jfs_dtree.c

jfs-2.4-1.1.7.tar.gz jfs 2.4－1.1.7 源码

/*
 *   Copyright (C) International Business Machines Corp., 2000-2004
 *
 *   This program is free software;  you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation; either version 2 of the License, or 
 *   (at your option) any later version.
 * 
 *   This program is distributed in the hope that it will be useful,
 *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
 *   the GNU General Public License for more details.
 *
 *   You should have received a copy of the GNU General Public License
 *   along with this program;  if not, write to the Free Software 
 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */

/*
 *	jfs_dtree.c: directory B+-tree manager
 *
 * B+-tree with variable length key directory:
 *
 * each directory page is structured as an array of 32-byte
 * directory entry slots initialized as a freelist
 * to avoid search/compaction of free space at insertion.
 * when an entry is inserted, a number of slots are allocated
 * from the freelist as required to store variable length data
 * of the entry; when the entry is deleted, slots of the entry
 * are returned to freelist.
 *
 * leaf entry stores full name as key and file serial number
 * (aka inode number) as data.
 * internal/router entry stores sufffix compressed name
 * as key and simple extent descriptor as data.
 *
 * each directory page maintains a sorted entry index table
 * which stores the start slot index of sorted entries
 * to allow binary search on the table.
 *
 * directory starts as a root/leaf page in on-disk inode
 * inline data area.
 * when it becomes full, it starts a leaf of a external extent
 * of length of 1 block. each time the first leaf becomes full,
 * it is extended rather than split (its size is doubled),
 * until its length becoms 4 KBytes, from then the extent is split
 * with new 4 Kbyte extent when it becomes full
 * to reduce external fragmentation of small directories.
 *
 * blah, blah, blah, for linear scan of directory in pieces by
 * readdir().
 *
 *
 *	case-insensitive directory file system
 *
 * names are stored in case-sensitive way in leaf entry.
 * but stored, searched and compared in case-insensitive (uppercase) order
 * (i.e., both search key and entry key are folded for search/compare):
 * (note that case-sensitive order is BROKEN in storage, e.g.,
 *  sensitive: Ad, aB, aC, aD -> insensitive: aB, aC, aD, Ad
 *
 *  entries which folds to the same key makes up a equivalent class
 *  whose members are stored as contiguous cluster (may cross page boundary)
 *  but whose order is arbitrary and acts as duplicate, e.g.,
 *  abc, Abc, aBc, abC)
 *
 * once match is found at leaf, requires scan forward/backward
 * either for, in case-insensitive search, duplicate
 * or for, in case-sensitive search, for exact match
 *
 * router entry must be created/stored in case-insensitive way
 * in internal entry:
 * (right most key of left page and left most key of right page
 * are folded, and its suffix compression is propagated as router
 * key in parent)
 * (e.g., if split occurs <abc> and <aBd>, <ABD> trather than <aB>
 * should be made the router key for the split)
 *
 * case-insensitive search:
 *
 * 	fold search key;
 *
 *	case-insensitive search of B-tree:
 *	for internal entry, router key is already folded;
 *	for leaf entry, fold the entry key before comparison.
 *
 *	if (leaf entry case-insensitive match found)
 *		if (next entry satisfies case-insensitive match)
 *			return EDUPLICATE;
 *		if (prev entry satisfies case-insensitive match)
 *			return EDUPLICATE;
 *		return match;
 *	else
 *		return no match;
 *
 * 	serialization:
 * target directory inode lock is being held on entry/exit
 * of all main directory service routines.
 *
 *	log based recovery:
 */

#include <linux/fs.h>
#include "jfs_incore.h"
#include "jfs_superblock.h"
#include "jfs_filsys.h"
#include "jfs_metapage.h"
#include "jfs_dmap.h"
#include "jfs_unicode.h"
#include "jfs_debug.h"

/* dtree split parameter */
struct dtsplit {
	struct metapage *mp;
	s16 index;
	s16 nslot;
	struct component_name *key;
	ddata_t *data;
	struct pxdlist *pxdlist;
};

#define DT_PAGE(IP, MP) BT_PAGE(IP, MP, dtpage_t, i_dtroot)

/* get page buffer for specified block address */
#define DT_GETPAGE(IP, BN, MP, SIZE, P, RC)\
{\
	BT_GETPAGE(IP, BN, MP, dtpage_t, SIZE, P, RC, i_dtroot)\
	if (!(RC))\
	{\
		if (((P)->header.nextindex > (((BN)==0)?DTROOTMAXSLOT:(P)->header.maxslot)) ||\
		    ((BN) && ((P)->header.maxslot > DTPAGEMAXSLOT)))\
		{\
			BT_PUTPAGE(MP);\
			jfs_error((IP)->i_sb, "DT_GETPAGE: dtree page corrupt");\
			MP = NULL;\
			RC = -EIO;\
		}\
	}\
}

/* for consistency */
#define DT_PUTPAGE(MP) BT_PUTPAGE(MP)

#define DT_GETSEARCH(IP, LEAF, BN, MP, P, INDEX) \
	BT_GETSEARCH(IP, LEAF, BN, MP, dtpage_t, P, INDEX, i_dtroot)

/*
 * forward references
 */
static int dtSplitUp(tid_t tid, struct inode *ip,
		     struct dtsplit * split, struct btstack * btstack);

static int dtSplitPage(tid_t tid, struct inode *ip, struct dtsplit * split,
		       struct metapage ** rmpp, dtpage_t ** rpp, pxd_t * rxdp);

static int dtExtendPage(tid_t tid, struct inode *ip,
			struct dtsplit * split, struct btstack * btstack);

static int dtSplitRoot(tid_t tid, struct inode *ip,
		       struct dtsplit * split, struct metapage ** rmpp);

static int dtDeleteUp(tid_t tid, struct inode *ip, struct metapage * fmp,
		      dtpage_t * fp, struct btstack * btstack);

static int dtRelink(tid_t tid, struct inode *ip, dtpage_t * p);

static int dtReadFirst(struct inode *ip, struct btstack * btstack);

static int dtReadNext(struct inode *ip,
		      loff_t * offset, struct btstack * btstack);

static int dtCompare(struct component_name * key, dtpage_t * p, int si);

static int ciCompare(struct component_name * key, dtpage_t * p, int si,
		     int flag);

static void dtGetKey(dtpage_t * p, int i, struct component_name * key,
		     int flag);

static int ciGetLeafPrefixKey(dtpage_t * lp, int li, dtpage_t * rp,
			      int ri, struct component_name * key, int flag);

static void dtInsertEntry(dtpage_t * p, int index, struct component_name * key,
			  ddata_t * data, struct dt_lock **);

static void dtMoveEntry(dtpage_t * sp, int si, dtpage_t * dp,
			struct dt_lock ** sdtlock, struct dt_lock ** ddtlock,
			int do_index);

static void dtDeleteEntry(dtpage_t * p, int fi, struct dt_lock ** dtlock);

static void dtTruncateEntry(dtpage_t * p, int ti, struct dt_lock ** dtlock);

static void dtLinelockFreelist(dtpage_t * p, int m, struct dt_lock ** dtlock);

#define ciToUpper(c)	UniStrupr((c)->name)

/*
 *	read_index_page()
 *
 *	Reads a page of a directory's index table.
 *	Having metadata mapped into the directory inode's address space
 *	presents a multitude of problems.  We avoid this by mapping to
 *	the absolute address space outside of the *_metapage routines
 */
static struct metapage *read_index_page(struct inode *inode, s64 blkno)
{
	int rc;
	s64 xaddr;
	int xflag;
	s32 xlen;

	rc = xtLookup(inode, blkno, 1, &xflag, &xaddr, &xlen, 1);
	if (rc || (xlen == 0))
		return NULL;

	return read_metapage(inode, xaddr, PSIZE, 1);
}

/*
 *	get_index_page()
 *
 *	Same as get_index_page(), but get's a new page without reading
 */
static struct metapage *get_index_page(struct inode *inode, s64 blkno)
{
	int rc;
	s64 xaddr;
	int xflag;
	s32 xlen;

	rc = xtLookup(inode, blkno, 1, &xflag, &xaddr, &xlen, 1);
	if (rc || (xlen == 0))
		return NULL;

	return get_metapage(inode, xaddr, PSIZE, 1);
}

/*
 *	find_index()
 *
 *	Returns dtree page containing directory table entry for specified
 *	index and pointer to its entry.
 *
 *	mp must be released by caller.
 */
static struct dir_table_slot *find_index(struct inode *ip, u32 index,
					 struct metapage ** mp, s64 *lblock)
{
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	s64 blkno;
	s64 offset;
	int page_offset;
	struct dir_table_slot *slot;
	static int maxWarnings = 10;

	if (index < 2) {
		if (maxWarnings) {
			jfs_warn("find_entry called with index = %d", index);
			maxWarnings--;
		}
		return 0;
	}

	if (index >= jfs_ip->next_index) {
		jfs_warn("find_entry called with index >= next_index");
		return 0;
	}

	if (jfs_ip->next_index <= (MAX_INLINE_DIRTABLE_ENTRY + 1)) {
		/*
		 * Inline directory table
		 */
		*mp = 0;
		slot = &jfs_ip->i_dirtable[index - 2];
	} else {
		offset = (index - 2) * sizeof(struct dir_table_slot);
		page_offset = offset & (PSIZE - 1);
		blkno = ((offset + 1) >> L2PSIZE) <<
		    JFS_SBI(ip->i_sb)->l2nbperpage;

		if (*mp && (*lblock != blkno)) {
			release_metapage(*mp);
			*mp = 0;
		}
		if (*mp == 0) {
			*lblock = blkno;
			*mp = read_index_page(ip, blkno);
		}
		if (*mp == 0) {
			jfs_err("free_index: error reading directory table");
			return 0;
		}

		slot =
		    (struct dir_table_slot *) ((char *) (*mp)->data +
					       page_offset);
	}
	return slot;
}

static inline void lock_index(tid_t tid, struct inode *ip, struct metapage * mp,
			      u32 index)
{
	struct tlock *tlck;
	struct linelock *llck;
	struct lv *lv;

	tlck = txLock(tid, ip, mp, tlckDATA);
	llck = (struct linelock *) tlck->lock;

	if (llck->index >= llck->maxcnt)
		llck = txLinelock(llck);
	lv = &llck->lv[llck->index];

	/*
	 *      Linelock slot size is twice the size of directory table
	 *      slot size.  512 entries per page.
	 */
	lv->offset = ((index - 2) & 511) >> 1;
	lv->length = 1;
	llck->index++;
}

/*
 *	add_index()
 *
 *	Adds an entry to the directory index table.  This is used to provide
 *	each directory entry with a persistent index in which to resume
 *	directory traversals
 */
static u32 add_index(tid_t tid, struct inode *ip, s64 bn, int slot)
{
	struct super_block *sb = ip->i_sb;
	struct jfs_sb_info *sbi = JFS_SBI(sb);
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	u64 blkno;
	struct dir_table_slot *dirtab_slot;
	u32 index;
	struct linelock *llck;
	struct lv *lv;
	struct metapage *mp;
	s64 offset;
	uint page_offset;
	struct tlock *tlck;
	s64 xaddr;

	ASSERT(DO_INDEX(ip));

	if (jfs_ip->next_index < 2) {
		jfs_warn("add_index: next_index = %d.  Resetting!",
			   jfs_ip->next_index);
		jfs_ip->next_index = 2;
	}

	index = jfs_ip->next_index++;

	if (index <= MAX_INLINE_DIRTABLE_ENTRY) {
		/*
		 * i_size reflects size of index table, or 8 bytes per entry.
		 */
		ip->i_size = (loff_t) (index - 1) << 3;

		/*
		 * dir table fits inline within inode
		 */
		dirtab_slot = &jfs_ip->i_dirtable[index-2];
		dirtab_slot->flag = DIR_INDEX_VALID;
		dirtab_slot->slot = slot;
		DTSaddress(dirtab_slot, bn);

		set_cflag(COMMIT_Dirtable, ip);

		return index;
	}
	if (index == (MAX_INLINE_DIRTABLE_ENTRY + 1)) {
		struct dir_table_slot temp_table[12];

		/*
		 * It's time to move the inline table to an external
		 * page and begin to build the xtree
		 */
		if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr))
			goto clean_up;	/* No space */

		/*
		 * Save the table, we're going to overwrite it with the
		 * xtree root
		 */
		memcpy(temp_table, &jfs_ip->i_dirtable, sizeof(temp_table));

		/*
		 * Initialize empty x-tree
		 */
		xtInitRoot(tid, ip);

		/*
		 * Allocate the first block & add it to the xtree
		 */
		if (xtInsert(tid, ip, 0, 0, sbi->nbperpage, &xaddr, 0)) {
			/* This really shouldn't fail */
			jfs_warn("add_index: xtInsert failed!");
			memcpy(&jfs_ip->i_dirtable, temp_table,
			       sizeof (temp_table));
			goto clean_up;
		}
		ip->i_size = PSIZE;
		ip->i_blocks += LBLK2PBLK(sb, sbi->nbperpage);

		if ((mp = get_index_page(ip, 0)) == 0) {
			jfs_err("add_index: get_metapage failed!");
			xtTruncate(tid, ip, 0, COMMIT_PWMAP);
			memcpy(&jfs_ip->i_dirtable, temp_table,
			       sizeof (temp_table));
			goto clean_up;
		}
		tlck = txLock(tid, ip, mp, tlckDATA);
		llck = (struct linelock *) & tlck->lock;
		ASSERT(llck->index == 0);
		lv = &llck->lv[0];

		lv->offset = 0;
		lv->length = 6;	/* tlckDATA slot size is 16 bytes */
		llck->index++;

		memcpy(mp->data, temp_table, sizeof(temp_table));

		mark_metapage_dirty(mp);
		release_metapage(mp);

		/*
		 * Logging is now directed by xtree tlocks
		 */
		clear_cflag(COMMIT_Dirtable, ip);
	}

	offset = (index - 2) * sizeof(struct dir_table_slot);
	page_offset = offset & (PSIZE - 1);
	blkno = ((offset + 1) >> L2PSIZE) << sbi->l2nbperpage;
	if (page_offset == 0) {
		/*
		 * This will be the beginning of a new page
		 */
		xaddr = 0;
		if (xtInsert(tid, ip, 0, blkno, sbi->nbperpage, &xaddr, 0)) {
			jfs_warn("add_index: xtInsert failed!");
			goto clean_up;
		}
		ip->i_size += PSIZE;
		ip->i_blocks += LBLK2PBLK(sb, sbi->nbperpage);

		if ((mp = get_index_page(ip, blkno)))
			memset(mp->data, 0, PSIZE);	/* Just looks better */
		else
			xtTruncate(tid, ip, offset, COMMIT_PWMAP);
	} else
		mp = read_index_page(ip, blkno);

	if (mp == 0) {
		jfs_err("add_index: get/read_metapage failed!");
		goto clean_up;
	}

	lock_index(tid, ip, mp, index);

	dirtab_slot =
	    (struct dir_table_slot *) ((char *) mp->data + page_offset);
	dirtab_slot->flag = DIR_INDEX_VALID;
	dirtab_slot->slot = slot;
	DTSaddress(dirtab_slot, bn);

	mark_metapage_dirty(mp);
	release_metapage(mp);

	return index;

      clean_up:

	jfs_ip->next_index--;

	return 0;
}

/*
 *	free_index()
 *
 *	Marks an entry to the directory index table as free.
 */
static void free_index(tid_t tid, struct inode *ip, u32 index, u32 next)
{
	struct dir_table_slot *dirtab_slot;
	s64 lblock;
	struct metapage *mp = 0;

	dirtab_slot = find_index(ip, index, &mp, &lblock);

	if (dirtab_slot == 0)
		return;

	dirtab_slot->flag = DIR_INDEX_FREE;
	dirtab_slot->slot = dirtab_slot->addr1 = 0;
	dirtab_slot->addr2 = cpu_to_le32(next);

	if (mp) {
		lock_index(tid, ip, mp, index);
		mark_metapage_dirty(mp);
		release_metapage(mp);
	} else
		set_cflag(COMMIT_Dirtable, ip);
}

/*
 *	modify_index()
 *
 *	Changes an entry in the directory index table
 */
static void modify_index(tid_t tid, struct inode *ip, u32 index, s64 bn,
			 int slot, struct metapage ** mp, u64 *lblock)
{
	struct dir_table_slot *dirtab_slot;

	dirtab_slot = find_index(ip, index, mp, lblock);

	if (dirtab_slot == 0)
		return;

	DTSaddress(dirtab_slot, bn);
	dirtab_slot->slot = slot;

	if (*mp) {
		lock_index(tid, ip, *mp, index);
		mark_metapage_dirty(*mp);
	} else
		set_cflag(COMMIT_Dirtable, ip);
}

/*
 *	read_index()
 *
 *	reads a directory table slot
 */
static int read_index(struct inode *ip, u32 index,
		     struct dir_table_slot * dirtab_slot)
{
	s64 lblock;
	struct metapage *mp = 0;
	struct dir_table_slot *slot;

	slot = find_index(ip, index, &mp, &lblock);
	if (slot == 0) {
		return -EIO;