Home | History | Annotate | Line # | Download | only in lfs
lfs_subr.c revision 1.80
      1  1.80  dholland /*	$NetBSD: lfs_subr.c,v 1.80 2013/07/28 01:05:52 dholland Exp $	*/
      2   1.2       cgd 
      3   1.8  perseant /*-
      4  1.31  perseant  * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc.
      5   1.8  perseant  * All rights reserved.
      6   1.8  perseant  *
      7   1.8  perseant  * This code is derived from software contributed to The NetBSD Foundation
      8   1.8  perseant  * by Konrad E. Schroder <perseant (at) hhhh.org>.
      9   1.8  perseant  *
     10   1.8  perseant  * Redistribution and use in source and binary forms, with or without
     11   1.8  perseant  * modification, are permitted provided that the following conditions
     12   1.8  perseant  * are met:
     13   1.8  perseant  * 1. Redistributions of source code must retain the above copyright
     14   1.8  perseant  *    notice, this list of conditions and the following disclaimer.
     15   1.8  perseant  * 2. Redistributions in binary form must reproduce the above copyright
     16   1.8  perseant  *    notice, this list of conditions and the following disclaimer in the
     17   1.8  perseant  *    documentation and/or other materials provided with the distribution.
     18   1.8  perseant  *
     19   1.8  perseant  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     20   1.8  perseant  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     21   1.8  perseant  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     22   1.8  perseant  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     23   1.8  perseant  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     24   1.8  perseant  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     25   1.8  perseant  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     26   1.8  perseant  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     27   1.8  perseant  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     28   1.8  perseant  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     29   1.8  perseant  * POSSIBILITY OF SUCH DAMAGE.
     30   1.8  perseant  */
     31   1.1   mycroft /*
     32   1.1   mycroft  * Copyright (c) 1991, 1993
     33   1.1   mycroft  *	The Regents of the University of California.  All rights reserved.
     34   1.1   mycroft  *
     35   1.1   mycroft  * Redistribution and use in source and binary forms, with or without
     36   1.1   mycroft  * modification, are permitted provided that the following conditions
     37   1.1   mycroft  * are met:
     38   1.1   mycroft  * 1. Redistributions of source code must retain the above copyright
     39   1.1   mycroft  *    notice, this list of conditions and the following disclaimer.
     40   1.1   mycroft  * 2. Redistributions in binary form must reproduce the above copyright
     41   1.1   mycroft  *    notice, this list of conditions and the following disclaimer in the
     42   1.1   mycroft  *    documentation and/or other materials provided with the distribution.
     43  1.43       agc  * 3. Neither the name of the University nor the names of its contributors
     44   1.1   mycroft  *    may be used to endorse or promote products derived from this software
     45   1.1   mycroft  *    without specific prior written permission.
     46   1.1   mycroft  *
     47   1.1   mycroft  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     48   1.1   mycroft  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     49   1.1   mycroft  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     50   1.1   mycroft  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     51   1.1   mycroft  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     52   1.1   mycroft  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     53   1.1   mycroft  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     54   1.1   mycroft  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     55   1.1   mycroft  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     56   1.1   mycroft  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     57   1.1   mycroft  * SUCH DAMAGE.
     58   1.1   mycroft  *
     59   1.6      fvdl  *	@(#)lfs_subr.c	8.4 (Berkeley) 5/8/95
     60   1.1   mycroft  */
     61  1.20     lukem 
     62  1.20     lukem #include <sys/cdefs.h>
     63  1.80  dholland __KERNEL_RCSID(0, "$NetBSD: lfs_subr.c,v 1.80 2013/07/28 01:05:52 dholland Exp $");
     64   1.1   mycroft 
     65   1.1   mycroft #include <sys/param.h>
     66   1.3  christos #include <sys/systm.h>
     67   1.1   mycroft #include <sys/namei.h>
     68   1.1   mycroft #include <sys/vnode.h>
     69   1.1   mycroft #include <sys/buf.h>
     70   1.1   mycroft #include <sys/mount.h>
     71   1.1   mycroft #include <sys/malloc.h>
     72   1.1   mycroft #include <sys/proc.h>
     73  1.61  perseant #include <sys/kauth.h>
     74   1.1   mycroft 
     75  1.78  dholland #include <ufs/lfs/ulfs_inode.h>
     76   1.1   mycroft #include <ufs/lfs/lfs.h>
     77  1.80  dholland #include <ufs/lfs/lfs_kernel.h>
     78   1.1   mycroft #include <ufs/lfs/lfs_extern.h>
     79   1.1   mycroft 
     80  1.31  perseant #include <uvm/uvm.h>
     81  1.31  perseant 
     82  1.50  perseant #ifdef DEBUG
     83  1.53  christos const char *lfs_res_names[LFS_NB_COUNT] = {
     84  1.31  perseant 	"summary",
     85  1.31  perseant 	"superblock",
     86  1.52  perseant 	"file block",
     87  1.31  perseant 	"cluster",
     88  1.31  perseant 	"clean",
     89  1.52  perseant 	"blkiov",
     90  1.31  perseant };
     91  1.31  perseant #endif
     92  1.31  perseant 
     93  1.31  perseant int lfs_res_qty[LFS_NB_COUNT] = {
     94  1.31  perseant 	LFS_N_SUMMARIES,
     95  1.31  perseant 	LFS_N_SBLOCKS,
     96  1.31  perseant 	LFS_N_IBLOCKS,
     97  1.31  perseant 	LFS_N_CLUSTERS,
     98  1.31  perseant 	LFS_N_CLEAN,
     99  1.52  perseant 	LFS_N_BLKIOV,
    100  1.31  perseant };
    101  1.31  perseant 
    102  1.31  perseant void
    103  1.31  perseant lfs_setup_resblks(struct lfs *fs)
    104  1.31  perseant {
    105  1.31  perseant 	int i, j;
    106  1.31  perseant 	int maxbpp;
    107  1.31  perseant 
    108  1.51  perseant 	ASSERT_NO_SEGLOCK(fs);
    109  1.31  perseant 	fs->lfs_resblk = (res_t *)malloc(LFS_N_TOTAL * sizeof(res_t), M_SEGMENT,
    110  1.33  perseant 					  M_WAITOK);
    111  1.31  perseant 	for (i = 0; i < LFS_N_TOTAL; i++) {
    112  1.31  perseant 		fs->lfs_resblk[i].inuse = 0;
    113  1.31  perseant 		fs->lfs_resblk[i].p = NULL;
    114  1.31  perseant 	}
    115  1.31  perseant 	for (i = 0; i < LFS_RESHASH_WIDTH; i++)
    116  1.31  perseant 		LIST_INIT(fs->lfs_reshash + i);
    117  1.31  perseant 
    118  1.31  perseant 	/*
    119  1.31  perseant 	 * These types of allocations can be larger than a page,
    120  1.31  perseant 	 * so we can't use the pool subsystem for them.
    121  1.31  perseant 	 */
    122  1.31  perseant 	for (i = 0, j = 0; j < LFS_N_SUMMARIES; j++, i++)
    123  1.34  perseant 		fs->lfs_resblk[i].size = fs->lfs_sumsize;
    124  1.31  perseant 	for (j = 0; j < LFS_N_SBLOCKS; j++, i++)
    125  1.34  perseant 		fs->lfs_resblk[i].size = LFS_SBPAD;
    126  1.31  perseant 	for (j = 0; j < LFS_N_IBLOCKS; j++, i++)
    127  1.34  perseant 		fs->lfs_resblk[i].size = fs->lfs_bsize;
    128  1.31  perseant 	for (j = 0; j < LFS_N_CLUSTERS; j++, i++)
    129  1.34  perseant 		fs->lfs_resblk[i].size = MAXPHYS;
    130  1.31  perseant 	for (j = 0; j < LFS_N_CLEAN; j++, i++)
    131  1.34  perseant 		fs->lfs_resblk[i].size = MAXPHYS;
    132  1.52  perseant 	for (j = 0; j < LFS_N_BLKIOV; j++, i++)
    133  1.52  perseant 		fs->lfs_resblk[i].size = LFS_MARKV_MAXBLKCNT * sizeof(BLOCK_INFO);
    134  1.34  perseant 
    135  1.34  perseant 	for (i = 0; i < LFS_N_TOTAL; i++) {
    136  1.34  perseant 		fs->lfs_resblk[i].p = malloc(fs->lfs_resblk[i].size,
    137  1.34  perseant 					     M_SEGMENT, M_WAITOK);
    138  1.34  perseant 	}
    139  1.31  perseant 
    140  1.31  perseant 	/*
    141  1.31  perseant 	 * Initialize pools for small types (XXX is BPP small?)
    142  1.31  perseant 	 */
    143  1.46    simonb 	pool_init(&fs->lfs_clpool, sizeof(struct lfs_cluster), 0, 0, 0,
    144  1.68        ad 		"lfsclpl", &pool_allocator_nointr, IPL_NONE);
    145  1.46    simonb 	pool_init(&fs->lfs_segpool, sizeof(struct segment), 0, 0, 0,
    146  1.68        ad 		"lfssegpool", &pool_allocator_nointr, IPL_NONE);
    147  1.31  perseant 	maxbpp = ((fs->lfs_sumsize - SEGSUM_SIZE(fs)) / sizeof(int32_t) + 2);
    148  1.79  christos 	maxbpp = MIN(maxbpp, lfs_segsize(fs) / fs->lfs_fsize + 2);
    149  1.46    simonb 	pool_init(&fs->lfs_bpppool, maxbpp * sizeof(struct buf *), 0, 0, 0,
    150  1.68        ad 		"lfsbpppl", &pool_allocator_nointr, IPL_NONE);
    151  1.31  perseant }
    152  1.31  perseant 
    153  1.31  perseant void
    154  1.31  perseant lfs_free_resblks(struct lfs *fs)
    155  1.31  perseant {
    156  1.31  perseant 	int i;
    157  1.31  perseant 
    158  1.31  perseant 	pool_destroy(&fs->lfs_bpppool);
    159  1.31  perseant 	pool_destroy(&fs->lfs_segpool);
    160  1.31  perseant 	pool_destroy(&fs->lfs_clpool);
    161  1.31  perseant 
    162  1.72        ad 	mutex_enter(&lfs_lock);
    163  1.31  perseant 	for (i = 0; i < LFS_N_TOTAL; i++) {
    164  1.39  perseant 		while (fs->lfs_resblk[i].inuse)
    165  1.72        ad 			mtsleep(&fs->lfs_resblk, PRIBIO + 1, "lfs_free", 0,
    166  1.72        ad 				&lfs_lock);
    167  1.31  perseant 		if (fs->lfs_resblk[i].p != NULL)
    168  1.31  perseant 			free(fs->lfs_resblk[i].p, M_SEGMENT);
    169  1.31  perseant 	}
    170  1.31  perseant 	free(fs->lfs_resblk, M_SEGMENT);
    171  1.72        ad 	mutex_exit(&lfs_lock);
    172  1.31  perseant }
    173  1.31  perseant 
    174  1.31  perseant static unsigned int
    175  1.31  perseant lfs_mhash(void *vp)
    176  1.31  perseant {
    177  1.31  perseant 	return (unsigned int)(((unsigned long)vp) >> 2) % LFS_RESHASH_WIDTH;
    178  1.31  perseant }
    179  1.31  perseant 
    180  1.31  perseant /*
    181  1.31  perseant  * Return memory of the given size for the given purpose, or use one of a
    182  1.31  perseant  * number of spare last-resort buffers, if malloc returns NULL.
    183  1.49     perry  */
    184  1.31  perseant void *
    185  1.31  perseant lfs_malloc(struct lfs *fs, size_t size, int type)
    186  1.31  perseant {
    187  1.31  perseant 	struct lfs_res_blk *re;
    188  1.31  perseant 	void *r;
    189  1.31  perseant 	int i, s, start;
    190  1.31  perseant 	unsigned int h;
    191  1.31  perseant 
    192  1.51  perseant 	ASSERT_MAYBE_SEGLOCK(fs);
    193  1.34  perseant 	r = NULL;
    194  1.34  perseant 
    195  1.31  perseant 	/* If no mem allocated for this type, it just waits */
    196  1.34  perseant 	if (lfs_res_qty[type] == 0) {
    197  1.34  perseant 		r = malloc(size, M_SEGMENT, M_WAITOK);
    198  1.34  perseant 		return r;
    199  1.34  perseant 	}
    200  1.31  perseant 
    201  1.31  perseant 	/* Otherwise try a quick malloc, and if it works, great */
    202  1.34  perseant 	if ((r = malloc(size, M_SEGMENT, M_NOWAIT)) != NULL) {
    203  1.31  perseant 		return r;
    204  1.34  perseant 	}
    205  1.31  perseant 
    206  1.31  perseant 	/*
    207  1.31  perseant 	 * If malloc returned NULL, we are forced to use one of our
    208  1.31  perseant 	 * reserve blocks.  We have on hand at least one summary block,
    209  1.31  perseant 	 * at least one cluster block, at least one superblock,
    210  1.31  perseant 	 * and several indirect blocks.
    211  1.31  perseant 	 */
    212  1.51  perseant 
    213  1.72        ad 	mutex_enter(&lfs_lock);
    214  1.31  perseant 	/* skip over blocks of other types */
    215  1.31  perseant 	for (i = 0, start = 0; i < type; i++)
    216  1.31  perseant 		start += lfs_res_qty[i];
    217  1.31  perseant 	while (r == NULL) {
    218  1.31  perseant 		for (i = 0; i < lfs_res_qty[type]; i++) {
    219  1.31  perseant 			if (fs->lfs_resblk[start + i].inuse == 0) {
    220  1.31  perseant 				re = fs->lfs_resblk + start + i;
    221  1.31  perseant 				re->inuse = 1;
    222  1.31  perseant 				r = re->p;
    223  1.34  perseant 				KASSERT(re->size >= size);
    224  1.31  perseant 				h = lfs_mhash(r);
    225  1.31  perseant 				s = splbio();
    226  1.31  perseant 				LIST_INSERT_HEAD(&fs->lfs_reshash[h], re, res);
    227  1.31  perseant 				splx(s);
    228  1.72        ad 				mutex_exit(&lfs_lock);
    229  1.31  perseant 				return r;
    230  1.31  perseant 			}
    231  1.31  perseant 		}
    232  1.51  perseant 		DLOG((DLOG_MALLOC, "sleeping on %s (%d)\n",
    233  1.51  perseant 		      lfs_res_names[type], lfs_res_qty[type]));
    234  1.72        ad 		mtsleep(&fs->lfs_resblk, PVM, "lfs_malloc", 0,
    235  1.72        ad 			&lfs_lock);
    236  1.51  perseant 		DLOG((DLOG_MALLOC, "done sleeping on %s\n",
    237  1.51  perseant 		      lfs_res_names[type]));
    238  1.31  perseant 	}
    239  1.31  perseant 	/* NOTREACHED */
    240  1.72        ad 	mutex_exit(&lfs_lock);
    241  1.31  perseant 	return r;
    242  1.31  perseant }
    243  1.31  perseant 
    244  1.31  perseant void
    245  1.65  christos lfs_free(struct lfs *fs, void *p, int type)
    246  1.31  perseant {
    247  1.31  perseant 	int s;
    248  1.31  perseant 	unsigned int h;
    249  1.31  perseant 	res_t *re;
    250  1.32      yamt #ifdef DEBUG
    251  1.32      yamt 	int i;
    252  1.32      yamt #endif
    253  1.31  perseant 
    254  1.51  perseant 	ASSERT_MAYBE_SEGLOCK(fs);
    255  1.31  perseant 	h = lfs_mhash(p);
    256  1.72        ad 	mutex_enter(&lfs_lock);
    257  1.31  perseant 	s = splbio();
    258  1.31  perseant 	LIST_FOREACH(re, &fs->lfs_reshash[h], res) {
    259  1.31  perseant 		if (re->p == p) {
    260  1.32      yamt 			KASSERT(re->inuse == 1);
    261  1.31  perseant 			LIST_REMOVE(re, res);
    262  1.31  perseant 			re->inuse = 0;
    263  1.31  perseant 			wakeup(&fs->lfs_resblk);
    264  1.31  perseant 			splx(s);
    265  1.72        ad 			mutex_exit(&lfs_lock);
    266  1.31  perseant 			return;
    267  1.31  perseant 		}
    268  1.31  perseant 	}
    269  1.32      yamt #ifdef DEBUG
    270  1.32      yamt 	for (i = 0; i < LFS_N_TOTAL; i++) {
    271  1.32      yamt 		if (fs->lfs_resblk[i].p == p)
    272  1.34  perseant 			panic("lfs_free: inconsistent reserved block");
    273  1.32      yamt 	}
    274  1.32      yamt #endif
    275  1.31  perseant 	splx(s);
    276  1.72        ad 	mutex_exit(&lfs_lock);
    277  1.51  perseant 
    278  1.31  perseant 	/*
    279  1.31  perseant 	 * If we didn't find it, free it.
    280  1.31  perseant 	 */
    281  1.31  perseant 	free(p, M_SEGMENT);
    282  1.31  perseant }
    283   1.1   mycroft 
    284   1.1   mycroft /*
    285   1.1   mycroft  * lfs_seglock --
    286   1.1   mycroft  *	Single thread the segment writer.
    287   1.1   mycroft  */
    288  1.31  perseant int
    289  1.18  perseant lfs_seglock(struct lfs *fs, unsigned long flags)
    290   1.1   mycroft {
    291   1.1   mycroft 	struct segment *sp;
    292  1.49     perry 
    293  1.72        ad 	mutex_enter(&lfs_lock);
    294   1.7   thorpej 	if (fs->lfs_seglock) {
    295  1.58  perseant 		if (fs->lfs_lockpid == curproc->p_pid &&
    296  1.58  perseant 		    fs->lfs_locklwp == curlwp->l_lid) {
    297   1.1   mycroft 			++fs->lfs_seglock;
    298   1.1   mycroft 			fs->lfs_sp->seg_flags |= flags;
    299  1.74   mlelstv 			mutex_exit(&lfs_lock);
    300  1.31  perseant 			return 0;
    301  1.38  perseant 		} else if (flags & SEGM_PAGEDAEMON) {
    302  1.72        ad 			mutex_exit(&lfs_lock);
    303  1.31  perseant 			return EWOULDBLOCK;
    304  1.51  perseant 		} else {
    305  1.51  perseant 			while (fs->lfs_seglock) {
    306  1.72        ad 				(void)mtsleep(&fs->lfs_seglock, PRIBIO + 1,
    307  1.72        ad 					"lfs_seglock", 0, &lfs_lock);
    308  1.51  perseant 			}
    309  1.51  perseant 		}
    310   1.7   thorpej 	}
    311  1.49     perry 
    312   1.1   mycroft 	fs->lfs_seglock = 1;
    313   1.1   mycroft 	fs->lfs_lockpid = curproc->p_pid;
    314  1.58  perseant 	fs->lfs_locklwp = curlwp->l_lid;
    315  1.72        ad 	mutex_exit(&lfs_lock);
    316  1.36  perseant 	fs->lfs_cleanind = 0;
    317  1.36  perseant 
    318  1.51  perseant #ifdef DEBUG
    319  1.51  perseant 	LFS_ENTER_LOG("seglock", __FILE__, __LINE__, 0, flags, curproc->p_pid);
    320  1.51  perseant #endif
    321  1.27  perseant 	/* Drain fragment size changes out */
    322  1.66        ad 	rw_enter(&fs->lfs_fraglock, RW_WRITER);
    323  1.27  perseant 
    324  1.31  perseant 	sp = fs->lfs_sp = pool_get(&fs->lfs_segpool, PR_WAITOK);
    325  1.31  perseant 	sp->bpp = pool_get(&fs->lfs_bpppool, PR_WAITOK);
    326   1.1   mycroft 	sp->seg_flags = flags;
    327   1.1   mycroft 	sp->vp = NULL;
    328  1.26  perseant 	sp->seg_iocount = 0;
    329   1.1   mycroft 	(void) lfs_initseg(fs);
    330  1.49     perry 
    331   1.1   mycroft 	/*
    332   1.1   mycroft 	 * Keep a cumulative count of the outstanding I/O operations.  If the
    333   1.1   mycroft 	 * disk drive catches up with us it could go to zero before we finish,
    334   1.1   mycroft 	 * so we artificially increment it by one until we've scheduled all of
    335   1.1   mycroft 	 * the writes we intend to do.
    336   1.1   mycroft 	 */
    337  1.72        ad 	mutex_enter(&lfs_lock);
    338   1.1   mycroft 	++fs->lfs_iocount;
    339  1.77  perseant 	fs->lfs_startseg = fs->lfs_curseg;
    340  1.72        ad 	mutex_exit(&lfs_lock);
    341  1.31  perseant 	return 0;
    342   1.1   mycroft }
    343   1.8  perseant 
    344  1.31  perseant static void lfs_unmark_dirop(struct lfs *);
    345  1.31  perseant 
    346  1.31  perseant static void
    347  1.31  perseant lfs_unmark_dirop(struct lfs *fs)
    348  1.31  perseant {
    349  1.31  perseant 	struct inode *ip, *nip;
    350  1.31  perseant 	struct vnode *vp;
    351  1.40  perseant 	int doit;
    352  1.31  perseant 
    353  1.51  perseant 	ASSERT_NO_SEGLOCK(fs);
    354  1.72        ad 	mutex_enter(&lfs_lock);
    355  1.40  perseant 	doit = !(fs->lfs_flags & LFS_UNDIROP);
    356  1.40  perseant 	if (doit)
    357  1.40  perseant 		fs->lfs_flags |= LFS_UNDIROP;
    358  1.51  perseant 	if (!doit) {
    359  1.72        ad 		mutex_exit(&lfs_lock);
    360  1.40  perseant 		return;
    361  1.51  perseant 	}
    362  1.40  perseant 
    363  1.31  perseant 	for (ip = TAILQ_FIRST(&fs->lfs_dchainhd); ip != NULL; ip = nip) {
    364  1.31  perseant 		nip = TAILQ_NEXT(ip, i_lfs_dchain);
    365  1.31  perseant 		vp = ITOV(ip);
    366  1.77  perseant 		if ((ip->i_flag & (IN_ADIROP | IN_CDIROP)) == IN_CDIROP) {
    367  1.31  perseant 			--lfs_dirvcount;
    368  1.59  perseant 			--fs->lfs_dirvcount;
    369  1.71        ad 			vp->v_uflag &= ~VU_DIROP;
    370  1.31  perseant 			TAILQ_REMOVE(&fs->lfs_dchainhd, ip, i_lfs_dchain);
    371  1.31  perseant 			wakeup(&lfs_dirvcount);
    372  1.31  perseant 			fs->lfs_unlockvp = vp;
    373  1.72        ad 			mutex_exit(&lfs_lock);
    374  1.76   hannken 			vrele(vp);
    375  1.72        ad 			mutex_enter(&lfs_lock);
    376  1.31  perseant 			fs->lfs_unlockvp = NULL;
    377  1.77  perseant 			ip->i_flag &= ~IN_CDIROP;
    378  1.72        ad 		}
    379  1.31  perseant 	}
    380  1.40  perseant 
    381  1.40  perseant 	fs->lfs_flags &= ~LFS_UNDIROP;
    382  1.51  perseant 	wakeup(&fs->lfs_flags);
    383  1.72        ad 	mutex_exit(&lfs_lock);
    384  1.31  perseant }
    385  1.31  perseant 
    386  1.31  perseant static void
    387  1.31  perseant lfs_auto_segclean(struct lfs *fs)
    388  1.31  perseant {
    389  1.48  perseant 	int i, error, s, waited;
    390  1.31  perseant 
    391  1.51  perseant 	ASSERT_SEGLOCK(fs);
    392  1.31  perseant 	/*
    393  1.31  perseant 	 * Now that we've swapped lfs_activesb, but while we still
    394  1.31  perseant 	 * hold the segment lock, run through the segment list marking
    395  1.31  perseant 	 * the empty ones clean.
    396  1.31  perseant 	 * XXX - do we really need to do them all at once?
    397  1.31  perseant 	 */
    398  1.48  perseant 	waited = 0;
    399  1.31  perseant 	for (i = 0; i < fs->lfs_nseg; i++) {
    400  1.31  perseant 		if ((fs->lfs_suflags[0][i] &
    401  1.31  perseant 		     (SEGUSE_ACTIVE | SEGUSE_DIRTY | SEGUSE_EMPTY)) ==
    402  1.31  perseant 		    (SEGUSE_DIRTY | SEGUSE_EMPTY) &&
    403  1.31  perseant 		    (fs->lfs_suflags[1][i] &
    404  1.31  perseant 		     (SEGUSE_ACTIVE | SEGUSE_DIRTY | SEGUSE_EMPTY)) ==
    405  1.31  perseant 		    (SEGUSE_DIRTY | SEGUSE_EMPTY)) {
    406  1.31  perseant 
    407  1.48  perseant 			/* Make sure the sb is written before we clean */
    408  1.72        ad 			mutex_enter(&lfs_lock);
    409  1.48  perseant 			s = splbio();
    410  1.48  perseant 			while (waited == 0 && fs->lfs_sbactive)
    411  1.72        ad 				mtsleep(&fs->lfs_sbactive, PRIBIO+1, "lfs asb",
    412  1.72        ad 					0, &lfs_lock);
    413  1.48  perseant 			splx(s);
    414  1.72        ad 			mutex_exit(&lfs_lock);
    415  1.48  perseant 			waited = 1;
    416  1.48  perseant 
    417  1.31  perseant 			if ((error = lfs_do_segclean(fs, i)) != 0) {
    418  1.50  perseant 				DLOG((DLOG_CLEAN, "lfs_auto_segclean: lfs_do_segclean returned %d for seg %d\n", error, i));
    419  1.31  perseant 			}
    420  1.31  perseant 		}
    421  1.31  perseant 		fs->lfs_suflags[1 - fs->lfs_activesb][i] =
    422  1.31  perseant 			fs->lfs_suflags[fs->lfs_activesb][i];
    423  1.31  perseant 	}
    424  1.31  perseant }
    425  1.31  perseant 
    426   1.1   mycroft /*
    427   1.1   mycroft  * lfs_segunlock --
    428   1.1   mycroft  *	Single thread the segment writer.
    429   1.1   mycroft  */
    430   1.1   mycroft void
    431  1.18  perseant lfs_segunlock(struct lfs *fs)
    432   1.1   mycroft {
    433   1.1   mycroft 	struct segment *sp;
    434   1.1   mycroft 	unsigned long sync, ckp;
    435  1.22  perseant 	struct buf *bp;
    436  1.40  perseant 	int do_unmark_dirop = 0;
    437  1.49     perry 
    438  1.15  perseant 	sp = fs->lfs_sp;
    439  1.15  perseant 
    440  1.72        ad 	mutex_enter(&lfs_lock);
    441  1.72        ad 	KASSERT(LFS_SEGLOCK_HELD(fs));
    442  1.15  perseant 	if (fs->lfs_seglock == 1) {
    443  1.77  perseant 		if ((sp->seg_flags & (SEGM_PROT | SEGM_CLEAN)) == 0)
    444  1.40  perseant 			do_unmark_dirop = 1;
    445  1.72        ad 		mutex_exit(&lfs_lock);
    446   1.1   mycroft 		sync = sp->seg_flags & SEGM_SYNC;
    447   1.1   mycroft 		ckp = sp->seg_flags & SEGM_CKP;
    448  1.57  perseant 
    449  1.57  perseant 		/* We should have a segment summary, and nothing else */
    450  1.57  perseant 		KASSERT(sp->cbpp == sp->bpp + 1);
    451  1.57  perseant 
    452  1.57  perseant 		/* Free allocated segment summary */
    453  1.79  christos 		fs->lfs_offset -= lfs_btofsb(fs, fs->lfs_sumsize);
    454  1.57  perseant 		bp = *sp->bpp;
    455  1.57  perseant 		lfs_freebuf(fs, bp);
    456   1.8  perseant 
    457  1.31  perseant 		pool_put(&fs->lfs_bpppool, sp->bpp);
    458  1.18  perseant 		sp->bpp = NULL;
    459  1.36  perseant 
    460  1.36  perseant 		/*
    461  1.36  perseant 		 * If we're not sync, we're done with sp, get rid of it.
    462  1.36  perseant 		 * Otherwise, we keep a local copy around but free
    463  1.36  perseant 		 * fs->lfs_sp so another process can use it (we have to
    464  1.36  perseant 		 * wait but they don't have to wait for us).
    465  1.36  perseant 		 */
    466  1.26  perseant 		if (!sync)
    467  1.31  perseant 			pool_put(&fs->lfs_segpool, sp);
    468  1.18  perseant 		fs->lfs_sp = NULL;
    469   1.1   mycroft 
    470   1.1   mycroft 		/*
    471   1.1   mycroft 		 * If the I/O count is non-zero, sleep until it reaches zero.
    472   1.1   mycroft 		 * At the moment, the user's process hangs around so we can
    473   1.1   mycroft 		 * sleep.
    474   1.1   mycroft 		 */
    475  1.72        ad 		mutex_enter(&lfs_lock);
    476  1.63  christos 		if (--fs->lfs_iocount == 0) {
    477  1.44      yamt 			LFS_DEBUG_COUNTLOCKED("lfs_segunlock");
    478  1.63  christos 		}
    479  1.36  perseant 		if (fs->lfs_iocount <= 1)
    480  1.22  perseant 			wakeup(&fs->lfs_iocount);
    481  1.72        ad 		mutex_exit(&lfs_lock);
    482   1.1   mycroft 		/*
    483  1.26  perseant 		 * If we're not checkpointing, we don't have to block
    484  1.26  perseant 		 * other processes to wait for a synchronous write
    485  1.26  perseant 		 * to complete.
    486  1.26  perseant 		 */
    487  1.26  perseant 		if (!ckp) {
    488  1.51  perseant #ifdef DEBUG
    489  1.51  perseant 			LFS_ENTER_LOG("segunlock_std", __FILE__, __LINE__, 0, 0, curproc->p_pid);
    490  1.51  perseant #endif
    491  1.72        ad 			mutex_enter(&lfs_lock);
    492  1.26  perseant 			--fs->lfs_seglock;
    493  1.26  perseant 			fs->lfs_lockpid = 0;
    494  1.58  perseant 			fs->lfs_locklwp = 0;
    495  1.72        ad 			mutex_exit(&lfs_lock);
    496  1.26  perseant 			wakeup(&fs->lfs_seglock);
    497  1.26  perseant 		}
    498  1.26  perseant 		/*
    499   1.1   mycroft 		 * We let checkpoints happen asynchronously.  That means
    500   1.1   mycroft 		 * that during recovery, we have to roll forward between
    501   1.1   mycroft 		 * the two segments described by the first and second
    502   1.1   mycroft 		 * superblocks to make sure that the checkpoint described
    503   1.1   mycroft 		 * by a superblock completed.
    504   1.1   mycroft 		 */
    505  1.72        ad 		mutex_enter(&lfs_lock);
    506  1.74   mlelstv 		while (ckp && sync && fs->lfs_iocount) {
    507  1.72        ad 			(void)mtsleep(&fs->lfs_iocount, PRIBIO + 1,
    508  1.72        ad 				      "lfs_iocount", 0, &lfs_lock);
    509  1.74   mlelstv 			DLOG((DLOG_SEG, "sleeping on iocount %x == %d\n", fs, fs->lfs_iocount));
    510  1.74   mlelstv 		}
    511  1.26  perseant 		while (sync && sp->seg_iocount) {
    512  1.72        ad 			(void)mtsleep(&sp->seg_iocount, PRIBIO + 1,
    513  1.72        ad 				     "seg_iocount", 0, &lfs_lock);
    514  1.50  perseant 			DLOG((DLOG_SEG, "sleeping on iocount %x == %d\n", sp, sp->seg_iocount));
    515  1.26  perseant 		}
    516  1.72        ad 		mutex_exit(&lfs_lock);
    517  1.26  perseant 		if (sync)
    518  1.31  perseant 			pool_put(&fs->lfs_segpool, sp);
    519  1.36  perseant 
    520   1.1   mycroft 		if (ckp) {
    521   1.1   mycroft 			fs->lfs_nactive = 0;
    522   1.8  perseant 			/* If we *know* everything's on disk, write both sbs */
    523  1.33  perseant 			/* XXX should wait for this one	 */
    524  1.21       chs 			if (sync)
    525  1.31  perseant 				lfs_writesuper(fs, fs->lfs_sboffs[fs->lfs_activesb]);
    526  1.31  perseant 			lfs_writesuper(fs, fs->lfs_sboffs[1 - fs->lfs_activesb]);
    527  1.48  perseant 			if (!(fs->lfs_ivnode->v_mount->mnt_iflag & IMNT_UNMOUNT)) {
    528  1.35  perseant 				lfs_auto_segclean(fs);
    529  1.48  perseant 				/* If sync, we can clean the remainder too */
    530  1.48  perseant 				if (sync)
    531  1.48  perseant 					lfs_auto_segclean(fs);
    532  1.48  perseant 			}
    533   1.8  perseant 			fs->lfs_activesb = 1 - fs->lfs_activesb;
    534  1.51  perseant #ifdef DEBUG
    535  1.51  perseant 			LFS_ENTER_LOG("segunlock_ckp", __FILE__, __LINE__, 0, 0, curproc->p_pid);
    536  1.51  perseant #endif
    537  1.72        ad 			mutex_enter(&lfs_lock);
    538  1.26  perseant 			--fs->lfs_seglock;
    539  1.26  perseant 			fs->lfs_lockpid = 0;
    540  1.58  perseant 			fs->lfs_locklwp = 0;
    541  1.72        ad 			mutex_exit(&lfs_lock);
    542  1.26  perseant 			wakeup(&fs->lfs_seglock);
    543   1.1   mycroft 		}
    544  1.27  perseant 		/* Reenable fragment size changes */
    545  1.66        ad 		rw_exit(&fs->lfs_fraglock);
    546  1.40  perseant 		if (do_unmark_dirop)
    547  1.40  perseant 			lfs_unmark_dirop(fs);
    548   1.1   mycroft 	} else if (fs->lfs_seglock == 0) {
    549  1.72        ad 		mutex_exit(&lfs_lock);
    550   1.1   mycroft 		panic ("Seglock not held");
    551   1.1   mycroft 	} else {
    552   1.1   mycroft 		--fs->lfs_seglock;
    553  1.72        ad 		mutex_exit(&lfs_lock);
    554   1.1   mycroft 	}
    555  1.41      yamt }
    556  1.41      yamt 
    557  1.41      yamt /*
    558  1.69  perseant  * Drain dirops and start writer.
    559  1.69  perseant  *
    560  1.69  perseant  * No simple_locks are held when we enter and none are held when we return.
    561  1.41      yamt  */
    562  1.41      yamt int
    563  1.41      yamt lfs_writer_enter(struct lfs *fs, const char *wmesg)
    564  1.41      yamt {
    565  1.41      yamt 	int error = 0;
    566  1.41      yamt 
    567  1.51  perseant 	ASSERT_MAYBE_SEGLOCK(fs);
    568  1.72        ad 	mutex_enter(&lfs_lock);
    569  1.41      yamt 
    570  1.41      yamt 	/* disallow dirops during flush */
    571  1.41      yamt 	fs->lfs_writer++;
    572  1.41      yamt 
    573  1.41      yamt 	while (fs->lfs_dirops > 0) {
    574  1.49     perry 		++fs->lfs_diropwait;
    575  1.72        ad 		error = mtsleep(&fs->lfs_writer, PRIBIO+1, wmesg, 0,
    576  1.72        ad 				&lfs_lock);
    577  1.49     perry 		--fs->lfs_diropwait;
    578  1.41      yamt 	}
    579  1.41      yamt 
    580  1.41      yamt 	if (error)
    581  1.41      yamt 		fs->lfs_writer--;
    582  1.41      yamt 
    583  1.72        ad 	mutex_exit(&lfs_lock);
    584  1.41      yamt 
    585  1.41      yamt 	return error;
    586  1.41      yamt }
    587  1.41      yamt 
    588  1.41      yamt void
    589  1.41      yamt lfs_writer_leave(struct lfs *fs)
    590  1.41      yamt {
    591  1.67   thorpej 	bool dowakeup;
    592  1.41      yamt 
    593  1.51  perseant 	ASSERT_MAYBE_SEGLOCK(fs);
    594  1.72        ad 	mutex_enter(&lfs_lock);
    595  1.41      yamt 	dowakeup = !(--fs->lfs_writer);
    596  1.72        ad 	mutex_exit(&lfs_lock);
    597  1.41      yamt 	if (dowakeup)
    598  1.41      yamt 		wakeup(&fs->lfs_dirops);
    599   1.1   mycroft }
    600  1.57  perseant 
    601  1.57  perseant /*
    602  1.57  perseant  * Unlock, wait for the cleaner, then relock to where we were before.
    603  1.57  perseant  * To be used only at a fairly high level, to address a paucity of free
    604  1.57  perseant  * segments propagated back from lfs_gop_write().
    605  1.57  perseant  */
    606  1.57  perseant void
    607  1.57  perseant lfs_segunlock_relock(struct lfs *fs)
    608  1.57  perseant {
    609  1.57  perseant 	int n = fs->lfs_seglock;
    610  1.57  perseant 	u_int16_t seg_flags;
    611  1.61  perseant 	CLEANERINFO *cip;
    612  1.61  perseant 	struct buf *bp;
    613  1.57  perseant 
    614  1.57  perseant 	if (n == 0)
    615  1.57  perseant 		return;
    616  1.57  perseant 
    617  1.57  perseant 	/* Write anything we've already gathered to disk */
    618  1.57  perseant 	lfs_writeseg(fs, fs->lfs_sp);
    619  1.57  perseant 
    620  1.61  perseant 	/* Tell cleaner */
    621  1.61  perseant 	LFS_CLEANERINFO(cip, fs, bp);
    622  1.61  perseant 	cip->flags |= LFS_CLEANER_MUST_CLEAN;
    623  1.61  perseant 	LFS_SYNC_CLEANERINFO(cip, fs, bp, 1);
    624  1.61  perseant 
    625  1.57  perseant 	/* Save segment flags for later */
    626  1.57  perseant 	seg_flags = fs->lfs_sp->seg_flags;
    627  1.57  perseant 
    628  1.57  perseant 	fs->lfs_sp->seg_flags |= SEGM_PROT; /* Don't unmark dirop nodes */
    629  1.57  perseant 	while(fs->lfs_seglock)
    630  1.57  perseant 		lfs_segunlock(fs);
    631  1.57  perseant 
    632  1.57  perseant 	/* Wait for the cleaner */
    633  1.60  perseant 	lfs_wakeup_cleaner(fs);
    634  1.72        ad 	mutex_enter(&lfs_lock);
    635  1.57  perseant 	while (LFS_STARVED_FOR_SEGS(fs))
    636  1.72        ad 		mtsleep(&fs->lfs_avail, PRIBIO, "relock", 0,
    637  1.72        ad 			&lfs_lock);
    638  1.72        ad 	mutex_exit(&lfs_lock);
    639  1.57  perseant 
    640  1.57  perseant 	/* Put the segment lock back the way it was. */
    641  1.57  perseant 	while(n--)
    642  1.57  perseant 		lfs_seglock(fs, seg_flags);
    643  1.57  perseant 
    644  1.61  perseant 	/* Cleaner can relax now */
    645  1.61  perseant 	LFS_CLEANERINFO(cip, fs, bp);
    646  1.61  perseant 	cip->flags &= ~LFS_CLEANER_MUST_CLEAN;
    647  1.61  perseant 	LFS_SYNC_CLEANERINFO(cip, fs, bp, 1);
    648  1.61  perseant 
    649  1.57  perseant 	return;
    650  1.57  perseant }
    651  1.60  perseant 
    652  1.60  perseant /*
    653  1.60  perseant  * Wake up the cleaner, provided that nowrap is not set.
    654  1.60  perseant  */
    655  1.60  perseant void
    656  1.60  perseant lfs_wakeup_cleaner(struct lfs *fs)
    657  1.60  perseant {
    658  1.60  perseant 	if (fs->lfs_nowrap > 0)
    659  1.60  perseant 		return;
    660  1.60  perseant 
    661  1.60  perseant 	wakeup(&fs->lfs_nextseg);
    662  1.60  perseant 	wakeup(&lfs_allclean_wakeup);
    663  1.60  perseant }
    664