jfs_imap.c

linux 内核源代码

	}

	ip->i_mapping->a_ops = &jfs_metapage_aops;
	mapping_set_gfp_mask(ip->i_mapping, GFP_NOFS);

	/* Allocations to metadata inodes should not affect quotas */
	ip->i_flags |= S_NOQUOTA;

	if ((inum == FILESYSTEM_I) && (JFS_IP(ip)->ipimap == sbi->ipaimap)) {
		sbi->gengen = le32_to_cpu(dp->di_gengen);
		sbi->inostamp = le32_to_cpu(dp->di_inostamp);
	}

	/* release the page */
	release_metapage(mp);

	hlist_add_head(&ip->i_hash, &aggregate_hash);

	return (ip);
}

/*
 * NAME:	diWriteSpecial()
 *
 * FUNCTION:	Write the special inode to disk
 *
 * PARAMETERS:
 *	ip - special inode
 *	secondary - 1 if secondary aggregate inode table
 *
 * RETURN VALUES: none
 */

void diWriteSpecial(struct inode *ip, int secondary)
{
	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
	uint address;
	struct dinode *dp;
	ino_t inum = ip->i_ino;
	struct metapage *mp;

	if (secondary)
		address = addressPXD(&sbi->ait2) >> sbi->l2nbperpage;
	else
		address = AITBL_OFF >> L2PSIZE;

	ASSERT(inum < INOSPEREXT);

	address += inum >> 3;	/* 8 inodes per 4K page */

	/* read the page of fixed disk inode (AIT) in raw mode */
	mp = read_metapage(ip, address << sbi->l2nbperpage, PSIZE, 1);
	if (mp == NULL) {
		jfs_err("diWriteSpecial: failed to read aggregate inode "
			"extent!");
		return;
	}

	/* get the pointer to the disk inode of interest */
	dp = (struct dinode *) (mp->data);
	dp += inum % 8;		/* 8 inodes per 4K page */

	/* copy on-disk inode to in-memory inode */
	copy_to_dinode(dp, ip);
	memcpy(&dp->di_xtroot, &JFS_IP(ip)->i_xtroot, 288);

	if (inum == FILESYSTEM_I)
		dp->di_gengen = cpu_to_le32(sbi->gengen);

	/* write the page */
	write_metapage(mp);
}

/*
 * NAME:	diFreeSpecial()
 *
 * FUNCTION:	Free allocated space for special inode
 */
void diFreeSpecial(struct inode *ip)
{
	if (ip == NULL) {
		jfs_err("diFreeSpecial called with NULL ip!");
		return;
	}
	filemap_write_and_wait(ip->i_mapping);
	truncate_inode_pages(ip->i_mapping, 0);
	iput(ip);
}



/*
 * NAME:	diWrite()
 *
 * FUNCTION:	write the on-disk inode portion of the in-memory inode
 *		to its corresponding on-disk inode.
 *
 *		on entry, the specifed incore inode should itself
 *		specify the disk inode number corresponding to the
 *		incore inode (i.e. i_number should be initialized).
 *
 *		the inode contains the inode extent address for the disk
 *		inode.  with the inode extent address in hand, the
 *		page of the extent that contains the disk inode is
 *		read and the disk inode portion of the incore inode
 *		is copied to the disk inode.
 *
 * PARAMETERS:
 *	tid -  transacation id
 *	ip  -  pointer to incore inode to be written to the inode extent.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 */
int diWrite(tid_t tid, struct inode *ip)
{
	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	int rc = 0;
	s32 ino;
	struct dinode *dp;
	s64 blkno;
	int block_offset;
	int inodes_left;
	struct metapage *mp;
	unsigned long pageno;
	int rel_inode;
	int dioffset;
	struct inode *ipimap;
	uint type;
	lid_t lid;
	struct tlock *ditlck, *tlck;
	struct linelock *dilinelock, *ilinelock;
	struct lv *lv;
	int n;

	ipimap = jfs_ip->ipimap;

	ino = ip->i_ino & (INOSPERIAG - 1);

	if (!addressPXD(&(jfs_ip->ixpxd)) ||
	    (lengthPXD(&(jfs_ip->ixpxd)) !=
	     JFS_IP(ipimap)->i_imap->im_nbperiext)) {
		jfs_error(ip->i_sb, "diWrite: ixpxd invalid");
		return -EIO;
	}

	/*
	 * read the page of disk inode containing the specified inode:
	 */
	/* compute the block address of the page */
	blkno = INOPBLK(&(jfs_ip->ixpxd), ino, sbi->l2nbperpage);

	rel_inode = (ino & (INOSPERPAGE - 1));
	pageno = blkno >> sbi->l2nbperpage;

	if ((block_offset = ((u32) blkno & (sbi->nbperpage - 1)))) {
		/*
		 * OS/2 didn't always align inode extents on page boundaries
		 */
		inodes_left =
		    (sbi->nbperpage - block_offset) << sbi->l2niperblk;

		if (rel_inode < inodes_left)
			rel_inode += block_offset << sbi->l2niperblk;
		else {
			pageno += 1;
			rel_inode -= inodes_left;
		}
	}
	/* read the page of disk inode */
      retry:
	mp = read_metapage(ipimap, pageno << sbi->l2nbperpage, PSIZE, 1);
	if (mp == 0)
		return -EIO;

	/* get the pointer to the disk inode */
	dp = (struct dinode *) mp->data;
	dp += rel_inode;

	dioffset = (ino & (INOSPERPAGE - 1)) << L2DISIZE;

	/*
	 * acquire transaction lock on the on-disk inode;
	 * N.B. tlock is acquired on ipimap not ip;
	 */
	if ((ditlck =
	     txLock(tid, ipimap, mp, tlckINODE | tlckENTRY)) == NULL)
		goto retry;
	dilinelock = (struct linelock *) & ditlck->lock;

	/*
	 * copy btree root from in-memory inode to on-disk inode
	 *
	 * (tlock is taken from inline B+-tree root in in-memory
	 * inode when the B+-tree root is updated, which is pointed
	 * by jfs_ip->blid as well as being on tx tlock list)
	 *
	 * further processing of btree root is based on the copy
	 * in in-memory inode, where txLog() will log from, and,
	 * for xtree root, txUpdateMap() will update map and reset
	 * XAD_NEW bit;
	 */

	if (S_ISDIR(ip->i_mode) && (lid = jfs_ip->xtlid)) {
		/*
		 * This is the special xtree inside the directory for storing
		 * the directory table
		 */
		xtpage_t *p, *xp;
		xad_t *xad;

		jfs_ip->xtlid = 0;
		tlck = lid_to_tlock(lid);
		assert(tlck->type & tlckXTREE);
		tlck->type |= tlckBTROOT;
		tlck->mp = mp;
		ilinelock = (struct linelock *) & tlck->lock;

		/*
		 * copy xtree root from inode to dinode:
		 */
		p = &jfs_ip->i_xtroot;
		xp = (xtpage_t *) &dp->di_dirtable;
		lv = ilinelock->lv;
		for (n = 0; n < ilinelock->index; n++, lv++) {
			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
			       lv->length << L2XTSLOTSIZE);
		}

		/* reset on-disk (metadata page) xtree XAD_NEW bit */
		xad = &xp->xad[XTENTRYSTART];
		for (n = XTENTRYSTART;
		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
	}

	if ((lid = jfs_ip->blid) == 0)
		goto inlineData;
	jfs_ip->blid = 0;

	tlck = lid_to_tlock(lid);
	type = tlck->type;
	tlck->type |= tlckBTROOT;
	tlck->mp = mp;
	ilinelock = (struct linelock *) & tlck->lock;

	/*
	 *	regular file: 16 byte (XAD slot) granularity
	 */
	if (type & tlckXTREE) {
		xtpage_t *p, *xp;
		xad_t *xad;

		/*
		 * copy xtree root from inode to dinode:
		 */
		p = &jfs_ip->i_xtroot;
		xp = &dp->di_xtroot;
		lv = ilinelock->lv;
		for (n = 0; n < ilinelock->index; n++, lv++) {
			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
			       lv->length << L2XTSLOTSIZE);
		}

		/* reset on-disk (metadata page) xtree XAD_NEW bit */
		xad = &xp->xad[XTENTRYSTART];
		for (n = XTENTRYSTART;
		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
	}
	/*
	 *	directory: 32 byte (directory entry slot) granularity
	 */
	else if (type & tlckDTREE) {
		dtpage_t *p, *xp;

		/*
		 * copy dtree root from inode to dinode:
		 */
		p = (dtpage_t *) &jfs_ip->i_dtroot;
		xp = (dtpage_t *) & dp->di_dtroot;
		lv = ilinelock->lv;
		for (n = 0; n < ilinelock->index; n++, lv++) {
			memcpy(&xp->slot[lv->offset], &p->slot[lv->offset],
			       lv->length << L2DTSLOTSIZE);
		}
	} else {
		jfs_err("diWrite: UFO tlock");
	}

      inlineData:
	/*
	 * copy inline symlink from in-memory inode to on-disk inode
	 */
	if (S_ISLNK(ip->i_mode) && ip->i_size < IDATASIZE) {
		lv = & dilinelock->lv[dilinelock->index];
		lv->offset = (dioffset + 2 * 128) >> L2INODESLOTSIZE;
		lv->length = 2;
		memcpy(&dp->di_fastsymlink, jfs_ip->i_inline, IDATASIZE);
		dilinelock->index++;
	}
	/*
	 * copy inline data from in-memory inode to on-disk inode:
	 * 128 byte slot granularity
	 */
	if (test_cflag(COMMIT_Inlineea, ip)) {
		lv = & dilinelock->lv[dilinelock->index];
		lv->offset = (dioffset + 3 * 128) >> L2INODESLOTSIZE;
		lv->length = 1;
		memcpy(&dp->di_inlineea, jfs_ip->i_inline_ea, INODESLOTSIZE);
		dilinelock->index++;

		clear_cflag(COMMIT_Inlineea, ip);
	}

	/*
	 *	lock/copy inode base: 128 byte slot granularity
	 */
	lv = & dilinelock->lv[dilinelock->index];
	lv->offset = dioffset >> L2INODESLOTSIZE;
	copy_to_dinode(dp, ip);
	if (test_and_clear_cflag(COMMIT_Dirtable, ip)) {
		lv->length = 2;
		memcpy(&dp->di_dirtable, &jfs_ip->i_dirtable, 96);
	} else
		lv->length = 1;
	dilinelock->index++;

	/* release the buffer holding the updated on-disk inode.
	 * the buffer will be later written by commit processing.
	 */
	write_metapage(mp);

	return (rc);
}


/*
 * NAME:	diFree(ip)
 *
 * FUNCTION:	free a specified inode from the inode working map
 *		for a fileset or aggregate.
 *
 *		if the inode to be freed represents the first (only)
 *		free inode within the iag, the iag will be placed on
 *		the ag free inode list.
 *
 *		freeing the inode will cause the inode extent to be
 *		freed if the inode is the only allocated inode within
 *		the extent.  in this case all the disk resource backing
 *		up the inode extent will be freed. in addition, the iag
 *		will be placed on the ag extent free list if the extent
 *		is the first free extent in the iag.  if freeing the
 *		extent also means that no free inodes will exist for
 *		the iag, the iag will also be removed from the ag free
 *		inode list.
 *
 *		the iag describing the inode will be freed if the extent
 *		is to be freed and it is the only backed extent within
 *		the iag.  in this case, the iag will be removed from the
 *		ag free extent list and ag free inode list and placed on
 *		the inode map's free iag list.
 *
 *		a careful update approach is used to provide consistency
 *		in the face of updates to multiple buffers.  under this
 *		approach, all required buffers are obtained before making
 *		any updates and are held until all updates are complete.
 *
 * PARAMETERS:
 *	ip	- inode to be freed.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 */
int diFree(struct inode *ip)
{
	int rc;
	ino_t inum = ip->i_ino;
	struct iag *iagp, *aiagp, *biagp, *ciagp, *diagp;
	struct metapage *mp, *amp, *bmp, *cmp, *dmp;
	int iagno, ino, extno, bitno, sword, agno;
	int back, fwd;
	u32 bitmap, mask;
	struct inode *ipimap = JFS_SBI(ip->i_sb)->ipimap;
	struct inomap *imap = JFS_IP(ipimap)->i_imap;
	pxd_t freepxd;
	tid_t tid;
	struct inode *iplist[3];
	struct tlock *tlck;
	struct pxd_lock *pxdlock;

	/*
	 * This is just to suppress compiler warnings.  The same logic that
	 * references these variables is used to initialize them.
	 */
	aiagp = biagp = ciagp = diagp = NULL;

	/* get the iag number containing the inode.
	 */
	iagno = INOTOIAG(inum);

	/* make sure that the iag is contained within
	 * the map.
	 */
	if (iagno >= imap->im_nextiag) {
		print_hex_dump(KERN_ERR, "imap: ", DUMP_PREFIX_ADDRESS, 16, 4,
			       imap, 32, 0);
		jfs_error(ip->i_sb,
			  "diFree: inum = %d, iagno = %d, nextiag = %d",
			  (uint) inum, iagno, imap->im_nextiag);
		return -EIO;
	}

	/* get the allocation group for this ino.
	 */
	agno = JFS_IP(ip)->agno;

	/* Lock the AG specific inode map information
	 */
	AG_LOCK(imap, agno);

	/* Obtain read lock in imap inode.  Don't release it until we have
	 * read all of the IAG's that we are going to.
	 */
	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);

	/* read the iag.
	 */
	if ((rc = diIAGRead(imap, iagno, &mp))) {
		IREAD_UNLOCK(ipimap);
		AG_UNLOCK(imap, agno);
		return (rc);
	}
	iagp = (struct iag *) mp->data;

	/* get the inode number and extent number of the inode within
	 * the iag and the inode number within the extent.
	 */
	ino = inum & (INOSPERIAG - 1);
	extno = ino >> L2INOSPEREXT;
	bitno = ino & (INOSPEREXT - 1);
	mask = HIGHORDER >> bitno;

	if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
		jfs_error(ip->i_sb,
			  "diFree: wmap shows inode already free");
	}

	if (!addressPXD(&iagp->inoext[extno])) {
		release_metapage(mp);
		IREAD_UNLOCK(ipimap);
		AG_UNLOCK(imap, agno);
		jfs_error(ip->i_sb, "diFree: invalid inoext");
		return -EIO;
	}

	/* compute the bitmap for the extent reflecting the freed inode.
	 */
	bitmap = le32_to_cpu(iagp->wmap[extno]) & ~mask;

	if (imap->im_agctl[agno].numfree > imap->im_agctl[agno].numinos) {
		release_metapage(mp);
		IREAD_UNLOCK(ipimap);
		AG_UNLOCK(imap, agno);
		jfs_error(ip->i_sb, "diFree: numfree > numinos");
		return -EIO;
	}
	/*
	 *	inode extent still has some inodes or below low water mark:
	 *	keep the inode extent;
	 */
	if (bitmap ||
	    imap->im_agctl[agno].numfree < 96 ||
	    (imap->im_agctl[agno].numfree < 288 &&
	     (((imap->im_agctl[agno].numfree * 100) /
	       imap->im_agctl[agno].numinos) <= 25))) {
		/* if the iag currently has no free inodes (i.e.,