending, 0); + INIT_LIST_HEAD(&sbi->s_freed_data_list[0]); + INIT_LIST_HEAD(&sbi->s_freed_data_list[1]); + INIT_LIST_HEAD(&sbi->s_discard_list); + INIT_WORK(&sbi->s_discard_work, ext4_discard_work); + atomic_set(&sbi->s_retry_alloc_pending, 0); sbi->s_mb_max_to_scan = MB_DEFAULT_MAX_TO_SCAN; sbi->s_mb_min_to_scan = MB_DEFAULT_MIN_TO_SCAN; sbi->s_mb_stats = MB_DEFAULT_STATS; sbi->s_mb_stream_request = MB_DEFAULT_STREAM_THRESHOLD; sbi->s_mb_order2_reqs = MB_DEFAULT_ORDER2_REQS; + sbi->s_mb_best_avail_max_trim_order = MB_DEFAULT_BEST_AVAIL_TRIM_ORDER; + /* * The default group preallocation is 512, which for 4k block * sizes translates to 2 megabytes. However for bigalloc file @@ -2760,13 +3802,22 @@ int ext4_mb_init(struct super_block *sb) */ if (sbi->s_stripe > 1) { sbi->s_mb_group_prealloc = roundup( - sbi->s_mb_group_prealloc, sbi->s_stripe); + sbi->s_mb_group_prealloc, EXT4_NUM_B2C(sbi, sbi->s_stripe)); + } + + sbi->s_mb_nr_global_goals = umin(num_possible_cpus(), + DIV_ROUND_UP(sbi->s_groups_count, 4)); + sbi->s_mb_last_groups = kcalloc(sbi->s_mb_nr_global_goals, + sizeof(ext4_group_t), GFP_KERNEL); + if (sbi->s_mb_last_groups == NULL) { + ret = -ENOMEM; + goto out; } sbi->s_locality_groups = alloc_percpu(struct ext4_locality_group); if (sbi->s_locality_groups == NULL) { ret = -ENOMEM; - goto out; + goto out_free_last_groups; } for_each_possible_cpu(i) { struct ext4_locality_group *lg; @@ -2777,6 +3828,10 @@ int ext4_mb_init(struct super_block *sb) spin_lock_init(&lg->lg_prealloc_lock); } + if (bdev_nonrot(sb->s_bdev)) + sbi->s_mb_max_linear_groups = 0; + else + sbi->s_mb_max_linear_groups = MB_DEFAULT_LINEAR_LIMIT; /* init file for buddy data */ ret = ext4_mb_init_backend(sb); if (ret != 0) @@ -2787,7 +3842,12 @@ int ext4_mb_init(struct super_block *sb) out_free_locality_groups: free_percpu(sbi->s_locality_groups); sbi->s_locality_groups = NULL; +out_free_last_groups: + kfree(sbi->s_mb_last_groups); + sbi->s_mb_last_groups = NULL; out: + ext4_mb_avg_fragment_size_destroy(sbi); + ext4_mb_largest_free_orders_destroy(sbi); kfree(sbi->s_mb_offsets); sbi->s_mb_offsets = NULL; kfree(sbi->s_mb_maxs); @@ -2811,7 +3871,7 @@ static int ext4_mb_cleanup_pa(struct ext4_group_info *grp) return count; } -int ext4_mb_release(struct super_block *sb) +void ext4_mb_release(struct super_block *sb) { ext4_group_t ngroups = ext4_get_groups_count(sb); ext4_group_t i; @@ -2821,10 +3881,20 @@ int ext4_mb_release(struct super_block *sb) struct kmem_cache *cachep = get_groupinfo_cache(sb->s_blocksize_bits); int count; + if (test_opt(sb, DISCARD)) { + /* + * wait the discard work to drain all of ext4_free_data + */ + flush_work(&sbi->s_discard_work); + WARN_ON_ONCE(!list_empty(&sbi->s_discard_list)); + } + if (sbi->s_group_info) { for (i = 0; i < ngroups; i++) { cond_resched(); grinfo = ext4_get_group_info(sb, i); + if (!grinfo) + continue; mb_group_bb_bitmap_free(grinfo); ext4_lock_group(sb, i); count = ext4_mb_cleanup_pa(grinfo); @@ -2844,6 +3914,8 @@ int ext4_mb_release(struct super_block *sb) kvfree(group_info); rcu_read_unlock(); } + ext4_mb_avg_fragment_size_destroy(sbi); + ext4_mb_largest_free_orders_destroy(sbi); kfree(sbi->s_mb_offsets); kfree(sbi->s_mb_maxs); iput(sbi->s_buddy_cache); @@ -2854,17 +3926,18 @@ int ext4_mb_release(struct super_block *sb) atomic_read(&sbi->s_bal_reqs), atomic_read(&sbi->s_bal_success)); ext4_msg(sb, KERN_INFO, - "mballoc: %u extents scanned, %u goal hits, " + "mballoc: %u extents scanned, %u groups scanned, %u goal hits, " "%u 2^N hits, %u breaks, %u lost", atomic_read(&sbi->s_bal_ex_scanned), + atomic_read(&sbi->s_bal_groups_scanned), atomic_read(&sbi->s_bal_goals), atomic_read(&sbi->s_bal_2orders), atomic_read(&sbi->s_bal_breaks), atomic_read(&sbi->s_mb_lost_chunks)); ext4_msg(sb, KERN_INFO, - "mballoc: %lu generated and it took %Lu", - sbi->s_mb_buddies_generated, - sbi->s_mb_generation_time); + "mballoc: %u generated and it took %llu", + atomic_read(&sbi->s_mb_buddies_generated), + atomic64_read(&sbi->s_mb_generation_time)); ext4_msg(sb, KERN_INFO, "mballoc: %u preallocated, %u discarded", atomic_read(&sbi->s_mb_preallocated), @@ -2872,13 +3945,11 @@ int ext4_mb_release(struct super_block *sb) } free_percpu(sbi->s_locality_groups); - - return 0; + kfree(sbi->s_mb_last_groups); } static inline int ext4_issue_discard(struct super_block *sb, - ext4_group_t block_group, ext4_grpblk_t cluster, int count, - struct bio **biop) + ext4_group_t block_group, ext4_grpblk_t cluster, int count) { ext4_fsblk_t discard_block; @@ -2887,13 +3958,8 @@ static inline int ext4_issue_discard(struct super_block *sb, count = EXT4_C2B(EXT4_SB(sb), count); trace_ext4_discard_blocks(sb, (unsigned long long) discard_block, count); - if (biop) { - return __blkdev_issue_discard(sb->s_bdev, - (sector_t)discard_block << (sb->s_blocksize_bits - 9), - (sector_t)count << (sb->s_blocksize_bits - 9), - GFP_NOFS, 0, biop); - } else - return sb_issue_discard(sb, discard_block, count, GFP_NOFS, 0); + + return sb_issue_discard(sb, discard_block, count, GFP_NOFS, 0); } static void ext4_free_data_in_buddy(struct super_block *sb, @@ -2901,7 +3967,7 @@ static void ext4_free_data_in_buddy(struct super_block *sb, { struct ext4_buddy e4b; struct ext4_group_info *db; - int err, count = 0, count2 = 0; + int err, count = 0; mb_debug(sb, "gonna free %u blocks in group %u (0x%p):", entry->efd_count, entry->efd_group, entry); @@ -2910,14 +3976,10 @@ static void ext4_free_data_in_buddy(struct super_block *sb, /* we expect to find existing buddy because it's pinned */ BUG_ON(err != 0); - spin_lock(&EXT4_SB(sb)->s_md_lock); - EXT4_SB(sb)->s_mb_free_pending -= entry->efd_count; - spin_unlock(&EXT4_SB(sb)->s_md_lock); - + atomic_sub(entry->efd_count, &EXT4_SB(sb)->s_mb_free_pending); db = e4b.bd_info; /* there are blocks to put in buddy to make them really free */ count += entry->efd_count; - count2++; ext4_lock_group(sb, entry->efd_group); /* Take it out of per group rb tree */ rb_erase(&entry->efd_node, &(db->bb_free_root)); @@ -2926,25 +3988,20 @@ static void ext4_free_data_in_buddy(struct super_block *sb, /* * Clear the trimmed flag for the group so that the next * ext4_trim_fs can trim it. - * If the volume is mounted with -o discard, online discard - * is supported and the free blocks will be trimmed online. */ - if (!test_opt(sb, DISCARD)) - EXT4_MB_GRP_CLEAR_TRIMMED(db); + EXT4_MB_GRP_CLEAR_TRIMMED(db); if (!db->bb_free_root.rb_node) { /* No more items in the per group rb tree * balance refcounts from ext4_mb_free_metadata() */ - put_page(e4b.bd_buddy_page); - put_page(e4b.bd_bitmap_page); + folio_put(e4b.bd_buddy_folio); + folio_put(e4b.bd_bitmap_folio); } ext4_unlock_group(sb, entry->efd_group); - kmem_cache_free(ext4_free_data_cachep, entry); ext4_mb_unload_buddy(&e4b); - mb_debug(sb, "freed %d blocks in %d structures\n", count, - count2); + mb_debug(sb, "freed %d blocks in 1 structures\n", count); } /* @@ -2955,48 +4012,26 @@ void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid) { struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_free_data *entry, *tmp; - struct bio *discard_bio = NULL; - struct list_head freed_data_list; - struct list_head *cut_pos = NULL; - int err; + LIST_HEAD(freed_data_list); + struct list_head *s_freed_head = &sbi->s_freed_data_list[commit_tid & 1]; + bool wake; - INIT_LIST_HEAD(&freed_data_list); + list_replace_init(s_freed_head, &freed_data_list); - spin_lock(&sbi->s_md_lock); - list_for_each_entry(entry, &sbi->s_freed_data_list, efd_list) { - if (entry->efd_tid != commit_tid) - break; - cut_pos = &entry->efd_list; - } - if (cut_pos) - list_cut_position(&freed_data_list, &sbi->s_freed_data_list, - cut_pos); - spin_unlock(&sbi->s_md_lock); + list_for_each_entry(entry, &freed_data_list, efd_list) + ext4_free_data_in_buddy(sb, entry); if (test_opt(sb, DISCARD)) { - list_for_each_entry(entry, &freed_data_list, efd_list) { - err = ext4_issue_discard(sb, entry->efd_group, - entry->efd_start_cluster, - entry->efd_count, - &discard_bio); - if (err && err != -EOPNOTSUPP) { - ext4_msg(sb, KERN_WARNING, "discard request in" - " group:%d block:%d count:%d failed" - " with %d", entry->efd_group, - entry->efd_start_cluster, - entry->efd_count, err); - } else if (err == -EOPNOTSUPP) - break; - } - - if (discard_bio) { - submit_bio_wait(discard_bio); - bio_put(discard_bio); - } + spin_lock(&sbi->s_md_lock); + wake = list_empty(&sbi->s_discard_list); + list_splice_tail(&freed_data_list, &sbi->s_discard_list); + spin_unlock(&sbi->s_md_lock); + if (wake) + queue_work(system_dfl_wq, &sbi->s_discard_work); + } else { + list_for_each_entry_safe(entry, tmp, &freed_data_list, efd_list) + kmem_cache_free(ext4_free_data_cachep, entry); } - - list_for_each_entry_safe(entry, tmp, &freed_data_list, efd_list) - ext4_free_data_in_buddy(sb, entry); } int __init ext4_init_mballoc(void) @@ -3039,6 +4074,111 @@ void ext4_exit_mballoc(void) ext4_groupinfo_destroy_slabs(); } +#define EXT4_MB_BITMAP_MARKED_CHECK 0x0001 +#define EXT4_MB_SYNC_UPDATE 0x0002 +static int +ext4_mb_mark_context(handle_t *handle, struct super_block *sb, bool state, + ext4_group_t group, ext4_grpblk_t blkoff, + ext4_grpblk_t len, int flags, ext4_grpblk_t *ret_changed) +{ + struct ext4_sb_info *sbi = EXT4_SB(sb); + struct buffer_head *bitmap_bh = NULL; + struct ext4_group_desc *gdp; + struct buffer_head *gdp_bh; + int err; + unsigned int i, already, changed = len; + + KUNIT_STATIC_STUB_REDIRECT(ext4_mb_mark_context, + handle, sb, state, group, blkoff, len, + flags, ret_changed); + + if (ret_changed) + *ret_changed = 0; + bitmap_bh = ext4_read_block_bitmap(sb, group); + if (IS_ERR(bitmap_bh)) + return PTR_ERR(bitmap_bh); + + if (handle) { + BUFFER_TRACE(bitmap_bh, "getting write access"); + err = ext4_journal_get_write_access(handle, sb, bitmap_bh, + EXT4_JTR_NONE); + if (err) + goto out_err; + } + + err = -EIO; + gdp = ext4_get_group_desc(sb, group, &gdp_bh); + if (!gdp) + goto out_err; + + if (handle) { + BUFFER_TRACE(gdp_bh, "get_write_access"); + err = ext4_journal_get_write_access(handle, sb, gdp_bh, + EXT4_JTR_NONE); + if (err) + goto out_err; + } + + ext4_lock_group(sb, group); + if (ext4_has_group_desc_csum(sb) && + (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))) { + gdp->bg_flags &= cpu_to_le16(~EXT4_BG_BLOCK_UNINIT); + ext4_free_group_clusters_set(sb, gdp, + ext4_free_clusters_after_init(sb, group, gdp)); + } + + if (flags & EXT4_MB_BITMAP_MARKED_CHECK) { + already = 0; + for (i = 0; i < len; i++) + if (mb_test_bit(blkoff + i, bitmap_bh->b_data) == + state) + already++; + changed = len - already; + } + + if (state) { + mb_set_bits(bitmap_bh->b_data, blkoff, len); + ext4_free_group_clusters_set(sb, gdp, + ext4_free_group_clusters(sb, gdp) - changed); + } else { + mb_clear_bits(bitmap_bh->b_data, blkoff, len); + ext4_free_group_clusters_set(sb, gdp, + ext4_free_group_clusters(sb, gdp) + changed); + } + + ext4_block_bitmap_csum_set(sb, gdp, bitmap_bh); + ext4_group_desc_csum_set(sb, group, gdp); + ext4_unlock_group(sb, group); + if (ret_changed) + *ret_changed = changed; + + if (sbi->s_log_groups_per_flex) { + ext4_group_t flex_group = ext4_flex_group(sbi, group); + struct flex_groups *fg = sbi_array_rcu_deref(sbi, + s_flex_groups, flex_group); + + if (state) + atomic64_sub(changed, &fg->free_clusters); + else + atomic64_add(changed, &fg->free_clusters); + } + + err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); + if (err) + goto out_err; + err = ext4_handle_dirty_metadata(handle, NULL, gdp_bh); + if (err) + goto out_err; + + if (flags & EXT4_MB_SYNC_UPDATE) { + sync_dirty_buffer(bitmap_bh); + sync_dirty_buffer(gdp_bh); + } + +out_err: + brelse(bitmap_bh); + return err; +} /* * Check quota and mark chosen space (ac->ac_b_ex) non-free in bitmaps @@ -3048,13 +4188,13 @@ static noinline_for_stack int ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac, handle_t *handle, unsigned int reserv_clstrs) { - struct buffer_head *bitmap_bh = NULL; struct ext4_group_desc *gdp; - struct buffer_head *gdp_bh; struct ext4_sb_info *sbi; struct super_block *sb; ext4_fsblk_t block; int err, len; + int flags = 0; + ext4_grpblk_t changed; BUG_ON(ac->ac_status != AC_STATUS_FOUND); BUG_ON(ac->ac_b_ex.fe_len <= 0); @@ -3062,76 +4202,44 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac, sb = ac->ac_sb; sbi = EXT4_SB(sb); - bitmap_bh = ext4_read_block_bitmap(sb, ac->ac_b_ex.fe_group); - if (IS_ERR(bitmap_bh)) { - err = PTR_ERR(bitmap_bh); - bitmap_bh = NULL; - goto out_err; - } - - BUFFER_TRACE(bitmap_bh, "getting write access"); - err = ext4_journal_get_write_access(handle, bitmap_bh); - if (err) - goto out_err; - - err = -EIO; - gdp = ext4_get_group_desc(sb, ac->ac_b_ex.fe_group, &gdp_bh); + gdp = ext4_get_group_desc(sb, ac->ac_b_ex.fe_group, NULL); if (!gdp) - goto out_err; - + return -EIO; ext4_debug("using block group %u(%d)\n", ac->ac_b_ex.fe_group, ext4_free_group_clusters(sb, gdp)); - BUFFER_TRACE(gdp_bh, "get_write_access"); - err = ext4_journal_get_write_access(handle, gdp_bh); - if (err) - goto out_err; - block = ext4_grp_offs_to_block(sb, &ac->ac_b_ex); - len = EXT4_C2B(sbi, ac->ac_b_ex.fe_len); - if (!ext4_data_block_valid(sbi, block, len)) { + if (!ext4_inode_block_valid(ac->ac_inode, block, len)) { ext4_error(sb, "Allocating blocks %llu-%llu which overlap " "fs metadata", block, block+len); /* File system mounted not to panic on error * Fix the bitmap and return EFSCORRUPTED * We leak some of the blocks here. */ - ext4_lock_group(sb, ac->ac_b_ex.fe_group); - ext4_set_bits(bitmap_bh->b_data, ac->ac_b_ex.fe_start, - ac->ac_b_ex.fe_len); - ext4_unlock_group(sb, ac->ac_b_ex.fe_group); - err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); + err = ext4_mb_mark_context(handle, sb, true, + ac->ac_b_ex.fe_group, + ac->ac_b_ex.fe_start, + ac->ac_b_ex.fe_len, + 0, NULL); if (!err) err = -EFSCORRUPTED; - goto out_err; + return err; } - ext4_lock_group(sb, ac->ac_b_ex.fe_group); #ifdef AGGRESSIVE_CHECK - { - int i; - for (i = 0; i < ac->ac_b_ex.fe_len; i++) { - BUG_ON(mb_test_bit(ac->ac_b_ex.fe_start + i, - bitmap_bh->b_data)); - } - } + flags |= EXT4_MB_BITMAP_MARKED_CHECK; #endif - ext4_set_bits(bitmap_bh->b_data, ac->ac_b_ex.fe_start, - ac->ac_b_ex.fe_len); - if (ext4_has_group_desc_csum(sb) && - (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))) { - gdp->bg_flags &= cpu_to_le16(~EXT4_BG_BLOCK_UNINIT); - ext4_free_group_clusters_set(sb, gdp, - ext4_free_clusters_after_init(sb, - ac->ac_b_ex.fe_group, gdp)); - } - len = ext4_free_group_clusters(sb, gdp) - ac->ac_b_ex.fe_len; - ext4_free_group_clusters_set(sb, gdp, len); - ext4_block_bitmap_csum_set(sb, ac->ac_b_ex.fe_group, gdp, bitmap_bh); - ext4_group_desc_csum_set(sb, ac->ac_b_ex.fe_group, gdp); + err = ext4_mb_mark_context(handle, sb, true, ac->ac_b_ex.fe_group, + ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len, + flags, &changed); - ext4_unlock_group(sb, ac->ac_b_ex.fe_group); + if (err && changed == 0) + return err; + +#ifdef AGGRESSIVE_CHECK + BUG_ON(changed != ac->ac_b_ex.fe_len); +#endif percpu_counter_sub(&sbi->s_freeclusters_counter, ac->ac_b_ex.fe_len); /* * Now reduce the dirty block count also. Should not go negative @@ -3141,22 +4249,56 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac, percpu_counter_sub(&sbi->s_dirtyclusters_counter, reserv_clstrs); - if (sbi->s_log_groups_per_flex) { - ext4_group_t flex_group = ext4_flex_group(sbi, - ac->ac_b_ex.fe_group); - atomic64_sub(ac->ac_b_ex.fe_len, - &sbi_array_rcu_deref(sbi, s_flex_groups, - flex_group)->free_clusters); - } + return err; +} - err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); - if (err) - goto out_err; - err = ext4_handle_dirty_metadata(handle, NULL, gdp_bh); +/* + * Idempotent helper for Ext4 fast commit replay path to set the state of + * blocks in bitmaps and update counters. + */ +void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, + int len, bool state) +{ + struct ext4_sb_info *sbi = EXT4_SB(sb); + ext4_group_t group; + ext4_grpblk_t blkoff; + int err = 0; + unsigned int clen, thisgrp_len; -out_err: - brelse(bitmap_bh); - return err; + while (len > 0) { + ext4_get_group_no_and_offset(sb, block, &group, &blkoff); + + /* + * Check to see if we are freeing blocks across a group + * boundary. + * In case of flex_bg, this can happen that (block, len) may + * span across more than one group. In that case we need to + * get the corresponding group metadata to work with. + * For this we have goto again loop. + */ + thisgrp_len = min_t(unsigned int, (unsigned int)len, + EXT4_BLOCKS_PER_GROUP(sb) - EXT4_C2B(sbi, blkoff)); + clen = EXT4_NUM_B2C(sbi, thisgrp_len); + + if (!ext4_sb_block_valid(sb, NULL, block, thisgrp_len)) { + ext4_error(sb, "Marking blocks in system zone - " + "Block = %llu, len = %u", + block, thisgrp_len); + break; + } + + err = ext4_mb_mark_context(NULL, sb, state, + group, blkoff, clen, + EXT4_MB_BITMAP_MARKED_CHECK | + EXT4_MB_SYNC_UPDATE, + NULL); + if (err) + break; + + block += thisgrp_len; + len -= thisgrp_len; + BUG_ON(len < 0); + } } /* @@ -3179,6 +4321,197 @@ static void ext4_mb_normalize_group_request(struct ext4_allocation_context *ac) } /* + * This function returns the next element to look at during inode + * PA rbtree walk. We assume that we have held the inode PA rbtree lock + * (ei->i_prealloc_lock) + * + * new_start The start of the range we want to compare + * cur_start The existing start that we are comparing against + * node The node of the rb_tree + */ +static inline struct rb_node* +ext4_mb_pa_rb_next_iter(ext4_lblk_t new_start, ext4_lblk_t cur_start, struct rb_node *node) +{ + if (new_start < cur_start) + return node->rb_left; + else + return node->rb_right; +} + +static inline void +ext4_mb_pa_assert_overlap(struct ext4_allocation_context *ac, + ext4_lblk_t start, loff_t end) +{ + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); + struct ext4_prealloc_space *tmp_pa; + ext4_lblk_t tmp_pa_start; + loff_t tmp_pa_end; + struct rb_node *iter; + + read_lock(&ei->i_prealloc_lock); + for (iter = ei->i_prealloc_node.rb_node; iter; + iter = ext4_mb_pa_rb_next_iter(start, tmp_pa_start, iter)) { + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + tmp_pa_start = tmp_pa->pa_lstart; + tmp_pa_end = pa_logical_end(sbi, tmp_pa); + + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) + BUG_ON(!(start >= tmp_pa_end || end <= tmp_pa_start)); + spin_unlock(&tmp_pa->pa_lock); + } + read_unlock(&ei->i_prealloc_lock); +} + +/* + * Given an allocation context "ac" and a range "start", "end", check + * and adjust boundaries if the range overlaps with any of the existing + * preallocatoins stored in the corresponding inode of the allocation context. + * + * Parameters: + * ac allocation context + * start start of the new range + * end end of the new range + */ +static inline void +ext4_mb_pa_adjust_overlap(struct ext4_allocation_context *ac, + ext4_lblk_t *start, loff_t *end) +{ + struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + struct ext4_prealloc_space *tmp_pa = NULL, *left_pa = NULL, *right_pa = NULL; + struct rb_node *iter; + ext4_lblk_t new_start, tmp_pa_start, right_pa_start = -1; + loff_t new_end, tmp_pa_end, left_pa_end = -1; + + new_start = *start; + new_end = *end; + + /* + * Adjust the normalized range so that it doesn't overlap with any + * existing preallocated blocks(PAs). Make sure to hold the rbtree lock + * so it doesn't change underneath us. + */ + read_lock(&ei->i_prealloc_lock); + + /* Step 1: find any one immediate neighboring PA of the normalized range */ + for (iter = ei->i_prealloc_node.rb_node; iter; + iter = ext4_mb_pa_rb_next_iter(ac->ac_o_ex.fe_logical, + tmp_pa_start, iter)) { + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + tmp_pa_start = tmp_pa->pa_lstart; + tmp_pa_end = pa_logical_end(sbi, tmp_pa); + + /* PA must not overlap original request */ + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) + BUG_ON(!(ac->ac_o_ex.fe_logical >= tmp_pa_end || + ac->ac_o_ex.fe_logical < tmp_pa_start)); + spin_unlock(&tmp_pa->pa_lock); + } + + /* + * Step 2: check if the found PA is left or right neighbor and + * get the other neighbor + */ + if (tmp_pa) { + if (tmp_pa->pa_lstart < ac->ac_o_ex.fe_logical) { + struct rb_node *tmp; + + left_pa = tmp_pa; + tmp = rb_next(&left_pa->pa_node.inode_node); + if (tmp) { + right_pa = rb_entry(tmp, + struct ext4_prealloc_space, + pa_node.inode_node); + } + } else { + struct rb_node *tmp; + + right_pa = tmp_pa; + tmp = rb_prev(&right_pa->pa_node.inode_node); + if (tmp) { + left_pa = rb_entry(tmp, + struct ext4_prealloc_space, + pa_node.inode_node); + } + } + } + + /* Step 3: get the non deleted neighbors */ + if (left_pa) { + for (iter = &left_pa->pa_node.inode_node;; + iter = rb_prev(iter)) { + if (!iter) { + left_pa = NULL; + break; + } + + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + left_pa = tmp_pa; + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) { + spin_unlock(&tmp_pa->pa_lock); + break; + } + spin_unlock(&tmp_pa->pa_lock); + } + } + + if (right_pa) { + for (iter = &right_pa->pa_node.inode_node;; + iter = rb_next(iter)) { + if (!iter) { + right_pa = NULL; + break; + } + + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + right_pa = tmp_pa; + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) { + spin_unlock(&tmp_pa->pa_lock); + break; + } + spin_unlock(&tmp_pa->pa_lock); + } + } + + if (left_pa) { + left_pa_end = pa_logical_end(sbi, left_pa); + BUG_ON(left_pa_end > ac->ac_o_ex.fe_logical); + } + + if (right_pa) { + right_pa_start = right_pa->pa_lstart; + BUG_ON(right_pa_start <= ac->ac_o_ex.fe_logical); + } + + /* Step 4: trim our normalized range to not overlap with the neighbors */ + if (left_pa) { + if (left_pa_end > new_start) + new_start = left_pa_end; + } + + if (right_pa) { + if (right_pa_start < new_end) + new_end = right_pa_start; + } + read_unlock(&ei->i_prealloc_lock); + + /* XXX: extra loop to check we really don't overlap preallocations */ + ext4_mb_pa_assert_overlap(ac, new_start, new_end); + + *start = new_start; + *end = new_end; +} + +/* * Normalization means making request better in terms of * size and alignment */ @@ -3187,13 +4520,11 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac, struct ext4_allocation_request *ar) { struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + struct ext4_super_block *es = sbi->s_es; int bsbits, max; - ext4_lblk_t end; - loff_t size, start_off; + loff_t size, start_off, end; loff_t orig_size __maybe_unused; ext4_lblk_t start; - struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); - struct ext4_prealloc_space *pa; /* do normalize only data requests, metadata requests do not need preallocation */ @@ -3218,7 +4549,7 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac, /* first, let's learn actual file size * given current request is allocated */ - size = ac->ac_o_ex.fe_logical + EXT4_C2B(sbi, ac->ac_o_ex.fe_len); + size = extent_logical_end(sbi, &ac->ac_o_ex); size = size << bsbits; if (size < i_size_read(ac->ac_inode)) size = i_size_read(ac->ac_inode); @@ -3255,19 +4586,32 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac, start_off = ((loff_t)ac->ac_o_ex.fe_logical >> (22 - bsbits)) << 22; size = 4 * 1024 * 1024; - } else if (NRL_CHECK_SIZE(ac->ac_o_ex.fe_len, + } else if (NRL_CHECK_SIZE(EXT4_C2B(sbi, ac->ac_o_ex.fe_len), (8<<20)>>bsbits, max, 8 * 1024)) { start_off = ((loff_t)ac->ac_o_ex.fe_logical >> (23 - bsbits)) << 23; size = 8 * 1024 * 1024; } else { start_off = (loff_t) ac->ac_o_ex.fe_logical << bsbits; - size = (loff_t) EXT4_C2B(EXT4_SB(ac->ac_sb), + size = (loff_t) EXT4_C2B(sbi, ac->ac_o_ex.fe_len) << bsbits; } size = size >> bsbits; start = start_off >> bsbits; + /* + * For tiny groups (smaller than 8MB) the chosen allocation + * alignment may be larger than group size. Make sure the + * alignment does not move allocation to a different group which + * makes mballoc fail assertions later. + */ + start = max(start, rounddown(ac->ac_o_ex.fe_logical, + (ext4_lblk_t)EXT4_BLOCKS_PER_GROUP(ac->ac_sb))); + + /* avoid unnecessary preallocation that may trigger assertions */ + if (start + size > EXT_MAX_BLOCKS) + size = EXT_MAX_BLOCKS - start; + /* don't cover already allocated blocks in selected range */ if (ar->pleft && start <= ar->lleft) { size -= ar->lleft + 1 - start; @@ -3285,62 +4629,26 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac, end = start + size; - /* check we don't cross already preallocated blocks */ - rcu_read_lock(); - list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) { - ext4_lblk_t pa_end; - - if (pa->pa_deleted) - continue; - spin_lock(&pa->pa_lock); - if (pa->pa_deleted) { - spin_unlock(&pa->pa_lock); - continue; - } + ext4_mb_pa_adjust_overlap(ac, &start, &end); - pa_end = pa->pa_lstart + EXT4_C2B(EXT4_SB(ac->ac_sb), - pa->pa_len); - - /* PA must not overlap original request */ - BUG_ON(!(ac->ac_o_ex.fe_logical >= pa_end || - ac->ac_o_ex.fe_logical < pa->pa_lstart)); - - /* skip PAs this normalized request doesn't overlap with */ - if (pa->pa_lstart >= end || pa_end <= start) { - spin_unlock(&pa->pa_lock); - continue; - } - BUG_ON(pa->pa_lstart <= start && pa_end >= end); - - /* adjust start or end to be adjacent to this pa */ - if (pa_end <= ac->ac_o_ex.fe_logical) { - BUG_ON(pa_end < start); - start = pa_end; - } else if (pa->pa_lstart > ac->ac_o_ex.fe_logical) { - BUG_ON(pa->pa_lstart > end); - end = pa->pa_lstart; - } - spin_unlock(&pa->pa_lock); - } - rcu_read_unlock(); size = end - start; - /* XXX: extra loop to check we really don't overlap preallocations */ - rcu_read_lock(); - list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) { - ext4_lblk_t pa_end; - - spin_lock(&pa->pa_lock); - if (pa->pa_deleted == 0) { - pa_end = pa->pa_lstart + EXT4_C2B(EXT4_SB(ac->ac_sb), - pa->pa_len); - BUG_ON(!(start >= pa_end || end <= pa->pa_lstart)); - } - spin_unlock(&pa->pa_lock); - } - rcu_read_unlock(); - - if (start + size <= ac->ac_o_ex.fe_logical && + /* + * In this function "start" and "size" are normalized for better + * alignment and length such that we could preallocate more blocks. + * This normalization is done such that original request of + * ac->ac_o_ex.fe_logical & fe_len should always lie within "start" and + * "size" boundaries. + * (Note fe_len can be relaxed since FS block allocation API does not + * provide gurantee on number of contiguous blocks allocation since that + * depends upon free space left, etc). + * In case of inode pa, later we use the allocated blocks + * [pa_pstart + fe_logical - pa_lstart, fe_len/size] from the preallocated + * range of goal/best blocks [start, size] to put it at the + * ac_o_ex.fe_logical extent of this inode. + * (See ext4_mb_use_inode_pa() for more details) + */ + if (start + size <= ac->ac_o_ex.fe_logical || start > ac->ac_o_ex.fe_logical) { ext4_msg(ac->ac_sb, KERN_ERR, "start %lu, size %lu, fe_logical %lu", @@ -3356,20 +4664,24 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac, * placement or satisfy big request as is */ ac->ac_g_ex.fe_logical = start; ac->ac_g_ex.fe_len = EXT4_NUM_B2C(sbi, size); + ac->ac_orig_goal_len = ac->ac_g_ex.fe_len; /* define goal start in order to merge */ - if (ar->pright && (ar->lright == (start + size))) { + if (ar->pright && (ar->lright == (start + size)) && + ar->pright >= size && + ar->pright - size >= le32_to_cpu(es->s_first_data_block)) { /* merge to the right */ ext4_get_group_no_and_offset(ac->ac_sb, ar->pright - size, - &ac->ac_f_ex.fe_group, - &ac->ac_f_ex.fe_start); + &ac->ac_g_ex.fe_group, + &ac->ac_g_ex.fe_start); ac->ac_flags |= EXT4_MB_HINT_TRY_GOAL; } - if (ar->pleft && (ar->lleft + 1 == start)) { + if (ar->pleft && (ar->lleft + 1 == start) && + ar->pleft + 1 < ext4_blocks_count(es)) { /* merge to the left */ ext4_get_group_no_and_offset(ac->ac_sb, ar->pleft + 1, - &ac->ac_f_ex.fe_group, - &ac->ac_f_ex.fe_start); + &ac->ac_g_ex.fe_group, + &ac->ac_g_ex.fe_start); ac->ac_flags |= EXT4_MB_HINT_TRY_GOAL; } @@ -3381,15 +4693,25 @@ static void ext4_mb_collect_stats(struct ext4_allocation_context *ac) { struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); - if (sbi->s_mb_stats && ac->ac_g_ex.fe_len > 1) { + if (sbi->s_mb_stats && ac->ac_g_ex.fe_len >= 1) { atomic_inc(&sbi->s_bal_reqs); atomic_add(ac->ac_b_ex.fe_len, &sbi->s_bal_allocated); if (ac->ac_b_ex.fe_len >= ac->ac_o_ex.fe_len) atomic_inc(&sbi->s_bal_success); + atomic_add(ac->ac_found, &sbi->s_bal_ex_scanned); + for (int i=0; i<EXT4_MB_NUM_CRS; i++) { + atomic_add(ac->ac_cX_found[i], &sbi->s_bal_cX_ex_scanned[i]); + } + + atomic_add(ac->ac_groups_scanned, &sbi->s_bal_groups_scanned); if (ac->ac_g_ex.fe_start == ac->ac_b_ex.fe_start && ac->ac_g_ex.fe_group == ac->ac_b_ex.fe_group) atomic_inc(&sbi->s_bal_goals); + /* did we allocate as much as normalizer originally wanted? */ + if (ac->ac_f_ex.fe_len == ac->ac_orig_goal_len) + atomic_inc(&sbi->s_bal_len_goals); + if (ac->ac_found > sbi->s_mb_max_to_scan) atomic_inc(&sbi->s_bal_breaks); } @@ -3416,15 +4738,14 @@ static void ext4_discard_allocated_blocks(struct ext4_allocation_context *ac) if (ac->ac_f_ex.fe_len == 0) return; err = ext4_mb_load_buddy(ac->ac_sb, ac->ac_f_ex.fe_group, &e4b); - if (err) { + if (WARN_RATELIMIT(err, + "ext4: mb_load_buddy failed (%d)", err)) /* * This should never happen since we pin the - * pages in the ext4_allocation_context so + * folios in the ext4_allocation_context so * ext4_mb_load_buddy() should never fail. */ - WARN(1, "mb_load_buddy failed (%d)", err); return; - } ext4_lock_group(ac->ac_sb, ac->ac_f_ex.fe_group); mb_free_blocks(ac->ac_inode, &e4b, ac->ac_f_ex.fe_start, ac->ac_f_ex.fe_len); @@ -3432,8 +4753,11 @@ static void ext4_discard_allocated_blocks(struct ext4_allocation_context *ac) ext4_mb_unload_buddy(&e4b); return; } - if (pa->pa_type == MB_INODE_PA) + if (pa->pa_type == MB_INODE_PA) { + spin_lock(&pa->pa_lock); pa->pa_free += ac->ac_b_ex.fe_len; + spin_unlock(&pa->pa_lock); + } } /* @@ -3461,6 +4785,7 @@ static void ext4_mb_use_inode_pa(struct ext4_allocation_context *ac, BUG_ON(start < pa->pa_pstart); BUG_ON(end > pa->pa_pstart + EXT4_C2B(sbi, pa->pa_len)); BUG_ON(pa->pa_free < len); + BUG_ON(ac->ac_b_ex.fe_len <= 0); pa->pa_free -= len; mb_debug(ac->ac_sb, "use %llu/%d from inode pa %p\n", start, len, pa); @@ -3481,14 +4806,14 @@ static void ext4_mb_use_group_pa(struct ext4_allocation_context *ac, ac->ac_status = AC_STATUS_FOUND; ac->ac_pa = pa; - /* we don't correct pa_pstart or pa_plen here to avoid + /* we don't correct pa_pstart or pa_len here to avoid * possible race when the group is being loaded concurrently * instead we correct pa later, after blocks are marked * in on-disk bitmap -- see ext4_mb_release_context() * Other CPUs are prevented from allocating from this pa by lg_mutex */ mb_debug(ac->ac_sb, "use %u/%u from group pa %p\n", - pa->pa_lstart-len, len, pa); + pa->pa_lstart, len, pa); } /* @@ -3521,6 +4846,37 @@ ext4_mb_check_group_pa(ext4_fsblk_t goal_block, } /* + * check if found pa meets EXT4_MB_HINT_GOAL_ONLY + */ +static bool +ext4_mb_pa_goal_check(struct ext4_allocation_context *ac, + struct ext4_prealloc_space *pa) +{ + struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); + ext4_fsblk_t start; + + if (likely(!(ac->ac_flags & EXT4_MB_HINT_GOAL_ONLY))) + return true; + + /* + * If EXT4_MB_HINT_GOAL_ONLY is set, ac_g_ex will not be adjusted + * in ext4_mb_normalize_request and will keep same with ac_o_ex + * from ext4_mb_initialize_context. Choose ac_g_ex here to keep + * consistent with ext4_mb_find_by_goal. + */ + start = pa->pa_pstart + + (ac->ac_g_ex.fe_logical - pa->pa_lstart); + if (ext4_grp_offs_to_block(ac->ac_sb, &ac->ac_g_ex) != start) + return false; + + if (ac->ac_g_ex.fe_len > pa->pa_len - + EXT4_B2C(sbi, ac->ac_g_ex.fe_logical - pa->pa_lstart)) + return false; + + return true; +} + +/* * search goal blocks in preallocated space */ static noinline_for_stack bool @@ -3530,43 +4886,158 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac) int order, i; struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); struct ext4_locality_group *lg; - struct ext4_prealloc_space *pa, *cpa = NULL; + struct ext4_prealloc_space *tmp_pa = NULL, *cpa = NULL; + struct rb_node *iter; ext4_fsblk_t goal_block; /* only data can be preallocated */ if (!(ac->ac_flags & EXT4_MB_HINT_DATA)) return false; - /* first, try per-file preallocation */ - rcu_read_lock(); - list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) { + /* + * first, try per-file preallocation by searching the inode pa rbtree. + * + * Here, we can't do a direct traversal of the tree because + * ext4_mb_discard_group_preallocation() can paralelly mark the pa + * deleted and that can cause direct traversal to skip some entries. + */ + read_lock(&ei->i_prealloc_lock); - /* all fields in this condition don't change, - * so we can skip locking for them */ - if (ac->ac_o_ex.fe_logical < pa->pa_lstart || - ac->ac_o_ex.fe_logical >= (pa->pa_lstart + - EXT4_C2B(sbi, pa->pa_len))) - continue; + if (RB_EMPTY_ROOT(&ei->i_prealloc_node)) { + goto try_group_pa; + } - /* non-extent files can't have physical blocks past 2^32 */ - if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)) && - (pa->pa_pstart + EXT4_C2B(sbi, pa->pa_len) > - EXT4_MAX_BLOCK_FILE_PHYS)) - continue; + /* + * Step 1: Find a pa with logical start immediately adjacent to the + * original logical start. This could be on the left or right. + * + * (tmp_pa->pa_lstart never changes so we can skip locking for it). + */ + for (iter = ei->i_prealloc_node.rb_node; iter; + iter = ext4_mb_pa_rb_next_iter(ac->ac_o_ex.fe_logical, + tmp_pa->pa_lstart, iter)) { + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + } - /* found preallocated blocks, use them */ - spin_lock(&pa->pa_lock); - if (pa->pa_deleted == 0 && pa->pa_free) { - atomic_inc(&pa->pa_count); - ext4_mb_use_inode_pa(ac, pa); - spin_unlock(&pa->pa_lock); - ac->ac_criteria = 10; - rcu_read_unlock(); - return true; + /* + * Step 2: The adjacent pa might be to the right of logical start, find + * the left adjacent pa. After this step we'd have a valid tmp_pa whose + * logical start is towards the left of original request's logical start + */ + if (tmp_pa->pa_lstart > ac->ac_o_ex.fe_logical) { + struct rb_node *tmp; + tmp = rb_prev(&tmp_pa->pa_node.inode_node); + + if (tmp) { + tmp_pa = rb_entry(tmp, struct ext4_prealloc_space, + pa_node.inode_node); + } else { + /* + * If there is no adjacent pa to the left then finding + * an overlapping pa is not possible hence stop searching + * inode pa tree + */ + goto try_group_pa; } - spin_unlock(&pa->pa_lock); } - rcu_read_unlock(); + + BUG_ON(!(tmp_pa && tmp_pa->pa_lstart <= ac->ac_o_ex.fe_logical)); + + /* + * Step 3: If the left adjacent pa is deleted, keep moving left to find + * the first non deleted adjacent pa. After this step we should have a + * valid tmp_pa which is guaranteed to be non deleted. + */ + for (iter = &tmp_pa->pa_node.inode_node;; iter = rb_prev(iter)) { + if (!iter) { + /* + * no non deleted left adjacent pa, so stop searching + * inode pa tree + */ + goto try_group_pa; + } + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) { + /* + * We will keep holding the pa_lock from + * this point on because we don't want group discard + * to delete this pa underneath us. Since group + * discard is anyways an ENOSPC operation it + * should be okay for it to wait a few more cycles. + */ + break; + } else { + spin_unlock(&tmp_pa->pa_lock); + } + } + + BUG_ON(!(tmp_pa && tmp_pa->pa_lstart <= ac->ac_o_ex.fe_logical)); + BUG_ON(tmp_pa->pa_deleted == 1); + + /* + * Step 4: We now have the non deleted left adjacent pa. Only this + * pa can possibly satisfy the request hence check if it overlaps + * original logical start and stop searching if it doesn't. + */ + if (ac->ac_o_ex.fe_logical >= pa_logical_end(sbi, tmp_pa)) { + spin_unlock(&tmp_pa->pa_lock); + goto try_group_pa; + } + + /* non-extent files can't have physical blocks past 2^32 */ + if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)) && + (tmp_pa->pa_pstart + EXT4_C2B(sbi, tmp_pa->pa_len) > + EXT4_MAX_BLOCK_FILE_PHYS)) { + /* + * Since PAs don't overlap, we won't find any other PA to + * satisfy this. + */ + spin_unlock(&tmp_pa->pa_lock); + goto try_group_pa; + } + + if (tmp_pa->pa_free && likely(ext4_mb_pa_goal_check(ac, tmp_pa))) { + atomic_inc(&tmp_pa->pa_count); + ext4_mb_use_inode_pa(ac, tmp_pa); + spin_unlock(&tmp_pa->pa_lock); + read_unlock(&ei->i_prealloc_lock); + return true; + } else { + /* + * We found a valid overlapping pa but couldn't use it because + * it had no free blocks. This should ideally never happen + * because: + * + * 1. When a new inode pa is added to rbtree it must have + * pa_free > 0 since otherwise we won't actually need + * preallocation. + * + * 2. An inode pa that is in the rbtree can only have it's + * pa_free become zero when another thread calls: + * ext4_mb_new_blocks + * ext4_mb_use_preallocated + * ext4_mb_use_inode_pa + * + * 3. Further, after the above calls make pa_free == 0, we will + * immediately remove it from the rbtree in: + * ext4_mb_new_blocks + * ext4_mb_release_context + * ext4_mb_put_pa + * + * 4. Since the pa_free becoming 0 and pa_free getting removed + * from tree both happen in ext4_mb_new_blocks, which is always + * called with i_data_sem held for data allocations, we can be + * sure that another process will never see a pa in rbtree with + * pa_free == 0. + */ + WARN_ON_ONCE(tmp_pa->pa_free == 0); + } + spin_unlock(&tmp_pa->pa_lock); +try_group_pa: + read_unlock(&ei->i_prealloc_lock); /* can we use group allocation? */ if (!(ac->ac_flags & EXT4_MB_HINT_GROUP_ALLOC)) @@ -3588,52 +5059,27 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac) */ for (i = order; i < PREALLOC_TB_SIZE; i++) { rcu_read_lock(); - list_for_each_entry_rcu(pa, &lg->lg_prealloc_list[i], - pa_inode_list) { - spin_lock(&pa->pa_lock); - if (pa->pa_deleted == 0 && - pa->pa_free >= ac->ac_o_ex.fe_len) { + list_for_each_entry_rcu(tmp_pa, &lg->lg_prealloc_list[i], + pa_node.lg_list) { + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0 && + tmp_pa->pa_free >= ac->ac_o_ex.fe_len) { cpa = ext4_mb_check_group_pa(goal_block, - pa, cpa); + tmp_pa, cpa); } - spin_unlock(&pa->pa_lock); + spin_unlock(&tmp_pa->pa_lock); } rcu_read_unlock(); } if (cpa) { ext4_mb_use_group_pa(ac, cpa); - ac->ac_criteria = 20; return true; } return false; } /* - * the function goes through all block freed in the group - * but not yet committed and marks them used in in-core bitmap. - * buddy must be generated from this bitmap - * Need to be called with the ext4 group lock held - */ -static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap, - ext4_group_t group) -{ - struct rb_node *n; - struct ext4_group_info *grp; - struct ext4_free_data *entry; - - grp = ext4_get_group_info(sb, group); - n = rb_first(&(grp->bb_free_root)); - - while (n) { - entry = rb_entry(n, struct ext4_free_data, efd_node); - ext4_set_bits(bitmap, entry->efd_start_cluster, entry->efd_count); - n = rb_next(n); - } - return; -} - -/* * the function goes through all preallocation in this group and marks them * used in in-core bitmap. buddy must be generated from this bitmap * Need to be called with ext4 group lock held @@ -3650,6 +5096,9 @@ void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap, int preallocated = 0; int len; + if (!grp) + return; + /* all form of preallocation discards first load group, * so the only competing code is preallocation use. * we don't need any locking here @@ -3668,22 +5117,48 @@ void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap, if (unlikely(len == 0)) continue; BUG_ON(groupnr != group); - ext4_set_bits(bitmap, start, len); + mb_set_bits(bitmap, start, len); preallocated += len; } mb_debug(sb, "preallocated %d for group %u\n", preallocated, group); } -static void ext4_mb_pa_callback(struct rcu_head *head) +static void ext4_mb_mark_pa_deleted(struct super_block *sb, + struct ext4_prealloc_space *pa) { - struct ext4_prealloc_space *pa; - pa = container_of(head, struct ext4_prealloc_space, u.pa_rcu); + struct ext4_inode_info *ei; + + if (pa->pa_deleted) { + ext4_warning(sb, "deleted pa, type:%d, pblk:%llu, lblk:%u, len:%d\n", + pa->pa_type, pa->pa_pstart, pa->pa_lstart, + pa->pa_len); + return; + } + + pa->pa_deleted = 1; + if (pa->pa_type == MB_INODE_PA) { + ei = EXT4_I(pa->pa_inode); + atomic_dec(&ei->i_prealloc_active); + } +} + +static inline void ext4_mb_pa_free(struct ext4_prealloc_space *pa) +{ + BUG_ON(!pa); BUG_ON(atomic_read(&pa->pa_count)); BUG_ON(pa->pa_deleted == 0); kmem_cache_free(ext4_pspace_cachep, pa); } +static void ext4_mb_pa_callback(struct rcu_head *head) +{ + struct ext4_prealloc_space *pa; + + pa = container_of(head, struct ext4_prealloc_space, u.pa_rcu); + ext4_mb_pa_free(pa); +} + /* * drops a reference to preallocated space descriptor * if this was the last reference and the space is consumed @@ -3693,6 +5168,7 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac, { ext4_group_t grp; ext4_fsblk_t grp_blk; + struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); /* in this short window concurrent discard can set pa_deleted */ spin_lock(&pa->pa_lock); @@ -3706,7 +5182,7 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac, return; } - pa->pa_deleted = 1; + ext4_mb_mark_pa_deleted(sb, pa); spin_unlock(&pa->pa_lock); grp_blk = pa->pa_pstart; @@ -3737,11 +5213,42 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac, list_del(&pa->pa_group_list); ext4_unlock_group(sb, grp); - spin_lock(pa->pa_obj_lock); - list_del_rcu(&pa->pa_inode_list); - spin_unlock(pa->pa_obj_lock); + if (pa->pa_type == MB_INODE_PA) { + write_lock(pa->pa_node_lock.inode_lock); + rb_erase(&pa->pa_node.inode_node, &ei->i_prealloc_node); + write_unlock(pa->pa_node_lock.inode_lock); + ext4_mb_pa_free(pa); + } else { + spin_lock(pa->pa_node_lock.lg_lock); + list_del_rcu(&pa->pa_node.lg_list); + spin_unlock(pa->pa_node_lock.lg_lock); + call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback); + } +} + +static void ext4_mb_pa_rb_insert(struct rb_root *root, struct rb_node *new) +{ + struct rb_node **iter = &root->rb_node, *parent = NULL; + struct ext4_prealloc_space *iter_pa, *new_pa; + ext4_lblk_t iter_start, new_start; + + while (*iter) { + iter_pa = rb_entry(*iter, struct ext4_prealloc_space, + pa_node.inode_node); + new_pa = rb_entry(new, struct ext4_prealloc_space, + pa_node.inode_node); + iter_start = iter_pa->pa_lstart; + new_start = new_pa->pa_lstart; + + parent = *iter; + if (new_start < iter_start) + iter = &((*iter)->rb_left); + else + iter = &((*iter)->rb_right); + } - call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback); + rb_link_node(new, parent, iter); + rb_insert_color(new, root); } /* @@ -3764,50 +5271,59 @@ ext4_mb_new_inode_pa(struct ext4_allocation_context *ac) pa = ac->ac_pa; - if (ac->ac_b_ex.fe_len < ac->ac_g_ex.fe_len) { - int winl; - int wins; - int win; - int offs; + if (ac->ac_b_ex.fe_len < ac->ac_orig_goal_len) { + struct ext4_free_extent ex = { + .fe_logical = ac->ac_g_ex.fe_logical, + .fe_len = ac->ac_orig_goal_len, + }; + loff_t orig_goal_end = extent_logical_end(sbi, &ex); + loff_t o_ex_end = extent_logical_end(sbi, &ac->ac_o_ex); - /* we can't allocate as much as normalizer wants. - * so, found space must get proper lstart - * to cover original request */ + /* + * We can't allocate as much as normalizer wants, so we try + * to get proper lstart to cover the original request, except + * when the goal doesn't cover the original request as below: + * + * orig_ex:2045/2055(10), isize:8417280 -> normalized:0/2048 + * best_ex:0/200(200) -> adjusted: 1848/2048(200) + */ BUG_ON(ac->ac_g_ex.fe_logical > ac->ac_o_ex.fe_logical); BUG_ON(ac->ac_g_ex.fe_len < ac->ac_o_ex.fe_len); - /* we're limited by original request in that - * logical block must be covered any way - * winl is window we can move our chunk within */ - winl = ac->ac_o_ex.fe_logical - ac->ac_g_ex.fe_logical; + /* + * Use the below logic for adjusting best extent as it keeps + * fragmentation in check while ensuring logical range of best + * extent doesn't overflow out of goal extent: + * + * 1. Check if best ex can be kept at end of goal (before + * cr_best_avail trimmed it) and still cover original start + * 2. Else, check if best ex can be kept at start of goal and + * still cover original end + * 3. Else, keep the best ex at start of original request. + */ + ex.fe_len = ac->ac_b_ex.fe_len; - /* also, we should cover whole original request */ - wins = EXT4_C2B(sbi, ac->ac_b_ex.fe_len - ac->ac_o_ex.fe_len); + ex.fe_logical = orig_goal_end - EXT4_C2B(sbi, ex.fe_len); + if (ac->ac_o_ex.fe_logical >= ex.fe_logical) + goto adjust_bex; - /* the smallest one defines real window */ - win = min(winl, wins); + ex.fe_logical = ac->ac_g_ex.fe_logical; + if (o_ex_end <= extent_logical_end(sbi, &ex)) + goto adjust_bex; - offs = ac->ac_o_ex.fe_logical % - EXT4_C2B(sbi, ac->ac_b_ex.fe_len); - if (offs && offs < win) - win = offs; + ex.fe_logical = ac->ac_o_ex.fe_logical; +adjust_bex: + ac->ac_b_ex.fe_logical = ex.fe_logical; - ac->ac_b_ex.fe_logical = ac->ac_o_ex.fe_logical - - EXT4_NUM_B2C(sbi, win); BUG_ON(ac->ac_o_ex.fe_logical < ac->ac_b_ex.fe_logical); - BUG_ON(ac->ac_o_ex.fe_len > ac->ac_b_ex.fe_len); + BUG_ON(extent_logical_end(sbi, &ex) > orig_goal_end); } - /* preallocation can change ac_b_ex, thus we store actually - * allocated blocks for history */ - ac->ac_f_ex = ac->ac_b_ex; - pa->pa_lstart = ac->ac_b_ex.fe_logical; pa->pa_pstart = ext4_grp_offs_to_block(sb, &ac->ac_b_ex); pa->pa_len = ac->ac_b_ex.fe_len; pa->pa_free = pa->pa_len; spin_lock_init(&pa->pa_lock); - INIT_LIST_HEAD(&pa->pa_inode_list); INIT_LIST_HEAD(&pa->pa_group_list); pa->pa_deleted = 0; pa->pa_type = MB_INODE_PA; @@ -3816,20 +5332,23 @@ ext4_mb_new_inode_pa(struct ext4_allocation_context *ac) pa->pa_len, pa->pa_lstart); trace_ext4_mb_new_inode_pa(ac, pa); - ext4_mb_use_inode_pa(ac, pa); atomic_add(pa->pa_free, &sbi->s_mb_preallocated); + ext4_mb_use_inode_pa(ac, pa); ei = EXT4_I(ac->ac_inode); grp = ext4_get_group_info(sb, ac->ac_b_ex.fe_group); + if (!grp) + return; - pa->pa_obj_lock = &ei->i_prealloc_lock; + pa->pa_node_lock.inode_lock = &ei->i_prealloc_lock; pa->pa_inode = ac->ac_inode; list_add(&pa->pa_group_list, &grp->bb_prealloc_list); - spin_lock(pa->pa_obj_lock); - list_add_rcu(&pa->pa_inode_list, &ei->i_prealloc_list); - spin_unlock(pa->pa_obj_lock); + write_lock(pa->pa_node_lock.inode_lock); + ext4_mb_pa_rb_insert(&ei->i_prealloc_node, &pa->pa_node.inode_node); + write_unlock(pa->pa_node_lock.inode_lock); + atomic_inc(&ei->i_prealloc_active); } /* @@ -3851,16 +5370,12 @@ ext4_mb_new_group_pa(struct ext4_allocation_context *ac) pa = ac->ac_pa; - /* preallocation can change ac_b_ex, thus we store actually - * allocated blocks for history */ - ac->ac_f_ex = ac->ac_b_ex; - pa->pa_pstart = ext4_grp_offs_to_block(sb, &ac->ac_b_ex); pa->pa_lstart = pa->pa_pstart; pa->pa_len = ac->ac_b_ex.fe_len; pa->pa_free = pa->pa_len; spin_lock_init(&pa->pa_lock); - INIT_LIST_HEAD(&pa->pa_inode_list); + INIT_LIST_HEAD(&pa->pa_node.lg_list); INIT_LIST_HEAD(&pa->pa_group_list); pa->pa_deleted = 0; pa->pa_type = MB_GROUP_PA; @@ -3873,10 +5388,12 @@ ext4_mb_new_group_pa(struct ext4_allocation_context *ac) atomic_add(pa->pa_free, &EXT4_SB(sb)->s_mb_preallocated); grp = ext4_get_group_info(sb, ac->ac_b_ex.fe_group); + if (!grp) + return; lg = ac->ac_lg; BUG_ON(lg == NULL); - pa->pa_obj_lock = &lg->lg_prealloc_lock; + pa->pa_node_lock.lg_lock = &lg->lg_prealloc_lock; pa->pa_inode = NULL; list_add(&pa->pa_group_list, &grp->bb_prealloc_list); @@ -3903,7 +5420,7 @@ static void ext4_mb_new_preallocation(struct ext4_allocation_context *ac) * the caller MUST hold group/inode locks. * TODO: optimize the case when there are no in-core structures yet */ -static noinline_for_stack int +static noinline_for_stack void ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh, struct ext4_prealloc_space *pa) { @@ -3953,11 +5470,9 @@ ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh, */ } atomic_add(free, &sbi->s_mb_discarded); - - return 0; } -static noinline_for_stack int +static noinline_for_stack void ext4_mb_release_group_pa(struct ext4_buddy *e4b, struct ext4_prealloc_space *pa) { @@ -3968,12 +5483,14 @@ ext4_mb_release_group_pa(struct ext4_buddy *e4b, trace_ext4_mb_release_group_pa(sb, pa); BUG_ON(pa->pa_deleted == 0); ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit); - BUG_ON(group != e4b->bd_group && pa->pa_len != 0); + if (unlikely(group != e4b->bd_group && pa->pa_len != 0)) { + ext4_warning(sb, "bad group: expected %u, group %u, pa_start %llu", + e4b->bd_group, group, pa->pa_pstart); + return; + } mb_free_blocks(pa->pa_inode, e4b, bit, pa->pa_len); atomic_add(pa->pa_len, &EXT4_SB(sb)->s_mb_discarded); trace_ext4_mballoc_discard(sb, NULL, group, bit, pa->pa_len); - - return 0; } /* @@ -3987,17 +5504,19 @@ ext4_mb_release_group_pa(struct ext4_buddy *e4b, */ static noinline_for_stack int ext4_mb_discard_group_preallocations(struct super_block *sb, - ext4_group_t group, int needed) + ext4_group_t group, int *busy) { struct ext4_group_info *grp = ext4_get_group_info(sb, group); struct buffer_head *bitmap_bh = NULL; struct ext4_prealloc_space *pa, *tmp; - struct list_head list; + LIST_HEAD(list); struct ext4_buddy e4b; + struct ext4_inode_info *ei; int err; - int busy = 0; int free = 0; + if (!grp) + return 0; mb_debug(sb, "discard preallocation for group %u\n", group); if (list_empty(&grp->bb_prealloc_list)) goto out_dbg; @@ -4019,19 +5538,13 @@ ext4_mb_discard_group_preallocations(struct super_block *sb, goto out_dbg; } - if (needed == 0) - needed = EXT4_CLUSTERS_PER_GROUP(sb) + 1; - - INIT_LIST_HEAD(&list); -repeat: ext4_lock_group(sb, group); - this_cpu_inc(discard_pa_seq); list_for_each_entry_safe(pa, tmp, &grp->bb_prealloc_list, pa_group_list) { spin_lock(&pa->pa_lock); if (atomic_read(&pa->pa_count)) { spin_unlock(&pa->pa_lock); - busy = 1; + *busy = 1; continue; } if (pa->pa_deleted) { @@ -4040,7 +5553,10 @@ repeat: } /* seems this one can be freed ... */ - pa->pa_deleted = 1; + ext4_mb_mark_pa_deleted(sb, pa); + + if (!free) + this_cpu_inc(discard_pa_seq); /* we can trust pa_free ... */ free += pa->pa_free; @@ -4051,40 +5567,32 @@ repeat: list_add(&pa->u.pa_tmp_list, &list); } - /* if we still need more blocks and some PAs were used, try again */ - if (free < needed && busy) { - busy = 0; - ext4_unlock_group(sb, group); - cond_resched(); - goto repeat; - } - - /* found anything to free? */ - if (list_empty(&list)) { - BUG_ON(free != 0); - mb_debug(sb, "Someone else may have freed PA for this group %u\n", - group); - goto out; - } - /* now free all selected PAs */ list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) { /* remove from object (inode or locality group) */ - spin_lock(pa->pa_obj_lock); - list_del_rcu(&pa->pa_inode_list); - spin_unlock(pa->pa_obj_lock); + if (pa->pa_type == MB_GROUP_PA) { + spin_lock(pa->pa_node_lock.lg_lock); + list_del_rcu(&pa->pa_node.lg_list); + spin_unlock(pa->pa_node_lock.lg_lock); + } else { + write_lock(pa->pa_node_lock.inode_lock); + ei = EXT4_I(pa->pa_inode); + rb_erase(&pa->pa_node.inode_node, &ei->i_prealloc_node); + write_unlock(pa->pa_node_lock.inode_lock); + } + + list_del(&pa->u.pa_tmp_list); - if (pa->pa_type == MB_GROUP_PA) + if (pa->pa_type == MB_GROUP_PA) { ext4_mb_release_group_pa(&e4b, pa); - else + call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback); + } else { ext4_mb_release_inode_pa(&e4b, bitmap_bh, pa); - - list_del(&pa->u.pa_tmp_list); - call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback); + ext4_mb_pa_free(pa); + } } -out: ext4_unlock_group(sb, group); ext4_mb_unload_buddy(&e4b); put_bh(bitmap_bh); @@ -4110,34 +5618,37 @@ void ext4_discard_preallocations(struct inode *inode) struct buffer_head *bitmap_bh = NULL; struct ext4_prealloc_space *pa, *tmp; ext4_group_t group = 0; - struct list_head list; + LIST_HEAD(list); struct ext4_buddy e4b; + struct rb_node *iter; int err; - if (!S_ISREG(inode->i_mode)) { - /*BUG_ON(!list_empty(&ei->i_prealloc_list));*/ + if (!S_ISREG(inode->i_mode)) + return; + + if (EXT4_SB(sb)->s_mount_state & EXT4_FC_REPLAY) return; - } mb_debug(sb, "discard preallocation for inode %lu\n", inode->i_ino); - trace_ext4_discard_preallocations(inode); - - INIT_LIST_HEAD(&list); + trace_ext4_discard_preallocations(inode, + atomic_read(&ei->i_prealloc_active)); repeat: /* first, collect all pa's in the inode */ - spin_lock(&ei->i_prealloc_lock); - while (!list_empty(&ei->i_prealloc_list)) { - pa = list_entry(ei->i_prealloc_list.next, - struct ext4_prealloc_space, pa_inode_list); - BUG_ON(pa->pa_obj_lock != &ei->i_prealloc_lock); + write_lock(&ei->i_prealloc_lock); + for (iter = rb_first(&ei->i_prealloc_node); iter; + iter = rb_next(iter)) { + pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + BUG_ON(pa->pa_node_lock.inode_lock != &ei->i_prealloc_lock); + spin_lock(&pa->pa_lock); if (atomic_read(&pa->pa_count)) { /* this shouldn't happen often - nobody should * use preallocation while we're discarding it */ spin_unlock(&pa->pa_lock); - spin_unlock(&ei->i_prealloc_lock); + write_unlock(&ei->i_prealloc_lock); ext4_msg(sb, KERN_ERR, "uh-oh! used pa while discarding"); WARN_ON(1); @@ -4146,16 +5657,16 @@ repeat: } if (pa->pa_deleted == 0) { - pa->pa_deleted = 1; + ext4_mb_mark_pa_deleted(sb, pa); spin_unlock(&pa->pa_lock); - list_del_rcu(&pa->pa_inode_list); + rb_erase(&pa->pa_node.inode_node, &ei->i_prealloc_node); list_add(&pa->u.pa_tmp_list, &list); continue; } /* someone is deleting pa right now */ spin_unlock(&pa->pa_lock); - spin_unlock(&ei->i_prealloc_lock); + write_unlock(&ei->i_prealloc_lock); /* we have to wait here because pa_deleted * doesn't mean pa is already unlinked from @@ -4172,7 +5683,7 @@ repeat: schedule_timeout_uninterruptible(HZ); goto repeat; } - spin_unlock(&ei->i_prealloc_lock); + write_unlock(&ei->i_prealloc_lock); list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) { BUG_ON(pa->pa_type != MB_INODE_PA); @@ -4204,7 +5715,7 @@ repeat: put_bh(bitmap_bh); list_del(&pa->u.pa_tmp_list); - call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback); + ext4_mb_pa_free(pa); } } @@ -4221,14 +5732,20 @@ static int ext4_mb_pa_alloc(struct ext4_allocation_context *ac) return 0; } -static void ext4_mb_pa_free(struct ext4_allocation_context *ac) +static void ext4_mb_pa_put_free(struct ext4_allocation_context *ac) { struct ext4_prealloc_space *pa = ac->ac_pa; BUG_ON(!pa); ac->ac_pa = NULL; WARN_ON(!atomic_dec_and_test(&pa->pa_count)); - kmem_cache_free(ext4_pspace_cachep, pa); + /* + * current function is only called due to an error or due to + * len of found blocks < len of requested blocks hence the PA has not + * been added to grp->bb_prealloc_list. So we don't need to lock it + */ + pa->pa_deleted = 1; + ext4_mb_pa_free(pa); } #ifdef CONFIG_EXT4_DEBUG @@ -4236,7 +5753,7 @@ static inline void ext4_mb_show_pa(struct super_block *sb) { ext4_group_t i, ngroups; - if (EXT4_SB(sb)->s_mount_flags & EXT4_MF_FS_ABORTED) + if (ext4_emergency_state(sb)) return; ngroups = ext4_get_groups_count(sb); @@ -4246,6 +5763,9 @@ static inline void ext4_mb_show_pa(struct super_block *sb) struct ext4_prealloc_space *pa; ext4_grpblk_t start; struct list_head *cur; + + if (!grp) + continue; ext4_lock_group(sb, i); list_for_each(cur, &grp->bb_prealloc_list) { pa = list_entry(cur, struct ext4_prealloc_space, @@ -4267,7 +5787,7 @@ static void ext4_mb_show_ac(struct ext4_allocation_context *ac) { struct super_block *sb = ac->ac_sb; - if (EXT4_SB(sb)->s_mount_flags & EXT4_MF_FS_ABORTED) + if (ext4_emergency_state(sb)) return; mb_debug(sb, "Can't allocate:" @@ -4291,17 +5811,19 @@ static void ext4_mb_show_ac(struct ext4_allocation_context *ac) (unsigned long)ac->ac_b_ex.fe_logical, (int)ac->ac_criteria); mb_debug(sb, "%u found", ac->ac_found); + mb_debug(sb, "used pa: %s, ", str_yes_no(ac->ac_pa)); + if (ac->ac_pa) + mb_debug(sb, "pa_type %s\n", ac->ac_pa->pa_type == MB_GROUP_PA ? + "group pa" : "inode pa"); ext4_mb_show_pa(sb); } #else static inline void ext4_mb_show_pa(struct super_block *sb) { - return; } static inline void ext4_mb_show_ac(struct ext4_allocation_context *ac) { ext4_mb_show_pa(ac->ac_sb); - return; } #endif @@ -4317,6 +5839,7 @@ static void ext4_mb_group_or_file(struct ext4_allocation_context *ac) struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); int bsbits = ac->ac_sb->s_blocksize_bits; loff_t size, isize; + bool inode_pa_eligible, group_pa_eligible; if (!(ac->ac_flags & EXT4_MB_HINT_DATA)) return; @@ -4324,25 +5847,27 @@ static void ext4_mb_group_or_file(struct ext4_allocation_context *ac) if (unlikely(ac->ac_flags & EXT4_MB_HINT_GOAL_ONLY)) return; - size = ac->ac_o_ex.fe_logical + EXT4_C2B(sbi, ac->ac_o_ex.fe_len); + group_pa_eligible = sbi->s_mb_group_prealloc > 0; + inode_pa_eligible = true; + size = extent_logical_end(sbi, &ac->ac_o_ex); isize = (i_size_read(ac->ac_inode) + ac->ac_sb->s_blocksize - 1) >> bsbits; + /* No point in using inode preallocation for closed files */ if ((size == isize) && !ext4_fs_is_busy(sbi) && - !inode_is_open_for_write(ac->ac_inode)) { - ac->ac_flags |= EXT4_MB_HINT_NOPREALLOC; - return; - } + !inode_is_open_for_write(ac->ac_inode)) + inode_pa_eligible = false; - if (sbi->s_mb_group_prealloc <= 0) { - ac->ac_flags |= EXT4_MB_STREAM_ALLOC; - return; - } - - /* don't use group allocation for large files */ size = max(size, isize); - if (size > sbi->s_mb_stream_request) { - ac->ac_flags |= EXT4_MB_STREAM_ALLOC; + /* Don't use group allocation for large files */ + if (size > sbi->s_mb_stream_request) + group_pa_eligible = false; + + if (!group_pa_eligible) { + if (inode_pa_eligible) + ac->ac_flags |= EXT4_MB_STREAM_ALLOC; + else + ac->ac_flags |= EXT4_MB_HINT_NOPREALLOC; return; } @@ -4361,7 +5886,7 @@ static void ext4_mb_group_or_file(struct ext4_allocation_context *ac) mutex_lock(&ac->ac_lg->lg_mutex); } -static noinline_for_stack int +static noinline_for_stack void ext4_mb_initialize_context(struct ext4_allocation_context *ac, struct ext4_allocation_request *ar) { @@ -4397,9 +5922,10 @@ ext4_mb_initialize_context(struct ext4_allocation_context *ac, ac->ac_o_ex.fe_start = block; ac->ac_o_ex.fe_len = len; ac->ac_g_ex = ac->ac_o_ex; + ac->ac_orig_goal_len = ac->ac_g_ex.fe_len; ac->ac_flags = ar->flags; - /* we have to define context: we'll we work with a file or + /* we have to define context: we'll work with a file or * locality group. this is a policy, actually */ ext4_mb_group_or_file(ac); @@ -4410,8 +5936,6 @@ ext4_mb_initialize_context(struct ext4_allocation_context *ac, (unsigned) ar->lleft, (unsigned) ar->pleft, (unsigned) ar->lright, (unsigned) ar->pright, inode_is_open_for_write(ar->inode) ? "" : "non-"); - return 0; - } static noinline_for_stack void @@ -4421,16 +5945,14 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb, { ext4_group_t group = 0; struct ext4_buddy e4b; - struct list_head discard_list; + LIST_HEAD(discard_list); struct ext4_prealloc_space *pa, *tmp; mb_debug(sb, "discard locality group preallocation\n"); - INIT_LIST_HEAD(&discard_list); - spin_lock(&lg->lg_prealloc_lock); list_for_each_entry_rcu(pa, &lg->lg_prealloc_list[order], - pa_inode_list, + pa_node.lg_list, lockdep_is_held(&lg->lg_prealloc_lock)) { spin_lock(&pa->pa_lock); if (atomic_read(&pa->pa_count)) { @@ -4450,10 +5972,10 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb, BUG_ON(pa->pa_type != MB_GROUP_PA); /* seems this one can be freed ... */ - pa->pa_deleted = 1; + ext4_mb_mark_pa_deleted(sb, pa); spin_unlock(&pa->pa_lock); - list_del_rcu(&pa->pa_inode_list); + list_del_rcu(&pa->pa_node.lg_list); list_add(&pa->u.pa_tmp_list, &discard_list); total_entries--; @@ -4514,7 +6036,7 @@ static void ext4_mb_add_n_trim(struct ext4_allocation_context *ac) /* Add the prealloc space to lg */ spin_lock(&lg->lg_prealloc_lock); list_for_each_entry_rcu(tmp_pa, &lg->lg_prealloc_list[order], - pa_inode_list, + pa_node.lg_list, lockdep_is_held(&lg->lg_prealloc_lock)) { spin_lock(&tmp_pa->pa_lock); if (tmp_pa->pa_deleted) { @@ -4523,8 +6045,8 @@ static void ext4_mb_add_n_trim(struct ext4_allocation_context *ac) } if (!added && pa->pa_free < tmp_pa->pa_free) { /* Add to the tail of the previous entry */ - list_add_tail_rcu(&pa->pa_inode_list, - &tmp_pa->pa_inode_list); + list_add_tail_rcu(&pa->pa_node.lg_list, + &tmp_pa->pa_node.lg_list); added = 1; /* * we want to count the total @@ -4535,23 +6057,20 @@ static void ext4_mb_add_n_trim(struct ext4_allocation_context *ac) lg_prealloc_count++; } if (!added) - list_add_tail_rcu(&pa->pa_inode_list, + list_add_tail_rcu(&pa->pa_node.lg_list, &lg->lg_prealloc_list[order]); spin_unlock(&lg->lg_prealloc_lock); /* Now trim the list to be not more than 8 elements */ - if (lg_prealloc_count > 8) { + if (lg_prealloc_count > 8) ext4_mb_discard_lg_preallocations(sb, lg, order, lg_prealloc_count); - return; - } - return ; } /* * release all resource we used in allocation */ -static int ext4_mb_release_context(struct ext4_allocation_context *ac) +static void ext4_mb_release_context(struct ext4_allocation_context *ac) { struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb); struct ext4_prealloc_space *pa = ac->ac_pa; @@ -4564,44 +6083,54 @@ static int ext4_mb_release_context(struct ext4_allocation_context *ac) pa->pa_free -= ac->ac_b_ex.fe_len; pa->pa_len -= ac->ac_b_ex.fe_len; spin_unlock(&pa->pa_lock); + + /* + * We want to add the pa to the right bucket. + * Remove it from the list and while adding + * make sure the list to which we are adding + * doesn't grow big. + */ + if (likely(pa->pa_free)) { + spin_lock(pa->pa_node_lock.lg_lock); + list_del_rcu(&pa->pa_node.lg_list); + spin_unlock(pa->pa_node_lock.lg_lock); + ext4_mb_add_n_trim(ac); + } } - } - if (pa) { - /* - * We want to add the pa to the right bucket. - * Remove it from the list and while adding - * make sure the list to which we are adding - * doesn't grow big. - */ - if ((pa->pa_type == MB_GROUP_PA) && likely(pa->pa_free)) { - spin_lock(pa->pa_obj_lock); - list_del_rcu(&pa->pa_inode_list); - spin_unlock(pa->pa_obj_lock); - ext4_mb_add_n_trim(ac); - } + ext4_mb_put_pa(ac, ac->ac_sb, pa); } - if (ac->ac_bitmap_page) - put_page(ac->ac_bitmap_page); - if (ac->ac_buddy_page) - put_page(ac->ac_buddy_page); + if (ac->ac_bitmap_folio) + folio_put(ac->ac_bitmap_folio); + if (ac->ac_buddy_folio) + folio_put(ac->ac_buddy_folio); if (ac->ac_flags & EXT4_MB_HINT_GROUP_ALLOC) mutex_unlock(&ac->ac_lg->lg_mutex); ext4_mb_collect_stats(ac); - return 0; } static int ext4_mb_discard_preallocations(struct super_block *sb, int needed) { ext4_group_t i, ngroups = ext4_get_groups_count(sb); int ret; - int freed = 0; + int freed = 0, busy = 0; + int retry = 0; trace_ext4_mb_discard_preallocations(sb, needed); + + if (needed == 0) + needed = EXT4_CLUSTERS_PER_GROUP(sb) + 1; + repeat: for (i = 0; i < ngroups && needed > 0; i++) { - ret = ext4_mb_discard_group_preallocations(sb, i, needed); + ret = ext4_mb_discard_group_preallocations(sb, i, &busy); freed += ret; needed -= ret; + cond_resched(); + } + + if (needed > 0 && busy && ++retry < 3) { + busy = 0; + goto repeat; } return freed; @@ -4627,11 +6156,79 @@ static bool ext4_mb_discard_preallocations_should_retry(struct super_block *sb, } out_dbg: - mb_debug(sb, "freed %d, retry ? %s\n", freed, ret ? "yes" : "no"); + mb_debug(sb, "freed %d, retry ? %s\n", freed, str_yes_no(ret)); return ret; } /* + * Simple allocator for Ext4 fast commit replay path. It searches for blocks + * linearly starting at the goal block and also excludes the blocks which + * are going to be in use after fast commit replay. + */ +static ext4_fsblk_t +ext4_mb_new_blocks_simple(struct ext4_allocation_request *ar, int *errp) +{ + struct buffer_head *bitmap_bh; + struct super_block *sb = ar->inode->i_sb; + struct ext4_sb_info *sbi = EXT4_SB(sb); + ext4_group_t group, nr; + ext4_grpblk_t blkoff; + ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb); + ext4_grpblk_t i = 0; + ext4_fsblk_t goal, block; + struct ext4_super_block *es = sbi->s_es; + + goal = ar->goal; + if (goal < le32_to_cpu(es->s_first_data_block) || + goal >= ext4_blocks_count(es)) + goal = le32_to_cpu(es->s_first_data_block); + + ar->len = 0; + ext4_get_group_no_and_offset(sb, goal, &group, &blkoff); + for (nr = ext4_get_groups_count(sb); nr > 0; nr--) { + bitmap_bh = ext4_read_block_bitmap(sb, group); + if (IS_ERR(bitmap_bh)) { + *errp = PTR_ERR(bitmap_bh); + pr_warn("Failed to read block bitmap\n"); + return 0; + } + + while (1) { + i = mb_find_next_zero_bit(bitmap_bh->b_data, max, + blkoff); + if (i >= max) + break; + if (ext4_fc_replay_check_excluded(sb, + ext4_group_first_block_no(sb, group) + + EXT4_C2B(sbi, i))) { + blkoff = i + 1; + } else + break; + } + brelse(bitmap_bh); + if (i < max) + break; + + if (++group >= ext4_get_groups_count(sb)) + group = 0; + + blkoff = 0; + } + + if (i >= max) { + *errp = -ENOSPC; + return 0; + } + + block = ext4_group_first_block_no(sb, group) + EXT4_C2B(sbi, i); + ext4_mb_mark_bb(sb, block, 1, true); + ar->len = 1; + + *errp = 0; + return block; +} + +/* * Main entry point into mballoc to allocate blocks * it tries to use preallocation first, then falls back * to usual allocation @@ -4645,6 +6242,7 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle, ext4_fsblk_t block = 0; unsigned int inquota = 0; unsigned int reserv_clstrs = 0; + int retries = 0; u64 seq; might_sleep(); @@ -4652,6 +6250,8 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle, sbi = EXT4_SB(sb); trace_ext4_request_blocks(ar); + if (sbi->s_mount_state & EXT4_FC_REPLAY) + return ext4_mb_new_blocks_simple(ar, errp); /* Allow to use superuser reservation for quota file */ if (ext4_is_quota_file(ar->inode)) @@ -4701,14 +6301,10 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle, goto out; } - *errp = ext4_mb_initialize_context(ac, ar); - if (*errp) { - ar->len = 0; - goto out; - } + ext4_mb_initialize_context(ac, ar); ac->ac_op = EXT4_MB_HISTORY_PREALLOC; - seq = *this_cpu_ptr(&discard_pa_seq); + seq = this_cpu_read(discard_pa_seq); if (!ext4_mb_use_preallocated(ac)) { ac->ac_op = EXT4_MB_HISTORY_ALLOC; ext4_mb_normalize_request(ac, ar); @@ -4727,13 +6323,13 @@ repeat: * So we have to free this pa here itself. */ if (*errp) { - ext4_mb_pa_free(ac); + ext4_mb_pa_put_free(ac); ext4_discard_allocated_blocks(ac); goto errout; } if (ac->ac_status == AC_STATUS_FOUND && ac->ac_o_ex.fe_len >= ac->ac_f_ex.fe_len) - ext4_mb_pa_free(ac); + ext4_mb_pa_put_free(ac); } if (likely(ac->ac_status == AC_STATUS_FOUND)) { *errp = ext4_mb_mark_diskspace_used(ac, handle, reserv_clstrs); @@ -4745,26 +6341,26 @@ repeat: ar->len = ac->ac_b_ex.fe_len; } } else { - if (ext4_mb_discard_preallocations_should_retry(sb, ac, &seq)) + if (++retries < 3 && + ext4_mb_discard_preallocations_should_retry(sb, ac, &seq)) goto repeat; /* * If block allocation fails then the pa allocated above * needs to be freed here itself. */ - ext4_mb_pa_free(ac); + ext4_mb_pa_put_free(ac); *errp = -ENOSPC; } -errout: if (*errp) { +errout: ac->ac_b_ex.fe_len = 0; ar->len = 0; ext4_mb_show_ac(ac); } ext4_mb_release_context(ac); + kmem_cache_free(ext4_ac_cachep, ac); out: - if (ac) - kmem_cache_free(ext4_ac_cachep, ac); if (inquota && ar->len < inquota) dquot_free_block(ar->inode, EXT4_C2B(sbi, inquota - ar->len)); if (!ar->len) { @@ -4784,47 +6380,83 @@ out: * are contiguous, AND the extents were freed by the same transaction, * AND the blocks are associated with the same group. */ -static void ext4_try_merge_freed_extent(struct ext4_sb_info *sbi, - struct ext4_free_data *entry, - struct ext4_free_data *new_entry, - struct rb_root *entry_rb_root) +static inline bool +ext4_freed_extents_can_be_merged(struct ext4_free_data *entry1, + struct ext4_free_data *entry2) { - if ((entry->efd_tid != new_entry->efd_tid) || - (entry->efd_group != new_entry->efd_group)) - return; - if (entry->efd_start_cluster + entry->efd_count == - new_entry->efd_start_cluster) { - new_entry->efd_start_cluster = entry->efd_start_cluster; - new_entry->efd_count += entry->efd_count; - } else if (new_entry->efd_start_cluster + new_entry->efd_count == - entry->efd_start_cluster) { - new_entry->efd_count += entry->efd_count; - } else - return; + if (entry1->efd_tid != entry2->efd_tid) + return false; + if (entry1->efd_start_cluster + entry1->efd_count != + entry2->efd_start_cluster) + return false; + if (WARN_ON_ONCE(entry1->efd_group != entry2->efd_group)) + return false; + return true; +} + +static inline void +ext4_merge_freed_extents(struct ext4_sb_info *sbi, struct rb_root *root, + struct ext4_free_data *entry1, + struct ext4_free_data *entry2) +{ + entry1->efd_count += entry2->efd_count; spin_lock(&sbi->s_md_lock); - list_del(&entry->efd_list); + list_del(&entry2->efd_list); spin_unlock(&sbi->s_md_lock); - rb_erase(&entry->efd_node, entry_rb_root); - kmem_cache_free(ext4_free_data_cachep, entry); + rb_erase(&entry2->efd_node, root); + kmem_cache_free(ext4_free_data_cachep, entry2); } -static noinline_for_stack int +static inline void +ext4_try_merge_freed_extent_prev(struct ext4_sb_info *sbi, struct rb_root *root, + struct ext4_free_data *entry) +{ + struct ext4_free_data *prev; + struct rb_node *node; + + node = rb_prev(&entry->efd_node); + if (!node) + return; + + prev = rb_entry(node, struct ext4_free_data, efd_node); + if (ext4_freed_extents_can_be_merged(prev, entry)) + ext4_merge_freed_extents(sbi, root, prev, entry); +} + +static inline void +ext4_try_merge_freed_extent_next(struct ext4_sb_info *sbi, struct rb_root *root, + struct ext4_free_data *entry) +{ + struct ext4_free_data *next; + struct rb_node *node; + + node = rb_next(&entry->efd_node); + if (!node) + return; + + next = rb_entry(node, struct ext4_free_data, efd_node); + if (ext4_freed_extents_can_be_merged(entry, next)) + ext4_merge_freed_extents(sbi, root, entry, next); +} + +static noinline_for_stack void ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, struct ext4_free_data *new_entry) { ext4_group_t group = e4b->bd_group; ext4_grpblk_t cluster; ext4_grpblk_t clusters = new_entry->efd_count; - struct ext4_free_data *entry; + struct ext4_free_data *entry = NULL; struct ext4_group_info *db = e4b->bd_info; struct super_block *sb = e4b->bd_sb; struct ext4_sb_info *sbi = EXT4_SB(sb); - struct rb_node **n = &db->bb_free_root.rb_node, *node; + struct rb_root *root = &db->bb_free_root; + struct rb_node **n = &root->rb_node; struct rb_node *parent = NULL, *new_node; BUG_ON(!ext4_handle_valid(handle)); - BUG_ON(e4b->bd_bitmap_page == NULL); - BUG_ON(e4b->bd_buddy_page == NULL); + BUG_ON(e4b->bd_bitmap_folio == NULL); + BUG_ON(e4b->bd_buddy_folio == NULL); new_node = &new_entry->efd_node; cluster = new_entry->efd_start_cluster; @@ -4835,8 +6467,8 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, * otherwise we'll refresh it from * on-disk bitmap and lose not-yet-available * blocks */ - get_page(e4b->bd_buddy_page); - get_page(e4b->bd_bitmap_page); + folio_get(e4b->bd_buddy_folio); + folio_get(e4b->bd_bitmap_folio); } while (*n) { parent = *n; @@ -4850,137 +6482,94 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b, ext4_group_first_block_no(sb, group) + EXT4_C2B(sbi, cluster), "Block already on to-be-freed list"); - return 0; + kmem_cache_free(ext4_free_data_cachep, new_entry); + return; } } - rb_link_node(new_node, parent, n); - rb_insert_color(new_node, &db->bb_free_root); + atomic_add(clusters, &sbi->s_mb_free_pending); + if (!entry) + goto insert; - /* Now try to see the extent can be merged to left and right */ - node = rb_prev(new_node); - if (node) { - entry = rb_entry(node, struct ext4_free_data, efd_node); - ext4_try_merge_freed_extent(sbi, entry, new_entry, - &(db->bb_free_root)); + /* Now try to see the extent can be merged to prev and next */ + if (ext4_freed_extents_can_be_merged(new_entry, entry)) { + entry->efd_start_cluster = cluster; + entry->efd_count += new_entry->efd_count; + kmem_cache_free(ext4_free_data_cachep, new_entry); + ext4_try_merge_freed_extent_prev(sbi, root, entry); + return; } - - node = rb_next(new_node); - if (node) { - entry = rb_entry(node, struct ext4_free_data, efd_node); - ext4_try_merge_freed_extent(sbi, entry, new_entry, - &(db->bb_free_root)); + if (ext4_freed_extents_can_be_merged(entry, new_entry)) { + entry->efd_count += new_entry->efd_count; + kmem_cache_free(ext4_free_data_cachep, new_entry); + ext4_try_merge_freed_extent_next(sbi, root, entry); + return; } +insert: + rb_link_node(new_node, parent, n); + rb_insert_color(new_node, root); spin_lock(&sbi->s_md_lock); - list_add_tail(&new_entry->efd_list, &sbi->s_freed_data_list); - sbi->s_mb_free_pending += clusters; + list_add_tail(&new_entry->efd_list, &sbi->s_freed_data_list[new_entry->efd_tid & 1]); spin_unlock(&sbi->s_md_lock); - return 0; +} + +static void ext4_free_blocks_simple(struct inode *inode, ext4_fsblk_t block, + unsigned long count) +{ + struct super_block *sb = inode->i_sb; + ext4_group_t group; + ext4_grpblk_t blkoff; + + ext4_get_group_no_and_offset(sb, block, &group, &blkoff); + ext4_mb_mark_context(NULL, sb, false, group, blkoff, count, + EXT4_MB_BITMAP_MARKED_CHECK | + EXT4_MB_SYNC_UPDATE, + NULL); } /** - * ext4_free_blocks() -- Free given blocks and update quota + * ext4_mb_clear_bb() -- helper function for freeing blocks. + * Used by ext4_free_blocks() * @handle: handle for this transaction * @inode: inode - * @bh: optional buffer of the block to be freed * @block: starting physical block to be freed * @count: number of blocks to be freed * @flags: flags used by ext4_free_blocks */ -void ext4_free_blocks(handle_t *handle, struct inode *inode, - struct buffer_head *bh, ext4_fsblk_t block, - unsigned long count, int flags) +static void ext4_mb_clear_bb(handle_t *handle, struct inode *inode, + ext4_fsblk_t block, unsigned long count, + int flags) { - struct buffer_head *bitmap_bh = NULL; struct super_block *sb = inode->i_sb; - struct ext4_group_desc *gdp; + struct ext4_group_info *grp; unsigned int overflow; ext4_grpblk_t bit; - struct buffer_head *gd_bh; ext4_group_t block_group; struct ext4_sb_info *sbi; struct ext4_buddy e4b; unsigned int count_clusters; int err = 0; - int ret; - - might_sleep(); - if (bh) { - if (block) - BUG_ON(block != bh->b_blocknr); - else - block = bh->b_blocknr; - } + int mark_flags = 0; + ext4_grpblk_t changed; sbi = EXT4_SB(sb); - if (!(flags & EXT4_FREE_BLOCKS_VALIDATED) && - !ext4_data_block_valid(sbi, block, count)) { - ext4_error(sb, "Freeing blocks not in datazone - " - "block = %llu, count = %lu", block, count); - goto error_return; - } - - ext4_debug("freeing block %llu\n", block); - trace_ext4_free_blocks(inode, block, count, flags); - - if (bh && (flags & EXT4_FREE_BLOCKS_FORGET)) { - BUG_ON(count > 1); - - ext4_forget(handle, flags & EXT4_FREE_BLOCKS_METADATA, - inode, bh, block); - } - - /* - * If the extent to be freed does not begin on a cluster - * boundary, we need to deal with partial clusters at the - * beginning and end of the extent. Normally we will free - * blocks at the beginning or the end unless we are explicitly - * requested to avoid doing so. - */ - overflow = EXT4_PBLK_COFF(sbi, block); - if (overflow) { - if (flags & EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER) { - overflow = sbi->s_cluster_ratio - overflow; - block += overflow; - if (count > overflow) - count -= overflow; - else - return; - } else { - block -= overflow; - count += overflow; - } - } - overflow = EXT4_LBLK_COFF(sbi, count); - if (overflow) { - if (flags & EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER) { - if (count > overflow) - count -= overflow; - else - return; - } else - count += sbi->s_cluster_ratio - overflow; - } - - if (!bh && (flags & EXT4_FREE_BLOCKS_FORGET)) { - int i; - int is_metadata = flags & EXT4_FREE_BLOCKS_METADATA; - for (i = 0; i < count; i++) { - cond_resched(); - if (is_metadata) - bh = sb_find_get_block(inode->i_sb, block + i); - ext4_forget(handle, is_metadata, inode, bh, block + i); - } + if (!(flags & EXT4_FREE_BLOCKS_VALIDATED) && + !ext4_inode_block_valid(inode, block, count)) { + ext4_error(sb, "Freeing blocks in system zone - " + "Block = %llu, count = %lu", block, count); + /* err = 0. ext4_std_error should be a no op */ + goto error_out; } + flags |= EXT4_FREE_BLOCKS_VALIDATED; do_more: overflow = 0; ext4_get_group_no_and_offset(sb, block, &block_group, &bit); - if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT( - ext4_get_group_info(sb, block_group)))) + grp = ext4_get_group_info(sb, block_group); + if (unlikely(!grp || EXT4_MB_GRP_BBITMAP_CORRUPT(grp))) return; /* @@ -4991,61 +6580,39 @@ do_more: overflow = EXT4_C2B(sbi, bit) + count - EXT4_BLOCKS_PER_GROUP(sb); count -= overflow; + /* The range changed so it's no longer validated */ + flags &= ~EXT4_FREE_BLOCKS_VALIDATED; } count_clusters = EXT4_NUM_B2C(sbi, count); - bitmap_bh = ext4_read_block_bitmap(sb, block_group); - if (IS_ERR(bitmap_bh)) { - err = PTR_ERR(bitmap_bh); - bitmap_bh = NULL; - goto error_return; - } - gdp = ext4_get_group_desc(sb, block_group, &gd_bh); - if (!gdp) { - err = -EIO; - goto error_return; - } + trace_ext4_mballoc_free(sb, inode, block_group, bit, count_clusters); - if (in_range(ext4_block_bitmap(sb, gdp), block, count) || - in_range(ext4_inode_bitmap(sb, gdp), block, count) || - in_range(block, ext4_inode_table(sb, gdp), - sbi->s_itb_per_group) || - in_range(block + count - 1, ext4_inode_table(sb, gdp), - sbi->s_itb_per_group)) { + /* __GFP_NOFAIL: retry infinitely, ignore TIF_MEMDIE and memcg limit. */ + err = ext4_mb_load_buddy_gfp(sb, block_group, &e4b, + GFP_NOFS|__GFP_NOFAIL); + if (err) + goto error_out; + if (!(flags & EXT4_FREE_BLOCKS_VALIDATED) && + !ext4_inode_block_valid(inode, block, count)) { ext4_error(sb, "Freeing blocks in system zone - " "Block = %llu, count = %lu", block, count); /* err = 0. ext4_std_error should be a no op */ - goto error_return; + goto error_clean; } - BUFFER_TRACE(bitmap_bh, "getting write access"); - err = ext4_journal_get_write_access(handle, bitmap_bh); - if (err) - goto error_return; - - /* - * We are about to modify some metadata. Call the journal APIs - * to unshare ->b_data if a currently-committing transaction is - * using it - */ - BUFFER_TRACE(gd_bh, "get_write_access"); - err = ext4_journal_get_write_access(handle, gd_bh); - if (err) - goto error_return; #ifdef AGGRESSIVE_CHECK - { - int i; - for (i = 0; i < count_clusters; i++) - BUG_ON(!mb_test_bit(bit + i, bitmap_bh->b_data)); - } + mark_flags |= EXT4_MB_BITMAP_MARKED_CHECK; #endif - trace_ext4_mballoc_free(sb, inode, block_group, bit, count_clusters); + err = ext4_mb_mark_context(handle, sb, false, block_group, bit, + count_clusters, mark_flags, &changed); - /* __GFP_NOFAIL: retry infinitely, ignore TIF_MEMDIE and memcg limit. */ - err = ext4_mb_load_buddy_gfp(sb, block_group, &e4b, - GFP_NOFS|__GFP_NOFAIL); - if (err) - goto error_return; + + if (err && changed == 0) + goto error_clean; + +#ifdef AGGRESSIVE_CHECK + BUG_ON(changed != count_clusters); +#endif /* * We need to make sure we don't reuse the freed block until after the @@ -5069,42 +6636,32 @@ do_more: new_entry->efd_tid = handle->h_transaction->t_tid; ext4_lock_group(sb, block_group); - mb_clear_bits(bitmap_bh->b_data, bit, count_clusters); ext4_mb_free_metadata(handle, &e4b, new_entry); } else { - /* need to update group_info->bb_free and bitmap - * with group lock held. generate_buddy look at - * them with group lock_held - */ if (test_opt(sb, DISCARD)) { - err = ext4_issue_discard(sb, block_group, bit, count, - NULL); - if (err && err != -EOPNOTSUPP) + err = ext4_issue_discard(sb, block_group, bit, + count_clusters); + /* + * Ignore EOPNOTSUPP error. This is consistent with + * what happens when using journal. + */ + if (err == -EOPNOTSUPP) + err = 0; + if (err) ext4_msg(sb, KERN_WARNING, "discard request in" - " group:%d block:%d count:%lu failed" + " group:%u block:%d count:%lu failed" " with %d", block_group, bit, count, err); - } else - EXT4_MB_GRP_CLEAR_TRIMMED(e4b.bd_info); + } + + EXT4_MB_GRP_CLEAR_TRIMMED(e4b.bd_info); ext4_lock_group(sb, block_group); - mb_clear_bits(bitmap_bh->b_data, bit, count_clusters); mb_free_blocks(inode, &e4b, bit, count_clusters); } - ret = ext4_free_group_clusters(sb, gdp) + count_clusters; - ext4_free_group_clusters_set(sb, gdp, ret); - ext4_block_bitmap_csum_set(sb, block_group, gdp, bitmap_bh); - ext4_group_desc_csum_set(sb, block_group, gdp); ext4_unlock_group(sb, block_group); - if (sbi->s_log_groups_per_flex) { - ext4_group_t flex_group = ext4_flex_group(sbi, block_group); - atomic64_add(count_clusters, - &sbi_array_rcu_deref(sbi, s_flex_groups, - flex_group)->free_clusters); - } - /* * on a bigalloc file system, defer the s_freeclusters_counter * update to the caller (ext4_remove_space and friends) so they @@ -5117,28 +6674,122 @@ do_more: count_clusters); } - ext4_mb_unload_buddy(&e4b); - - /* We dirtied the bitmap block */ - BUFFER_TRACE(bitmap_bh, "dirtied bitmap block"); - err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); - - /* And the group descriptor block */ - BUFFER_TRACE(gd_bh, "dirtied group descriptor block"); - ret = ext4_handle_dirty_metadata(handle, NULL, gd_bh); - if (!err) - err = ret; - if (overflow && !err) { block += count; count = overflow; - put_bh(bitmap_bh); + ext4_mb_unload_buddy(&e4b); + /* The range changed so it's no longer validated */ + flags &= ~EXT4_FREE_BLOCKS_VALIDATED; goto do_more; } -error_return: - brelse(bitmap_bh); + +error_clean: + ext4_mb_unload_buddy(&e4b); +error_out: ext4_std_error(sb, err); - return; +} + +/** + * ext4_free_blocks() -- Free given blocks and update quota + * @handle: handle for this transaction + * @inode: inode + * @bh: optional buffer of the block to be freed + * @block: starting physical block to be freed + * @count: number of blocks to be freed + * @flags: flags used by ext4_free_blocks + */ +void ext4_free_blocks(handle_t *handle, struct inode *inode, + struct buffer_head *bh, ext4_fsblk_t block, + unsigned long count, int flags) +{ + struct super_block *sb = inode->i_sb; + unsigned int overflow; + struct ext4_sb_info *sbi; + + sbi = EXT4_SB(sb); + + if (bh) { + if (block) + BUG_ON(block != bh->b_blocknr); + else + block = bh->b_blocknr; + } + + if (sbi->s_mount_state & EXT4_FC_REPLAY) { + ext4_free_blocks_simple(inode, block, EXT4_NUM_B2C(sbi, count)); + return; + } + + might_sleep(); + + if (!(flags & EXT4_FREE_BLOCKS_VALIDATED) && + !ext4_inode_block_valid(inode, block, count)) { + ext4_error(sb, "Freeing blocks not in datazone - " + "block = %llu, count = %lu", block, count); + return; + } + flags |= EXT4_FREE_BLOCKS_VALIDATED; + + ext4_debug("freeing block %llu\n", block); + trace_ext4_free_blocks(inode, block, count, flags); + + if (bh && (flags & EXT4_FREE_BLOCKS_FORGET)) { + BUG_ON(count > 1); + + ext4_forget(handle, flags & EXT4_FREE_BLOCKS_METADATA, + inode, bh, block); + } + + /* + * If the extent to be freed does not begin on a cluster + * boundary, we need to deal with partial clusters at the + * beginning and end of the extent. Normally we will free + * blocks at the beginning or the end unless we are explicitly + * requested to avoid doing so. + */ + overflow = EXT4_PBLK_COFF(sbi, block); + if (overflow) { + if (flags & EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER) { + overflow = sbi->s_cluster_ratio - overflow; + block += overflow; + if (count > overflow) + count -= overflow; + else + return; + } else { + block -= overflow; + count += overflow; + } + /* The range changed so it's no longer validated */ + flags &= ~EXT4_FREE_BLOCKS_VALIDATED; + } + overflow = EXT4_LBLK_COFF(sbi, count); + if (overflow) { + if (flags & EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER) { + if (count > overflow) + count -= overflow; + else + return; + } else + count += sbi->s_cluster_ratio - overflow; + /* The range changed so it's no longer validated */ + flags &= ~EXT4_FREE_BLOCKS_VALIDATED; + } + + if (!bh && (flags & EXT4_FREE_BLOCKS_FORGET)) { + int i; + int is_metadata = flags & EXT4_FREE_BLOCKS_METADATA; + + for (i = 0; i < count; i++) { + cond_resched(); + if (is_metadata) + bh = sb_find_get_block_nonatomic(inode->i_sb, + block + i); + ext4_forget(handle, is_metadata, inode, bh, block + i); + } + } + + ext4_mb_clear_bb(handle, inode, block, count, flags); } /** @@ -5153,23 +6804,19 @@ error_return: int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, ext4_fsblk_t block, unsigned long count) { - struct buffer_head *bitmap_bh = NULL; - struct buffer_head *gd_bh; ext4_group_t block_group; ext4_grpblk_t bit; - unsigned int i; - struct ext4_group_desc *desc; struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_buddy e4b; - int err = 0, ret, free_clusters_count; - ext4_grpblk_t clusters_freed; + int err = 0; ext4_fsblk_t first_cluster = EXT4_B2C(sbi, block); ext4_fsblk_t last_cluster = EXT4_B2C(sbi, block + count - 1); unsigned long cluster_count = last_cluster - first_cluster + 1; + ext4_grpblk_t changed; ext4_debug("Adding block(s) %llu-%llu\n", block, block + count - 1); - if (count == 0) + if (cluster_count == 0) return 0; ext4_get_group_no_and_offset(sb, block, &block_group, &bit); @@ -5181,102 +6828,39 @@ int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, ext4_warning(sb, "too many blocks added to group %u", block_group); err = -EINVAL; - goto error_return; + goto error_out; } - bitmap_bh = ext4_read_block_bitmap(sb, block_group); - if (IS_ERR(bitmap_bh)) { - err = PTR_ERR(bitmap_bh); - bitmap_bh = NULL; - goto error_return; - } - - desc = ext4_get_group_desc(sb, block_group, &gd_bh); - if (!desc) { - err = -EIO; - goto error_return; - } + err = ext4_mb_load_buddy(sb, block_group, &e4b); + if (err) + goto error_out; - if (in_range(ext4_block_bitmap(sb, desc), block, count) || - in_range(ext4_inode_bitmap(sb, desc), block, count) || - in_range(block, ext4_inode_table(sb, desc), sbi->s_itb_per_group) || - in_range(block + count - 1, ext4_inode_table(sb, desc), - sbi->s_itb_per_group)) { + if (!ext4_sb_block_valid(sb, NULL, block, count)) { ext4_error(sb, "Adding blocks in system zones - " "Block = %llu, count = %lu", block, count); err = -EINVAL; - goto error_return; + goto error_clean; } - BUFFER_TRACE(bitmap_bh, "getting write access"); - err = ext4_journal_get_write_access(handle, bitmap_bh); - if (err) - goto error_return; - - /* - * We are about to modify some metadata. Call the journal APIs - * to unshare ->b_data if a currently-committing transaction is - * using it - */ - BUFFER_TRACE(gd_bh, "get_write_access"); - err = ext4_journal_get_write_access(handle, gd_bh); - if (err) - goto error_return; - - for (i = 0, clusters_freed = 0; i < cluster_count; i++) { - BUFFER_TRACE(bitmap_bh, "clear bit"); - if (!mb_test_bit(bit + i, bitmap_bh->b_data)) { - ext4_error(sb, "bit already cleared for block %llu", - (ext4_fsblk_t)(block + i)); - BUFFER_TRACE(bitmap_bh, "bit already cleared"); - } else { - clusters_freed++; - } - } + err = ext4_mb_mark_context(handle, sb, false, block_group, bit, + cluster_count, EXT4_MB_BITMAP_MARKED_CHECK, + &changed); + if (err && changed == 0) + goto error_clean; - err = ext4_mb_load_buddy(sb, block_group, &e4b); - if (err) - goto error_return; + if (changed != cluster_count) + ext4_error(sb, "bit already cleared in group %u", block_group); - /* - * need to update group_info->bb_free and bitmap - * with group lock held. generate_buddy look at - * them with group lock_held - */ ext4_lock_group(sb, block_group); - mb_clear_bits(bitmap_bh->b_data, bit, cluster_count); mb_free_blocks(NULL, &e4b, bit, cluster_count); - free_clusters_count = clusters_freed + - ext4_free_group_clusters(sb, desc); - ext4_free_group_clusters_set(sb, desc, free_clusters_count); - ext4_block_bitmap_csum_set(sb, block_group, desc, bitmap_bh); - ext4_group_desc_csum_set(sb, block_group, desc); ext4_unlock_group(sb, block_group); percpu_counter_add(&sbi->s_freeclusters_counter, - clusters_freed); - - if (sbi->s_log_groups_per_flex) { - ext4_group_t flex_group = ext4_flex_group(sbi, block_group); - atomic64_add(clusters_freed, - &sbi_array_rcu_deref(sbi, s_flex_groups, - flex_group)->free_clusters); - } + changed); +error_clean: ext4_mb_unload_buddy(&e4b); - - /* We dirtied the bitmap block */ - BUFFER_TRACE(bitmap_bh, "dirtied bitmap block"); - err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); - - /* And the group descriptor block */ - BUFFER_TRACE(gd_bh, "dirtied group descriptor block"); - ret = ext4_handle_dirty_metadata(handle, NULL, gd_bh); - if (!err) - err = ret; - -error_return: - brelse(bitmap_bh); +error_out: ext4_std_error(sb, err); return err; } @@ -5286,19 +6870,19 @@ error_return: * @sb: super block for the file system * @start: starting block of the free extent in the alloc. group * @count: number of blocks to TRIM - * @group: alloc. group we are working with * @e4b: ext4 buddy for the group * * Trim "count" blocks starting at "start" in the "group". To assure that no * one will allocate those blocks, mark it as used in buddy bitmap. This must * be called with under the group lock. */ -static int ext4_trim_extent(struct super_block *sb, int start, int count, - ext4_group_t group, struct ext4_buddy *e4b) +static int ext4_trim_extent(struct super_block *sb, + int start, int count, struct ext4_buddy *e4b) __releases(bitlock) __acquires(bitlock) { struct ext4_free_extent ex; + ext4_group_t group = e4b->bd_group; int ret = 0; trace_ext4_trim_extent(sb, group, start, count); @@ -5315,12 +6899,92 @@ __acquires(bitlock) */ mb_mark_used(e4b, &ex); ext4_unlock_group(sb, group); - ret = ext4_issue_discard(sb, group, start, count, NULL); + ret = ext4_issue_discard(sb, group, start, count); ext4_lock_group(sb, group); mb_free_blocks(NULL, e4b, start, ex.fe_len); return ret; } +static ext4_grpblk_t ext4_last_grp_cluster(struct super_block *sb, + ext4_group_t grp) +{ + unsigned long nr_clusters_in_group; + + if (grp < (ext4_get_groups_count(sb) - 1)) + nr_clusters_in_group = EXT4_CLUSTERS_PER_GROUP(sb); + else + nr_clusters_in_group = (ext4_blocks_count(EXT4_SB(sb)->s_es) - + ext4_group_first_block_no(sb, grp)) + >> EXT4_CLUSTER_BITS(sb); + + return nr_clusters_in_group - 1; +} + +static bool ext4_trim_interrupted(void) +{ + return fatal_signal_pending(current) || freezing(current); +} + +static int ext4_try_to_trim_range(struct super_block *sb, + struct ext4_buddy *e4b, ext4_grpblk_t start, + ext4_grpblk_t max, ext4_grpblk_t minblocks) +__acquires(ext4_group_lock_ptr(sb, e4b->bd_group)) +__releases(ext4_group_lock_ptr(sb, e4b->bd_group)) +{ + ext4_grpblk_t next, count, free_count, last, origin_start; + bool set_trimmed = false; + void *bitmap; + + if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) + return 0; + + last = ext4_last_grp_cluster(sb, e4b->bd_group); + bitmap = e4b->bd_bitmap; + if (start == 0 && max >= last) + set_trimmed = true; + origin_start = start; + start = max(e4b->bd_info->bb_first_free, start); + count = 0; + free_count = 0; + + while (start <= max) { + start = mb_find_next_zero_bit(bitmap, max + 1, start); + if (start > max) + break; + + next = mb_find_next_bit(bitmap, last + 1, start); + if (origin_start == 0 && next >= last) + set_trimmed = true; + + if ((next - start) >= minblocks) { + int ret = ext4_trim_extent(sb, start, next - start, e4b); + + if (ret && ret != -EOPNOTSUPP) + return count; + count += next - start; + } + free_count += next - start; + start = next + 1; + + if (ext4_trim_interrupted()) + return count; + + if (need_resched()) { + ext4_unlock_group(sb, e4b->bd_group); + cond_resched(); + ext4_lock_group(sb, e4b->bd_group); + } + + if ((e4b->bd_info->bb_free - free_count) < minblocks) + break; + } + + if (set_trimmed) + EXT4_MB_GRP_SET_TRIMMED(e4b->bd_info); + + return count; +} + /** * ext4_trim_all_free -- function to trim all free space in alloc. group * @sb: super block for file system @@ -5329,25 +6993,18 @@ __acquires(bitlock) * @max: last group block to examine * @minblocks: minimum extent block count * - * ext4_trim_all_free walks through group's buddy bitmap searching for free - * extents. When the free block is found, ext4_trim_extent is called to TRIM - * the extent. - * - * * ext4_trim_all_free walks through group's block bitmap searching for free * extents. When the free extent is found, mark it as used in group buddy * bitmap. Then issue a TRIM command on this extent and free the extent in - * the group buddy bitmap. This is done until whole group is scanned. + * the group buddy bitmap. */ static ext4_grpblk_t ext4_trim_all_free(struct super_block *sb, ext4_group_t group, ext4_grpblk_t start, ext4_grpblk_t max, ext4_grpblk_t minblocks) { - void *bitmap; - ext4_grpblk_t next, count = 0, free_count = 0; struct ext4_buddy e4b; - int ret = 0; + int ret; trace_ext4_trim_all_free(sb, group, start, max); @@ -5357,58 +7014,20 @@ ext4_trim_all_free(struct super_block *sb, ext4_group_t group, ret, group); return ret; } - bitmap = e4b.bd_bitmap; ext4_lock_group(sb, group); - if (EXT4_MB_GRP_WAS_TRIMMED(e4b.bd_info) && - minblocks >= atomic_read(&EXT4_SB(sb)->s_last_trim_minblks)) - goto out; - - start = (e4b.bd_info->bb_first_free > start) ? - e4b.bd_info->bb_first_free : start; - - while (start <= max) { - start = mb_find_next_zero_bit(bitmap, max + 1, start); - if (start > max) - break; - next = mb_find_next_bit(bitmap, max + 1, start); - - if ((next - start) >= minblocks) { - ret = ext4_trim_extent(sb, start, - next - start, group, &e4b); - if (ret && ret != -EOPNOTSUPP) - break; - ret = 0; - count += next - start; - } - free_count += next - start; - start = next + 1; - - if (fatal_signal_pending(current)) { - count = -ERESTARTSYS; - break; - } - - if (need_resched()) { - ext4_unlock_group(sb, group); - cond_resched(); - ext4_lock_group(sb, group); - } - if ((e4b.bd_info->bb_free - free_count) < minblocks) - break; - } + if (!EXT4_MB_GRP_WAS_TRIMMED(e4b.bd_info) || + minblocks < EXT4_SB(sb)->s_last_trim_minblks) + ret = ext4_try_to_trim_range(sb, &e4b, start, max, minblocks); + else + ret = 0; - if (!ret) { - ret = count; - EXT4_MB_GRP_SET_TRIMMED(e4b.bd_info); - } -out: ext4_unlock_group(sb, group); ext4_mb_unload_buddy(&e4b); ext4_debug("trimmed %d blocks in the group %d\n", - count, group); + ret, group); return ret; } @@ -5427,6 +7046,7 @@ out: */ int ext4_trim_fs(struct super_block *sb, struct fstrim_range *range) { + unsigned int discard_granularity = bdev_discard_granularity(sb->s_bdev); struct ext4_group_info *grp; ext4_group_t group, first_group, last_group; ext4_grpblk_t cnt = 0, first_cluster, last_cluster; @@ -5445,7 +7065,14 @@ int ext4_trim_fs(struct super_block *sb, struct fstrim_range *range) start >= max_blks || range->len < sb->s_blocksize) return -EINVAL; - if (end >= max_blks) + /* No point to try to trim less than discard granularity */ + if (range->minlen < discard_granularity) { + minlen = EXT4_NUM_B2C(EXT4_SB(sb), + discard_granularity >> sb->s_blocksize_bits); + if (minlen > EXT4_CLUSTERS_PER_GROUP(sb)) + goto out; + } + if (end >= max_blks - 1) end = max_blks - 1; if (end <= first_data_blk) goto out; @@ -5462,7 +7089,11 @@ int ext4_trim_fs(struct super_block *sb, struct fstrim_range *range) end = EXT4_CLUSTERS_PER_GROUP(sb) - 1; for (group = first_group; group <= last_group; group++) { + if (ext4_trim_interrupted()) + break; grp = ext4_get_group_info(sb, group); + if (!grp) + continue; /* We only do this if the grp has never been initialized */ if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) { ret = ext4_mb_init_group(sb, group, GFP_NOFS); @@ -5478,10 +7109,9 @@ int ext4_trim_fs(struct super_block *sb, struct fstrim_range *range) */ if (group == last_group) end = last_cluster; - if (grp->bb_free >= minlen) { cnt = ext4_trim_all_free(sb, group, first_cluster, - end, minlen); + end, minlen); if (cnt < 0) { ret = cnt; break; @@ -5497,7 +7127,7 @@ int ext4_trim_fs(struct super_block *sb, struct fstrim_range *range) } if (!ret) - atomic_set(&EXT4_SB(sb)->s_last_trim_minblks, minlen); + EXT4_SB(sb)->s_last_trim_minblks = minlen; out: range->len = EXT4_C2B(EXT4_SB(sb), trimmed) << sb->s_blocksize_bits; @@ -5509,13 +7139,14 @@ int ext4_mballoc_query_range( struct super_block *sb, ext4_group_t group, - ext4_grpblk_t start, + ext4_grpblk_t first, ext4_grpblk_t end, + ext4_mballoc_query_range_fn meta_formatter, ext4_mballoc_query_range_fn formatter, void *priv) { void *bitmap; - ext4_grpblk_t next; + ext4_grpblk_t start, next; struct ext4_buddy e4b; int error; @@ -5526,11 +7157,19 @@ ext4_mballoc_query_range( ext4_lock_group(sb, group); - start = (e4b.bd_info->bb_first_free > start) ? - e4b.bd_info->bb_first_free : start; + start = max(e4b.bd_info->bb_first_free, first); if (end >= EXT4_CLUSTERS_PER_GROUP(sb)) end = EXT4_CLUSTERS_PER_GROUP(sb) - 1; - + if (meta_formatter && start != first) { + if (start > end) + start = end; + ext4_unlock_group(sb, group); + error = meta_formatter(sb, group, first, start - first, + priv); + if (error) + goto out_unload; + ext4_lock_group(sb, group); + } while (start <= end) { start = mb_find_next_zero_bit(bitmap, end + 1, start); if (start > end) @@ -5552,3 +7191,7 @@ out_unload: return error; } + +#ifdef CONFIG_EXT4_KUNIT_TESTS +#include "mballoc-test.c" +#endif |