/* * Allocate an inode on disk. * Mode is used to tell whether the new inode will need space, and whether * it is a directory. * * The arguments IO_agbp and alloc_done are defined to work within * the constraint of one allocation per transaction. * xfs_dialloc() is designed to be called twice if it has to do an * allocation to make more free inodes. On the first call, * IO_agbp should be set to NULL. If an inode is available, * i.e., xfs_dialloc() did not need to do an allocation, an inode * number is returned. In this case, IO_agbp would be set to the * current ag_buf and alloc_done set to false. * If an allocation needed to be done, xfs_dialloc would return * the current ag_buf in IO_agbp and set alloc_done to true. * The caller should then commit the current transaction, allocate a new * transaction, and call xfs_dialloc() again, passing in the previous * value of IO_agbp. IO_agbp should be held across the transactions. * Since the agbp is locked across the two calls, the second call is * guaranteed to have a free inode available. * * Once we successfully pick an inode its number is returned and the * on-disk data structures are updated. The inode itself is not read * in, since doing so would break ordering constraints with xfs_reclaim. */ int xfs_dialloc( xfs_trans_t *tp, /* transaction pointer */ xfs_ino_t parent, /* parent inode (directory) */ mode_t mode, /* mode bits for new inode */ int okalloc, /* ok to allocate more space */ xfs_buf_t **IO_agbp, /* in/out ag header's buffer */ boolean_t *alloc_done, /* true if we needed to replenish inode freelist */ xfs_ino_t *inop) /* inode number allocated */ { xfs_agnumber_t agcount; /* number of allocation groups */ xfs_buf_t *agbp; /* allocation group header's buffer */ xfs_agnumber_t agno; /* allocation group number */ xfs_agi_t *agi; /* allocation group header structure */ xfs_btree_cur_t *cur; /* inode allocation btree cursor */ int error; /* error return value */ int i; /* result code */ int ialloced; /* inode allocation status */ int noroom = 0; /* no space for inode blk allocation */ xfs_ino_t ino; /* fs-relative inode to be returned */ /* REFERENCED */ int j; /* result code */ xfs_mount_t *mp; /* file system mount structure */ int offset; /* index of inode in chunk */ xfs_agino_t pagino; /* parent's a.g. relative inode # */ xfs_agnumber_t pagno; /* parent's allocation group number */ xfs_inobt_rec_t rec; /* inode allocation record */ xfs_agnumber_t tagno; /* testing allocation group number */ xfs_btree_cur_t *tcur; /* temp cursor */ xfs_inobt_rec_t trec; /* temp inode allocation record */ if (*IO_agbp == NULL) { /* * We do not have an agbp, so select an initial allocation * group for inode allocation. */ agbp = xfs_ialloc_ag_select(tp, parent, mode, okalloc); /* * Couldn't find an allocation group satisfying the * criteria, give up. */ if (!agbp) { *inop = NULLFSINO; return 0; } agi = XFS_BUF_TO_AGI(agbp); ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC); } else { /* * Continue where we left off before. In this case, we * know that the allocation group has free inodes. */ agbp = *IO_agbp; agi = XFS_BUF_TO_AGI(agbp); ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC); ASSERT(be32_to_cpu(agi->agi_freecount) > 0); } mp = tp->t_mountp; agcount = mp->m_sb.sb_agcount; agno = be32_to_cpu(agi->agi_seqno); tagno = agno; pagno = XFS_INO_TO_AGNO(mp, parent); pagino = XFS_INO_TO_AGINO(mp, parent); /* * If we have already hit the ceiling of inode blocks then clear * okalloc so we scan all available agi structures for a free * inode. */ if (mp->m_maxicount && mp->m_sb.sb_icount + XFS_IALLOC_INODES(mp) > mp->m_maxicount) { noroom = 1; okalloc = 0; } /* * Loop until we find an allocation group that either has free inodes * or in which we can allocate some inodes. Iterate through the * allocation groups upward, wrapping at the end. */ *alloc_done = B_FALSE; while (!agi->agi_freecount) { /* * Don't do anything if we're not supposed to allocate * any blocks, just go on to the next ag. */ if (okalloc) { /* * Try to allocate some new inodes in the allocation * group. */ if ((error = xfs_ialloc_ag_alloc(tp, agbp, &ialloced))) { xfs_trans_brelse(tp, agbp); if (error == ENOSPC) { *inop = NULLFSINO; return 0; } else return error; } if (ialloced) { /* * We successfully allocated some inodes, return * the current context to the caller so that it * can commit the current transaction and call * us again where we left off. */ ASSERT(be32_to_cpu(agi->agi_freecount) > 0); *alloc_done = B_TRUE; *IO_agbp = agbp; *inop = NULLFSINO; return 0; } } /* * If it failed, give up on this ag. */ xfs_trans_brelse(tp, agbp); /* * Go on to the next ag: get its ag header. */ nextag: if (++tagno == agcount) tagno = 0; if (tagno == agno) { *inop = NULLFSINO; return noroom ? ENOSPC : 0; } down_read(&mp->m_peraglock); if (mp->m_perag[tagno].pagi_inodeok == 0) { up_read(&mp->m_peraglock); goto nextag; } error = xfs_ialloc_read_agi(mp, tp, tagno, &agbp); up_read(&mp->m_peraglock); if (error) goto nextag; agi = XFS_BUF_TO_AGI(agbp); ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC); } /* * Here with an allocation group that has a free inode. * Reset agno since we may have chosen a new ag in the * loop above. */ agno = tagno; *IO_agbp = NULL; cur = xfs_btree_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno), XFS_BTNUM_INO, (xfs_inode_t *)0, 0); /* * If pagino is 0 (this is the root inode allocation) use newino. * This must work because we've just allocated some. */ if (!pagino) pagino = be32_to_cpu(agi->agi_newino); #ifdef DEBUG if (cur->bc_nlevels == 1) { int freecount = 0; if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i))) goto error0; XFS_WANT_CORRUPTED_GOTO(i == 1, error0); do { if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &i))) goto error0; XFS_WANT_CORRUPTED_GOTO(i == 1, error0); freecount += rec.ir_freecount; if ((error = xfs_inobt_increment(cur, 0, &i))) goto error0; } while (i == 1); ASSERT(freecount == be32_to_cpu(agi->agi_freecount) || XFS_FORCED_SHUTDOWN(mp)); } #endif /* * If in the same a.g. as the parent, try to get near the parent. */ if (pagno == agno) { if ((error = xfs_inobt_lookup_le(cur, pagino, 0, 0, &i))) goto error0; if (i != 0 && (error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &j)) == 0 && j == 1 && rec.ir_freecount > 0) { /* * Found a free inode in the same chunk * as parent, done. */ } /* * In the same a.g. as parent, but parent's chunk is full. */ else { int doneleft; /* done, to the left */ int doneright; /* done, to the right */ if (error) goto error0; ASSERT(i == 1); ASSERT(j == 1); /* * Duplicate the cursor, search left & right * simultaneously. */ if ((error = xfs_btree_dup_cursor(cur, &tcur))) goto error0; /* * Search left with tcur, back up 1 record. */ if ((error = xfs_inobt_decrement(tcur, 0, &i))) goto error1; doneleft = !i; if (!doneleft) { if ((error = xfs_inobt_get_rec(tcur, &trec.ir_startino, &trec.ir_freecount, &trec.ir_free, &i))) goto error1; XFS_WANT_CORRUPTED_GOTO(i == 1, error1); } /* * Search right with cur, go forward 1 record. */ if ((error = xfs_inobt_increment(cur, 0, &i))) goto error1; doneright = !i; if (!doneright) { if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &i))) goto error1; XFS_WANT_CORRUPTED_GOTO(i == 1, error1); } /* * Loop until we find the closest inode chunk * with a free one. */ while (!doneleft || !doneright) { int useleft; /* using left inode chunk this time */ /* * Figure out which block is closer, * if both are valid. */ if (!doneleft && !doneright) useleft = pagino - (trec.ir_startino + XFS_INODES_PER_CHUNK - 1) < rec.ir_startino - pagino; else useleft = !doneleft; /* * If checking the left, does it have * free inodes? */ if (useleft && trec.ir_freecount) { /* * Yes, set it up as the chunk to use. */ rec = trec; xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR); cur = tcur; break; } /* * If checking the right, does it have * free inodes? */ if (!useleft && rec.ir_freecount) { /* * Yes, it's already set up. */ xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR); break; } /* * If used the left, get another one * further left. */ if (useleft) { if ((error = xfs_inobt_decrement(tcur, 0, &i))) goto error1; doneleft = !i; if (!doneleft) { if ((error = xfs_inobt_get_rec( tcur, &trec.ir_startino, &trec.ir_freecount, &trec.ir_free, &i))) goto error1; XFS_WANT_CORRUPTED_GOTO(i == 1, error1); } } /* * If used the right, get another one * further right. */ else { if ((error = xfs_inobt_increment(cur, 0, &i))) goto error1; doneright = !i; if (!doneright) { if ((error = xfs_inobt_get_rec( cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &i))) goto error1; XFS_WANT_CORRUPTED_GOTO(i == 1, error1); } } } ASSERT(!doneleft || !doneright); } } /* * In a different a.g. from the parent. * See if the most recently allocated block has any free. */ else if (be32_to_cpu(agi->agi_newino) != NULLAGINO) { if ((error = xfs_inobt_lookup_eq(cur, be32_to_cpu(agi->agi_newino), 0, 0, &i))) goto error0; if (i == 1 && (error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &j)) == 0 && j == 1 && rec.ir_freecount > 0) { /* * The last chunk allocated in the group still has * a free inode. */ } /* * None left in the last group, search the whole a.g. */ else { if (error) goto error0; if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i))) goto error0; ASSERT(i == 1); for (;;) { if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &i))) goto error0; XFS_WANT_CORRUPTED_GOTO(i == 1, error0); if (rec.ir_freecount > 0) break; if ((error = xfs_inobt_increment(cur, 0, &i))) goto error0; XFS_WANT_CORRUPTED_GOTO(i == 1, error0); } } } offset = XFS_IALLOC_FIND_FREE(&rec.ir_free); ASSERT(offset >= 0); ASSERT(offset < XFS_INODES_PER_CHUNK); ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) % XFS_INODES_PER_CHUNK) == 0); ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset); XFS_INOBT_CLR_FREE(&rec, offset); rec.ir_freecount--; if ((error = xfs_inobt_update(cur, rec.ir_startino, rec.ir_freecount, rec.ir_free))) goto error0; be32_add(&agi->agi_freecount, -1); xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT); down_read(&mp->m_peraglock); mp->m_perag[tagno].pagi_freecount--; up_read(&mp->m_peraglock); #ifdef DEBUG if (cur->bc_nlevels == 1) { int freecount = 0; if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i))) goto error0; do { if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount, &rec.ir_free, &i))) goto error0; XFS_WANT_CORRUPTED_GOTO(i == 1, error0); freecount += rec.ir_freecount; if ((error = xfs_inobt_increment(cur, 0, &i))) goto error0; } while (i == 1); ASSERT(freecount == be32_to_cpu(agi->agi_freecount) || XFS_FORCED_SHUTDOWN(mp)); } #endif xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR); xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1); *inop = ino; return 0; error1: xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR); error0: xfs_btree_del_cursor(cur, XFS_BTREE_ERROR); return error; }
/* * Remove an entry from a block format directory. * If that makes the block small enough to fit in shortform, transform it. */ int /* error */ xfs_dir2_block_removename( xfs_da_args_t *args) /* directory operation args */ { xfs_dir2_block_t *block; /* block structure */ xfs_dir2_leaf_entry_t *blp; /* block leaf pointer */ xfs_dabuf_t *bp; /* block buffer */ xfs_dir2_block_tail_t *btp; /* block tail */ xfs_dir2_data_entry_t *dep; /* block data entry */ xfs_inode_t *dp; /* incore inode */ int ent; /* block leaf entry index */ int error; /* error return value */ xfs_mount_t *mp; /* filesystem mount point */ int needlog; /* need to log block header */ int needscan; /* need to fixup bestfree */ xfs_dir2_sf_hdr_t sfh; /* shortform header */ int size; /* shortform size */ xfs_trans_t *tp; /* transaction pointer */ xfs_dir2_trace_args("block_removename", args); /* * Look up the entry in the block. Gets the buffer and entry index. * It will always be there, the vnodeops level does a lookup first. */ if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent))) { return error; } dp = args->dp; tp = args->trans; mp = dp->i_mount; block = bp->data; btp = xfs_dir2_block_tail_p(mp, block); blp = xfs_dir2_block_leaf_p(btp); /* * Point to the data entry using the leaf entry. */ dep = (xfs_dir2_data_entry_t *) ((char *)block + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(blp[ent].address))); /* * Mark the data entry's space free. */ needlog = needscan = 0; xfs_dir2_data_make_free(tp, bp, (xfs_dir2_data_aoff_t)((char *)dep - (char *)block), xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan); /* * Fix up the block tail. */ be32_add(&btp->stale, 1); xfs_dir2_block_log_tail(tp, bp); /* * Remove the leaf entry by marking it stale. */ blp[ent].address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR); xfs_dir2_block_log_leaf(tp, bp, ent, ent); /* * Fix up bestfree, log the header if necessary. */ if (needscan) xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog); if (needlog) xfs_dir2_data_log_header(tp, bp); xfs_dir2_data_check(dp, bp); /* * See if the size as a shortform is good enough. */ if ((size = xfs_dir2_block_sfsize(dp, block, &sfh)) > XFS_IFORK_DSIZE(dp)) { xfs_da_buf_done(bp); return 0; } /* * If it works, do the conversion. */ return xfs_dir2_block_to_sf(args, bp, size, &sfh); }
/* * Allocate new inodes in the allocation group specified by agbp. * Return 0 for success, else error code. */ STATIC int /* error code or 0 */ xfs_ialloc_ag_alloc( xfs_trans_t *tp, /* transaction pointer */ xfs_buf_t *agbp, /* alloc group buffer */ int *alloc) { xfs_agi_t *agi; /* allocation group header */ xfs_alloc_arg_t args; /* allocation argument structure */ int blks_per_cluster; /* fs blocks per inode cluster */ xfs_btree_cur_t *cur; /* inode btree cursor */ xfs_daddr_t d; /* disk addr of buffer */ int error; xfs_buf_t *fbuf; /* new free inodes' buffer */ xfs_dinode_t *free; /* new free inode structure */ int i; /* inode counter */ int j; /* block counter */ int nbufs; /* num bufs of new inodes */ xfs_agino_t newino; /* new first inode's number */ xfs_agino_t newlen; /* new number of inodes */ int ninodes; /* num inodes per buf */ xfs_agino_t thisino; /* current inode number, for loop */ int version; /* inode version number to use */ int isaligned = 0; /* inode allocation at stripe unit */ /* boundary */ args.tp = tp; args.mp = tp->t_mountp; /* * Locking will ensure that we don't have two callers in here * at one time. */ newlen = XFS_IALLOC_INODES(args.mp); if (args.mp->m_maxicount && args.mp->m_sb.sb_icount + newlen > args.mp->m_maxicount) return XFS_ERROR(ENOSPC); args.minlen = args.maxlen = XFS_IALLOC_BLOCKS(args.mp); /* * First try to allocate inodes contiguous with the last-allocated * chunk of inodes. If the filesystem is striped, this will fill * an entire stripe unit with inodes. */ agi = XFS_BUF_TO_AGI(agbp); newino = be32_to_cpu(agi->agi_newino); args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) + XFS_IALLOC_BLOCKS(args.mp); if (likely(newino != NULLAGINO && (args.agbno < be32_to_cpu(agi->agi_length)))) { args.fsbno = XFS_AGB_TO_FSB(args.mp, be32_to_cpu(agi->agi_seqno), args.agbno); args.type = XFS_ALLOCTYPE_THIS_BNO; args.mod = args.total = args.wasdel = args.isfl = args.userdata = args.minalignslop = 0; args.prod = 1; args.alignment = 1; /* * Allow space for the inode btree to split. */ args.minleft = XFS_IN_MAXLEVELS(args.mp) - 1; if ((error = xfs_alloc_vextent(&args))) return error; } else args.fsbno = NULLFSBLOCK; if (unlikely(args.fsbno == NULLFSBLOCK)) { /* * Set the alignment for the allocation. * If stripe alignment is turned on then align at stripe unit * boundary. * If the cluster size is smaller than a filesystem block * then we're doing I/O for inodes in filesystem block size * pieces, so don't need alignment anyway. */ isaligned = 0; if (args.mp->m_sinoalign) { ASSERT(!(args.mp->m_flags & XFS_MOUNT_NOALIGN)); args.alignment = args.mp->m_dalign; isaligned = 1; } else if (XFS_SB_VERSION_HASALIGN(&args.mp->m_sb) && args.mp->m_sb.sb_inoalignmt >= XFS_B_TO_FSBT(args.mp, XFS_INODE_CLUSTER_SIZE(args.mp))) args.alignment = args.mp->m_sb.sb_inoalignmt; else args.alignment = 1; /* * Need to figure out where to allocate the inode blocks. * Ideally they should be spaced out through the a.g. * For now, just allocate blocks up front. */ args.agbno = be32_to_cpu(agi->agi_root); args.fsbno = XFS_AGB_TO_FSB(args.mp, be32_to_cpu(agi->agi_seqno), args.agbno); /* * Allocate a fixed-size extent of inodes. */ args.type = XFS_ALLOCTYPE_NEAR_BNO; args.mod = args.total = args.wasdel = args.isfl = args.userdata = args.minalignslop = 0; args.prod = 1; /* * Allow space for the inode btree to split. */ args.minleft = XFS_IN_MAXLEVELS(args.mp) - 1; if ((error = xfs_alloc_vextent(&args))) return error; } /* * If stripe alignment is turned on, then try again with cluster * alignment. */ if (isaligned && args.fsbno == NULLFSBLOCK) { args.type = XFS_ALLOCTYPE_NEAR_BNO; args.agbno = be32_to_cpu(agi->agi_root); args.fsbno = XFS_AGB_TO_FSB(args.mp, be32_to_cpu(agi->agi_seqno), args.agbno); if (XFS_SB_VERSION_HASALIGN(&args.mp->m_sb) && args.mp->m_sb.sb_inoalignmt >= XFS_B_TO_FSBT(args.mp, XFS_INODE_CLUSTER_SIZE(args.mp))) args.alignment = args.mp->m_sb.sb_inoalignmt; else args.alignment = 1; if ((error = xfs_alloc_vextent(&args))) return error; } if (args.fsbno == NULLFSBLOCK) { *alloc = 0; return 0; } ASSERT(args.len == args.minlen); /* * Convert the results. */ newino = XFS_OFFBNO_TO_AGINO(args.mp, args.agbno, 0); /* * Loop over the new block(s), filling in the inodes. * For small block sizes, manipulate the inodes in buffers * which are multiples of the blocks size. */ if (args.mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(args.mp)) { blks_per_cluster = 1; nbufs = (int)args.len; ninodes = args.mp->m_sb.sb_inopblock; } else { blks_per_cluster = XFS_INODE_CLUSTER_SIZE(args.mp) / args.mp->m_sb.sb_blocksize; nbufs = (int)args.len / blks_per_cluster; ninodes = blks_per_cluster * args.mp->m_sb.sb_inopblock; } /* * Figure out what version number to use in the inodes we create. * If the superblock version has caught up to the one that supports * the new inode format, then use the new inode version. Otherwise * use the old version so that old kernels will continue to be * able to use the file system. */ if (XFS_SB_VERSION_HASNLINK(&args.mp->m_sb)) version = XFS_DINODE_VERSION_2; else version = XFS_DINODE_VERSION_1; for (j = 0; j < nbufs; j++) { /* * Get the block. */ d = XFS_AGB_TO_DADDR(args.mp, be32_to_cpu(agi->agi_seqno), args.agbno + (j * blks_per_cluster)); fbuf = xfs_trans_get_buf(tp, args.mp->m_ddev_targp, d, args.mp->m_bsize * blks_per_cluster, XFS_BUF_LOCK); ASSERT(fbuf); ASSERT(!XFS_BUF_GETERROR(fbuf)); /* * Set initial values for the inodes in this buffer. */ xfs_biozero(fbuf, 0, ninodes << args.mp->m_sb.sb_inodelog); for (i = 0; i < ninodes; i++) { free = XFS_MAKE_IPTR(args.mp, fbuf, i); INT_SET(free->di_core.di_magic, ARCH_CONVERT, XFS_DINODE_MAGIC); INT_SET(free->di_core.di_version, ARCH_CONVERT, version); INT_SET(free->di_next_unlinked, ARCH_CONVERT, NULLAGINO); xfs_ialloc_log_di(tp, fbuf, i, XFS_DI_CORE_BITS | XFS_DI_NEXT_UNLINKED); } xfs_trans_inode_alloc_buf(tp, fbuf); } be32_add(&agi->agi_count, newlen); be32_add(&agi->agi_freecount, newlen); down_read(&args.mp->m_peraglock); args.mp->m_perag[be32_to_cpu(agi->agi_seqno)].pagi_freecount += newlen; up_read(&args.mp->m_peraglock); agi->agi_newino = cpu_to_be32(newino); /* * Insert records describing the new inode chunk into the btree. */ cur = xfs_btree_init_cursor(args.mp, tp, agbp, be32_to_cpu(agi->agi_seqno), XFS_BTNUM_INO, (xfs_inode_t *)0, 0); for (thisino = newino; thisino < newino + newlen; thisino += XFS_INODES_PER_CHUNK) { if ((error = xfs_inobt_lookup_eq(cur, thisino, XFS_INODES_PER_CHUNK, XFS_INOBT_ALL_FREE, &i))) { xfs_btree_del_cursor(cur, XFS_BTREE_ERROR); return error; } ASSERT(i == 0); if ((error = xfs_inobt_insert(cur, &i))) { xfs_btree_del_cursor(cur, XFS_BTREE_ERROR); return error; } ASSERT(i == 1); } xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR); /* * Log allocation group header fields */ xfs_ialloc_log_agi(tp, agbp, XFS_AGI_COUNT | XFS_AGI_FREECOUNT | XFS_AGI_NEWINO); /* * Modify/log superblock values for inode count and inode free count. */ xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, (long)newlen); xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, (long)newlen); *alloc = 1; return 0; }
/* * Add an entry to a block directory. */ int /* error */ xfs_dir2_block_addname( xfs_da_args_t *args) /* directory op arguments */ { xfs_dir2_data_free_t *bf; /* bestfree table in block */ xfs_dir2_block_t *block; /* directory block structure */ xfs_dir2_leaf_entry_t *blp; /* block leaf entries */ xfs_dabuf_t *bp; /* buffer for block */ xfs_dir2_block_tail_t *btp; /* block tail */ int compact; /* need to compact leaf ents */ xfs_dir2_data_entry_t *dep; /* block data entry */ xfs_inode_t *dp; /* directory inode */ xfs_dir2_data_unused_t *dup; /* block unused entry */ int error; /* error return value */ xfs_dir2_data_unused_t *enddup=NULL; /* unused at end of data */ xfs_dahash_t hash; /* hash value of found entry */ int high; /* high index for binary srch */ int highstale; /* high stale index */ int lfloghigh=0; /* last final leaf to log */ int lfloglow=0; /* first final leaf to log */ int len; /* length of the new entry */ int low; /* low index for binary srch */ int lowstale; /* low stale index */ int mid=0; /* midpoint for binary srch */ xfs_mount_t *mp; /* filesystem mount point */ int needlog; /* need to log header */ int needscan; /* need to rescan freespace */ __be16 *tagp; /* pointer to tag value */ xfs_trans_t *tp; /* transaction structure */ xfs_dir2_trace_args("block_addname", args); dp = args->dp; tp = args->trans; mp = dp->i_mount; /* * Read the (one and only) directory block into dabuf bp. */ if ((error = xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, &bp, XFS_DATA_FORK))) { return error; } ASSERT(bp != NULL); block = bp->data; /* * Check the magic number, corrupted if wrong. */ if (unlikely(be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC)) { XFS_CORRUPTION_ERROR("xfs_dir2_block_addname", XFS_ERRLEVEL_LOW, mp, block); xfs_da_brelse(tp, bp); return XFS_ERROR(EFSCORRUPTED); } len = xfs_dir2_data_entsize(args->namelen); /* * Set up pointers to parts of the block. */ bf = block->hdr.bestfree; btp = xfs_dir2_block_tail_p(mp, block); blp = xfs_dir2_block_leaf_p(btp); /* * No stale entries? Need space for entry and new leaf. */ if (!btp->stale) { /* * Tag just before the first leaf entry. */ tagp = (__be16 *)blp - 1; /* * Data object just before the first leaf entry. */ enddup = (xfs_dir2_data_unused_t *)((char *)block + be16_to_cpu(*tagp)); /* * If it's not free then can't do this add without cleaning up: * the space before the first leaf entry needs to be free so it * can be expanded to hold the pointer to the new entry. */ if (be16_to_cpu(enddup->freetag) != XFS_DIR2_DATA_FREE_TAG) dup = enddup = NULL; /* * Check out the biggest freespace and see if it's the same one. */ else { dup = (xfs_dir2_data_unused_t *) ((char *)block + be16_to_cpu(bf[0].offset)); if (dup == enddup) { /* * It is the biggest freespace, is it too small * to hold the new leaf too? */ if (be16_to_cpu(dup->length) < len + (uint)sizeof(*blp)) { /* * Yes, we use the second-largest * entry instead if it works. */ if (be16_to_cpu(bf[1].length) >= len) dup = (xfs_dir2_data_unused_t *) ((char *)block + be16_to_cpu(bf[1].offset)); else dup = NULL; } } else { /* * Not the same free entry, * just check its length. */ if (be16_to_cpu(dup->length) < len) { dup = NULL; } } } compact = 0; } /* * If there are stale entries we'll use one for the leaf. * Is the biggest entry enough to avoid compaction? */ else if (be16_to_cpu(bf[0].length) >= len) { dup = (xfs_dir2_data_unused_t *) ((char *)block + be16_to_cpu(bf[0].offset)); compact = 0; } /* * Will need to compact to make this work. */ else { /* * Tag just before the first leaf entry. */ tagp = (__be16 *)blp - 1; /* * Data object just before the first leaf entry. */ dup = (xfs_dir2_data_unused_t *)((char *)block + be16_to_cpu(*tagp)); /* * If it's not free then the data will go where the * leaf data starts now, if it works at all. */ if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) { if (be16_to_cpu(dup->length) + (be32_to_cpu(btp->stale) - 1) * (uint)sizeof(*blp) < len) dup = NULL; } else if ((be32_to_cpu(btp->stale) - 1) * (uint)sizeof(*blp) < len) dup = NULL; else dup = (xfs_dir2_data_unused_t *)blp; compact = 1; } /* * If this isn't a real add, we're done with the buffer. */ if (args->justcheck) xfs_da_brelse(tp, bp); /* * If we don't have space for the new entry & leaf ... */ if (!dup) { /* * Not trying to actually do anything, or don't have * a space reservation: return no-space. */ if (args->justcheck || args->total == 0) return XFS_ERROR(ENOSPC); /* * Convert to the next larger format. * Then add the new entry in that format. */ error = xfs_dir2_block_to_leaf(args, bp); xfs_da_buf_done(bp); if (error) return error; return xfs_dir2_leaf_addname(args); } /* * Just checking, and it would work, so say so. */ if (args->justcheck) return 0; needlog = needscan = 0; /* * If need to compact the leaf entries, do it now. * Leave the highest-numbered stale entry stale. * XXX should be the one closest to mid but mid is not yet computed. */ if (compact) { int fromidx; /* source leaf index */ int toidx; /* target leaf index */ for (fromidx = toidx = be32_to_cpu(btp->count) - 1, highstale = lfloghigh = -1; fromidx >= 0; fromidx--) { if (be32_to_cpu(blp[fromidx].address) == XFS_DIR2_NULL_DATAPTR) { if (highstale == -1) highstale = toidx; else { if (lfloghigh == -1) lfloghigh = toidx; continue; } } if (fromidx < toidx) blp[toidx] = blp[fromidx]; toidx--; } lfloglow = toidx + 1 - (be32_to_cpu(btp->stale) - 1); lfloghigh -= be32_to_cpu(btp->stale) - 1; be32_add(&btp->count, -(be32_to_cpu(btp->stale) - 1)); xfs_dir2_data_make_free(tp, bp, (xfs_dir2_data_aoff_t)((char *)blp - (char *)block), (xfs_dir2_data_aoff_t)((be32_to_cpu(btp->stale) - 1) * sizeof(*blp)), &needlog, &needscan); blp += be32_to_cpu(btp->stale) - 1; btp->stale = cpu_to_be32(1); /* * If we now need to rebuild the bestfree map, do so. * This needs to happen before the next call to use_free. */ if (needscan) { xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog); needscan = 0; } } /* * Set leaf logging boundaries to impossible state. * For the no-stale case they're set explicitly. */ else if (btp->stale) { lfloglow = be32_to_cpu(btp->count); lfloghigh = -1; } /* * Find the slot that's first lower than our hash value, -1 if none. */ for (low = 0, high = be32_to_cpu(btp->count) - 1; low <= high; ) { mid = (low + high) >> 1; if ((hash = be32_to_cpu(blp[mid].hashval)) == args->hashval) break; if (hash < args->hashval) low = mid + 1; else high = mid - 1; } while (mid >= 0 && be32_to_cpu(blp[mid].hashval) >= args->hashval) { mid--; } /* * No stale entries, will use enddup space to hold new leaf. */ if (!btp->stale) { /* * Mark the space needed for the new leaf entry, now in use. */ xfs_dir2_data_use_free(tp, bp, enddup, (xfs_dir2_data_aoff_t) ((char *)enddup - (char *)block + be16_to_cpu(enddup->length) - sizeof(*blp)), (xfs_dir2_data_aoff_t)sizeof(*blp), &needlog, &needscan); /* * Update the tail (entry count). */ be32_add(&btp->count, 1); /* * If we now need to rebuild the bestfree map, do so. * This needs to happen before the next call to use_free. */ if (needscan) { xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog); needscan = 0; } /* * Adjust pointer to the first leaf entry, we're about to move * the table up one to open up space for the new leaf entry. * Then adjust our index to match. */ blp--; mid++; if (mid) memmove(blp, &blp[1], mid * sizeof(*blp)); lfloglow = 0; lfloghigh = mid; } /* * Use a stale leaf for our new entry. */ else { for (lowstale = mid; lowstale >= 0 && be32_to_cpu(blp[lowstale].address) != XFS_DIR2_NULL_DATAPTR; lowstale--) continue; for (highstale = mid + 1; highstale < be32_to_cpu(btp->count) && be32_to_cpu(blp[highstale].address) != XFS_DIR2_NULL_DATAPTR && (lowstale < 0 || mid - lowstale > highstale - mid); highstale++) continue; /* * Move entries toward the low-numbered stale entry. */ if (lowstale >= 0 && (highstale == be32_to_cpu(btp->count) || mid - lowstale <= highstale - mid)) { if (mid - lowstale) memmove(&blp[lowstale], &blp[lowstale + 1], (mid - lowstale) * sizeof(*blp)); lfloglow = MIN(lowstale, lfloglow); lfloghigh = MAX(mid, lfloghigh); } /* * Move entries toward the high-numbered stale entry. */ else { ASSERT(highstale < be32_to_cpu(btp->count)); mid++; if (highstale - mid) memmove(&blp[mid + 1], &blp[mid], (highstale - mid) * sizeof(*blp)); lfloglow = MIN(mid, lfloglow); lfloghigh = MAX(highstale, lfloghigh); } be32_add(&btp->stale, -1); } /* * Point to the new data entry. */ dep = (xfs_dir2_data_entry_t *)dup; /* * Fill in the leaf entry. */ blp[mid].hashval = cpu_to_be32(args->hashval); blp[mid].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp, (char *)dep - (char *)block)); xfs_dir2_block_log_leaf(tp, bp, lfloglow, lfloghigh); /* * Mark space for the data entry used. */ xfs_dir2_data_use_free(tp, bp, dup, (xfs_dir2_data_aoff_t)((char *)dup - (char *)block), (xfs_dir2_data_aoff_t)len, &needlog, &needscan); /* * Create the new data entry. */ dep->inumber = cpu_to_be64(args->inumber); dep->namelen = args->namelen; memcpy(dep->name, args->name, args->namelen); tagp = xfs_dir2_data_entry_tag_p(dep); *tagp = cpu_to_be16((char *)dep - (char *)block); /* * Clean up the bestfree array and log the header, tail, and entry. */ if (needscan) xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog); if (needlog) xfs_dir2_data_log_header(tp, bp); xfs_dir2_block_log_tail(tp, bp); xfs_dir2_data_log_entry(tp, bp, dep); xfs_dir2_data_check(dp, bp); xfs_da_buf_done(bp); return 0; }