-Index: linux-2.6.9-42.0.10.EL_lustre.1.4.10/include/linux/ext3_fs_sb.h
+Index: linux-2.6.9-full/include/linux/ext3_fs.h
===================================================================
---- linux-2.6.9-42.0.10.EL_lustre.1.4.10.orig/include/linux/ext3_fs_sb.h 2007-06-14 13:59:04.000000000 +0200
-+++ linux-2.6.9-42.0.10.EL_lustre.1.4.10/include/linux/ext3_fs_sb.h 2007-06-14 14:16:57.000000000 +0200
-@@ -81,6 +81,61 @@ struct ext3_sb_info {
- char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
- int s_jquota_fmt; /* Format of quota to use */
- #endif
-+
-+ /* for buddy allocator */
-+ struct ext3_group_info ***s_group_info;
-+ struct inode *s_buddy_cache;
-+ long s_blocks_reserved;
-+ spinlock_t s_reserve_lock;
-+ struct list_head s_active_transaction;
-+ struct list_head s_closed_transaction;
-+ struct list_head s_committed_transaction;
-+ spinlock_t s_md_lock;
-+ tid_t s_last_transaction;
-+ unsigned short *s_mb_offsets, *s_mb_maxs;
-+
-+ /* tunables */
-+ unsigned long s_mb_factor;
-+ unsigned long s_stripe;
-+ unsigned long s_mb_stream_request;
-+ unsigned long s_mb_max_to_scan;
-+ unsigned long s_mb_min_to_scan;
-+ unsigned long s_mb_max_groups_to_scan;
-+ unsigned long s_mb_stats;
-+ unsigned long s_mb_order2_reqs;
-+
-+ /* history to debug policy */
-+ struct ext3_mb_history *s_mb_history;
-+ int s_mb_history_cur;
-+ int s_mb_history_max;
-+ int s_mb_history_num;
-+ struct proc_dir_entry *s_mb_proc;
-+ spinlock_t s_mb_history_lock;
-+ int s_mb_history_filter;
-+
-+ /* stats for buddy allocator */
-+ spinlock_t s_mb_pa_lock;
-+ atomic_t s_bal_reqs; /* number of reqs with len > 1 */
-+ atomic_t s_bal_success; /* we found long enough chunks */
-+ atomic_t s_bal_allocated; /* in blocks */
-+ atomic_t s_bal_ex_scanned; /* total extents scanned */
-+ atomic_t s_bal_goals; /* goal hits */
-+ atomic_t s_bal_breaks; /* too long searches */
-+ atomic_t s_bal_2orders; /* 2^order hits */
-+ spinlock_t s_bal_lock;
-+ unsigned long s_mb_buddies_generated;
-+ unsigned long long s_mb_generation_time;
-+ atomic_t s_mb_lost_chunks;
-+ atomic_t s_mb_preallocated;
-+ atomic_t s_mb_discarded;
-+
-+ /* locality groups */
-+ struct ext3_locality_group *s_locality_groups;
-+
- };
-
-+#define EXT3_GROUP_INFO(sb, group) \
-+ EXT3_SB(sb)->s_group_info[(group) >> EXT3_DESC_PER_BLOCK_BITS(sb)] \
-+ [(group) & (EXT3_DESC_PER_BLOCK(sb) - 1)]
-+
- #endif /* _LINUX_EXT3_FS_SB */
-Index: linux-2.6.9-42.0.10.EL_lustre.1.4.10/include/linux/ext3_fs.h
-===================================================================
---- linux-2.6.9-42.0.10.EL_lustre.1.4.10.orig/include/linux/ext3_fs.h 2007-06-14 13:59:04.000000000 +0200
-+++ linux-2.6.9-42.0.10.EL_lustre.1.4.10/include/linux/ext3_fs.h 2007-06-14 14:16:57.000000000 +0200
+--- linux-2.6.9-full.orig/include/linux/ext3_fs.h 2007-06-08 23:44:08.000000000 +0400
++++ linux-2.6.9-full/include/linux/ext3_fs.h 2007-10-17 22:25:01.000000000 +0400
@@ -57,6 +57,30 @@ struct statfs;
#define ext3_debug(f, a...) do {} while (0)
#endif
/* inode.c */
extern int ext3_block_truncate_page(handle_t *, struct page *,
-Index: linux-2.6.9-42.0.10.EL_lustre.1.4.10/fs/ext3/super.c
+Index: linux-2.6.9-full/include/linux/ext3_fs_sb.h
===================================================================
---- linux-2.6.9-42.0.10.EL_lustre.1.4.10.orig/fs/ext3/super.c 2007-06-14 13:59:04.000000000 +0200
-+++ linux-2.6.9-42.0.10.EL_lustre.1.4.10/fs/ext3/super.c 2007-06-14 14:16:57.000000000 +0200
+--- linux-2.6.9-full.orig/include/linux/ext3_fs_sb.h 2007-06-08 23:44:07.000000000 +0400
++++ linux-2.6.9-full/include/linux/ext3_fs_sb.h 2007-10-17 22:25:01.000000000 +0400
+@@ -81,6 +81,61 @@ struct ext3_sb_info {
+ char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
+ int s_jquota_fmt; /* Format of quota to use */
+ #endif
++
++ /* for buddy allocator */
++ struct ext3_group_info ***s_group_info;
++ struct inode *s_buddy_cache;
++ long s_blocks_reserved;
++ spinlock_t s_reserve_lock;
++ struct list_head s_active_transaction;
++ struct list_head s_closed_transaction;
++ struct list_head s_committed_transaction;
++ spinlock_t s_md_lock;
++ tid_t s_last_transaction;
++ unsigned short *s_mb_offsets, *s_mb_maxs;
++
++ /* tunables */
++ unsigned long s_mb_factor;
++ unsigned long s_stripe;
++ unsigned long s_mb_stream_request;
++ unsigned long s_mb_max_to_scan;
++ unsigned long s_mb_min_to_scan;
++ unsigned long s_mb_max_groups_to_scan;
++ unsigned long s_mb_stats;
++ unsigned long s_mb_order2_reqs;
++
++ /* history to debug policy */
++ struct ext3_mb_history *s_mb_history;
++ int s_mb_history_cur;
++ int s_mb_history_max;
++ int s_mb_history_num;
++ struct proc_dir_entry *s_mb_proc;
++ spinlock_t s_mb_history_lock;
++ int s_mb_history_filter;
++
++ /* stats for buddy allocator */
++ spinlock_t s_mb_pa_lock;
++ atomic_t s_bal_reqs; /* number of reqs with len > 1 */
++ atomic_t s_bal_success; /* we found long enough chunks */
++ atomic_t s_bal_allocated; /* in blocks */
++ atomic_t s_bal_ex_scanned; /* total extents scanned */
++ atomic_t s_bal_goals; /* goal hits */
++ atomic_t s_bal_breaks; /* too long searches */
++ atomic_t s_bal_2orders; /* 2^order hits */
++ spinlock_t s_bal_lock;
++ unsigned long s_mb_buddies_generated;
++ unsigned long long s_mb_generation_time;
++ atomic_t s_mb_lost_chunks;
++ atomic_t s_mb_preallocated;
++ atomic_t s_mb_discarded;
++
++ /* locality groups */
++ struct ext3_locality_group *s_locality_groups;
++
+ };
+
++#define EXT3_GROUP_INFO(sb, group) \
++ EXT3_SB(sb)->s_group_info[(group) >> EXT3_DESC_PER_BLOCK_BITS(sb)] \
++ [(group) & (EXT3_DESC_PER_BLOCK(sb) - 1)]
++
+ #endif /* _LINUX_EXT3_FS_SB */
+Index: linux-2.6.9-full/fs/ext3/super.c
+===================================================================
+--- linux-2.6.9-full.orig/fs/ext3/super.c 2007-06-08 23:44:08.000000000 +0400
++++ linux-2.6.9-full/fs/ext3/super.c 2007-10-17 22:26:27.000000000 +0400
@@ -394,6 +394,7 @@ void ext3_put_super (struct super_block
struct ext3_super_block *es = sbi->s_es;
int i;
journal_destroy(sbi->s_journal);
@@ -463,6 +464,8 @@ static struct inode *ext3_alloc_inode(st
ei->vfs_inode.i_version = 1;
-
+
memset(&ei->i_cached_extent, 0, sizeof(ei->i_cached_extent));
+ INIT_LIST_HEAD(&ei->i_prealloc_list);
+ spin_lock_init(&ei->i_prealloc_lock);
}
int ext3_prep_san_write(struct inode *inode, long *blocks,
-Index: linux-2.6.9-42.0.10.EL_lustre.1.4.10/fs/ext3/mballoc.c
+Index: linux-2.6.9-full/fs/ext3/mballoc.c
===================================================================
---- /dev/null 1970-01-01 00:00:00.000000000 +0000
-+++ linux-2.6.9-42.0.10.EL_lustre.1.4.10/fs/ext3/mballoc.c 2007-06-14 14:19:27.000000000 +0200
-@@ -0,0 +1,4369 @@
+--- linux-2.6.9-full.orig/fs/ext3/mballoc.c 2007-10-17 21:59:51.072534980 +0400
++++ linux-2.6.9-full/fs/ext3/mballoc.c 2007-10-17 23:09:22.000000000 +0400
+@@ -0,0 +1,4404 @@
+/*
+ * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
+ * Written by Alex Tomas <alex@clusterfs.com>
+ ext2_set_bit(bit, addr);
+}
+
-+static inline void mb_set_bit_atomic(int bit, void *addr)
++static inline void mb_set_bit_atomic(spinlock_t *lock, int bit, void *addr)
+{
+ mb_correct_addr_and_bit(bit,addr);
-+ ext2_set_bit_atomic(NULL, bit, addr);
++ ext2_set_bit_atomic(lock, bit, addr);
+}
+
+static inline void mb_clear_bit(int bit, void *addr)
+ ext2_clear_bit(bit, addr);
+}
+
-+static inline void mb_clear_bit_atomic(int bit, void *addr)
++static inline void mb_clear_bit_atomic(spinlock_t *lock, int bit, void *addr)
+{
+ mb_correct_addr_and_bit(bit,addr);
-+ ext2_clear_bit_atomic(NULL, bit, addr);
++ ext2_clear_bit_atomic(lock, bit, addr);
+}
+
+static inline int mb_find_next_zero_bit(void *addr, int max, int start)
+ return 0;
+}
+
-+static inline void mb_clear_bits(void *bm, int cur, int len)
++static inline void mb_clear_bits(spinlock_t *lock, void *bm, int cur, int len)
+{
+ __u32 *addr;
+
+ cur += 32;
+ continue;
+ }
-+ mb_clear_bit_atomic(cur, bm);
++ mb_clear_bit_atomic(lock, cur, bm);
+ cur++;
+ }
+}
+
-+static inline void mb_set_bits(void *bm, int cur, int len)
++static inline void mb_set_bits(spinlock_t *lock, void *bm, int cur, int len)
+{
+ __u32 *addr;
+
+ cur += 32;
+ continue;
+ }
-+ mb_set_bit_atomic(cur, bm);
++ mb_set_bit_atomic(lock, cur, bm);
+ cur++;
+ }
+}
+ e3b->bd_info->bb_counters[ord]++;
+ }
+
-+ mb_set_bits(EXT3_MB_BITMAP(e3b), ex->fe_start, len0);
++ mb_set_bits(sb_bgl_lock(EXT3_SB(e3b->bd_sb), ex->fe_group),
++ EXT3_MB_BITMAP(e3b), ex->fe_start, len0);
+ mb_check_buddy(e3b);
+
+ return ret;
+ struct ext3_free_extent *gex = &ac->ac_g_ex;
+
+ BUG_ON(ex->fe_len <= 0);
-+ BUG_ON(ex->fe_len >= (1 << ac->ac_sb->s_blocksize_bits) * 8);
-+ BUG_ON(ex->fe_start >= (1 << ac->ac_sb->s_blocksize_bits) * 8);
++ BUG_ON(ex->fe_len >= EXT3_BLOCKS_PER_GROUP(ac->ac_sb));
++ BUG_ON(ex->fe_start >= EXT3_BLOCKS_PER_GROUP(ac->ac_sb));
+ BUG_ON(ac->ac_status != AC_STATUS_CONTINUE);
+
+ ac->ac_found++;
+ i = e3b->bd_info->bb_first_free;
+
+ while (free && ac->ac_status == AC_STATUS_CONTINUE) {
-+ i = mb_find_next_zero_bit(bitmap, sb->s_blocksize * 8, i);
-+ if (i >= sb->s_blocksize * 8) {
++ i = mb_find_next_zero_bit(bitmap, EXT3_BLOCKS_PER_GROUP(sb), i);
++ if (i >= EXT3_BLOCKS_PER_GROUP(sb)) {
+ BUG_ON(free != 0);
+ break;
+ }
+ i = (i - le32_to_cpu(sbi->s_es->s_first_data_block))
+ % EXT3_BLOCKS_PER_GROUP(sb);
+
-+ while (i < sb->s_blocksize * 8) {
++ while (i < EXT3_BLOCKS_PER_GROUP(sb)) {
+ if (!mb_test_bit(i, bitmap)) {
+ max = mb_find_extent(e3b, 0, i, sbi->s_stripe, &ex);
+ if (max >= sbi->s_stripe) {
+ spin_lock_init(&sbi->s_mb_history_lock);
+ i = sbi->s_mb_history_max * sizeof(struct ext3_mb_history);
+ sbi->s_mb_history = kmalloc(i, GFP_KERNEL);
-+ memset(sbi->s_mb_history, 0, i);
++ if (likely(sbi->s_mb_history != NULL))
++ memset(sbi->s_mb_history, 0, i);
+ /* if we can't allocate history, then we simple won't use it */
+}
+
+ struct ext3_sb_info *sbi = EXT3_SB(ac->ac_sb);
+ struct ext3_mb_history h;
+
-+ if (likely(sbi->s_mb_history == NULL))
++ if (unlikely(sbi->s_mb_history == NULL))
+ return;
+
+ if (!(ac->ac_op & sbi->s_mb_history_filter))
+ struct ext3_sb_info *sbi;
+ struct super_block *sb;
+ sector_t block;
-+ int len, err;
++ int err;
+
+ BUG_ON(ac->ac_status != AC_STATUS_FOUND);
+ BUG_ON(ac->ac_b_ex.fe_len <= 0);
+ ext3_debug("using block group %d(%d)\n", ac->ac_b_group.group,
+ gdp->bg_free_blocks_count);
+
-+ /* time to check quota, we can't do this before because
-+ * having quota spent on preallocated-unused-yet blocks
-+ * would be wrong */
-+ len = ac->ac_b_ex.fe_len;
-+ while (len && DQUOT_ALLOC_BLOCK(ac->ac_inode, len)) len--;
-+ if (ac->ac_b_ex.fe_len != len) {
-+ /* some blocks can't be allocated due to quota
-+ * we have to return them back */
-+ BUG();
-+ }
-+ err = -EDQUOT;
-+ if (len == 0)
-+ goto out_err;
-+
+ err = -EIO;
+ bitmap_bh = read_block_bitmap(sb, ac->ac_b_ex.fe_group);
+ if (!bitmap_bh)
+ }
+ }
+#endif
-+ mb_set_bits(bitmap_bh->b_data, ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len);
++ mb_set_bits(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group), bitmap_bh->b_data,
++ ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len);
+
+ spin_lock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
+ gdp->bg_free_blocks_count =
+ struct ext3_inode_info *ei = EXT3_I(ac->ac_inode);
+ loff_t start, end, size, orig_size, orig_start;
+ struct list_head *cur;
-+ int bsbits;
++ int bsbits, max;
+
+ /* do normalize only data requests, metadata requests
+ do not need preallocation */
+ if (size < i_size_read(ac->ac_inode))
+ size = i_size_read(ac->ac_inode);
+
++ /* max available blocks in a free group */
++ max = EXT3_BLOCKS_PER_GROUP(ac->ac_sb) - 1 - 1
++ - EXT3_SB(ac->ac_sb)->s_itb_per_group;
++
++#define NRL_CHECK_SIZE(req,size,max,bits) \
++ (req <= (size) || max <= ((size) >> bits))
++
+ /* first, try to predict filesize */
+ /* XXX: should this table be tunable? */
+ start = 0;
+ size = 512 * 1024;
+ } else if (size <= 1024 * 1024) {
+ size = 1024 * 1024;
-+ } else if (size < 4 * 1024 * 1024) {
++ } else if (NRL_CHECK_SIZE(size, 4 * 1024 * 1024, max, bsbits)) {
+ start = ac->ac_o_ex.fe_logical << bsbits;
+ start = (start / (1024 * 1024)) * (1024 * 1024);
+ size = 1024 * 1024;
-+ } else if (size < 8 * 1024 * 1024) {
++ } else if (NRL_CHECK_SIZE(size, 8 * 1024 * 1024, max, bsbits)) {
+ start = ac->ac_o_ex.fe_logical << bsbits;
+ start = (start / (4 * (1024 * 1024))) * 4 * (1024 * 1024);
+ size = 4 * 1024 * 1024;
-+ } else if (ac->ac_o_ex.fe_len < ((8 << 20) >> bsbits)) {
++ } else if(NRL_CHECK_SIZE(ac->ac_o_ex.fe_len,(8<<20)>>bsbits,max,bsbits)){
+ start = ac->ac_o_ex.fe_logical;
+ start = start << bsbits;
+ start = (start / (8 * (1024 * 1024))) * 8 * (1024 * 1024);
+ unsigned long pa_end;
+
+ pa = list_entry(cur, struct ext3_prealloc_space, pa_inode_list);
++
++ if (pa->pa_deleted)
++ continue;
++ spin_lock(&pa->pa_lock);
++ if (pa->pa_deleted) {
++ spin_unlock(&pa->pa_lock);
++ continue;
++ }
++
+ pa_end = pa->pa_lstart + pa->pa_len;
+
+ /* PA must not overlap original request */
+ ac->ac_o_ex.fe_logical < pa->pa_lstart));
+
+ /* skip PA normalized request doesn't overlap with */
-+ if (pa->pa_lstart >= end)
++ if (pa->pa_lstart >= end) {
++ spin_unlock(&pa->pa_lock);
+ continue;
-+ if (pa_end <= start)
++ }
++ if (pa_end <= start) {
++ spin_unlock(&pa->pa_lock);
+ continue;
++ }
+ BUG_ON(pa->pa_lstart <= start && pa_end >= end);
+
+ if (pa_end <= 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;
+ struct ext3_prealloc_space *pa;
+ unsigned long pa_end;
+ pa = list_entry(cur, struct ext3_prealloc_space, pa_inode_list);
-+ pa_end = pa->pa_lstart + pa->pa_len;
-+ BUG_ON(!(start >= pa_end || end <= pa->pa_lstart));
++ spin_lock(&pa->pa_lock);
++ if (pa->pa_deleted == 0) {
++ pa_end = pa->pa_lstart + pa->pa_len;
++ BUG_ON(!(start >= pa_end || end <= pa->pa_lstart));
++ }
++ spin_unlock(&pa->pa_lock);
+ }
+ rcu_read_unlock();
+
+ ext3_get_group_no_and_offset(sb, pa->pa_pstart, &groupnr, &start);
+ len = pa->pa_len;
+ spin_unlock(&pa->pa_lock);
-+ BUG_ON(groupnr != group);
-+ mb_set_bits(bitmap, start, len);
++ if (unlikely(len == 0))
++ continue;
++ BUG_ON(groupnr != group && len != 0);
++ mb_set_bits(sb_bgl_lock(EXT3_SB(sb), group), bitmap, start,len);
+ preallocated += len;
+ count++;
+ }
+ BUG_ON(ac->ac_status != AC_STATUS_FOUND);
+ BUG_ON(!S_ISREG(ac->ac_inode->i_mode));
+
-+ pa = kmem_cache_alloc(ext3_pspace_cachep, SLAB_NOFS);
++ pa = kmem_cache_alloc(ext3_pspace_cachep, GFP_NOFS);
+ if (pa == NULL)
+ return -ENOMEM;
+
+ BUG_ON(!S_ISREG(ac->ac_inode->i_mode));
+
+ BUG_ON(ext3_pspace_cachep == NULL);
-+ pa = kmem_cache_alloc(ext3_pspace_cachep, SLAB_NOFS);
++ pa = kmem_cache_alloc(ext3_pspace_cachep, GFP_NOFS);
+ if (pa == NULL)
+ return -ENOMEM;
+
+
+ BUG_ON(pa->pa_deleted == 0);
+ ext3_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
-+ BUG_ON(group != e3b->bd_group);
++ BUG_ON(group != e3b->bd_group && pa->pa_len != 0);
+ end = bit + pa->pa_len;
+
+ ac.ac_sb = sb;
+
+ BUG_ON(pa->pa_deleted == 0);
+ ext3_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
-+ BUG_ON(group != e3b->bd_group);
++ BUG_ON(group != e3b->bd_group && pa->pa_len != 0);
+ mb_free_blocks(pa->pa_inode, e3b, bit, pa->pa_len);
+ atomic_add(pa->pa_len, &EXT3_SB(sb)->s_mb_discarded);
+
+ }
+
+ /* if we still need more blocks and some PAs were used, try again */
-+ if (free < needed && busy)
++ if (free < needed && busy) {
++ ext3_unlock_group(sb, group);
+ goto repeat;
++ }
+
+ /* found anything to free? */
+ if (list_empty(&list)) {
+ struct ext3_allocation_context ac;
+ struct ext3_sb_info *sbi;
+ struct super_block *sb;
-+ unsigned long block;
-+ int err, freed;
++ unsigned long block = 0;
++ int freed, inquota;
+
+ sb = ar->inode->i_sb;
+ sbi = EXT3_SB(sb);
+ printk(KERN_ERR "EXT3-fs: multiblock request with "
+ "mballoc disabled!\n");
+ ar->len = 1;
-+ err = ext3_new_block_old(handle, ar->inode, ar->goal, errp);
-+ return err;
++ block = ext3_new_block_old(handle, ar->inode, ar->goal, errp);
++ return block;
++ }
++
++ while (ar->len && DQUOT_ALLOC_BLOCK(ar->inode, ar->len)) {
++ ar->flags |= EXT3_MB_HINT_NOPREALLOC;
++ ar->len--;
++ }
++ if (ar->len == 0) {
++ *errp = -EDQUOT;
++ return 0;
+ }
++ inquota = ar->len;
+
+ ext3_mb_poll_new_transaction(sb, handle);
+
-+ if ((err = ext3_mb_initialize_context(&ac, ar)))
-+ return err;
++ if ((*errp = ext3_mb_initialize_context(&ac, ar))) {
++ ar->len = 0;
++ goto out;
++ }
+
+ ac.ac_op = EXT3_MB_HISTORY_PREALLOC;
+ if (!ext3_mb_use_preallocated(&ac)) {
+ goto repeat;
+ *errp = -ENOSPC;
+ ac.ac_b_ex.fe_len = 0;
-+ block = 0;
++ ar->len = 0;
+ ext3_mb_show_ac(&ac);
+ }
+
+ ext3_mb_release_context(&ac);
+
++out:
++ if (ar->len < inquota)
++ DQUOT_FREE_BLOCK(ar->inode, inquota - ar->len);
++
+ return block;
+}
+EXPORT_SYMBOL(ext3_mb_new_blocks);
+ BUG_ON(!mb_test_bit(bit + i, bitmap_bh->b_data));
+ }
+#endif
-+ mb_clear_bits(bitmap_bh->b_data, bit, count);
++ mb_clear_bits(sb_bgl_lock(sbi, block_group), bitmap_bh->b_data, bit,
++ count);
+
+ /* We dirtied the bitmap block */
+ BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
git://git.whamcloud.com / fs / lustre-release.git / blobdiff