diff options
Diffstat (limited to 'fs/xfs/xfs_da_btree.c')
| -rw-r--r-- | fs/xfs/xfs_da_btree.c | 1501 |
1 files changed, 931 insertions, 570 deletions
diff --git a/fs/xfs/xfs_da_btree.c b/fs/xfs/xfs_da_btree.c index 4d7696a02418..9b26a99ebfe9 100644 --- a/fs/xfs/xfs_da_btree.c +++ b/fs/xfs/xfs_da_btree.c @@ -1,5 +1,6 @@ /* * Copyright (c) 2000-2005 Silicon Graphics, Inc. + * Copyright (c) 2013 Red Hat, Inc. * All Rights Reserved. * * This program is free software; you can redistribute it and/or @@ -38,6 +39,8 @@ #include "xfs_attr_leaf.h" #include "xfs_error.h" #include "xfs_trace.h" +#include "xfs_cksum.h" +#include "xfs_buf_item.h" /* * xfs_da_btree.c @@ -52,69 +55,195 @@ /* * Routines used for growing the Btree. */ -STATIC int xfs_da_root_split(xfs_da_state_t *state, +STATIC int xfs_da3_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *existing_root, xfs_da_state_blk_t *new_child); -STATIC int xfs_da_node_split(xfs_da_state_t *state, +STATIC int xfs_da3_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *existing_blk, xfs_da_state_blk_t *split_blk, xfs_da_state_blk_t *blk_to_add, int treelevel, int *result); -STATIC void xfs_da_node_rebalance(xfs_da_state_t *state, +STATIC void xfs_da3_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *node_blk_1, xfs_da_state_blk_t *node_blk_2); -STATIC void xfs_da_node_add(xfs_da_state_t *state, +STATIC void xfs_da3_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *old_node_blk, xfs_da_state_blk_t *new_node_blk); /* * Routines used for shrinking the Btree. */ -STATIC int xfs_da_root_join(xfs_da_state_t *state, +STATIC int xfs_da3_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk); -STATIC int xfs_da_node_toosmall(xfs_da_state_t *state, int *retval); -STATIC void xfs_da_node_remove(xfs_da_state_t *state, +STATIC int xfs_da3_node_toosmall(xfs_da_state_t *state, int *retval); +STATIC void xfs_da3_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk); -STATIC void xfs_da_node_unbalance(xfs_da_state_t *state, +STATIC void xfs_da3_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *src_node_blk, xfs_da_state_blk_t *dst_node_blk); /* * Utility routines. */ -STATIC uint xfs_da_node_lasthash(struct xfs_buf *bp, int *count); -STATIC int xfs_da_node_order(struct xfs_buf *node1_bp, - struct xfs_buf *node2_bp); -STATIC int xfs_da_blk_unlink(xfs_da_state_t *state, +STATIC int xfs_da3_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, xfs_da_state_blk_t *save_blk); -STATIC void xfs_da_state_kill_altpath(xfs_da_state_t *state); -static void -xfs_da_node_verify( + +kmem_zone_t *xfs_da_state_zone; /* anchor for state struct zone */ + +/* + * Allocate a dir-state structure. + * We don't put them on the stack since they're large. + */ +xfs_da_state_t * +xfs_da_state_alloc(void) +{ + return kmem_zone_zalloc(xfs_da_state_zone, KM_NOFS); +} + +/* + * Kill the altpath contents of a da-state structure. + */ +STATIC void +xfs_da_state_kill_altpath(xfs_da_state_t *state) +{ + int i; + + for (i = 0; i < state->altpath.active; i++) + state->altpath.blk[i].bp = NULL; + state->altpath.active = 0; +} + +/* + * Free a da-state structure. + */ +void +xfs_da_state_free(xfs_da_state_t *state) +{ + xfs_da_state_kill_altpath(state); +#ifdef DEBUG + memset((char *)state, 0, sizeof(*state)); +#endif /* DEBUG */ + kmem_zone_free(xfs_da_state_zone, state); +} + +void +xfs_da3_node_hdr_from_disk( + struct xfs_da3_icnode_hdr *to, + struct xfs_da_intnode *from) +{ + ASSERT(from->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) || + from->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)); + + if (from->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) { + struct xfs_da3_node_hdr *hdr3 = (struct xfs_da3_node_hdr *)from; + + to->forw = be32_to_cpu(hdr3->info.hdr.forw); + to->back = be32_to_cpu(hdr3->info.hdr.back); + to->magic = be16_to_cpu(hdr3->info.hdr.magic); + to->count = be16_to_cpu(hdr3->__count); + to->level = be16_to_cpu(hdr3->__level); + return; + } + to->forw = be32_to_cpu(from->hdr.info.forw); + to->back = be32_to_cpu(from->hdr.info.back); + to->magic = be16_to_cpu(from->hdr.info.magic); + to->count = be16_to_cpu(from->hdr.__count); + to->level = be16_to_cpu(from->hdr.__level); +} + +void +xfs_da3_node_hdr_to_disk( + struct xfs_da_intnode *to, + struct xfs_da3_icnode_hdr *from) +{ + ASSERT(from->magic == XFS_DA_NODE_MAGIC || + from->magic == XFS_DA3_NODE_MAGIC); + + if (from->magic == XFS_DA3_NODE_MAGIC) { + struct xfs_da3_node_hdr *hdr3 = (struct xfs_da3_node_hdr *)to; + + hdr3->info.hdr.forw = cpu_to_be32(from->forw); + hdr3->info.hdr.back = cpu_to_be32(from->back); + hdr3->info.hdr.magic = cpu_to_be16(from->magic); + hdr3->__count = cpu_to_be16(from->count); + hdr3->__level = cpu_to_be16(from->level); + return; + } + to->hdr.info.forw = cpu_to_be32(from->forw); + to->hdr.info.back = cpu_to_be32(from->back); + to->hdr.info.magic = cpu_to_be16(from->magic); + to->hdr.__count = cpu_to_be16(from->count); + to->hdr.__level = cpu_to_be16(from->level); +} + +static bool +xfs_da3_node_verify( struct xfs_buf *bp) { struct xfs_mount *mp = bp->b_target->bt_mount; - struct xfs_da_node_hdr *hdr = bp->b_addr; - int block_ok = 0; - - block_ok = hdr->info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC); - block_ok = block_ok && - be16_to_cpu(hdr->level) > 0 && - be16_to_cpu(hdr->count) > 0 ; - if (!block_ok) { - XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr); - xfs_buf_ioerror(bp, EFSCORRUPTED); + struct xfs_da_intnode *hdr = bp->b_addr; + struct xfs_da3_icnode_hdr ichdr; + + xfs_da3_node_hdr_from_disk(&ichdr, hdr); + + if (xfs_sb_version_hascrc(&mp->m_sb)) { + struct xfs_da3_node_hdr *hdr3 = bp->b_addr; + + if (ichdr.magic != XFS_DA3_NODE_MAGIC) + return false; + + if (!uuid_equal(&hdr3->info.uuid, &mp->m_sb.sb_uuid)) + return false; + if (be64_to_cpu(hdr3->info.blkno) != bp->b_bn) + return false; + } else { + if (ichdr.magic != XFS_DA_NODE_MAGIC) + return false; } + if (ichdr.level == 0) + return false; + if (ichdr.level > XFS_DA_NODE_MAXDEPTH) + return false; + if (ichdr.count == 0) + return false; + /* + * we don't know if the node is for and attribute or directory tree, + * so only fail if the count is outside both bounds + */ + if (ichdr.count > mp->m_dir_node_ents && + ichdr.count > mp->m_attr_node_ents) + return false; + + /* XXX: hash order check? */ + + return true; } static void -xfs_da_node_write_verify( +xfs_da3_node_write_verify( struct xfs_buf *bp) { - xfs_da_node_verify(bp); + struct xfs_mount *mp = bp->b_target->bt_mount; + struct xfs_buf_log_item *bip = bp->b_fspriv; + struct xfs_da3_node_hdr *hdr3 = bp->b_addr; + + if (!xfs_da3_node_verify(bp)) { + XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr); + xfs_buf_ioerror(bp, EFSCORRUPTED); + return; + } + + if (!xfs_sb_version_hascrc(&mp->m_sb)) + return; + + if (bip) + hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn); + + xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DA3_NODE_CRC_OFF); } /* @@ -124,40 +253,47 @@ xfs_da_node_write_verify( * format of the block being read. */ static void -xfs_da_node_read_verify( +xfs_da3_node_read_verify( struct xfs_buf *bp) { struct xfs_mount *mp = bp->b_target->bt_mount; struct xfs_da_blkinfo *info = bp->b_addr; switch (be16_to_cpu(info->magic)) { + case XFS_DA3_NODE_MAGIC: + if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length), + XFS_DA3_NODE_CRC_OFF)) + break; + /* fall through */ case XFS_DA_NODE_MAGIC: - xfs_da_node_verify(bp); - break; + if (!xfs_da3_node_verify(bp)) + break; + return; case XFS_ATTR_LEAF_MAGIC: - bp->b_ops = &xfs_attr_leaf_buf_ops; + bp->b_ops = &xfs_attr3_leaf_buf_ops; bp->b_ops->verify_read(bp); return; case XFS_DIR2_LEAFN_MAGIC: - bp->b_ops = &xfs_dir2_leafn_buf_ops; + case XFS_DIR3_LEAFN_MAGIC: + bp->b_ops = &xfs_dir3_leafn_buf_ops; bp->b_ops->verify_read(bp); return; default: - XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, - mp, info); - xfs_buf_ioerror(bp, EFSCORRUPTED); break; } + + /* corrupt block */ + XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr); + xfs_buf_ioerror(bp, EFSCORRUPTED); } -const struct xfs_buf_ops xfs_da_node_buf_ops = { - .verify_read = xfs_da_node_read_verify, - .verify_write = xfs_da_node_write_verify, +const struct xfs_buf_ops xfs_da3_node_buf_ops = { + .verify_read = xfs_da3_node_read_verify, + .verify_write = xfs_da3_node_write_verify, }; - int -xfs_da_node_read( +xfs_da3_node_read( struct xfs_trans *tp, struct xfs_inode *dp, xfs_dablk_t bno, @@ -165,8 +301,35 @@ xfs_da_node_read( struct xfs_buf **bpp, int which_fork) { - return xfs_da_read_buf(tp, dp, bno, mappedbno, bpp, - which_fork, &xfs_da_node_buf_ops); + int err; + + err = xfs_da_read_buf(tp, dp, bno, mappedbno, bpp, + which_fork, &xfs_da3_node_buf_ops); + if (!err && tp) { + struct xfs_da_blkinfo *info = (*bpp)->b_addr; + int type; + + switch (be16_to_cpu(info->magic)) { + case XFS_DA_NODE_MAGIC: + case XFS_DA3_NODE_MAGIC: + type = XFS_BLFT_DA_NODE_BUF; + break; + case XFS_ATTR_LEAF_MAGIC: + case XFS_ATTR3_LEAF_MAGIC: + type = XFS_BLFT_ATTR_LEAF_BUF; + break; + case XFS_DIR2_LEAFN_MAGIC: + case XFS_DIR3_LEAFN_MAGIC: + type = XFS_BLFT_DIR_LEAFN_BUF; + break; + default: + type = 0; + ASSERT(0); + break; + } + xfs_trans_buf_set_type(tp, *bpp, type); + } + return err; } /*======================================================================== @@ -177,33 +340,46 @@ xfs_da_node_read( * Create the initial contents of an intermediate node. */ int -xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level, - struct xfs_buf **bpp, int whichfork) +xfs_da3_node_create( + struct xfs_da_args *args, + xfs_dablk_t blkno, + int level, + struct xfs_buf **bpp, + int whichfork) { - xfs_da_intnode_t *node; - struct xfs_buf *bp; - int error; - xfs_trans_t *tp; + struct xfs_da_intnode *node; + struct xfs_trans *tp = args->trans; + struct xfs_mount *mp = tp->t_mountp; + struct xfs_da3_icnode_hdr ichdr = {0}; + struct xfs_buf *bp; + int error; trace_xfs_da_node_create(args); + ASSERT(level <= XFS_DA_NODE_MAXDEPTH); - tp = args->trans; error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork); if (error) return(error); - ASSERT(bp != NULL); + bp->b_ops = &xfs_da3_node_buf_ops; + xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF); node = bp->b_addr; - node->hdr.info.forw = 0; - node->hdr.info.back = 0; - node->hdr.info.magic = cpu_to_be16(XFS_DA_NODE_MAGIC); - node->hdr.info.pad = 0; - node->hdr.count = 0; - node->hdr.level = cpu_to_be16(level); + if (xfs_sb_version_hascrc(&mp->m_sb)) { + struct xfs_da3_node_hdr *hdr3 = bp->b_addr; + + ichdr.magic = XFS_DA3_NODE_MAGIC; + hdr3->info.blkno = cpu_to_be64(bp->b_bn); + hdr3->info.owner = cpu_to_be64(args->dp->i_ino); + uuid_copy(&hdr3->info.uuid, &mp->m_sb.sb_uuid); + } else { + ichdr.magic = XFS_DA_NODE_MAGIC; + } + ichdr.level = level; + + xfs_da3_node_hdr_to_disk(node, &ichdr); xfs_trans_log_buf(tp, bp, - XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr))); + XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node))); - bp->b_ops = &xfs_da_node_buf_ops; *bpp = bp; return(0); } @@ -213,12 +389,18 @@ xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level, * intermediate nodes, rebalance, etc. */ int /* error */ -xfs_da_split(xfs_da_state_t *state) +xfs_da3_split( + struct xfs_da_state *state) { - xfs_da_state_blk_t *oldblk, *newblk, *addblk; - xfs_da_intnode_t *node; - struct xfs_buf *bp; - int max, action, error, i; + struct xfs_da_state_blk *oldblk; + struct xfs_da_state_blk *newblk; + struct xfs_da_state_blk *addblk; + struct xfs_da_intnode *node; + struct xfs_buf *bp; + int max; + int action; + int error; + int i; trace_xfs_da_split(state->args); @@ -246,7 +428,7 @@ xfs_da_split(xfs_da_state_t *state) */ switch (oldblk->magic) { case XFS_ATTR_LEAF_MAGIC: - error = xfs_attr_leaf_split(state, oldblk, newblk); + error = xfs_attr3_leaf_split(state, oldblk, newblk); if ((error != 0) && (error != ENOSPC)) { return(error); /* GROT: attr is inconsistent */ } @@ -261,12 +443,12 @@ xfs_da_split(xfs_da_state_t *state) if (state->inleaf) { state->extraafter = 0; /* before newblk */ trace_xfs_attr_leaf_split_before(state->args); - error = xfs_attr_leaf_split(state, oldblk, + error = xfs_attr3_leaf_split(state, oldblk, &state->extrablk); } else { state->extraafter = 1; /* after newblk */ trace_xfs_attr_leaf_split_after(state->args); - error = xfs_attr_leaf_split(state, newblk, + error = xfs_attr3_leaf_split(state, newblk, &state->extrablk); } if (error) @@ -280,7 +462,7 @@ xfs_da_split(xfs_da_state_t *state) addblk = newblk; break; case XFS_DA_NODE_MAGIC: - error = xfs_da_node_split(state, oldblk, newblk, addblk, + error = xfs_da3_node_split(state, oldblk, newblk, addblk, max - i, &action); addblk->bp = NULL; if (error) @@ -298,7 +480,7 @@ xfs_da_split(xfs_da_state_t *state) /* * Update the btree to show the new hashval for this child. */ - xfs_da_fixhashpath(state, &state->path); + xfs_da3_fixhashpath(state, &state->path); } if (!addblk) return(0); @@ -308,7 +490,7 @@ xfs_da_split(xfs_da_state_t *state) */ ASSERT(state->path.active == 0); oldblk = &state->path.blk[0]; - error = xfs_da_root_split(state, oldblk, addblk); + error = xfs_da3_root_split(state, oldblk, addblk); if (error) { addblk->bp = NULL; return(error); /* GROT: dir is inconsistent */ @@ -319,8 +501,12 @@ xfs_da_split(xfs_da_state_t *state) * just got bumped because of the addition of a new root node. * There might be three blocks involved if a double split occurred, * and the original block 0 could be at any position in the list. + * + * Note: the magic numbers and sibling pointers are in the same + * physical place for both v2 and v3 headers (by design). Hence it + * doesn't matter which version of the xfs_da_intnode structure we use + * here as the result will be the same using either structure. */ - node = oldblk->bp->b_addr; if (node->hdr.info.forw) { if (be32_to_cpu(node->hdr.info.forw) == addblk->blkno) { @@ -359,18 +545,25 @@ xfs_da_split(xfs_da_state_t *state) * the EOF, extending the inode in process. */ STATIC int /* error */ -xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, - xfs_da_state_blk_t *blk2) +xfs_da3_root_split( + struct xfs_da_state *state, + struct xfs_da_state_blk *blk1, + struct xfs_da_state_blk *blk2) { - xfs_da_intnode_t *node, *oldroot; - xfs_da_args_t *args; - xfs_dablk_t blkno; - struct xfs_buf *bp; - int error, size; - xfs_inode_t *dp; - xfs_trans_t *tp; - xfs_mount_t *mp; - xfs_dir2_leaf_t *leaf; + struct xfs_da_intnode *node; + struct xfs_da_intnode *oldroot; + struct xfs_da_node_entry *btree; + struct xfs_da3_icnode_hdr nodehdr; + struct xfs_da_args *args; + struct xfs_buf *bp; + struct xfs_inode *dp; + struct xfs_trans *tp; + struct xfs_mount *mp; + struct xfs_dir2_leaf *leaf; + xfs_dablk_t blkno; + int level; + int error; + int size; trace_xfs_da_root_split(state->args); @@ -379,29 +572,65 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, * to a free space somewhere. */ args = state->args; - ASSERT(args != NULL); error = xfs_da_grow_inode(args, &blkno); if (error) - return(error); + return error; + dp = args->dp; tp = args->trans; mp = state->mp; error = xfs_da_get_buf(tp, dp, blkno, -1, &bp, args->whichfork); if (error) - return(error); - ASSERT(bp != NULL); + return error; node = bp->b_addr; oldroot = blk1->bp->b_addr; - if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)) { - size = (int)((char *)&oldroot->btree[be16_to_cpu(oldroot->hdr.count)] - - (char *)oldroot); + if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) || + oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) { + struct xfs_da3_icnode_hdr nodehdr; + + xfs_da3_node_hdr_from_disk(&nodehdr, oldroot); + btree = xfs_da3_node_tree_p(oldroot); + size = (int)((char *)&btree[nodehdr.count] - (char *)oldroot); + level = nodehdr.level; + + /* + * we are about to copy oldroot to bp, so set up the type + * of bp while we know exactly what it will be. + */ + xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF); } else { - ASSERT(oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)); + struct xfs_dir3_icleaf_hdr leafhdr; + struct xfs_dir2_leaf_entry *ents; + leaf = (xfs_dir2_leaf_t *)oldroot; - size = (int)((char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] - - (char *)leaf); + xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf); + ents = xfs_dir3_leaf_ents_p(leaf); + + ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC || + leafhdr.magic == XFS_DIR3_LEAFN_MAGIC); + size = (int)((char *)&ents[leafhdr.count] - (char *)leaf); + level = 0; + + /* + * we are about to copy oldroot to bp, so set up the type + * of bp while we know exactly what it will be. + */ + xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF); } + + /* + * we can copy most of the information in the node from one block to + * another, but for CRC enabled headers we have to make sure that the + * block specific identifiers are kept intact. We update the buffer + * directly for this. + */ memcpy(node, oldroot, size); + if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) || + oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) { + struct xfs_da3_intnode *node3 = (struct xfs_da3_intnode *)node; + + node3->hdr.info.blkno = cpu_to_be64(bp->b_bn); + } xfs_trans_log_buf(tp, bp, 0, size - 1); bp->b_ops = blk1->bp->b_ops; @@ -411,20 +640,25 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, /* * Set up the new root node. */ - error = xfs_da_node_create(args, + error = xfs_da3_node_create(args, (args->whichfork == XFS_DATA_FORK) ? mp->m_dirleafblk : 0, - be16_to_cpu(node->hdr.level) + 1, &bp, args->whichfork); + level + 1, &bp, args->whichfork); if (error) - return(error); + return error; + node = bp->b_addr; - node->btree[0].hashval = cpu_to_be32(blk1->hashval); - node->btree[0].before = cpu_to_be32(blk1->blkno); - node->btree[1].hashval = cpu_to_be32(blk2->hashval); - node->btree[1].before = cpu_to_be32(blk2->blkno); - node->hdr.count = cpu_to_be16(2); + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); + btree[0].hashval = cpu_to_be32(blk1->hashval); + btree[0].before = cpu_to_be32(blk1->blkno); + btree[1].hashval = cpu_to_be32(blk2->hashval); + btree[1].before = cpu_to_be32(blk2->blkno); + nodehdr.count = 2; + xfs_da3_node_hdr_to_disk(node, &nodehdr); #ifdef DEBUG - if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)) { + if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) || + oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) { ASSERT(blk1->blkno >= mp->m_dirleafblk && blk1->blkno < mp->m_dirfreeblk); ASSERT(blk2->blkno >= mp->m_dirleafblk && @@ -434,30 +668,34 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, /* Header is already logged by xfs_da_node_create */ xfs_trans_log_buf(tp, bp, - XFS_DA_LOGRANGE(node, node->btree, - sizeof(xfs_da_node_entry_t) * 2)); + XFS_DA_LOGRANGE(node, btree, sizeof(xfs_da_node_entry_t) * 2)); - return(0); + return 0; } /* * Split the node, rebalance, then add the new entry. */ STATIC int /* error */ -xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, - xfs_da_state_blk_t *newblk, - xfs_da_state_blk_t *addblk, - int treelevel, int *result) +xfs_da3_node_split( + struct xfs_da_state *state, + struct xfs_da_state_blk *oldblk, + struct xfs_da_state_blk *newblk, + struct xfs_da_state_blk *addblk, + int treelevel, + int *result) { - xfs_da_intnode_t *node; - xfs_dablk_t blkno; - int newcount, error; - int useextra; + struct xfs_da_intnode *node; + struct xfs_da3_icnode_hdr nodehdr; + xfs_dablk_t blkno; + int newcount; + int error; + int useextra; trace_xfs_da_node_split(state->args); node = oldblk->bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); + xfs_da3_node_hdr_from_disk(&nodehdr, node); /* * With V2 dirs the extra block is data or freespace. @@ -467,7 +705,7 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, /* * Do we have to split the node? */ - if ((be16_to_cpu(node->hdr.count) + newcount) > state->node_ents) { + if (nodehdr.count + newcount > state->node_ents) { /* * Allocate a new node, add to the doubly linked chain of * nodes, then move some of our excess entries into it. @@ -476,14 +714,14 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, if (error) return(error); /* GROT: dir is inconsistent */ - error = xfs_da_node_create(state->args, blkno, treelevel, + error = xfs_da3_node_create(state->args, blkno, treelevel, &newblk->bp, state->args->whichfork); if (error) return(error); /* GROT: dir is inconsistent */ newblk->blkno = blkno; newblk->magic = XFS_DA_NODE_MAGIC; - xfs_da_node_rebalance(state, oldblk, newblk); - error = xfs_da_blk_link(state, oldblk, newblk); + xfs_da3_node_rebalance(state, oldblk, newblk); + error = xfs_da3_blk_link(state, oldblk, newblk); if (error) return(error); *result = 1; @@ -495,7 +733,7 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, * Insert the new entry(s) into the correct block * (updating last hashval in the process). * - * xfs_da_node_add() inserts BEFORE the given index, + * xfs_da3_node_add() inserts BEFORE the given index, * and as a result of using node_lookup_int() we always * point to a valid entry (not after one), but a split * operation always results in a new block whose hashvals @@ -504,22 +742,23 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, * If we had double-split op below us, then add the extra block too. */ node = oldblk->bp->b_addr; - if (oldblk->index <= be16_to_cpu(node->hdr.count)) { + xfs_da3_node_hdr_from_disk(&nodehdr, node); + if (oldblk->index <= nodehdr.count) { oldblk->index++; - xfs_da_node_add(state, oldblk, addblk); + xfs_da3_node_add(state, oldblk, addblk); if (useextra) { if (state->extraafter) oldblk->index++; - xfs_da_node_add(state, oldblk, &state->extrablk); + xfs_da3_node_add(state, oldblk, &state->extrablk); state->extravalid = 0; } } else { newblk->index++; - xfs_da_node_add(state, newblk, addblk); + xfs_da3_node_add(state, newblk, addblk); if (useextra) { if (state->extraafter) newblk->index++; - xfs_da_node_add(state, newblk, &state->extrablk); + xfs_da3_node_add(state, newblk, &state->extrablk); state->extravalid = 0; } } @@ -534,33 +773,53 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, * NOTE: if blk2 is empty, then it will get the upper half of blk1. */ STATIC void -xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, - xfs_da_state_blk_t *blk2) +xfs_da3_node_rebalance( + struct xfs_da_state *state, + struct xfs_da_state_blk *blk1, + struct xfs_da_state_blk *blk2) { - xfs_da_intnode_t *node1, *node2, *tmpnode; - xfs_da_node_entry_t *btree_s, *btree_d; - int count, tmp; - xfs_trans_t *tp; + struct xfs_da_intnode *node1; + struct xfs_da_intnode *node2; + struct xfs_da_intnode *tmpnode; + struct xfs_da_node_entry *btree1; + struct xfs_da_node_entry *btree2; + struct xfs_da_node_entry *btree_s; + struct xfs_da_node_entry *btree_d; + struct xfs_da3_icnode_hdr nodehdr1; + struct xfs_da3_icnode_hdr nodehdr2; + struct xfs_trans *tp; + int count; + int tmp; + int swap = 0; trace_xfs_da_node_rebalance(state->args); node1 = blk1->bp->b_addr; node2 = blk2->bp->b_addr; + xfs_da3_node_hdr_from_disk(&nodehdr1, node1); + xfs_da3_node_hdr_from_disk(&nodehdr2, node2); + btree1 = xfs_da3_node_tree_p(node1); + btree2 = xfs_da3_node_tree_p(node2); + /* * Figure out how many entries need to move, and in which direction. * Swap the nodes around if that makes it simpler. */ - if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) && - ((be32_to_cpu(node2->btree[0].hashval) < be32_to_cpu(node1->btree[0].hashval)) || - (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) < - be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) { + if (nodehdr1.count > 0 && nodehdr2.count > 0 && + ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) || + (be32_to_cpu(btree2[nodehdr2.count - 1].hashval) < + be32_to_cpu(btree1[nodehdr1.count - 1].hashval)))) { tmpnode = node1; node1 = node2; node2 = tmpnode; + xfs_da3_node_hdr_from_disk(&nodehdr1, node1); + xfs_da3_node_hdr_from_disk(&nodehdr2, node2); + btree1 = xfs_da3_node_tree_p(node1); + btree2 = xfs_da3_node_tree_p(node2); + swap = 1; } - ASSERT(node1->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - ASSERT(node2->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - count = (be16_to_cpu(node1->hdr.count) - be16_to_cpu(node2->hdr.count)) / 2; + + count = (nodehdr1.count - nodehdr2.count) / 2; if (count == 0) return; tp = state->args->trans; @@ -571,10 +830,11 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, /* * Move elements in node2 up to make a hole. */ - if ((tmp = be16_to_cpu(node2->hdr.count)) > 0) { + tmp = nodehdr2.count; + if (tmp > 0) { tmp *= (uint)sizeof(xfs_da_node_entry_t); - btree_s = &node2->btree[0]; - btree_d = &node2->btree[count]; + btree_s = &btree2[0]; + btree_d = &btree2[count]; memmove(btree_d, btree_s, tmp); } @@ -582,12 +842,12 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, * Move the req'd B-tree elements from high in node1 to * low in node2. */ - be16_add_cpu(&node2->hdr.count, count); + nodehdr2.count += count; tmp = count * (uint)sizeof(xfs_da_node_entry_t); - btree_s = &node1->btree[be16_to_cpu(node1->hdr.count) - count]; - btree_d = &node2->btree[0]; + btree_s = &btree1[nodehdr1.count - count]; + btree_d = &btree2[0]; memcpy(btree_d, btree_s, tmp); - be16_add_cpu(&node1->hdr.count, -count); + nodehdr1.count -= count; } else { /* * Move the req'd B-tree elements from low in node2 to @@ -595,49 +855,60 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, */ count = -count; tmp = count * (uint)sizeof(xfs_da_node_entry_t); - btree_s = &node2->btree[0]; - btree_d = &node1->btree[be16_to_cpu(node1->hdr.count)]; + btree_s = &btree2[0]; + btree_d = &btree1[nodehdr1.count]; memcpy(btree_d, btree_s, tmp); - be16_add_cpu(&node1->hdr.count, count); + nodehdr1.count += count; + xfs_trans_log_buf(tp, blk1->bp, XFS_DA_LOGRANGE(node1, btree_d, tmp)); /* * Move elements in node2 down to fill the hole. */ - tmp = be16_to_cpu(node2->hdr.count) - count; + tmp = nodehdr2.count - count; tmp *= (uint)sizeof(xfs_da_node_entry_t); - btree_s = &node2->btree[count]; - btree_d = &node2->btree[0]; + btree_s = &btree2[count]; + btree_d = &btree2[0]; memmove(btree_d, btree_s, tmp); - be16_add_cpu(&node2->hdr.count, -count); + nodehdr2.count -= count; } /* * Log header of node 1 and all current bits of node 2. */ + xfs_da3_node_hdr_to_disk(node1, &nodehdr1); xfs_trans_log_buf(tp, blk1->bp, - XFS_DA_LOGRANGE(node1, &node1->hdr, sizeof(node1->hdr))); + XFS_DA_LOGRANGE(node1, &node1->hdr, + xfs_da3_node_hdr_size(node1))); + + xfs_da3_node_hdr_to_disk(node2, &nodehdr2); xfs_trans_log_buf(tp, blk2->bp, XFS_DA_LOGRANGE(node2, &node2->hdr, - sizeof(node2->hdr) + - sizeof(node2->btree[0]) * be16_to_cpu(node2->hdr.count))); + xfs_da3_node_hdr_size(node2) + + (sizeof(btree2[0]) * nodehdr2.count))); /* * Record the last hashval from each block for upward propagation. * (note: don't use the swapped node pointers) */ - node1 = blk1->bp->b_addr; - node2 = blk2->bp->b_addr; - blk1->hashval = be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval); - blk2->hashval = be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval); + if (swap) { + node1 = blk1->bp->b_addr; + node2 = blk2->bp->b_addr; + xfs_da3_node_hdr_from_disk(&nodehdr1, node1); + xfs_da3_node_hdr_from_disk(&nodehdr2, node2); + btree1 = xfs_da3_node_tree_p(node1); + btree2 = xfs_da3_node_tree_p(node2); + } + blk1->hashval = be32_to_cpu(btree1[nodehdr1.count - 1].hashval); + blk2->hashval = be32_to_cpu(btree2[nodehdr2.count - 1].hashval); /* * Adjust the expected index for insertion. */ - if (blk1->index >= be16_to_cpu(node1->hdr.count)) { - blk2->index = blk1->index - be16_to_cpu(node1->hdr.count); - blk1->index = be16_to_cpu(node1->hdr.count) + 1; /* make it invalid */ + if (blk1->index >= nodehdr1.count) { + blk2->index = blk1->index - nodehdr1.count; + blk1->index = nodehdr1.count + 1; /* make it invalid */ } } @@ -645,18 +916,23 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1, * Add a new entry to an intermediate node. */ STATIC void -xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, - xfs_da_state_blk_t *newblk) +xfs_da3_node_add( + struct xfs_da_state *state, + struct xfs_da_state_blk *oldblk, + struct xfs_da_state_blk *newblk) { - xfs_da_intnode_t *node; - xfs_da_node_entry_t *btree; - int tmp; + struct xfs_da_intnode *node; + struct xfs_da3_icnode_hdr nodehdr; + struct xfs_da_node_entry *btree; + int tmp; trace_xfs_da_node_add(state->args); node = oldblk->bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - ASSERT((oldblk->index >= 0) && (oldblk->index <= be16_to_cpu(node->hdr.count))); + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); + + ASSERT(oldblk->index >= 0 && oldblk->index <= nodehdr.count); ASSERT(newblk->blkno != 0); if (state->args->whichfork == XFS_DATA_FORK) ASSERT(newblk->blkno >= state->mp->m_dirleafblk && @@ -666,23 +942,25 @@ xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, * We may need to make some room before we insert the new node. */ tmp = 0; - btree = &node->btree[ oldblk->index ]; - if (oldblk->index < be16_to_cpu(node->hdr.count)) { - tmp = (be16_to_cpu(node->hdr.count) - oldblk->index) * (uint)sizeof(*btree); - memmove(btree + 1, btree, tmp); + if (oldblk->index < nodehdr.count) { + tmp = (nodehdr.count - oldblk->index) * (uint)sizeof(*btree); + memmove(&btree[oldblk->index + 1], &btree[oldblk->index], tmp); } - btree->hashval = cpu_to_be32(newblk->hashval); - btree->before = cpu_to_be32(newblk->blkno); + btree[oldblk->index].hashval = cpu_to_be32(newblk->hashval); + btree[oldblk->index].before = cpu_to_be32(newblk->blkno); xfs_trans_log_buf(state->args->trans, oldblk->bp, - XFS_DA_LOGRANGE(node, btree, tmp + sizeof(*btree))); - be16_add_cpu(&node->hdr.count, 1); + XFS_DA_LOGRANGE(node, &btree[oldblk->index], + tmp + sizeof(*btree))); + + nodehdr.count += 1; + xfs_da3_node_hdr_to_disk(node, &nodehdr); xfs_trans_log_buf(state->args->trans, oldblk->bp, - XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr))); + XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node))); /* * Copy the last hash value from the oldblk to propagate upwards. */ - oldblk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1 ].hashval); + oldblk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval); } /*======================================================================== @@ -694,14 +972,16 @@ xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk, * possibly deallocating that block, etc... */ int -xfs_da_join(xfs_da_state_t *state) +xfs_da3_join( + struct xfs_da_state *state) { - xfs_da_state_blk_t *drop_blk, *save_blk; - int action, error; + struct xfs_da_state_blk *drop_blk; + struct xfs_da_state_blk *save_blk; + int action = 0; + int error; trace_xfs_da_join(state->args); - action = 0; drop_blk = &state->path.blk[ state->path.active-1 ]; save_blk = &state->altpath.blk[ state->path.active-1 ]; ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC); @@ -722,12 +1002,12 @@ xfs_da_join(xfs_da_state_t *state) */ switch (drop_blk->magic) { case XFS_ATTR_LEAF_MAGIC: - error = xfs_attr_leaf_toosmall(state, &action); + error = xfs_attr3_leaf_toosmall(state, &action); if (error) return(error); if (action == 0) return(0); - xfs_attr_leaf_unbalance(state, drop_blk, save_blk); + xfs_attr3_leaf_unbalance(state, drop_blk, save_blk); break; case XFS_DIR2_LEAFN_MAGIC: error = xfs_dir2_leafn_toosmall(state, &action); @@ -742,18 +1022,18 @@ xfs_da_join(xfs_da_state_t *state) * Remove the offending node, fixup hashvals, * check for a toosmall neighbor. */ - xfs_da_node_remove(state, drop_blk); - xfs_da_fixhashpath(state, &state->path); - error = xfs_da_node_toosmall(state, &action); + xfs_da3_node_remove(state, drop_blk); + xfs_da3_fixhashpath(state, &state->path); + error = xfs_da3_node_toosmall(state, &action); if (error) return(error); if (action == 0) return 0; - xfs_da_node_unbalance(state, drop_blk, save_blk); + xfs_da3_node_unbalance(state, drop_blk, save_blk); break; } - xfs_da_fixhashpath(state, &state->altpath); - error = xfs_da_blk_unlink(state, drop_blk, save_blk); + xfs_da3_fixhashpath(state, &state->altpath); + error = xfs_da3_blk_unlink(state, drop_blk, save_blk); xfs_da_state_kill_altpath(state); if (error) return(error); @@ -768,9 +1048,9 @@ xfs_da_join(xfs_da_state_t *state) * we only have one entry in the root, make the child block * the new root. */ - xfs_da_node_remove(state, drop_blk); - xfs_da_fixhashpath(state, &state->path); - error = xfs_da_root_join(state, &state->path.blk[0]); + xfs_da3_node_remove(state, drop_blk); + xfs_da3_fixhashpath(state, &state->path); + error = xfs_da3_root_join(state, &state->path.blk[0]); return(error); } @@ -782,9 +1062,13 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level) if (level == 1) { ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) || - magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC)); - } else - ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); + magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) || + magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) || + magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC)); + } else { + ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC) || + magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)); + } ASSERT(!blkinfo->forw); ASSERT(!blkinfo->back); } @@ -797,52 +1081,61 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level) * the old root to block 0 as the new root node. */ STATIC int -xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk) +xfs_da3_root_join( + struct xfs_da_state *state, + struct xfs_da_state_blk *root_blk) { - xfs_da_intnode_t *oldroot; - xfs_da_args_t *args; - xfs_dablk_t child; - struct xfs_buf *bp; - int error; + struct xfs_da_intnode *oldroot; + struct xfs_da_args *args; + xfs_dablk_t child; + struct xfs_buf *bp; + struct xfs_da3_icnode_hdr oldroothdr; + struct xfs_da_node_entry *btree; + int error; trace_xfs_da_root_join(state->args); - args = state->args; - ASSERT(args != NULL); ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC); + + args = state->args; oldroot = root_blk->bp->b_addr; - ASSERT(oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - ASSERT(!oldroot->hdr.info.forw); - ASSERT(!oldroot->hdr.info.back); + xfs_da3_node_hdr_from_disk(&oldroothdr, oldroot); + ASSERT(oldroothdr.forw == 0); + ASSERT(oldroothdr.back == 0); /* * If the root has more than one child, then don't do anything. */ - if (be16_to_cpu(oldroot->hdr.count) > 1) - return(0); + if (oldroothdr.count > 1) + return 0; /* * Read in the (only) child block, then copy those bytes into * the root block's buffer and free the original child block. */ - child = be32_to_cpu(oldroot->btree[0].before); + btree = xfs_da3_node_tree_p(oldroot); + child = be32_to_cpu(btree[0].before); ASSERT(child != 0); - error = xfs_da_node_read(args->trans, args->dp, child, -1, &bp, + error = xfs_da3_node_read(args->trans, args->dp, child, -1, &bp, args->whichfork); if (error) - return(error); - ASSERT(bp != NULL); - xfs_da_blkinfo_onlychild_validate(bp->b_addr, - be16_to_cpu(oldroot->hdr.level)); + return error; + xfs_da_blkinfo_onlychild_validate(bp->b_addr, oldroothdr.level); /* * This could be copying a leaf back into the root block in the case of * there only being a single leaf block left in the tree. Hence we have * to update the b_ops pointer as well to match the buffer type change - * that could occur. + * that could occur. For dir3 blocks we also need to update the block + * number in the buffer header. */ memcpy(root_blk->bp->b_addr, bp->b_addr, state->blocksize); root_blk->bp->b_ops = bp->b_ops; + xfs_trans_buf_copy_type(root_blk->bp, bp); + if (oldroothdr.magic == XFS_DA3_NODE_MAGIC) { + struct xfs_da3_blkinfo *da3 = root_blk->bp->b_addr; + da3->blkno = cpu_to_be64(root_blk->bp->b_bn); + } xfs_trans_log_buf(args->trans, root_blk->bp, 0, state->blocksize - 1); error = xfs_da_shrink_inode(args, child, bp); return(error); @@ -858,14 +1151,21 @@ xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk) * If nothing can be done, return 0. */ STATIC int -xfs_da_node_toosmall(xfs_da_state_t *state, int *action) +xfs_da3_node_toosmall( + struct xfs_da_state *state, + int *action) { - xfs_da_intnode_t *node; - xfs_da_state_blk_t *blk; - xfs_da_blkinfo_t *info; - int count, forward, error, retval, i; - xfs_dablk_t blkno; - struct xfs_buf *bp; + struct xfs_da_intnode *node; + struct xfs_da_state_blk *blk; + struct xfs_da_blkinfo *info; + xfs_dablk_t blkno; + struct xfs_buf *bp; + struct xfs_da3_icnode_hdr nodehdr; + int count; + int forward; + int error; + int retval; + int i; trace_xfs_da_node_toosmall(state->args); @@ -876,10 +1176,9 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action) */ blk = &state->path.blk[ state->path.active-1 ]; info = blk->bp->b_addr; - ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); node = (xfs_da_intnode_t *)info; - count = be16_to_cpu(node->hdr.count); - if (count > (state->node_ents >> 1)) { + xfs_da3_node_hdr_from_disk(&nodehdr, node); + if (nodehdr.count > (state->node_ents >> 1)) { *action = 0; /* blk over 50%, don't try to join */ return(0); /* blk over 50%, don't try to join */ } @@ -890,14 +1189,14 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action) * coalesce it with a sibling block. We choose (arbitrarily) * to merge with the forward block unless it is NULL. */ - if (count == 0) { + if (nodehdr.count == 0) { /* * Make altpath point to the block we want to keep and * path point to the block we want to drop (this one). */ forward = (info->forw != 0); memcpy(&state->altpath, &state->path, sizeof(state->path)); - error = xfs_da_path_shift(state, &state->altpath, forward, + error = xfs_da3_path_shift(state, &state->altpath, forward, 0, &retval); if (error) return(error); @@ -916,35 +1215,34 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action) * We prefer coalescing with the lower numbered sibling so as * to shrink a directory over time. */ + count = state->node_ents; + count -= state->node_ents >> 2; + count -= nodehdr.count; + /* start with smaller blk num */ - forward = (be32_to_cpu(info->forw) < be32_to_cpu(info->back)); + forward = nodehdr.forw < nodehdr.back; for (i = 0; i < 2; forward = !forward, i++) { if (forward) - blkno = be32_to_cpu(info->forw); + blkno = nodehdr.forw; else - blkno = be32_to_cpu(info->back); + blkno = nodehdr.back; if (blkno == 0) continue; - error = xfs_da_node_read(state->args->trans, state->args->dp, + error = xfs_da3_node_read(state->args->trans, state->args->dp, blkno, -1, &bp, state->args->whichfork); if (error) return(error); - ASSERT(bp != NULL); - node = (xfs_da_intnode_t *)info; - count = state->node_ents; - count -= state->node_ents >> 2; - count -= be16_to_cpu(node->hdr.count); node = bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - count -= be16_to_cpu(node->hdr.count); + xfs_da3_node_hdr_from_disk(&nodehdr, node); xfs_trans_brelse(state->args->trans, bp); - if (count >= 0) + + if (count - nodehdr.count >= 0) break; /* fits with at least 25% to spare */ } if (i >= 2) { *action = 0; - return(0); + return 0; } /* @@ -953,28 +1251,42 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action) */ memcpy(&state->altpath, &state->path, sizeof(state->path)); if (blkno < blk->blkno) { - error = xfs_da_path_shift(state, &state->altpath, forward, + error = xfs_da3_path_shift(state, &state->altpath, forward, 0, &retval); - if (error) { - return(error); - } - if (retval) { - *action = 0; - return(0); - } } else { - error = xfs_da_path_shift(state, &state->path, forward, + error = xfs_da3_path_shift(state, &state->path, forward, 0, &retval); - if (error) { - return(error); - } - if (retval) { - *action = 0; - return(0); - } + } + if (error) + return error; + if (retval) { + *action = 0; + return 0; } *action = 1; - return(0); + return 0; +} + +/* + * Pick up the last hashvalue from an intermediate node. + */ +STATIC uint +xfs_da3_node_lasthash( + struct xfs_buf *bp, + int *count) +{ + struct xfs_da_intnode *node; + struct xfs_da_node_entry *btree; + struct xfs_da3_icnode_hdr nodehdr; + + node = bp->b_addr; + xfs_da3_node_hdr_from_disk(&nodehdr, node); + if (count) + *count = nodehdr.count; + if (!nodehdr.count) + return 0; + btree = xfs_da3_node_tree_p(node); + return be32_to_cpu(btree[nodehdr.count - 1].hashval); } /* @@ -982,13 +1294,16 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action) * when we stop making changes, return. */ void -xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path) +xfs_da3_fixhashpath( + struct xfs_da_state *state, + struct xfs_da_state_path *path) { - xfs_da_state_blk_t *blk; - xfs_da_intnode_t *node; - xfs_da_node_entry_t *btree; - xfs_dahash_t lasthash=0; - int level, count; + struct xfs_da_state_blk *blk; + struct xfs_da_intnode *node; + struct xfs_da_node_entry *btree; + xfs_dahash_t lasthash=0; + int level; + int count; trace_xfs_da_fixhashpath(state->args); @@ -1006,23 +1321,26 @@ xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path) return; break; case XFS_DA_NODE_MAGIC: - lasthash = xfs_da_node_lasthash(blk->bp, &count); + lasthash = xfs_da3_node_lasthash(blk->bp, &count); if (count == 0) return; break; } for (blk--, level--; level >= 0; blk--, level--) { + struct xfs_da3_icnode_hdr nodehdr; + node = blk->bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - btree = &node->btree[ blk->index ]; + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); if (be32_to_cpu(btree->hashval) == lasthash) break; blk->hashval = lasthash; - btree->hashval = cpu_to_be32(lasthash); + btree[blk->index].hashval = cpu_to_be32(lasthash); xfs_trans_log_buf(state->args->trans, blk->bp, - XFS_DA_LOGRANGE(node, btree, sizeof(*btree))); + XFS_DA_LOGRANGE(node, &btree[blk->index], + sizeof(*btree))); - lasthash = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval); + lasthash = be32_to_cpu(btree[nodehdr.count - 1].hashval); } } @@ -1030,104 +1348,120 @@ xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path) * Remove an entry from an intermediate node. */ STATIC void -xfs_da_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk) +xfs_da3_node_remove( + struct xfs_da_state *state, + struct xfs_da_state_blk *drop_blk) { - xfs_da_intnode_t *node; - xfs_da_node_entry_t *btree; - int tmp; + struct xfs_da_intnode *node; + struct xfs_da3_icnode_hdr nodehdr; + struct xfs_da_node_entry *btree; + int index; + int tmp; trace_xfs_da_node_remove(state->args); node = drop_blk->bp->b_addr; - ASSERT(drop_blk->index < be16_to_cpu(node->hdr.count)); + xfs_da3_node_hdr_from_disk(&nodehdr, node); + ASSERT(drop_blk->index < nodehdr.count); ASSERT(drop_blk->index >= 0); /* * Copy over the offending entry, or just zero it out. */ - btree = &node->btree[drop_blk->index]; - if (drop_blk->index < (be16_to_cpu(node->hdr.count)-1)) { - tmp = be16_to_cpu(node->hdr.count) - drop_blk->index - 1; + index = drop_blk->index; + btree = xfs_da3_node_tree_p(node); + if (index < nodehdr.count - 1) { + tmp = nodehdr.count - index - 1; tmp *= (uint)sizeof(xfs_da_node_entry_t); - memmove(btree, btree + 1, tmp); + memmove(&btree[index], &btree[index + 1], tmp); xfs_trans_log_buf(state->args->trans, drop_blk->bp, - XFS_DA_LOGRANGE(node, btree, tmp)); - btree = &node->btree[be16_to_cpu(node->hdr.count)-1]; + XFS_DA_LOGRANGE(node, &btree[index], tmp)); + index = nodehdr.count - 1; } - memset((char *)btree, 0, sizeof(xfs_da_node_entry_t)); + memset(&btree[index], 0, sizeof(xfs_da_node_entry_t)); xfs_trans_log_buf(state->args->trans, drop_blk->bp, - XFS_DA_LOGRANGE(node, btree, sizeof(*btree))); - be16_add_cpu(&node->hdr.count, -1); + XFS_DA_LOGRANGE(node, &btree[index], sizeof(btree[index]))); + nodehdr.count -= 1; + xfs_da3_node_hdr_to_disk(node, &nodehdr); xfs_trans_log_buf(state->args->trans, drop_blk->bp, - XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr))); + XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node))); /* * Copy the last hash value from the block to propagate upwards. */ - btree--; - drop_blk->hashval = be32_to_cpu(btree->hashval); + drop_blk->hashval = be32_to_cpu(btree[index - 1].hashval); } /* - * Unbalance the btree elements between two intermediate nodes, + * Unbalance the elements between two intermediate nodes, * move all Btree elements from one node into another. */ STATIC void -xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, - xfs_da_state_blk_t *save_blk) +xfs_da3_node_unbalance( + struct xfs_da_state *state, + struct xfs_da_state_blk *drop_blk, + struct xfs_da_state_blk *save_blk) { - xfs_da_intnode_t *drop_node, *save_node; - xfs_da_node_entry_t *btree; - int tmp; - xfs_trans_t *tp; + struct xfs_da_intnode *drop_node; + struct xfs_da_intnode *save_node; + struct xfs_da_node_entry *drop_btree; + struct xfs_da_node_entry *save_btree; + struct xfs_da3_icnode_hdr drop_hdr; + struct xfs_da3_icnode_hdr save_hdr; + struct xfs_trans *tp; + int sindex; + int tmp; trace_xfs_da_node_unbalance(state->args); drop_node = drop_blk->bp->b_addr; save_node = save_blk->bp->b_addr; - ASSERT(drop_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - ASSERT(save_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); + xfs_da3_node_hdr_from_disk(&drop_hdr, drop_node); + xfs_da3_node_hdr_from_disk(&save_hdr, save_node); + drop_btree = xfs_da3_node_tree_p(drop_node); + save_btree = xfs_da3_node_tree_p(save_node); tp = state->args->trans; /* * If the dying block has lower hashvals, then move all the * elements in the remaining block up to make a hole. */ - if ((be32_to_cpu(drop_node->btree[0].hashval) < be32_to_cpu(save_node->btree[ 0 ].hashval)) || - (be32_to_cpu(drop_node->btree[be16_to_cpu(drop_node->hdr.count)-1].hashval) < - be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval))) - { - btree = &save_node->btree[be16_to_cpu(drop_node->hdr.count)]; - tmp = be16_to_cpu(save_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t); - memmove(btree, &save_node->btree[0], tmp); - btree = &save_node->btree[0]; + if ((be32_to_cpu(drop_btree[0].hashval) < + be32_to_cpu(save_btree[0].hashval)) || + (be32_to_cpu(drop_btree[drop_hdr.count - 1].hashval) < + be32_to_cpu(save_btree[save_hdr.count - 1].hashval))) { + /* XXX: check this - is memmove dst correct? */ + tmp = save_hdr.count * sizeof(xfs_da_node_entry_t); + memmove(&save_btree[drop_hdr.count], &save_btree[0], tmp); + + sindex = 0; xfs_trans_log_buf(tp, save_blk->bp, - XFS_DA_LOGRANGE(save_node, btree, - (be16_to_cpu(save_node->hdr.count) + be16_to_cpu(drop_node->hdr.count)) * - sizeof(xfs_da_node_entry_t))); + XFS_DA_LOGRANGE(save_node, &save_btree[0], + (save_hdr.count + drop_hdr.count) * + sizeof(xfs_da_node_entry_t))); } else { - btree = &save_node->btree[be16_to_cpu(save_node->hdr.count)]; + sindex = save_hdr.count; xfs_trans_log_buf(tp, save_blk->bp, - XFS_DA_LOGRANGE(save_node, btree, - be16_to_cpu(drop_node->hdr.count) * - sizeof(xfs_da_node_entry_t))); + XFS_DA_LOGRANGE(save_node, &save_btree[sindex], + drop_hdr.count * sizeof(xfs_da_node_entry_t))); } /* * Move all the B-tree elements from drop_blk to save_blk. */ - tmp = be16_to_cpu(drop_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t); - memcpy(btree, &drop_node->btree[0], tmp); - be16_add_cpu(&save_node->hdr.count, be16_to_cpu(drop_node->hdr.count)); + tmp = drop_hdr.count * (uint)sizeof(xfs_da_node_entry_t); + memcpy(&save_btree[sindex], &drop_btree[0], tmp); + save_hdr.count += drop_hdr.count; + xfs_da3_node_hdr_to_disk(save_node, &save_hdr); xfs_trans_log_buf(tp, save_blk->bp, XFS_DA_LOGRANGE(save_node, &save_node->hdr, - sizeof(save_node->hdr))); + xfs_da3_node_hdr_size(save_node))); /* * Save the last hashval in the remaining block for upward propagation. */ - save_blk->hashval = be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval); + save_blk->hashval = be32_to_cpu(save_btree[save_hdr.count - 1].hashval); } /*======================================================================== @@ -1146,16 +1480,24 @@ xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, * pruned depth-first tree search. */ int /* error */ -xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) +xfs_da3_node_lookup_int( + struct xfs_da_state *state, + int *result) { - xfs_da_state_blk_t *blk; - xfs_da_blkinfo_t *curr; - xfs_da_intnode_t *node; - xfs_da_node_entry_t *btree; - xfs_dablk_t blkno; - int probe, span, max, error, retval; - xfs_dahash_t hashval, btreehashval; - xfs_da_args_t *args; + struct xfs_da_state_blk *blk; + struct xfs_da_blkinfo *curr; + struct xfs_da_intnode *node; + struct xfs_da_node_entry *btree; + struct xfs_da3_icnode_hdr nodehdr; + struct xfs_da_args *args; + xfs_dablk_t blkno; + xfs_dahash_t hashval; + xfs_dahash_t btreehashval; + int probe; + int span; + int max; + int error; + int retval; args = state->args; @@ -1171,7 +1513,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) * Read the next node down in the tree. */ blk->blkno = blkno; - error = xfs_da_node_read(args->trans, args->dp, blkno, + error = xfs_da3_node_read(args->trans, args->dp, blkno, -1, &blk->bp, args->whichfork); if (error) { blk->blkno = 0; @@ -1180,66 +1522,75 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) } curr = blk->bp->b_addr; blk->magic = be16_to_cpu(curr->magic); - ASSERT(blk->magic == XFS_DA_NODE_MAGIC || - blk->magic == XFS_DIR2_LEAFN_MAGIC || - blk->magic == XFS_ATTR_LEAF_MAGIC); + + if (blk->magic == XFS_ATTR_LEAF_MAGIC || + blk->magic == XFS_ATTR3_LEAF_MAGIC) { + blk->magic = XFS_ATTR_LEAF_MAGIC; + blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL); + break; + } + + if (blk->magic == XFS_DIR2_LEAFN_MAGIC || + blk->magic == XFS_DIR3_LEAFN_MAGIC) { + blk->magic = XFS_DIR2_LEAFN_MAGIC; + blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL); + break; + } + + blk->magic = XFS_DA_NODE_MAGIC; + /* * Search an intermediate node for a match. */ - if (blk->magic == XFS_DA_NODE_MAGIC) { - node = blk->bp->b_addr; - max = be16_to_cpu(node->hdr.count); - blk->hashval = be32_to_cpu(node->btree[max-1].hashval); + node = blk->bp->b_addr; + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); - /* - * Binary search. (note: small blocks will skip loop) - */ - probe = span = max / 2; - hashval = args->hashval; - for (btree = &node->btree[probe]; span > 4; - btree = &node->btree[probe]) { - span /= 2; - btreehashval = be32_to_cpu(btree->hashval); - if (btreehashval < hashval) - probe += span; - else if (btreehashval > hashval) - probe -= span; - else - break; - } - ASSERT((probe >= 0) && (probe < max)); - ASSERT((span <= 4) || (be32_to_cpu(btree->hashval) == hashval)); + max = nodehdr.count; + blk->hashval = be32_to_cpu(btree[max - 1].hashval); - /* - * Since we may have duplicate hashval's, find the first - * matching hashval in the node. - */ - while ((probe > 0) && (be32_to_cpu(btree->hashval) >= hashval)) { - btree--; - probe--; - } - while ((probe < max) && (be32_to_cpu(btree->hashval) < hashval)) { - btree++; - probe++; - } + /* + * Binary search. (note: small blocks will skip loop) + */ + probe = span = max / 2; + hashval = args->hashval; + while (span > 4) { + span /= 2; + btreehashval = be32_to_cpu(btree[probe].hashval); + if (btreehashval < hashval) + probe += span; + else if (btreehashval > hashval) + probe -= span; + else + break; + } + ASSERT((probe >= 0) && (probe < max)); + ASSERT((span <= 4) || + (be32_to_cpu(btree[probe].hashval) == hashval)); - /* - * Pick the right block to descend on. - */ - if (probe == max) { - blk->index = max-1; - blkno = be32_to_cpu(node->btree[max-1].before); - } else { - blk->index = probe; - blkno = be32_to_cpu(btree->before); - } - } else if (blk->magic == XFS_ATTR_LEAF_MAGIC) { - blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL); - break; - } else if (blk->magic == XFS_DIR2_LEAFN_MAGIC) { - blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL); - break; + /* + * Since we may have duplicate hashval's, find the first + * matching hashval in the node. + */ + while (probe > 0 && + be32_to_cpu(btree[probe].hashval) >= hashval) { + probe--; + } + while (probe < max && + be32_to_cpu(btree[probe].hashval) < hashval) { + probe++; + } + + /* + * Pick the right block to descend on. + */ + if (probe == max) { + blk->index = max - 1; + blkno = be32_to_cpu(btree[max - 1].before); + } else { + blk->index = probe; + blkno = be32_to_cpu(btree[probe].before); } } @@ -1254,7 +1605,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) retval = xfs_dir2_leafn_lookup_int(blk->bp, args, &blk->index, state); } else if (blk->magic == XFS_ATTR_LEAF_MAGIC) { - retval = xfs_attr_leaf_lookup_int(blk->bp, args); + retval = xfs_attr3_leaf_lookup_int(blk->bp, args); blk->index = args->index; args->blkno = blk->blkno; } else { @@ -1263,7 +1614,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) } if (((retval == ENOENT) || (retval == ENOATTR)) && (blk->hashval == args->hashval)) { - error = xfs_da_path_shift(state, &state->path, 1, 1, + error = xfs_da3_path_shift(state, &state->path, 1, 1, &retval); if (error) return(error); @@ -1285,16 +1636,52 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result) *========================================================================*/ /* + * Compare two intermediate nodes for "order". + */ +STATIC int +xfs_da3_node_order( + struct xfs_buf *node1_bp, + struct xfs_buf *node2_bp) +{ + struct xfs_da_intnode *node1; + struct xfs_da_intnode *node2; + struct xfs_da_node_entry *btree1; + struct xfs_da_node_entry *btree2; + struct xfs_da3_icnode_hdr node1hdr; + struct xfs_da3_icnode_hdr node2hdr; + + node1 = node1_bp->b_addr; + node2 = node2_bp->b_addr; + xfs_da3_node_hdr_from_disk(&node1hdr, node1); + xfs_da3_node_hdr_from_disk(&node2hdr, node2); + btree1 = xfs_da3_node_tree_p(node1); + btree2 = xfs_da3_node_tree_p(node2); + + if (node1hdr.count > 0 && node2hdr.count > 0 && + ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) || + (be32_to_cpu(btree2[node2hdr.count - 1].hashval) < + be32_to_cpu(btree1[node1hdr.count - 1].hashval)))) { + return 1; + } + return 0; +} + +/* * Link a new block into a doubly linked list of blocks (of whatever type). */ int /* error */ -xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, - xfs_da_state_blk_t *new_blk) +xfs_da3_blk_link( + struct xfs_da_state *state, + struct xfs_da_state_blk *old_blk, + struct xfs_da_state_blk *new_blk) { - xfs_da_blkinfo_t *old_info, *new_info, *tmp_info; - xfs_da_args_t *args; - int before=0, error; - struct xfs_buf *bp; + struct xfs_da_blkinfo *old_info; + struct xfs_da_blkinfo *new_info; + struct xfs_da_blkinfo *tmp_info; + struct xfs_da_args *args; + struct xfs_buf *bp; + int before = 0; + int error; /* * Set up environment. @@ -1306,9 +1693,6 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC || old_blk->magic == XFS_DIR2_LEAFN_MAGIC || old_blk->magic == XFS_ATTR_LEAF_MAGIC); - ASSERT(old_blk->magic == be16_to_cpu(old_info->magic)); - ASSERT(new_blk->magic == be16_to_cpu(new_info->magic)); - ASSERT(old_blk->magic == new_blk->magic); switch (old_blk->magic) { case XFS_ATTR_LEAF_MAGIC: @@ -1318,7 +1702,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, before = xfs_dir2_leafn_order(old_blk->bp, new_blk->bp); break; case XFS_DA_NODE_MAGIC: - before = xfs_da_node_order(old_blk->bp, new_blk->bp); + before = xfs_da3_node_order(old_blk->bp, new_blk->bp); break; } @@ -1333,14 +1717,14 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, new_info->forw = cpu_to_be32(old_blk->blkno); new_info->back = old_info->back; if (old_info->back) { - error = xfs_da_node_read(args->trans, args->dp, + error = xfs_da3_node_read(args->trans, args->dp, be32_to_cpu(old_info->back), -1, &bp, args->whichfork); if (error) return(error); ASSERT(bp != NULL); tmp_info = bp->b_addr; - ASSERT(be16_to_cpu(tmp_info->magic) == be16_to_cpu(old_info->magic)); + ASSERT(tmp_info->magic == old_info->magic); ASSERT(be32_to_cpu(tmp_info->forw) == old_blk->blkno); tmp_info->forw = cpu_to_be32(new_blk->blkno); xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1); @@ -1354,7 +1738,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, new_info->forw = old_info->forw; new_info->back = cpu_to_be32(old_blk->blkno); if (old_info->forw) { - error = xfs_da_node_read(args->trans, args->dp, + error = xfs_da3_node_read(args->trans, args->dp, be32_to_cpu(old_info->forw), -1, &bp, args->whichfork); if (error) @@ -1375,59 +1759,20 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk, } /* - * Compare two intermediate nodes for "order". - */ -STATIC int -xfs_da_node_order( - struct xfs_buf *node1_bp, - struct xfs_buf *node2_bp) -{ - xfs_da_intnode_t *node1, *node2; - - node1 = node1_bp->b_addr; - node2 = node2_bp->b_addr; - ASSERT(node1->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) && - node2->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) && - ((be32_to_cpu(node2->btree[0].hashval) < - be32_to_cpu(node1->btree[0].hashval)) || - (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) < - be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) { - return(1); - } - return(0); -} - -/* - * Pick up the last hashvalue from an intermediate node. - */ -STATIC uint -xfs_da_node_lasthash( - struct xfs_buf *bp, - int *count) -{ - xfs_da_intnode_t *node; - - node = bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - if (count) - *count = be16_to_cpu(node->hdr.count); - if (!node->hdr.count) - return(0); - return be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval); -} - -/* * Unlink a block from a doubly linked list of blocks. */ STATIC int /* error */ -xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, - xfs_da_state_blk_t *save_blk) +xfs_da3_blk_unlink( + struct xfs_da_state *state, + struct xfs_da_state_blk *drop_blk, + struct xfs_da_state_blk *save_blk) { - xfs_da_blkinfo_t *drop_info, *save_info, *tmp_info; - xfs_da_args_t *args; - struct xfs_buf *bp; - int error; + struct xfs_da_blkinfo *drop_info; + struct xfs_da_blkinfo *save_info; + struct xfs_da_blkinfo *tmp_info; + struct xfs_da_args *args; + struct xfs_buf *bp; + int error; /* * Set up environment. @@ -1439,8 +1784,6 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC || save_blk->magic == XFS_DIR2_LEAFN_MAGIC || save_blk->magic == XFS_ATTR_LEAF_MAGIC); - ASSERT(save_blk->magic == be16_to_cpu(save_info->magic)); - ASSERT(drop_blk->magic == be16_to_cpu(drop_info->magic)); ASSERT(save_blk->magic == drop_blk->magic); ASSERT((be32_to_cpu(save_info->forw) == drop_blk->blkno) || (be32_to_cpu(save_info->back) == drop_blk->blkno)); @@ -1454,7 +1797,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, trace_xfs_da_unlink_back(args); save_info->back = drop_info->back; if (drop_info->back) { - error = xfs_da_node_read(args->trans, args->dp, + error = xfs_da3_node_read(args->trans, args->dp, be32_to_cpu(drop_info->back), -1, &bp, args->whichfork); if (error) @@ -1471,7 +1814,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, trace_xfs_da_unlink_forward(args); save_info->forw = drop_info->forw; if (drop_info->forw) { - error = xfs_da_node_read(args->trans, args->dp, + error = xfs_da3_node_read(args->trans, args->dp, be32_to_cpu(drop_info->forw), -1, &bp, args->whichfork); if (error) @@ -1499,15 +1842,22 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk, * the new bottom and the root. */ int /* error */ -xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path, - int forward, int release, int *result) +xfs_da3_path_shift( + struct xfs_da_state *state, + struct xfs_da_state_path *path, + int forward, + int release, + int *result) { - xfs_da_state_blk_t *blk; - xfs_da_blkinfo_t *info; - xfs_da_intnode_t *node; - xfs_da_args_t *args; - xfs_dablk_t blkno=0; - int level, error; + struct xfs_da_state_blk *blk; + struct xfs_da_blkinfo *info; + struct xfs_da_intnode *node; + struct xfs_da_args *args; + struct xfs_da_node_entry *btree; + struct xfs_da3_icnode_hdr nodehdr; + xfs_dablk_t blkno = 0; + int level; + int error; trace_xfs_da_path_shift(state->args); @@ -1522,16 +1872,17 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path, ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH)); level = (path->active-1) - 1; /* skip bottom layer in path */ for (blk = &path->blk[level]; level >= 0; blk--, level--) { - ASSERT(blk->bp != NULL); node = blk->bp->b_addr; - ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); - if (forward && (blk->index < be16_to_cpu(node->hdr.count)-1)) { + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); + + if (forward && (blk->index < nodehdr.count - 1)) { blk->index++; - blkno = be32_to_cpu(node->btree[blk->index].before); + blkno = be32_to_cpu(btree[blk->index].before); break; } else if (!forward && (blk->index > 0)) { blk->index--; - blkno = be32_to_cpu(node->btree[blk->index].before); + blkno = be32_to_cpu(btree[blk->index].before); break; } } @@ -1557,45 +1908,60 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path, * Read the next child block. */ blk->blkno = blkno; - error = xfs_da_node_read(args->trans, args->dp, blkno, -1, + error = xfs_da3_node_read(args->trans, args->dp, blkno, -1, &blk->bp, args->whichfork); if (error) return(error); - ASSERT(blk->bp != NULL); info = blk->bp->b_addr; ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) || + info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) || info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) || - info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC)); - blk->magic = be16_to_cpu(info->magic); - if (blk->magic == XFS_DA_NODE_MAGIC) { + info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) || + info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) || + info->magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC)); + + + /* + * Note: we flatten the magic number to a single type so we + * don't have to compare against crc/non-crc types elsewhere. + */ + switch (be16_to_cpu(info->magic)) { + case XFS_DA_NODE_MAGIC: + case XFS_DA3_NODE_MAGIC: + blk->magic = XFS_DA_NODE_MAGIC; node = (xfs_da_intnode_t *)info; - blk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval); + xfs_da3_node_hdr_from_disk(&nodehdr, node); + btree = xfs_da3_node_tree_p(node); + blk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval); if (forward) blk->index = 0; else - blk->index = be16_to_cpu(node->hdr.count)-1; - blkno = be32_to_cpu(node->btree[blk->index].before); - } else { + blk->index = nodehdr.count - 1; + blkno = be32_to_cpu(btree[blk->index].before); + break; + case XFS_ATTR_LEAF_MAGIC: + case XFS_ATTR3_LEAF_MAGIC: + blk->magic = XFS_ATTR_LEAF_MAGIC; ASSERT(level == path->active-1); blk->index = 0; - switch(blk->magic) { - case XFS_ATTR_LEAF_MAGIC: - blk->hashval = xfs_attr_leaf_lasthash(blk->bp, - NULL); - break; - case XFS_DIR2_LEAFN_MAGIC: - blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, - NULL); - break; - default: - ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC || - blk->magic == XFS_DIR2_LEAFN_MAGIC); - break; - } + blk->hashval = xfs_attr_leaf_lasthash(blk->bp, + NULL); + break; + case XFS_DIR2_LEAFN_MAGIC: + case XFS_DIR3_LEAFN_MAGIC: + blk->magic = XFS_DIR2_LEAFN_MAGIC; + ASSERT(level == path->active-1); + blk->index = 0; + blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, + NULL); + break; + default: + ASSERT(0); + break; } } *result = 0; - return(0); + return 0; } @@ -1782,22 +2148,36 @@ xfs_da_grow_inode( * a bmap btree split to do that. */ STATIC int -xfs_da_swap_lastblock( - xfs_da_args_t *args, - xfs_dablk_t *dead_blknop, - struct xfs_buf **dead_bufp) +xfs_da3_swap_lastblock( + struct xfs_da_args *args, + xfs_dablk_t *dead_blknop, + struct xfs_buf **dead_bufp) { - xfs_dablk_t dead_blkno, last_blkno, sib_blkno, par_blkno; - struct xfs_buf *dead_buf, *last_buf, *sib_buf, *par_buf; - xfs_fileoff_t lastoff; - xfs_inode_t *ip; - xfs_trans_t *tp; - xfs_mount_t *mp; - int error, w, entno, level, dead_level; - xfs_da_blkinfo_t *dead_info, *sib_info; - xfs_da_intnode_t *par_node, *dead_node; - xfs_dir2_leaf_t *dead_leaf2; - xfs_dahash_t dead_hash; + struct xfs_da_blkinfo *dead_info; + struct xfs_da_blkinfo *sib_info; + struct xfs_da_intnode *par_node; + struct xfs_da_intnode *dead_node; + struct xfs_dir2_leaf *dead_leaf2; + struct xfs_da_node_entry *btree; + struct xfs_da3_icnode_hdr par_hdr; + struct xfs_inode *ip; + struct xfs_trans *tp; + struct xfs_mount *mp; + struct xfs_buf *dead_buf; + struct xfs_buf *last_buf; + struct xfs_buf *sib_buf; + struct xfs_buf *par_buf; + xfs_dahash_t dead_hash; + xfs_fileoff_t lastoff; + xfs_dablk_t dead_blkno; + xfs_dablk_t last_blkno; + xfs_dablk_t sib_blkno; + xfs_dablk_t par_blkno; + int error; + int w; + int entno; + int level; + int dead_level; trace_xfs_da_swap_lastblock(args); @@ -1821,7 +2201,7 @@ xfs_da_swap_lastblock( * Read the last block in the btree space. */ last_blkno = (xfs_dablk_t)lastoff - mp->m_dirblkfsbs; - error = xfs_da_node_read(tp, ip, last_blkno, -1, &last_buf, w); + error = xfs_da3_node_read(tp, ip, last_blkno, -1, &last_buf, w); if (error) return error; /* @@ -1833,22 +2213,31 @@ xfs_da_swap_lastblock( /* * Get values from the moved block. */ - if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)) { + if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) || + dead_info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) { + struct xfs_dir3_icleaf_hdr leafhdr; + struct xfs_dir2_leaf_entry *ents; + dead_leaf2 = (xfs_dir2_leaf_t *)dead_info; + xfs_dir3_leaf_hdr_from_disk(&leafhdr, dead_leaf2); + ents = xfs_dir3_leaf_ents_p(dead_leaf2); dead_level = 0; - dead_hash = be32_to_cpu(dead_leaf2->ents[be16_to_cpu(dead_leaf2->hdr.count) - 1].hashval); + dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval); } else { - ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC)); + struct xfs_da3_icnode_hdr deadhdr; + dead_node = (xfs_da_intnode_t *)dead_info; - dead_level = be16_to_cpu(dead_node->hdr.level); - dead_hash = be32_to_cpu(dead_node->btree[be16_to_cpu(dead_node->hdr.count) - 1].hashval); + xfs_da3_node_hdr_from_disk(&deadhdr, dead_node); + btree = xfs_da3_node_tree_p(dead_node); + dead_level = deadhdr.level; + dead_hash = be32_to_cpu(btree[deadhdr.count - 1].hashval); } sib_buf = par_buf = NULL; /* * If the moved block has a left sibling, fix up the pointers. */ if ((sib_blkno = be32_to_cpu(dead_info->back))) { - error = xfs_da_node_read(tp, ip, sib_blkno, -1, &sib_buf, w); + error = xfs_da3_node_read(tp, ip, sib_blkno, -1, &sib_buf, w); if (error) goto done; sib_info = sib_buf->b_addr; @@ -1870,7 +2259,7 @@ xfs_da_swap_lastblock( * If the moved block has a right sibling, fix up the pointers. */ if ((sib_blkno = be32_to_cpu(dead_info->forw))) { - error = xfs_da_node_read(tp, ip, sib_blkno, -1, &sib_buf, w); + error = xfs_da3_node_read(tp, ip, sib_blkno, -1, &sib_buf, w); if (error) goto done; sib_info = sib_buf->b_addr; @@ -1894,31 +2283,31 @@ xfs_da_swap_lastblock( * Walk down the tree looking for the parent of the moved block. */ for (;;) { - error = xfs_da_node_read(tp, ip, par_blkno, -1, &par_buf, w); + error = xfs_da3_node_read(tp, ip, par_blkno, -1, &par_buf, w); if (error) goto done; par_node = par_buf->b_addr; - if (unlikely(par_node->hdr.info.magic != - cpu_to_be16(XFS_DA_NODE_MAGIC) || - (level >= 0 && level != be16_to_cpu(par_node->hdr.level) + 1))) { + xfs_da3_node_hdr_from_disk(&par_hdr, par_node); + if (level >= 0 && level != par_hdr.level + 1) { XFS_ERROR_REPORT("xfs_da_swap_lastblock(4)", XFS_ERRLEVEL_LOW, mp); error = XFS_ERROR(EFSCORRUPTED); goto done; } - level = be16_to_cpu(par_node->hdr.level); + level = par_hdr.level; + btree = xfs_da3_node_tree_p(par_node); for (entno = 0; - entno < be16_to_cpu(par_node->hdr.count) && - be32_to_cpu(par_node->btree[entno].hashval) < dead_hash; + entno < par_hdr.count && + be32_to_cpu(btree[entno].hashval) < dead_hash; entno++) continue; - if (unlikely(entno == be16_to_cpu(par_node->hdr.count))) { + if (entno == par_hdr.count) { XFS_ERROR_REPORT("xfs_da_swap_lastblock(5)", XFS_ERRLEVEL_LOW, mp); error = XFS_ERROR(EFSCORRUPTED); goto done; } - par_blkno = be32_to_cpu(par_node->btree[entno].before); + par_blkno = be32_to_cpu(btree[entno].before); if (level == dead_level + 1) break; xfs_trans_brelse(tp, par_buf); @@ -1930,13 +2319,13 @@ xfs_da_swap_lastblock( */ for (;;) { for (; - entno < be16_to_cpu(par_node->hdr.count) && - be32_to_cpu(par_node->btree[entno].before) != last_blkno; + entno < par_hdr.count && + be32_to_cpu(btree[entno].before) != last_blkno; entno++) continue; - if (entno < be16_to_cpu(par_node->hdr.count)) + if (entno < par_hdr.count) break; - par_blkno = be32_to_cpu(par_node->hdr.info.forw); + par_blkno = par_hdr.forw; xfs_trans_brelse(tp, par_buf); par_buf = NULL; if (unlikely(par_blkno == 0)) { @@ -1945,27 +2334,27 @@ xfs_da_swap_lastblock( error = XFS_ERROR(EFSCORRUPTED); goto done; } - error = xfs_da_node_read(tp, ip, par_blkno, -1, &par_buf, w); + error = xfs_da3_node_read(tp, ip, par_blkno, -1, &par_buf, w); if (error) goto done; par_node = par_buf->b_addr; - if (unlikely( - be16_to_cpu(par_node->hdr.level) != level || - par_node->hdr.info.magic != cpu_to_be16(XFS_DA_NODE_MAGIC))) { + xfs_da3_node_hdr_from_disk(&par_hdr, par_node); + if (par_hdr.level != level) { XFS_ERROR_REPORT("xfs_da_swap_lastblock(7)", XFS_ERRLEVEL_LOW, mp); error = XFS_ERROR(EFSCORRUPTED); goto done; } + btree = xfs_da3_node_tree_p(par_node); entno = 0; } /* * Update the parent entry pointing to the moved block. */ - par_node->btree[entno].before = cpu_to_be32(dead_blkno); + btree[entno].before = cpu_to_be32(dead_blkno); xfs_trans_log_buf(tp, par_buf, - XFS_DA_LOGRANGE(par_node, &par_node->btree[entno].before, - sizeof(par_node->btree[entno].before))); + XFS_DA_LOGRANGE(par_node, &btree[entno].before, + sizeof(btree[entno].before))); *dead_blknop = last_blkno; *dead_bufp = last_buf; return 0; @@ -2007,14 +2396,15 @@ xfs_da_shrink_inode( * Remove extents. If we get ENOSPC for a dir we have to move * the last block to the place we want to kill. */ - if ((error = xfs_bunmapi(tp, dp, dead_blkno, count, - xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA, - 0, args->firstblock, args->flist, - &done)) == ENOSPC) { + error = xfs_bunmapi(tp, dp, dead_blkno, count, + xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA, + 0, args->firstblock, args->flist, &done); + if (error == ENOSPC) { if (w != XFS_DATA_FORK) break; - if ((error = xfs_da_swap_lastblock(args, &dead_blkno, - &dead_buf))) + error = xfs_da3_swap_lastblock(args, &dead_blkno, + &dead_buf); + if (error) break; } else { break; @@ -2279,12 +2669,21 @@ xfs_da_read_buf( magic1 = be32_to_cpu(hdr->magic); if (unlikely( XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) && + (magic != XFS_DA3_NODE_MAGIC) && (magic != XFS_ATTR_LEAF_MAGIC) && + (magic != XFS_ATTR3_LEAF_MAGIC) && (magic != XFS_DIR2_LEAF1_MAGIC) && + (magic != XFS_DIR3_LEAF1_MAGIC) && (magic != XFS_DIR2_LEAFN_MAGIC) && + (magic != XFS_DIR3_LEAFN_MAGIC) && (magic1 != XFS_DIR2_BLOCK_MAGIC) && + (magic1 != XFS_DIR3_BLOCK_MAGIC) && (magic1 != XFS_DIR2_DATA_MAGIC) && - (free->hdr.magic != cpu_to_be32(XFS_DIR2_FREE_MAGIC)), + (magic1 != XFS_DIR3_DATA_MAGIC) && + (free->hdr.magic != + cpu_to_be32(XFS_DIR2_FREE_MAGIC)) && + (free->hdr.magic != + cpu_to_be32(XFS_DIR3_FREE_MAGIC)), mp, XFS_ERRTAG_DA_READ_BUF, XFS_RANDOM_DA_READ_BUF))) { trace_xfs_da_btree_corrupt(bp, _RET_IP_); @@ -2342,41 +2741,3 @@ out_free: return -1; return mappedbno; } - -kmem_zone_t *xfs_da_state_zone; /* anchor for state struct zone */ - -/* - * Allocate a dir-state structure. - * We don't put them on the stack since they're large. - */ -xfs_da_state_t * -xfs_da_state_alloc(void) -{ - return kmem_zone_zalloc(xfs_da_state_zone, KM_NOFS); -} - -/* - * Kill the altpath contents of a da-state structure. - */ -STATIC void -xfs_da_state_kill_altpath(xfs_da_state_t *state) -{ - int i; - - for (i = 0; i < state->altpath.active; i++) - state->altpath.blk[i].bp = NULL; - state->altpath.active = 0; -} - -/* - * Free a da-state structure. - */ -void -xfs_da_state_free(xfs_da_state_t *state) -{ - xfs_da_state_kill_altpath(state); -#ifdef DEBUG - memset((char *)state, 0, sizeof(*state)); -#endif /* DEBUG */ - kmem_zone_free(xfs_da_state_zone, state); -} |