Home | History | Annotate | Line # | Download | only in ffs
ffs_alloc.c revision 1.28
      1  1.28       agc /*	$NetBSD: ffs_alloc.c,v 1.28 2015/03/29 05:52:59 agc Exp $	*/
      2   1.1     lukem /* From: NetBSD: ffs_alloc.c,v 1.50 2001/09/06 02:16:01 lukem Exp */
      3   1.1     lukem 
      4   1.1     lukem /*
      5  1.12      fvdl  * Copyright (c) 2002 Networks Associates Technology, Inc.
      6  1.12      fvdl  * All rights reserved.
      7  1.12      fvdl  *
      8  1.12      fvdl  * This software was developed for the FreeBSD Project by Marshall
      9  1.12      fvdl  * Kirk McKusick and Network Associates Laboratories, the Security
     10  1.12      fvdl  * Research Division of Network Associates, Inc. under DARPA/SPAWAR
     11  1.12      fvdl  * contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS
     12  1.12      fvdl  * research program
     13  1.12      fvdl  *
     14   1.1     lukem  * Copyright (c) 1982, 1986, 1989, 1993
     15   1.1     lukem  *	The Regents of the University of California.  All rights reserved.
     16   1.1     lukem  *
     17   1.1     lukem  * Redistribution and use in source and binary forms, with or without
     18   1.1     lukem  * modification, are permitted provided that the following conditions
     19   1.1     lukem  * are met:
     20   1.1     lukem  * 1. Redistributions of source code must retain the above copyright
     21   1.1     lukem  *    notice, this list of conditions and the following disclaimer.
     22   1.1     lukem  * 2. Redistributions in binary form must reproduce the above copyright
     23   1.1     lukem  *    notice, this list of conditions and the following disclaimer in the
     24   1.1     lukem  *    documentation and/or other materials provided with the distribution.
     25  1.13       agc  * 3. Neither the name of the University nor the names of its contributors
     26   1.1     lukem  *    may be used to endorse or promote products derived from this software
     27   1.1     lukem  *    without specific prior written permission.
     28   1.1     lukem  *
     29   1.1     lukem  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     30   1.1     lukem  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     31   1.1     lukem  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     32   1.1     lukem  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     33   1.1     lukem  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     34   1.1     lukem  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     35   1.1     lukem  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     36   1.1     lukem  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     37   1.1     lukem  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     38   1.1     lukem  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     39   1.1     lukem  * SUCH DAMAGE.
     40   1.1     lukem  *
     41   1.1     lukem  *	@(#)ffs_alloc.c	8.19 (Berkeley) 7/13/95
     42   1.1     lukem  */
     43   1.2     lukem 
     44  1.14       jmc #if HAVE_NBTOOL_CONFIG_H
     45  1.14       jmc #include "nbtool_config.h"
     46  1.14       jmc #endif
     47  1.14       jmc 
     48   1.2     lukem #include <sys/cdefs.h>
     49   1.8        tv #if defined(__RCSID) && !defined(__lint)
     50  1.28       agc __RCSID("$NetBSD: ffs_alloc.c,v 1.28 2015/03/29 05:52:59 agc Exp $");
     51   1.2     lukem #endif	/* !__lint */
     52   1.1     lukem 
     53   1.1     lukem #include <sys/param.h>
     54   1.1     lukem #include <sys/time.h>
     55   1.1     lukem 
     56   1.1     lukem #include <errno.h>
     57   1.4     lukem 
     58   1.4     lukem #include "makefs.h"
     59   1.1     lukem 
     60   1.7     lukem #include <ufs/ufs/dinode.h>
     61   1.5     lukem #include <ufs/ufs/ufs_bswap.h>
     62   1.5     lukem #include <ufs/ffs/fs.h>
     63   1.1     lukem 
     64   1.1     lukem #include "ffs/buf.h"
     65   1.6     lukem #include "ffs/ufs_inode.h"
     66   1.1     lukem #include "ffs/ffs_extern.h"
     67   1.1     lukem 
     68   1.1     lukem 
     69   1.1     lukem static int scanc(u_int, const u_char *, const u_char *, int);
     70   1.1     lukem 
     71  1.11      fvdl static daddr_t ffs_alloccg(struct inode *, int, daddr_t, int);
     72  1.11      fvdl static daddr_t ffs_alloccgblk(struct inode *, struct buf *, daddr_t);
     73  1.11      fvdl static daddr_t ffs_hashalloc(struct inode *, int, daddr_t, int,
     74  1.11      fvdl 		     daddr_t (*)(struct inode *, int, daddr_t, int));
     75  1.12      fvdl static int32_t ffs_mapsearch(struct fs *, struct cg *, daddr_t, int);
     76   1.1     lukem 
     77   1.1     lukem /* in ffs_tables.c */
     78   1.1     lukem extern const int inside[], around[];
     79   1.1     lukem extern const u_char * const fragtbl[];
     80   1.1     lukem 
     81   1.1     lukem /*
     82   1.1     lukem  * Allocate a block in the file system.
     83   1.1     lukem  *
     84   1.1     lukem  * The size of the requested block is given, which must be some
     85   1.1     lukem  * multiple of fs_fsize and <= fs_bsize.
     86   1.1     lukem  * A preference may be optionally specified. If a preference is given
     87   1.1     lukem  * the following hierarchy is used to allocate a block:
     88   1.1     lukem  *   1) allocate the requested block.
     89   1.1     lukem  *   2) allocate a rotationally optimal block in the same cylinder.
     90   1.1     lukem  *   3) allocate a block in the same cylinder group.
     91   1.1     lukem  *   4) quadradically rehash into other cylinder groups, until an
     92   1.1     lukem  *      available block is located.
     93   1.1     lukem  * If no block preference is given the following hierarchy is used
     94   1.1     lukem  * to allocate a block:
     95   1.1     lukem  *   1) allocate a block in the cylinder group that contains the
     96   1.1     lukem  *      inode for the file.
     97   1.1     lukem  *   2) quadradically rehash into other cylinder groups, until an
     98   1.1     lukem  *      available block is located.
     99   1.1     lukem  */
    100   1.1     lukem int
    101  1.17  christos ffs_alloc(struct inode *ip, daddr_t lbn __unused, daddr_t bpref, int size,
    102  1.11      fvdl     daddr_t *bnp)
    103   1.1     lukem {
    104   1.1     lukem 	struct fs *fs = ip->i_fs;
    105  1.11      fvdl 	daddr_t bno;
    106   1.1     lukem 	int cg;
    107   1.1     lukem 
    108   1.1     lukem 	*bnp = 0;
    109  1.24  dholland 	if (size > fs->fs_bsize || ffs_fragoff(fs, size) != 0) {
    110   1.1     lukem 		errx(1, "ffs_alloc: bad size: bsize %d size %d",
    111   1.1     lukem 		    fs->fs_bsize, size);
    112   1.1     lukem 	}
    113   1.1     lukem 	if (size == fs->fs_bsize && fs->fs_cstotal.cs_nbfree == 0)
    114   1.1     lukem 		goto nospace;
    115   1.1     lukem 	if (bpref >= fs->fs_size)
    116   1.1     lukem 		bpref = 0;
    117   1.1     lukem 	if (bpref == 0)
    118   1.1     lukem 		cg = ino_to_cg(fs, ip->i_number);
    119   1.1     lukem 	else
    120   1.1     lukem 		cg = dtog(fs, bpref);
    121  1.11      fvdl 	bno = ffs_hashalloc(ip, cg, bpref, size, ffs_alloccg);
    122   1.1     lukem 	if (bno > 0) {
    123  1.15      fvdl 		DIP_ADD(ip, blocks, size / DEV_BSIZE);
    124   1.1     lukem 		*bnp = bno;
    125   1.1     lukem 		return (0);
    126   1.1     lukem 	}
    127   1.1     lukem nospace:
    128   1.1     lukem 	return (ENOSPC);
    129   1.1     lukem }
    130   1.1     lukem 
    131   1.1     lukem /*
    132   1.1     lukem  * Select the desired position for the next block in a file.  The file is
    133   1.1     lukem  * logically divided into sections. The first section is composed of the
    134   1.1     lukem  * direct blocks. Each additional section contains fs_maxbpg blocks.
    135   1.1     lukem  *
    136   1.1     lukem  * If no blocks have been allocated in the first section, the policy is to
    137   1.1     lukem  * request a block in the same cylinder group as the inode that describes
    138   1.1     lukem  * the file. If no blocks have been allocated in any other section, the
    139   1.1     lukem  * policy is to place the section in a cylinder group with a greater than
    140   1.1     lukem  * average number of free blocks.  An appropriate cylinder group is found
    141   1.1     lukem  * by using a rotor that sweeps the cylinder groups. When a new group of
    142   1.1     lukem  * blocks is needed, the sweep begins in the cylinder group following the
    143   1.1     lukem  * cylinder group from which the previous allocation was made. The sweep
    144   1.1     lukem  * continues until a cylinder group with greater than the average number
    145   1.1     lukem  * of free blocks is found. If the allocation is for the first block in an
    146   1.1     lukem  * indirect block, the information on the previous allocation is unavailable;
    147   1.1     lukem  * here a best guess is made based upon the logical block number being
    148   1.1     lukem  * allocated.
    149   1.1     lukem  *
    150   1.1     lukem  * If a section is already partially allocated, the policy is to
    151   1.1     lukem  * contiguously allocate fs_maxcontig blocks.  The end of one of these
    152   1.1     lukem  * contiguous blocks and the beginning of the next is physically separated
    153   1.1     lukem  * so that the disk head will be in transit between them for at least
    154   1.1     lukem  * fs_rotdelay milliseconds.  This is to allow time for the processor to
    155   1.1     lukem  * schedule another I/O transfer.
    156   1.1     lukem  */
    157  1.11      fvdl /* XXX ondisk32 */
    158  1.11      fvdl daddr_t
    159  1.12      fvdl ffs_blkpref_ufs1(struct inode *ip, daddr_t lbn, int indx, int32_t *bap)
    160   1.1     lukem {
    161   1.1     lukem 	struct fs *fs;
    162   1.1     lukem 	int cg;
    163   1.1     lukem 	int avgbfree, startcg;
    164   1.1     lukem 
    165   1.1     lukem 	fs = ip->i_fs;
    166   1.1     lukem 	if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
    167  1.24  dholland 		if (lbn < UFS_NDADDR + FFS_NINDIR(fs)) {
    168   1.1     lukem 			cg = ino_to_cg(fs, ip->i_number);
    169   1.1     lukem 			return (fs->fs_fpg * cg + fs->fs_frag);
    170   1.1     lukem 		}
    171   1.1     lukem 		/*
    172   1.1     lukem 		 * Find a cylinder with greater than average number of
    173   1.1     lukem 		 * unused data blocks.
    174   1.1     lukem 		 */
    175   1.1     lukem 		if (indx == 0 || bap[indx - 1] == 0)
    176   1.1     lukem 			startcg =
    177   1.1     lukem 			    ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
    178   1.1     lukem 		else
    179   1.1     lukem 			startcg = dtog(fs,
    180   1.1     lukem 				ufs_rw32(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + 1);
    181   1.1     lukem 		startcg %= fs->fs_ncg;
    182   1.1     lukem 		avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
    183   1.1     lukem 		for (cg = startcg; cg < fs->fs_ncg; cg++)
    184  1.12      fvdl 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree)
    185  1.12      fvdl 				return (fs->fs_fpg * cg + fs->fs_frag);
    186  1.12      fvdl 		for (cg = 0; cg <= startcg; cg++)
    187  1.12      fvdl 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree)
    188  1.12      fvdl 				return (fs->fs_fpg * cg + fs->fs_frag);
    189  1.12      fvdl 		return (0);
    190  1.12      fvdl 	}
    191  1.12      fvdl 	/*
    192  1.12      fvdl 	 * We just always try to lay things out contiguously.
    193  1.12      fvdl 	 */
    194  1.12      fvdl 	return ufs_rw32(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + fs->fs_frag;
    195  1.12      fvdl }
    196  1.12      fvdl 
    197  1.12      fvdl daddr_t
    198  1.19  christos ffs_blkpref_ufs2(struct inode *ip, daddr_t lbn, int indx, int64_t *bap)
    199  1.12      fvdl {
    200  1.12      fvdl 	struct fs *fs;
    201  1.12      fvdl 	int cg;
    202  1.12      fvdl 	int avgbfree, startcg;
    203  1.12      fvdl 
    204  1.12      fvdl 	fs = ip->i_fs;
    205  1.12      fvdl 	if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
    206  1.24  dholland 		if (lbn < UFS_NDADDR + FFS_NINDIR(fs)) {
    207  1.12      fvdl 			cg = ino_to_cg(fs, ip->i_number);
    208  1.12      fvdl 			return (fs->fs_fpg * cg + fs->fs_frag);
    209  1.12      fvdl 		}
    210  1.12      fvdl 		/*
    211  1.12      fvdl 		 * Find a cylinder with greater than average number of
    212  1.12      fvdl 		 * unused data blocks.
    213  1.12      fvdl 		 */
    214  1.12      fvdl 		if (indx == 0 || bap[indx - 1] == 0)
    215  1.12      fvdl 			startcg =
    216  1.12      fvdl 			    ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
    217  1.12      fvdl 		else
    218  1.12      fvdl 			startcg = dtog(fs,
    219  1.12      fvdl 				ufs_rw64(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + 1);
    220  1.12      fvdl 		startcg %= fs->fs_ncg;
    221  1.12      fvdl 		avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
    222  1.12      fvdl 		for (cg = startcg; cg < fs->fs_ncg; cg++)
    223   1.1     lukem 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
    224   1.1     lukem 				return (fs->fs_fpg * cg + fs->fs_frag);
    225   1.1     lukem 			}
    226  1.12      fvdl 		for (cg = 0; cg < startcg; cg++)
    227   1.1     lukem 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
    228   1.1     lukem 				return (fs->fs_fpg * cg + fs->fs_frag);
    229   1.1     lukem 			}
    230   1.1     lukem 		return (0);
    231   1.1     lukem 	}
    232   1.1     lukem 	/*
    233  1.12      fvdl 	 * We just always try to lay things out contiguously.
    234   1.1     lukem 	 */
    235  1.12      fvdl 	return ufs_rw64(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + fs->fs_frag;
    236   1.1     lukem }
    237   1.1     lukem 
    238   1.1     lukem /*
    239   1.1     lukem  * Implement the cylinder overflow algorithm.
    240   1.1     lukem  *
    241   1.1     lukem  * The policy implemented by this algorithm is:
    242   1.1     lukem  *   1) allocate the block in its requested cylinder group.
    243   1.1     lukem  *   2) quadradically rehash on the cylinder group number.
    244   1.1     lukem  *   3) brute force search for a free block.
    245   1.1     lukem  *
    246   1.1     lukem  * `size':	size for data blocks, mode for inodes
    247   1.1     lukem  */
    248   1.1     lukem /*VARARGS5*/
    249  1.11      fvdl static daddr_t
    250  1.11      fvdl ffs_hashalloc(struct inode *ip, int cg, daddr_t pref, int size,
    251  1.11      fvdl     daddr_t (*allocator)(struct inode *, int, daddr_t, int))
    252   1.1     lukem {
    253   1.1     lukem 	struct fs *fs;
    254  1.11      fvdl 	daddr_t result;
    255   1.1     lukem 	int i, icg = cg;
    256   1.1     lukem 
    257   1.1     lukem 	fs = ip->i_fs;
    258   1.1     lukem 	/*
    259   1.1     lukem 	 * 1: preferred cylinder group
    260   1.1     lukem 	 */
    261   1.1     lukem 	result = (*allocator)(ip, cg, pref, size);
    262   1.1     lukem 	if (result)
    263   1.1     lukem 		return (result);
    264   1.1     lukem 	/*
    265   1.1     lukem 	 * 2: quadratic rehash
    266   1.1     lukem 	 */
    267   1.1     lukem 	for (i = 1; i < fs->fs_ncg; i *= 2) {
    268   1.1     lukem 		cg += i;
    269   1.1     lukem 		if (cg >= fs->fs_ncg)
    270   1.1     lukem 			cg -= fs->fs_ncg;
    271   1.1     lukem 		result = (*allocator)(ip, cg, 0, size);
    272   1.1     lukem 		if (result)
    273   1.1     lukem 			return (result);
    274   1.1     lukem 	}
    275   1.1     lukem 	/*
    276   1.1     lukem 	 * 3: brute force search
    277   1.1     lukem 	 * Note that we start at i == 2, since 0 was checked initially,
    278   1.1     lukem 	 * and 1 is always checked in the quadratic rehash.
    279   1.1     lukem 	 */
    280   1.1     lukem 	cg = (icg + 2) % fs->fs_ncg;
    281   1.1     lukem 	for (i = 2; i < fs->fs_ncg; i++) {
    282   1.1     lukem 		result = (*allocator)(ip, cg, 0, size);
    283   1.1     lukem 		if (result)
    284   1.1     lukem 			return (result);
    285   1.1     lukem 		cg++;
    286   1.1     lukem 		if (cg == fs->fs_ncg)
    287   1.1     lukem 			cg = 0;
    288   1.1     lukem 	}
    289   1.1     lukem 	return (0);
    290   1.1     lukem }
    291   1.1     lukem 
    292   1.1     lukem /*
    293   1.1     lukem  * Determine whether a block can be allocated.
    294   1.1     lukem  *
    295   1.1     lukem  * Check to see if a block of the appropriate size is available,
    296   1.1     lukem  * and if it is, allocate it.
    297   1.1     lukem  */
    298  1.11      fvdl static daddr_t
    299  1.11      fvdl ffs_alloccg(struct inode *ip, int cg, daddr_t bpref, int size)
    300   1.1     lukem {
    301   1.1     lukem 	struct cg *cgp;
    302   1.1     lukem 	struct buf *bp;
    303  1.11      fvdl 	daddr_t bno, blkno;
    304   1.1     lukem 	int error, frags, allocsiz, i;
    305   1.1     lukem 	struct fs *fs = ip->i_fs;
    306   1.1     lukem 	const int needswap = UFS_FSNEEDSWAP(fs);
    307   1.1     lukem 
    308   1.1     lukem 	if (fs->fs_cs(fs, cg).cs_nbfree == 0 && size == fs->fs_bsize)
    309   1.1     lukem 		return (0);
    310  1.25  dholland 	error = bread(ip->i_devvp, FFS_FSBTODB(fs, cgtod(fs, cg)),
    311  1.28       agc 	    (int)fs->fs_cgsize, 0, &bp);
    312   1.1     lukem 	if (error) {
    313   1.1     lukem 		return (0);
    314   1.1     lukem 	}
    315   1.1     lukem 	cgp = (struct cg *)bp->b_data;
    316   1.1     lukem 	if (!cg_chkmagic(cgp, needswap) ||
    317   1.1     lukem 	    (cgp->cg_cs.cs_nbfree == 0 && size == fs->fs_bsize)) {
    318  1.21  christos 		brelse(bp, 0);
    319   1.1     lukem 		return (0);
    320   1.1     lukem 	}
    321   1.1     lukem 	if (size == fs->fs_bsize) {
    322   1.1     lukem 		bno = ffs_alloccgblk(ip, bp, bpref);
    323  1.22  christos 		bwrite(bp);
    324   1.1     lukem 		return (bno);
    325   1.1     lukem 	}
    326   1.1     lukem 	/*
    327   1.1     lukem 	 * check to see if any fragments are already available
    328   1.1     lukem 	 * allocsiz is the size which will be allocated, hacking
    329   1.1     lukem 	 * it down to a smaller size if necessary
    330   1.1     lukem 	 */
    331  1.26  dholland 	frags = ffs_numfrags(fs, size);
    332   1.1     lukem 	for (allocsiz = frags; allocsiz < fs->fs_frag; allocsiz++)
    333   1.1     lukem 		if (cgp->cg_frsum[allocsiz] != 0)
    334   1.1     lukem 			break;
    335   1.1     lukem 	if (allocsiz == fs->fs_frag) {
    336   1.1     lukem 		/*
    337   1.1     lukem 		 * no fragments were available, so a block will be
    338   1.1     lukem 		 * allocated, and hacked up
    339   1.1     lukem 		 */
    340   1.1     lukem 		if (cgp->cg_cs.cs_nbfree == 0) {
    341  1.21  christos 			brelse(bp, 0);
    342   1.1     lukem 			return (0);
    343   1.1     lukem 		}
    344   1.1     lukem 		bno = ffs_alloccgblk(ip, bp, bpref);
    345   1.1     lukem 		bpref = dtogd(fs, bno);
    346   1.1     lukem 		for (i = frags; i < fs->fs_frag; i++)
    347   1.1     lukem 			setbit(cg_blksfree(cgp, needswap), bpref + i);
    348   1.1     lukem 		i = fs->fs_frag - frags;
    349   1.1     lukem 		ufs_add32(cgp->cg_cs.cs_nffree, i, needswap);
    350   1.1     lukem 		fs->fs_cstotal.cs_nffree += i;
    351   1.1     lukem 		fs->fs_cs(fs, cg).cs_nffree += i;
    352   1.1     lukem 		fs->fs_fmod = 1;
    353   1.1     lukem 		ufs_add32(cgp->cg_frsum[i], 1, needswap);
    354   1.1     lukem 		bdwrite(bp);
    355   1.1     lukem 		return (bno);
    356   1.1     lukem 	}
    357   1.1     lukem 	bno = ffs_mapsearch(fs, cgp, bpref, allocsiz);
    358   1.1     lukem 	for (i = 0; i < frags; i++)
    359   1.1     lukem 		clrbit(cg_blksfree(cgp, needswap), bno + i);
    360   1.1     lukem 	ufs_add32(cgp->cg_cs.cs_nffree, -frags, needswap);
    361   1.1     lukem 	fs->fs_cstotal.cs_nffree -= frags;
    362   1.1     lukem 	fs->fs_cs(fs, cg).cs_nffree -= frags;
    363   1.1     lukem 	fs->fs_fmod = 1;
    364   1.1     lukem 	ufs_add32(cgp->cg_frsum[allocsiz], -1, needswap);
    365   1.1     lukem 	if (frags != allocsiz)
    366   1.1     lukem 		ufs_add32(cgp->cg_frsum[allocsiz - frags], 1, needswap);
    367   1.1     lukem 	blkno = cg * fs->fs_fpg + bno;
    368   1.1     lukem 	bdwrite(bp);
    369   1.1     lukem 	return blkno;
    370   1.1     lukem }
    371   1.1     lukem 
    372   1.1     lukem /*
    373   1.1     lukem  * Allocate a block in a cylinder group.
    374   1.1     lukem  *
    375   1.1     lukem  * This algorithm implements the following policy:
    376   1.1     lukem  *   1) allocate the requested block.
    377   1.1     lukem  *   2) allocate a rotationally optimal block in the same cylinder.
    378   1.1     lukem  *   3) allocate the next available block on the block rotor for the
    379   1.1     lukem  *      specified cylinder group.
    380   1.1     lukem  * Note that this routine only allocates fs_bsize blocks; these
    381   1.1     lukem  * blocks may be fragmented by the routine that allocates them.
    382   1.1     lukem  */
    383  1.11      fvdl static daddr_t
    384  1.11      fvdl ffs_alloccgblk(struct inode *ip, struct buf *bp, daddr_t bpref)
    385   1.1     lukem {
    386   1.1     lukem 	struct cg *cgp;
    387  1.12      fvdl 	daddr_t blkno;
    388  1.12      fvdl 	int32_t bno;
    389   1.1     lukem 	struct fs *fs = ip->i_fs;
    390   1.1     lukem 	const int needswap = UFS_FSNEEDSWAP(fs);
    391  1.12      fvdl 	u_int8_t *blksfree;
    392   1.1     lukem 
    393   1.1     lukem 	cgp = (struct cg *)bp->b_data;
    394  1.12      fvdl 	blksfree = cg_blksfree(cgp, needswap);
    395   1.1     lukem 	if (bpref == 0 || dtog(fs, bpref) != ufs_rw32(cgp->cg_cgx, needswap)) {
    396   1.1     lukem 		bpref = ufs_rw32(cgp->cg_rotor, needswap);
    397  1.12      fvdl 	} else {
    398  1.27  dholland 		bpref = ffs_blknum(fs, bpref);
    399  1.12      fvdl 		bno = dtogd(fs, bpref);
    400   1.1     lukem 		/*
    401  1.12      fvdl 		 * if the requested block is available, use it
    402   1.1     lukem 		 */
    403  1.27  dholland 		if (ffs_isblock(fs, blksfree, ffs_fragstoblks(fs, bno)))
    404  1.12      fvdl 			goto gotit;
    405   1.1     lukem 	}
    406   1.1     lukem 	/*
    407  1.12      fvdl 	 * Take the next available one in this cylinder group.
    408   1.1     lukem 	 */
    409   1.1     lukem 	bno = ffs_mapsearch(fs, cgp, bpref, (int)fs->fs_frag);
    410   1.1     lukem 	if (bno < 0)
    411   1.1     lukem 		return (0);
    412  1.12      fvdl 	cgp->cg_rotor = ufs_rw32(bno, needswap);
    413   1.1     lukem gotit:
    414  1.27  dholland 	blkno = ffs_fragstoblks(fs, bno);
    415  1.12      fvdl 	ffs_clrblock(fs, blksfree, (long)blkno);
    416   1.1     lukem 	ffs_clusteracct(fs, cgp, blkno, -1);
    417   1.1     lukem 	ufs_add32(cgp->cg_cs.cs_nbfree, -1, needswap);
    418   1.1     lukem 	fs->fs_cstotal.cs_nbfree--;
    419   1.1     lukem 	fs->fs_cs(fs, ufs_rw32(cgp->cg_cgx, needswap)).cs_nbfree--;
    420   1.1     lukem 	fs->fs_fmod = 1;
    421   1.1     lukem 	blkno = ufs_rw32(cgp->cg_cgx, needswap) * fs->fs_fpg + bno;
    422   1.1     lukem 	return (blkno);
    423   1.1     lukem }
    424   1.1     lukem 
    425   1.1     lukem /*
    426   1.1     lukem  * Free a block or fragment.
    427   1.1     lukem  *
    428   1.1     lukem  * The specified block or fragment is placed back in the
    429   1.1     lukem  * free map. If a fragment is deallocated, a possible
    430   1.1     lukem  * block reassembly is checked.
    431   1.1     lukem  */
    432   1.1     lukem void
    433  1.11      fvdl ffs_blkfree(struct inode *ip, daddr_t bno, long size)
    434   1.1     lukem {
    435   1.1     lukem 	struct cg *cgp;
    436   1.1     lukem 	struct buf *bp;
    437  1.12      fvdl 	int32_t fragno, cgbno;
    438   1.1     lukem 	int i, error, cg, blk, frags, bbase;
    439   1.1     lukem 	struct fs *fs = ip->i_fs;
    440   1.1     lukem 	const int needswap = UFS_FSNEEDSWAP(fs);
    441   1.1     lukem 
    442  1.24  dholland 	if (size > fs->fs_bsize || ffs_fragoff(fs, size) != 0 ||
    443  1.27  dholland 	    ffs_fragnum(fs, bno) + ffs_numfrags(fs, size) > fs->fs_frag) {
    444  1.11      fvdl 		errx(1, "blkfree: bad size: bno %lld bsize %d size %ld",
    445  1.11      fvdl 		    (long long)bno, fs->fs_bsize, size);
    446   1.1     lukem 	}
    447   1.1     lukem 	cg = dtog(fs, bno);
    448  1.12      fvdl 	if (bno >= fs->fs_size) {
    449  1.16  christos 		warnx("bad block %lld, ino %llu", (long long)bno,
    450  1.16  christos 		    (unsigned long long)ip->i_number);
    451   1.1     lukem 		return;
    452   1.1     lukem 	}
    453  1.25  dholland 	error = bread(ip->i_devvp, FFS_FSBTODB(fs, cgtod(fs, cg)),
    454  1.28       agc 	    (int)fs->fs_cgsize, 0, &bp);
    455   1.1     lukem 	if (error) {
    456  1.21  christos 		brelse(bp, 0);
    457   1.1     lukem 		return;
    458   1.1     lukem 	}
    459   1.1     lukem 	cgp = (struct cg *)bp->b_data;
    460   1.1     lukem 	if (!cg_chkmagic(cgp, needswap)) {
    461  1.21  christos 		brelse(bp, 0);
    462   1.1     lukem 		return;
    463   1.1     lukem 	}
    464  1.12      fvdl 	cgbno = dtogd(fs, bno);
    465   1.1     lukem 	if (size == fs->fs_bsize) {
    466  1.27  dholland 		fragno = ffs_fragstoblks(fs, cgbno);
    467  1.12      fvdl 		if (!ffs_isfreeblock(fs, cg_blksfree(cgp, needswap), fragno)) {
    468  1.11      fvdl 			errx(1, "blkfree: freeing free block %lld",
    469  1.11      fvdl 			    (long long)bno);
    470   1.1     lukem 		}
    471  1.12      fvdl 		ffs_setblock(fs, cg_blksfree(cgp, needswap), fragno);
    472  1.12      fvdl 		ffs_clusteracct(fs, cgp, fragno, 1);
    473   1.1     lukem 		ufs_add32(cgp->cg_cs.cs_nbfree, 1, needswap);
    474   1.1     lukem 		fs->fs_cstotal.cs_nbfree++;
    475   1.1     lukem 		fs->fs_cs(fs, cg).cs_nbfree++;
    476   1.1     lukem 	} else {
    477  1.27  dholland 		bbase = cgbno - ffs_fragnum(fs, cgbno);
    478   1.1     lukem 		/*
    479   1.1     lukem 		 * decrement the counts associated with the old frags
    480   1.1     lukem 		 */
    481   1.1     lukem 		blk = blkmap(fs, cg_blksfree(cgp, needswap), bbase);
    482   1.1     lukem 		ffs_fragacct(fs, blk, cgp->cg_frsum, -1, needswap);
    483   1.1     lukem 		/*
    484   1.1     lukem 		 * deallocate the fragment
    485   1.1     lukem 		 */
    486  1.26  dholland 		frags = ffs_numfrags(fs, size);
    487   1.1     lukem 		for (i = 0; i < frags; i++) {
    488  1.12      fvdl 			if (isset(cg_blksfree(cgp, needswap), cgbno + i)) {
    489  1.11      fvdl 				errx(1, "blkfree: freeing free frag: block %lld",
    490  1.12      fvdl 				    (long long)(cgbno + i));
    491   1.1     lukem 			}
    492  1.12      fvdl 			setbit(cg_blksfree(cgp, needswap), cgbno + i);
    493   1.1     lukem 		}
    494   1.1     lukem 		ufs_add32(cgp->cg_cs.cs_nffree, i, needswap);
    495   1.1     lukem 		fs->fs_cstotal.cs_nffree += i;
    496   1.1     lukem 		fs->fs_cs(fs, cg).cs_nffree += i;
    497   1.1     lukem 		/*
    498   1.1     lukem 		 * add back in counts associated with the new frags
    499   1.1     lukem 		 */
    500   1.1     lukem 		blk = blkmap(fs, cg_blksfree(cgp, needswap), bbase);
    501   1.1     lukem 		ffs_fragacct(fs, blk, cgp->cg_frsum, 1, needswap);
    502   1.1     lukem 		/*
    503   1.1     lukem 		 * if a complete block has been reassembled, account for it
    504   1.1     lukem 		 */
    505  1.27  dholland 		fragno = ffs_fragstoblks(fs, bbase);
    506  1.12      fvdl 		if (ffs_isblock(fs, cg_blksfree(cgp, needswap), fragno)) {
    507   1.1     lukem 			ufs_add32(cgp->cg_cs.cs_nffree, -fs->fs_frag, needswap);
    508   1.1     lukem 			fs->fs_cstotal.cs_nffree -= fs->fs_frag;
    509   1.1     lukem 			fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag;
    510  1.12      fvdl 			ffs_clusteracct(fs, cgp, fragno, 1);
    511   1.1     lukem 			ufs_add32(cgp->cg_cs.cs_nbfree, 1, needswap);
    512   1.1     lukem 			fs->fs_cstotal.cs_nbfree++;
    513   1.1     lukem 			fs->fs_cs(fs, cg).cs_nbfree++;
    514   1.1     lukem 		}
    515   1.1     lukem 	}
    516   1.1     lukem 	fs->fs_fmod = 1;
    517   1.1     lukem 	bdwrite(bp);
    518   1.1     lukem }
    519   1.1     lukem 
    520   1.1     lukem 
    521   1.1     lukem static int
    522   1.1     lukem scanc(u_int size, const u_char *cp, const u_char table[], int mask)
    523   1.1     lukem {
    524   1.1     lukem 	const u_char *end = &cp[size];
    525   1.1     lukem 
    526   1.1     lukem 	while (cp < end && (table[*cp] & mask) == 0)
    527   1.1     lukem 		cp++;
    528   1.1     lukem 	return (end - cp);
    529   1.1     lukem }
    530   1.1     lukem 
    531   1.1     lukem /*
    532   1.1     lukem  * Find a block of the specified size in the specified cylinder group.
    533   1.1     lukem  *
    534   1.1     lukem  * It is a panic if a request is made to find a block if none are
    535   1.1     lukem  * available.
    536   1.1     lukem  */
    537  1.12      fvdl static int32_t
    538  1.11      fvdl ffs_mapsearch(struct fs *fs, struct cg *cgp, daddr_t bpref, int allocsiz)
    539   1.1     lukem {
    540  1.12      fvdl 	int32_t bno;
    541   1.1     lukem 	int start, len, loc, i;
    542   1.1     lukem 	int blk, field, subfield, pos;
    543   1.1     lukem 	int ostart, olen;
    544   1.1     lukem 	const int needswap = UFS_FSNEEDSWAP(fs);
    545   1.1     lukem 
    546   1.1     lukem 	/*
    547   1.1     lukem 	 * find the fragment by searching through the free block
    548   1.1     lukem 	 * map for an appropriate bit pattern
    549   1.1     lukem 	 */
    550   1.1     lukem 	if (bpref)
    551   1.1     lukem 		start = dtogd(fs, bpref) / NBBY;
    552   1.1     lukem 	else
    553   1.1     lukem 		start = ufs_rw32(cgp->cg_frotor, needswap) / NBBY;
    554   1.1     lukem 	len = howmany(fs->fs_fpg, NBBY) - start;
    555   1.1     lukem 	ostart = start;
    556   1.1     lukem 	olen = len;
    557   1.1     lukem 	loc = scanc((u_int)len,
    558   1.1     lukem 		(const u_char *)&cg_blksfree(cgp, needswap)[start],
    559   1.1     lukem 		(const u_char *)fragtbl[fs->fs_frag],
    560   1.1     lukem 		(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
    561   1.1     lukem 	if (loc == 0) {
    562   1.1     lukem 		len = start + 1;
    563   1.1     lukem 		start = 0;
    564   1.1     lukem 		loc = scanc((u_int)len,
    565   1.1     lukem 			(const u_char *)&cg_blksfree(cgp, needswap)[0],
    566   1.1     lukem 			(const u_char *)fragtbl[fs->fs_frag],
    567   1.1     lukem 			(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
    568   1.1     lukem 		if (loc == 0) {
    569   1.1     lukem 			errx(1,
    570   1.1     lukem     "ffs_alloccg: map corrupted: start %d len %d offset %d %ld",
    571   1.1     lukem 				ostart, olen,
    572   1.1     lukem 				ufs_rw32(cgp->cg_freeoff, needswap),
    573   1.1     lukem 				(long)cg_blksfree(cgp, needswap) - (long)cgp);
    574   1.1     lukem 			/* NOTREACHED */
    575   1.1     lukem 		}
    576   1.1     lukem 	}
    577   1.1     lukem 	bno = (start + len - loc) * NBBY;
    578   1.1     lukem 	cgp->cg_frotor = ufs_rw32(bno, needswap);
    579   1.1     lukem 	/*
    580   1.1     lukem 	 * found the byte in the map
    581   1.1     lukem 	 * sift through the bits to find the selected frag
    582   1.1     lukem 	 */
    583   1.1     lukem 	for (i = bno + NBBY; bno < i; bno += fs->fs_frag) {
    584   1.1     lukem 		blk = blkmap(fs, cg_blksfree(cgp, needswap), bno);
    585   1.1     lukem 		blk <<= 1;
    586   1.1     lukem 		field = around[allocsiz];
    587   1.1     lukem 		subfield = inside[allocsiz];
    588   1.1     lukem 		for (pos = 0; pos <= fs->fs_frag - allocsiz; pos++) {
    589   1.1     lukem 			if ((blk & field) == subfield)
    590   1.1     lukem 				return (bno + pos);
    591   1.1     lukem 			field <<= 1;
    592   1.1     lukem 			subfield <<= 1;
    593   1.1     lukem 		}
    594   1.1     lukem 	}
    595  1.11      fvdl 	errx(1, "ffs_alloccg: block not in map: bno %lld", (long long)bno);
    596   1.1     lukem 	return (-1);
    597   1.1     lukem }
    598