diff options
| author | cvs2svn <cvs2svn@FreeBSD.org> | 2002-07-03 13:01:42 +0000 |
|---|---|---|
| committer | cvs2svn <cvs2svn@FreeBSD.org> | 2002-07-03 13:01:42 +0000 |
| commit | db0a3b09965d55f9dee7a58fb04e31df42b6ae73 (patch) | |
| tree | 15796774b99f140ef7279f6e0a9e8762ca56f414 /sys/gnu/ext2fs | |
| parent | b8fd368d376570f335654782a15814d70c9d21c6 (diff) | |
| parent | 7194701a4d8d2a0fc75c1b41fcb2a9447ed3a4ca (diff) | |
Notes
Diffstat (limited to 'sys/gnu/ext2fs')
| -rw-r--r-- | sys/gnu/ext2fs/COPYRIGHT.INFO | 35 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/alpha-bitops.h | 255 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_alloc.c | 556 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_balloc.c | 314 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_extern.h | 107 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_fs.h | 572 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_fs_sb.h | 81 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_inode.c | 494 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_inode_cnv.c | 160 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_linux_balloc.c | 610 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_linux_ialloc.c | 515 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_lookup.c | 1078 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_readwrite.c | 308 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_subr.c | 128 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_vfsops.c | 1207 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/ext2_vnops.c | 1215 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/fs.h | 184 | ||||
| -rw-r--r-- | sys/gnu/ext2fs/i386-bitops.h | 162 |
18 files changed, 7981 insertions, 0 deletions
diff --git a/sys/gnu/ext2fs/COPYRIGHT.INFO b/sys/gnu/ext2fs/COPYRIGHT.INFO new file mode 100644 index 000000000000..699f519c02a8 --- /dev/null +++ b/sys/gnu/ext2fs/COPYRIGHT.INFO @@ -0,0 +1,35 @@ +$FreeBSD$ + +Most of the files in this directory are written by Godmar Back or modified +by him using the CSRG sources. Those files are covered by the Berkeley-style +copyright. However the following files are covered by GPL. Since the policy +of the FreeBSD project is to keep the files with the more restrictive +copyright in the gnu tree and it is a good idea to keep the filesystem code +all together, the EXT2FS in its entirety resides under the gnu tree. Note +that only the files below are under the GPL. In the eventuality that these +files are redesigned or rewritten, this tree can be moved back into the less +restrictive FreeBSD tree. + + ext2_bitmap.c (in the cvs attic) + ext2_fs.h + ext2_fs_i.h + ext2_fs_sb.h + ext2_linux_balloc.c + ext2_linux_ialloc.c + ext2_super.c (in the cvs attic) + ext2_vfsops.c (has some GPL'ed code from ext2_super.c) + i386-bitops.h + +PS. + THANKS GODMAR!!! + +Note that this port has been modified by John Dyson and others on +the FreeBSD team, and it is best to send the bug reports to the FreeBSD +team. If there are any non-FreeBSD specific bugs, fixes will be sent to +Godmar to help him fix the original code base. It is also our intention +to send Godmar any FreeBSD specific porting changes so that he can keep +control of his code.... + +John +dyson@freebsd.org + diff --git a/sys/gnu/ext2fs/alpha-bitops.h b/sys/gnu/ext2fs/alpha-bitops.h new file mode 100644 index 000000000000..3910da8a323b --- /dev/null +++ b/sys/gnu/ext2fs/alpha-bitops.h @@ -0,0 +1,255 @@ +/* $FreeBSD$ */ +#ifndef _ALPHA_BITOPS_H +#define _ALPHA_BITOPS_H + +/* + * Copyright 1994, Linus Torvalds. + */ + +/* + * These have to be done with inline assembly: that way the bit-setting + * is guaranteed to be atomic. All bit operations return 0 if the bit + * was cleared before the operation and != 0 if it was not. + * + * To get proper branch prediction for the main line, we must branch + * forward to code at the end of this object's .text section, then + * branch back to restart the operation. + * + * bit 0 is the LSB of addr; bit 64 is the LSB of (addr+1). + */ +static __inline unsigned int set_bit(unsigned long, volatile void *); +static __inline unsigned int clear_bit(unsigned long, volatile void *); +static __inline unsigned int change_bit(unsigned long, volatile void *); +static __inline unsigned int test_bit(int, volatile void *); +static __inline unsigned long ffz_b(unsigned long x); +static __inline unsigned long ffz(unsigned long); +/* static __inline int ffs(int); */ +static __inline void * memscan(void *, int, size_t); +#ifdef __alpha_cix__ +static __inline unsigned long hweight64(unsigned long); +#endif +static __inline unsigned long +find_next_zero_bit(void *, unsigned long, unsigned long); + +static __inline unsigned int set_bit(unsigned long nr, volatile void * addr) +{ + unsigned long oldbit; + unsigned long temp; + volatile unsigned int *m = ((volatile unsigned int *) addr) + (nr >> 5); + + __asm__ __volatile__( + "1: ldl_l %0,%1\n" + " and %0,%3,%2\n" + " bne %2,2f\n" + " xor %0,%3,%0\n" + " stl_c %0,%1\n" + " beq %0,3f\n" + "2:\n" + ".section .text2,\"ax\"\n" + "3: br 1b\n" + ".previous" + :"=&r" (temp), "=m" (*m), "=&r" (oldbit) + :"Ir" (1UL << (nr & 31)), "m" (*m)); + return oldbit; +} + +static __inline unsigned int clear_bit(unsigned long nr, volatile void * addr) +{ + unsigned long oldbit; + unsigned long temp; + volatile unsigned int *m = ((volatile unsigned int *) addr) + (nr >> 5); + + __asm__ __volatile__( + "1: ldl_l %0,%1\n" + " and %0,%3,%2\n" + " beq %2,2f\n" + " xor %0,%3,%0\n" + " stl_c %0,%1\n" + " beq %0,3f\n" + "2:\n" + ".section .text2,\"ax\"\n" + "3: br 1b\n" + ".previous" + :"=&r" (temp), "=m" (*m), "=&r" (oldbit) + :"Ir" (1UL << (nr & 31)), "m" (*m)); + return oldbit; +} + +static __inline unsigned int change_bit(unsigned long nr, volatile void * addr) +{ + unsigned long oldbit; + unsigned long temp; + volatile unsigned int *m = ((volatile unsigned int *) addr) + (nr >> 5); + + __asm__ __volatile__( + "1: ldl_l %0,%1\n" + " xor %0,%2,%0\n" + " stl_c %0,%1\n" + " beq %0,3f\n" + ".section .text2,\"ax\"\n" + "3: br 1b\n" + ".previous" + :"=&r" (temp), "=m" (*m), "=&r" (oldbit) + :"Ir" (1UL << (nr & 31)), "m" (*m)); + return oldbit; +} + +static __inline unsigned int test_bit(int nr, volatile void * addr) +{ + return 1UL & (((volatile int *) addr)[nr >> 5] >> (nr & 31)); +} + +/* + * ffz = Find First Zero in word. Undefined if no zero exists, + * so code should check against ~0UL first.. + * + * Do a binary search on the bits. Due to the nature of large + * constants on the alpha, it is worthwhile to split the search. + */ +static __inline unsigned long ffz_b(unsigned long x) +{ + unsigned long sum = 0; + + x = ~x & -~x; /* set first 0 bit, clear others */ + if (x & 0xF0) sum += 4; + if (x & 0xCC) sum += 2; + if (x & 0xAA) sum += 1; + + return sum; +} + +static __inline unsigned long ffz(unsigned long word) +{ +#ifdef __alpha_cix__ + /* Whee. EV6 can calculate it directly. */ + unsigned long result; + __asm__("ctlz %1,%0" : "=r"(result) : "r"(~word)); + return result; +#else + unsigned long bits, qofs, bofs; + + __asm__("cmpbge %1,%2,%0" : "=r"(bits) : "r"(word), "r"(~0UL)); + qofs = ffz_b(bits); + __asm__("extbl %1,%2,%0" : "=r"(bits) : "r"(word), "r"(qofs)); + bofs = ffz_b(bits); + + return qofs*8 + bofs; +#endif +} + +#ifdef __KERNEL__ +#if 0 +/* + * ffs: find first bit set. This is defined the same way as + * the libc and compiler builtin ffs routines, therefore + * differs in spirit from the above ffz (man ffs). + */ + +static __inline int ffs(int word) +{ + int result = ffz(~word); + return word ? result+1 : 0; +} +#endif + +/* + * hweightN: returns the hamming weight (i.e. the number + * of bits set) of a N-bit word + */ + +#ifdef __alpha_cix__ +/* Whee. EV6 can calculate it directly. */ +static __inline unsigned long hweight64(unsigned long w) +{ + unsigned long result; + __asm__("ctpop %1,%0" : "=r"(result) : "r"(w)); + return result; +} + +#define hweight32(x) hweight64((x) & 0xfffffffful) +#define hweight16(x) hweight64((x) & 0xfffful) +#define hweight8(x) hweight64((x) & 0xfful) +#else +#define hweight32(x) generic_hweight32(x) +#define hweight16(x) generic_hweight16(x) +#define hweight8(x) generic_hweight8(x) +#endif + +#endif /* __alpha_cix__ */ + +/* from lib/string.c */ +static __inline void * memscan(void * addr, int c, size_t size) +{ + unsigned char * p = (unsigned char *) addr; + + while (size) { + if (*p == c) + return (void *) p; + p++; + size--; + } + return (void *) p; +} + + +/* + * Find next zero bit in a bitmap reasonably efficiently.. + */ +static __inline unsigned long find_next_zero_bit(void * addr, unsigned long size, unsigned long offset) +{ + unsigned long * p = ((unsigned long *) addr) + (offset >> 6); + unsigned long result = offset & ~63UL; + unsigned long tmp; + + if (offset >= size) + return size; + size -= result; + offset &= 63UL; + if (offset) { + tmp = *(p++); + tmp |= ~0UL >> (64-offset); + if (size < 64) + goto found_first; + if (~tmp) + goto found_middle; + size -= 64; + result += 64; + } + while (size & ~63UL) { + if (~(tmp = *(p++))) + goto found_middle; + result += 64; + size -= 64; + } + if (!size) + return result; + tmp = *p; +found_first: + tmp |= ~0UL << size; +found_middle: + return result + ffz(tmp); +} + +/* + * The optimizer actually does good code for this case.. + */ +#define find_first_zero_bit(addr, size) \ + find_next_zero_bit((addr), (size), 0) + +#ifdef __KERNEL__ + +#define ext2_set_bit test_and_set_bit +#define ext2_clear_bit test_and_clear_bit +#define ext2_test_bit test_bit +#define ext2_find_first_zero_bit find_first_zero_bit +#define ext2_find_next_zero_bit find_next_zero_bit + +/* Bitmap functions for the minix filesystem. */ +#define minix_set_bit(nr,addr) test_and_set_bit(nr,addr) +#define minix_clear_bit(nr,addr) test_and_clear_bit(nr,addr) +#define minix_test_bit(nr,addr) test_bit(nr,addr) +#define minix_find_first_zero_bit(addr,size) find_first_zero_bit(addr,size) + +#endif /* __KERNEL__ */ + +#endif /* _ALPHA_BITOPS_H */ diff --git a/sys/gnu/ext2fs/ext2_alloc.c b/sys/gnu/ext2fs/ext2_alloc.c new file mode 100644 index 000000000000..b5336a534fed --- /dev/null +++ b/sys/gnu/ext2fs/ext2_alloc.c @@ -0,0 +1,556 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ext2_alloc.c 8.8 (Berkeley) 2/21/94 + * $FreeBSD$ + */ + +#include "opt_quota.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/conf.h> +#include <sys/vnode.h> +#include <sys/stat.h> +#include <sys/mount.h> +#include <sys/syslog.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufsmount.h> + +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> +#include <gnu/ext2fs/ext2_extern.h> + +static void ext2_fserr __P((struct ext2_sb_info *, u_int, char *)); + +/* + * Linux calls this functions at the following locations: + * (1) the inode is freed + * (2) a preallocation miss occurs + * (3) truncate is called + * (4) release_file is called and f_mode & 2 + * + * I call it in ext2_inactive, ext2_truncate, ext2_vfree and in (2) + * the call in vfree might be redundant + */ +void +ext2_discard_prealloc(ip) + struct inode * ip; +{ +#ifdef EXT2_PREALLOCATE + if (ip->i_prealloc_count) { + int i = ip->i_prealloc_count; + ip->i_prealloc_count = 0; + ext2_free_blocks (ITOV(ip)->v_mount, + ip->i_prealloc_block, + i); + } +#endif +} + +/* + * Allocate a block in the file system. + * + * this takes the framework from ffs_alloc. To implement the + * actual allocation, it calls ext2_new_block, the ported version + * of the same Linux routine. + * + * we note that this is always called in connection with ext2_blkpref + * + * preallocation is done as Linux does it + */ +int +ext2_alloc(ip, lbn, bpref, size, cred, bnp) + register struct inode *ip; + daddr_t lbn, bpref; + int size; + struct ucred *cred; + daddr_t *bnp; +{ + register struct ext2_sb_info *fs; + daddr_t bno; +#if QUOTA + int error; +#endif + + *bnp = 0; + fs = ip->i_e2fs; +#if DIAGNOSTIC + if ((u_int)size > fs->s_blocksize || blkoff(fs, size) != 0) { + printf("dev = %s, bsize = %lu, size = %d, fs = %s\n", + devtoname(ip->i_dev), fs->s_blocksize, size, fs->fs_fsmnt); + panic("ext2_alloc: bad size"); + } + if (cred == NOCRED) + panic("ext2_alloc: missing credential"); +#endif /* DIAGNOSTIC */ + if (size == fs->s_blocksize && fs->s_es->s_free_blocks_count == 0) + goto nospace; + if (cred->cr_uid != 0 && + fs->s_es->s_free_blocks_count < fs->s_es->s_r_blocks_count) + goto nospace; +#if QUOTA + if ((error = chkdq(ip, (long)btodb(size), cred, 0)) != 0) + return (error); +#endif + if (bpref >= fs->s_es->s_blocks_count) + bpref = 0; + /* call the Linux code */ +#ifdef EXT2_PREALLOCATE + /* To have a preallocation hit, we must + * - have at least one block preallocated + * - and our preferred block must have that block number or one below + */ + if (ip->i_prealloc_count && + (bpref == ip->i_prealloc_block || + bpref + 1 == ip->i_prealloc_block)) + { + bno = ip->i_prealloc_block++; + ip->i_prealloc_count--; + /* ext2_debug ("preallocation hit (%lu/%lu).\n", + ++alloc_hits, ++alloc_attempts); */ + + /* Linux gets, clears, and releases the buffer at this + point - we don't have to that; we leave it to the caller + */ + } else { + ext2_discard_prealloc (ip); + /* ext2_debug ("preallocation miss (%lu/%lu).\n", + alloc_hits, ++alloc_attempts); */ + if (S_ISREG(ip->i_mode)) + bno = ext2_new_block + (ITOV(ip)->v_mount, bpref, + &ip->i_prealloc_count, + &ip->i_prealloc_block); + else + bno = (daddr_t)ext2_new_block(ITOV(ip)->v_mount, + bpref, 0, 0); + } +#else + bno = (daddr_t)ext2_new_block(ITOV(ip)->v_mount, bpref, 0, 0); +#endif + + if (bno > 0) { + /* set next_alloc fields as done in block_getblk */ + ip->i_next_alloc_block = lbn; + ip->i_next_alloc_goal = bno; + + ip->i_blocks += btodb(size); + ip->i_flag |= IN_CHANGE | IN_UPDATE; + *bnp = bno; + return (0); + } +#if QUOTA + /* + * Restore user's disk quota because allocation failed. + */ + (void) chkdq(ip, (long)-btodb(size), cred, FORCE); +#endif +nospace: + ext2_fserr(fs, cred->cr_uid, "file system full"); + uprintf("\n%s: write failed, file system is full\n", fs->fs_fsmnt); + return (ENOSPC); +} + +/* + * Reallocate a sequence of blocks into a contiguous sequence of blocks. + * + * The vnode and an array of buffer pointers for a range of sequential + * logical blocks to be made contiguous is given. The allocator attempts + * to find a range of sequential blocks starting as close as possible to + * an fs_rotdelay offset from the end of the allocation for the logical + * block immediately preceeding the current range. If successful, the + * physical block numbers in the buffer pointers and in the inode are + * changed to reflect the new allocation. If unsuccessful, the allocation + * is left unchanged. The success in doing the reallocation is returned. + * Note that the error return is not reflected back to the user. Rather + * the previous block allocation will be used. + */ + +#ifdef FANCY_REALLOC +#include <sys/sysctl.h> +static int doasyncfree = 1; +#ifdef OPT_DEBUG +SYSCTL_INT(_debug, 14, doasyncfree, CTLFLAG_RW, &doasyncfree, 0, ""); +#endif /* OPT_DEBUG */ +#endif + +int +ext2_reallocblks(ap) + struct vop_reallocblks_args /* { + struct vnode *a_vp; + struct cluster_save *a_buflist; + } */ *ap; +{ +#ifndef FANCY_REALLOC +/* printf("ext2_reallocblks not implemented\n"); */ +return ENOSPC; +#else + + struct ext2_sb_info *fs; + struct inode *ip; + struct vnode *vp; + struct buf *sbp, *ebp; + daddr_t *bap, *sbap, *ebap; + struct cluster_save *buflist; + daddr_t start_lbn, end_lbn, soff, eoff, newblk, blkno; + struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp; + int i, len, start_lvl, end_lvl, pref, ssize; + + vp = ap->a_vp; + ip = VTOI(vp); + fs = ip->i_e2fs; +#ifdef UNKLAR + if (fs->fs_contigsumsize <= 0) + return (ENOSPC); +#endif + buflist = ap->a_buflist; + len = buflist->bs_nchildren; + start_lbn = buflist->bs_children[0]->b_lblkno; + end_lbn = start_lbn + len - 1; +#if DIAGNOSTIC + for (i = 1; i < len; i++) + if (buflist->bs_children[i]->b_lblkno != start_lbn + i) + panic("ext2_reallocblks: non-cluster"); +#endif + /* + * If the latest allocation is in a new cylinder group, assume that + * the filesystem has decided to move and do not force it back to + * the previous cylinder group. + */ + if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) != + dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno))) + return (ENOSPC); + if (ufs_getlbns(vp, start_lbn, start_ap, &start_lvl) || + ufs_getlbns(vp, end_lbn, end_ap, &end_lvl)) + return (ENOSPC); + /* + * Get the starting offset and block map for the first block. + */ + if (start_lvl == 0) { + sbap = &ip->i_db[0]; + soff = start_lbn; + } else { + idp = &start_ap[start_lvl - 1]; + if (bread(vp, idp->in_lbn, (int)fs->s_blocksize, NOCRED, &sbp)) { + brelse(sbp); + return (ENOSPC); + } + sbap = (daddr_t *)sbp->b_data; + soff = idp->in_off; + } + /* + * Find the preferred location for the cluster. + */ + pref = ext2_blkpref(ip, start_lbn, soff, sbap); + /* + * If the block range spans two block maps, get the second map. + */ + if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) { + ssize = len; + } else { +#if DIAGNOSTIC + if (start_ap[start_lvl-1].in_lbn == idp->in_lbn) + panic("ext2_reallocblk: start == end"); +#endif + ssize = len - (idp->in_off + 1); + if (bread(vp, idp->in_lbn, (int)fs->s_blocksize, NOCRED, &ebp)) + goto fail; + ebap = (daddr_t *)ebp->b_data; + } + /* + * Search the block map looking for an allocation of the desired size. + */ + if ((newblk = (daddr_t)ext2_hashalloc(ip, dtog(fs, pref), (long)pref, + len, (u_long (*)())ext2_clusteralloc)) == 0) + goto fail; + /* + * We have found a new contiguous block. + * + * First we have to replace the old block pointers with the new + * block pointers in the inode and indirect blocks associated + * with the file. + */ + blkno = newblk; + for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->s_frags_per_block) { + if (i == ssize) + bap = ebap; +#if DIAGNOSTIC + if (buflist->bs_children[i]->b_blkno != fsbtodb(fs, *bap)) + panic("ext2_reallocblks: alloc mismatch"); +#endif + *bap++ = blkno; + } + /* + * Next we must write out the modified inode and indirect blocks. + * For strict correctness, the writes should be synchronous since + * the old block values may have been written to disk. In practise + * they are almost never written, but if we are concerned about + * strict correctness, the `doasyncfree' flag should be set to zero. + * + * The test on `doasyncfree' should be changed to test a flag + * that shows whether the associated buffers and inodes have + * been written. The flag should be set when the cluster is + * started and cleared whenever the buffer or inode is flushed. + * We can then check below to see if it is set, and do the + * synchronous write only when it has been cleared. + */ + if (sbap != &ip->i_db[0]) { + if (doasyncfree) + bdwrite(sbp); + else + bwrite(sbp); + } else { + ip->i_flag |= IN_CHANGE | IN_UPDATE; + if (!doasyncfree) + UFS_UPDATE(vp, 1); + } + if (ssize < len) + if (doasyncfree) + bdwrite(ebp); + else + bwrite(ebp); + /* + * Last, free the old blocks and assign the new blocks to the buffers. + */ + for (blkno = newblk, i = 0; i < len; i++, blkno += fs->s_frags_per_block) { + ext2_blkfree(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno), + fs->s_blocksize); + buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno); + } + return (0); + +fail: + if (ssize < len) + brelse(ebp); + if (sbap != &ip->i_db[0]) + brelse(sbp); + return (ENOSPC); + +#endif /* FANCY_REALLOC */ +} + +/* + * Allocate an inode in the file system. + * + * we leave the actual allocation strategy to the (modified) + * ext2_new_inode(), to make sure we get the policies right + */ +int +ext2_valloc(pvp, mode, cred, vpp) + struct vnode *pvp; + int mode; + struct ucred *cred; + struct vnode **vpp; +{ + register struct inode *pip; + register struct ext2_sb_info *fs; + register struct inode *ip; + ino_t ino; + int i, error; + + *vpp = NULL; + pip = VTOI(pvp); + fs = pip->i_e2fs; + if (fs->s_es->s_free_inodes_count == 0) + goto noinodes; + + /* call the Linux routine - it returns the inode number only */ + ino = ext2_new_inode(pip, mode); + + if (ino == 0) + goto noinodes; + error = VFS_VGET(pvp->v_mount, ino, vpp); + if (error) { + UFS_VFREE(pvp, ino, mode); + return (error); + } + ip = VTOI(*vpp); + + /* + the question is whether using VGET was such good idea at all - + Linux doesn't read the old inode in when it's allocating a + new one. I will set at least i_size & i_blocks the zero. + */ + ip->i_mode = 0; + ip->i_size = 0; + ip->i_blocks = 0; + ip->i_flags = 0; + /* now we want to make sure that the block pointers are zeroed out */ + for (i = 0; i < NDADDR; i++) + ip->i_db[i] = 0; + for (i = 0; i < NIADDR; i++) + ip->i_ib[i] = 0; + + /* + * Set up a new generation number for this inode. + * XXX check if this makes sense in ext2 + */ + if (ip->i_gen == 0 || ++ip->i_gen == 0) + ip->i_gen = random() / 2 + 1; +/* +printf("ext2_valloc: allocated inode %d\n", ino); +*/ + return (0); +noinodes: + ext2_fserr(fs, cred->cr_uid, "out of inodes"); + uprintf("\n%s: create/symlink failed, no inodes free\n", fs->fs_fsmnt); + return (ENOSPC); +} + +/* + * Select the desired position for the next block in a file. + * + * we try to mimic what Remy does in inode_getblk/block_getblk + * + * we note: blocknr == 0 means that we're about to allocate either + * a direct block or a pointer block at the first level of indirection + * (In other words, stuff that will go in i_db[] or i_ib[]) + * + * blocknr != 0 means that we're allocating a block that is none + * of the above. Then, blocknr tells us the number of the block + * that will hold the pointer + */ +daddr_t +ext2_blkpref(ip, lbn, indx, bap, blocknr) + struct inode *ip; + daddr_t lbn; + int indx; + daddr_t *bap; + daddr_t blocknr; +{ + int tmp; + + /* if the next block is actually what we thought it is, + then set the goal to what we thought it should be + */ + if(ip->i_next_alloc_block == lbn) + return ip->i_next_alloc_goal; + + /* now check whether we were provided with an array that basically + tells us previous blocks to which we want to stay closeby + */ + if(bap) + for (tmp = indx - 1; tmp >= 0; tmp--) + if (bap[tmp]) + return bap[tmp]; + + /* else let's fall back to the blocknr, or, if there is none, + follow the rule that a block should be allocated near its inode + */ + return blocknr ? blocknr : + (daddr_t)(ip->i_block_group * + EXT2_BLOCKS_PER_GROUP(ip->i_e2fs)) + + ip->i_e2fs->s_es->s_first_data_block; +} + +/* + * Free a block or fragment. + * + * pass on to the Linux code + */ +void +ext2_blkfree(ip, bno, size) + register struct inode *ip; + daddr_t bno; + long size; +{ + register struct ext2_sb_info *fs; + + fs = ip->i_e2fs; + /* + * call Linux code with mount *, block number, count + */ + ext2_free_blocks(ITOV(ip)->v_mount, bno, size / fs->s_frag_size); +} + +/* + * Free an inode. + * + * the maintenance of the actual bitmaps is again up to the linux code + */ +int +ext2_vfree(pvp, ino, mode) + struct vnode *pvp; + ino_t ino; + int mode; +{ + register struct ext2_sb_info *fs; + register struct inode *pip; + register mode_t save_i_mode; + + pip = VTOI(pvp); + fs = pip->i_e2fs; + if ((u_int)ino >= fs->s_inodes_per_group * fs->s_groups_count) + panic("ext2_vfree: range: dev = (%d, %d), ino = %d, fs = %s", + major(pip->i_dev), minor(pip->i_dev), ino, fs->fs_fsmnt); + +/* ext2_debug("ext2_vfree (%d, %d) called\n", pip->i_number, mode); + */ + ext2_discard_prealloc(pip); + + /* we need to make sure that ext2_free_inode can adjust the + used_dir_counts in the group summary information - I'd + really like to know what the rationale behind this + 'set i_mode to zero to denote an unused inode' is + */ + save_i_mode = pip->i_mode; + pip->i_mode = mode; + ext2_free_inode(pip); + pip->i_mode = save_i_mode; + return (0); +} + +/* + * Fserr prints the name of a file system with an error diagnostic. + * + * The form of the error message is: + * fs: error message + */ +static void +ext2_fserr(fs, uid, cp) + struct ext2_sb_info *fs; + u_int uid; + char *cp; +{ + + log(LOG_ERR, "uid %d on %s: %s\n", uid, fs->fs_fsmnt, cp); +} diff --git a/sys/gnu/ext2fs/ext2_balloc.c b/sys/gnu/ext2fs/ext2_balloc.c new file mode 100644 index 000000000000..4ecb5dba6a67 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_balloc.c @@ -0,0 +1,314 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ffs_balloc.c 8.4 (Berkeley) 9/23/93 + * $FreeBSD$ + */ + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/buf.h> +#include <sys/lock.h> +#include <sys/ucred.h> +#include <sys/vnode.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufs_extern.h> + +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> +#include <gnu/ext2fs/ext2_extern.h> + +/* + * Balloc defines the structure of file system storage + * by allocating the physical blocks on a device given + * the inode and the logical block number in a file. + */ +int +ext2_balloc(ip, bn, size, cred, bpp, flags) + register struct inode *ip; + register daddr_t bn; + int size; + struct ucred *cred; + struct buf **bpp; + int flags; +{ + register struct ext2_sb_info *fs; + register daddr_t nb; + struct buf *bp, *nbp; + struct vnode *vp = ITOV(ip); + struct indir indirs[NIADDR + 2]; + daddr_t newb, lbn, *bap, pref; + int osize, nsize, num, i, error; +/* +ext2_debug("ext2_balloc called (%d, %d, %d)\n", + ip->i_number, (int)bn, (int)size); +*/ + *bpp = NULL; + if (bn < 0) + return (EFBIG); + fs = ip->i_e2fs; + lbn = bn; + + /* + * check if this is a sequential block allocation. + * If so, increment next_alloc fields to allow ext2_blkpref + * to make a good guess + */ + if (lbn == ip->i_next_alloc_block + 1) { + ip->i_next_alloc_block++; + ip->i_next_alloc_goal++; + } + + /* + * The first NDADDR blocks are direct blocks + */ + if (bn < NDADDR) { + nb = ip->i_db[bn]; + /* no new block is to be allocated, and no need to expand + the file */ + if (nb != 0 && ip->i_size >= (bn + 1) * fs->s_blocksize) { + error = bread(vp, bn, fs->s_blocksize, NOCRED, &bp); + if (error) { + brelse(bp); + return (error); + } + *bpp = bp; + return (0); + } + if (nb != 0) { + /* + * Consider need to reallocate a fragment. + */ + osize = fragroundup(fs, blkoff(fs, ip->i_size)); + nsize = fragroundup(fs, size); + if (nsize <= osize) { + error = bread(vp, bn, osize, NOCRED, &bp); + if (error) { + brelse(bp); + return (error); + } + } else { + /* Godmar thinks: this shouldn't happen w/o fragments */ + printf("nsize %d(%d) > osize %d(%d) nb %d\n", + (int)nsize, (int)size, (int)osize, + (int)ip->i_size, (int)nb); + panic( + "ext2_balloc: Something is terribly wrong"); +/* + * please note there haven't been any changes from here on - + * FFS seems to work. + */ + } + } else { + if (ip->i_size < (bn + 1) * fs->s_blocksize) + nsize = fragroundup(fs, size); + else + nsize = fs->s_blocksize; + error = ext2_alloc(ip, bn, + ext2_blkpref(ip, bn, (int)bn, &ip->i_db[0], 0), + nsize, cred, &newb); + if (error) + return (error); + bp = getblk(vp, bn, nsize, 0, 0); + bp->b_blkno = fsbtodb(fs, newb); + if (flags & B_CLRBUF) + vfs_bio_clrbuf(bp); + } + ip->i_db[bn] = dbtofsb(fs, bp->b_blkno); + ip->i_flag |= IN_CHANGE | IN_UPDATE; + *bpp = bp; + return (0); + } + /* + * Determine the number of levels of indirection. + */ + pref = 0; + if ((error = ufs_getlbns(vp, bn, indirs, &num)) != 0) + return(error); +#if DIAGNOSTIC + if (num < 1) + panic ("ext2_balloc: ufs_bmaparray returned indirect block"); +#endif + /* + * Fetch the first indirect block allocating if necessary. + */ + --num; + nb = ip->i_ib[indirs[0].in_off]; + if (nb == 0) { +#if 0 + pref = ext2_blkpref(ip, lbn, 0, (daddr_t *)0, 0); +#else + /* see the comment by ext2_blkpref. What we do here is + to pretend that it'd be good for a block holding indirect + pointers to be allocated near its predecessor in terms + of indirection, or the last direct block. + We shamelessly exploit the fact that i_ib immediately + follows i_db. + Godmar thinks it make sense to allocate i_ib[0] immediately + after i_db[11], but it's not utterly clear whether this also + applies to i_ib[1] and i_ib[0] + */ + + pref = ext2_blkpref(ip, lbn, indirs[0].in_off + + EXT2_NDIR_BLOCKS, &ip->i_db[0], 0); +#endif + if ((error = ext2_alloc(ip, lbn, pref, (int)fs->s_blocksize, + cred, &newb)) != 0) + return (error); + nb = newb; + bp = getblk(vp, indirs[1].in_lbn, fs->s_blocksize, 0, 0); + bp->b_blkno = fsbtodb(fs, newb); + vfs_bio_clrbuf(bp); + /* + * Write synchronously so that indirect blocks + * never point at garbage. + */ + if ((error = bwrite(bp)) != 0) { + ext2_blkfree(ip, nb, fs->s_blocksize); + return (error); + } + ip->i_ib[indirs[0].in_off] = newb; + ip->i_flag |= IN_CHANGE | IN_UPDATE; + } + /* + * Fetch through the indirect blocks, allocating as necessary. + */ + for (i = 1;;) { + error = bread(vp, + indirs[i].in_lbn, (int)fs->s_blocksize, NOCRED, &bp); + if (error) { + brelse(bp); + return (error); + } + bap = (daddr_t *)bp->b_data; + nb = bap[indirs[i].in_off]; + if (i == num) + break; + i += 1; + if (nb != 0) { + brelse(bp); + continue; + } + if (pref == 0) +#if 1 + /* see the comment above and by ext2_blkpref + * I think this implements Linux policy, but + * does it really make sense to allocate to + * block containing pointers together ? + * Also, will it ever succeed ? + */ + pref = ext2_blkpref(ip, lbn, indirs[i].in_off, bap, + bp->b_lblkno); +#else + pref = ext2_blkpref(ip, lbn, 0, (daddr_t *)0, 0); +#endif + if ((error = + ext2_alloc(ip, lbn, pref, (int)fs->s_blocksize, cred, &newb)) != 0) { + brelse(bp); + return (error); + } + nb = newb; + nbp = getblk(vp, indirs[i].in_lbn, fs->s_blocksize, 0, 0); + nbp->b_blkno = fsbtodb(fs, nb); + vfs_bio_clrbuf(nbp); + /* + * Write synchronously so that indirect blocks + * never point at garbage. + */ + if ((error = bwrite(nbp)) != 0) { + ext2_blkfree(ip, nb, fs->s_blocksize); + brelse(bp); + return (error); + } + bap[indirs[i - 1].in_off] = nb; + /* + * If required, write synchronously, otherwise use + * delayed write. + */ + if (flags & B_SYNC) { + bwrite(bp); + } else { + bdwrite(bp); + } + } + /* + * Get the data block, allocating if necessary. + */ + if (nb == 0) { + pref = ext2_blkpref(ip, lbn, indirs[i].in_off, &bap[0], + bp->b_lblkno); + if ((error = ext2_alloc(ip, + lbn, pref, (int)fs->s_blocksize, cred, &newb)) != 0) { + brelse(bp); + return (error); + } + nb = newb; + nbp = getblk(vp, lbn, fs->s_blocksize, 0, 0); + nbp->b_blkno = fsbtodb(fs, nb); + if (flags & B_CLRBUF) + vfs_bio_clrbuf(nbp); + bap[indirs[i].in_off] = nb; + /* + * If required, write synchronously, otherwise use + * delayed write. + */ + if (flags & B_SYNC) { + bwrite(bp); + } else { + bdwrite(bp); + } + *bpp = nbp; + return (0); + } + brelse(bp); + if (flags & B_CLRBUF) { + error = bread(vp, lbn, (int)fs->s_blocksize, NOCRED, &nbp); + if (error) { + brelse(nbp); + return (error); + } + } else { + nbp = getblk(vp, lbn, fs->s_blocksize, 0, 0); + nbp->b_blkno = fsbtodb(fs, nb); + } + *bpp = nbp; + return (0); +} diff --git a/sys/gnu/ext2fs/ext2_extern.h b/sys/gnu/ext2fs/ext2_extern.h new file mode 100644 index 000000000000..b37b3ec634ee --- /dev/null +++ b/sys/gnu/ext2fs/ext2_extern.h @@ -0,0 +1,107 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/*- + * Copyright (c) 1991, 1993, 1994 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ffs_extern.h 8.3 (Berkeley) 4/16/94 + * $FreeBSD$ + */ + +#ifndef _SYS_GNU_EXT2FS_EXT2_EXTERN_H_ +#define _SYS_GNU_EXT2FS_EXT2_EXTERN_H_ + +struct dinode; +struct ext2_inode; +struct inode; +struct mount; +struct vfsconf; +struct vnode; + +int ext2_alloc __P((struct inode *, + daddr_t, daddr_t, int, struct ucred *, daddr_t *)); +int ext2_balloc __P((struct inode *, + daddr_t, int, struct ucred *, struct buf **, int)); +int ext2_blkatoff __P((struct vnode *, off_t, char **, struct buf **)); +void ext2_blkfree __P((struct inode *, daddr_t, long)); +daddr_t ext2_blkpref __P((struct inode *, daddr_t, int, daddr_t *, daddr_t)); +int ext2_bmap __P((struct vop_bmap_args *)); +int ext2_init __P((struct vfsconf *)); +int ext2_reallocblks __P((struct vop_reallocblks_args *)); +int ext2_reclaim __P((struct vop_reclaim_args *)); +void ext2_setblock __P((struct ext2_sb_info *, u_char *, daddr_t)); +int ext2_truncate __P((struct vnode *, off_t, int, struct ucred *, struct proc *)); +int ext2_update __P((struct vnode *, int)); +int ext2_valloc __P((struct vnode *, int, struct ucred *, struct vnode **)); +int ext2_vfree __P((struct vnode *, ino_t, int)); +int ext2_lookup __P((struct vop_cachedlookup_args *)); +int ext2_readdir __P((struct vop_readdir_args *)); +void ext2_print_dinode __P((struct dinode *)); +void ext2_print_inode __P((struct inode *)); +int ext2_direnter __P((struct inode *, + struct vnode *, struct componentname *)); +int ext2_dirremove __P((struct vnode *, struct componentname *)); +int ext2_dirrewrite __P((struct inode *, + struct inode *, struct componentname *)); +int ext2_dirempty __P((struct inode *, ino_t, struct ucred *)); +int ext2_checkpath __P((struct inode *, struct inode *, struct ucred *)); +struct ext2_group_desc * get_group_desc __P((struct mount * , + unsigned int , struct buf ** )); +int ext2_group_sparse __P((int group)); +void ext2_discard_prealloc __P((struct inode *)); +int ext2_inactive __P((struct vop_inactive_args *)); +int ext2_new_block __P ((struct mount * mp, unsigned long goal, + u_int32_t * prealloc_count, + u_int32_t * prealloc_block)); +ino_t ext2_new_inode __P ((const struct inode * dir, int mode)); +unsigned long ext2_count_free __P((struct buf *map, unsigned int numchars)); +void ext2_free_blocks __P((struct mount * mp, unsigned long block, + unsigned long count)); +void ext2_free_inode __P((struct inode * inode)); +void ext2_ei2di __P((struct ext2_inode *ei, struct dinode *di)); +void ext2_di2ei __P((struct dinode *di, struct ext2_inode *ei)); +void mark_buffer_dirty __P((struct buf *bh)); + +/* + * This macro allows the ufs code to distinguish between an EXT2 and a + * non-ext2(FFS/LFS) vnode. + */ +#define IS_EXT2_VNODE(vp) (vp->v_mount->mnt_stat.f_type == MOUNT_EXT2FS) + +extern vop_t **ext2_vnodeop_p; +extern vop_t **ext2_specop_p; +extern vop_t **ext2_fifoop_p; + +#endif /* !_SYS_GNU_EXT2FS_EXT2_EXTERN_H_ */ diff --git a/sys/gnu/ext2fs/ext2_fs.h b/sys/gnu/ext2fs/ext2_fs.h new file mode 100644 index 000000000000..c449e2151b54 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_fs.h @@ -0,0 +1,572 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + * + * $FreeBSD$ + */ +/* + * linux/include/linux/ext2_fs.h + * + * Copyright (C) 1992, 1993, 1994, 1995 + * Remy Card (card@masi.ibp.fr) + * Laboratoire MASI - Institut Blaise Pascal + * Universite Pierre et Marie Curie (Paris VI) + * + * from + * + * linux/include/linux/minix_fs.h + * + * Copyright (C) 1991, 1992 Linus Torvalds + */ + +#ifndef _LINUX_EXT2_FS_H +#define _LINUX_EXT2_FS_H + +#include <sys/types.h> + +#define __u32 u_int32_t +#define u32 u_int32_t +#define __u16 u_int16_t +#define __u8 u_int8_t + +#define __s32 int32_t +#define __s16 int16_t +#define __s8 int8_t + +#define umode_t mode_t +#define loff_t off_t + +/* the Linux implementation of EXT2 stores some information about + * an inode in a ext2_inode_info structure which is part of the incore + * inode in Linux + * I decided to use the "spare" fields instead - we'll see how this + * works out + */ + +#define i_block_group i_spare[0] +#define i_next_alloc_block i_spare[1] +#define i_next_alloc_goal i_spare[2] +#define i_prealloc_block i_din.di_spare[0] +#define i_prealloc_count i_din.di_spare[1] + +/* + * The second extended filesystem constants/structures + */ + +/* + * Define EXT2FS_DEBUG to produce debug messages + */ +#undef EXT2FS_DEBUG + +/* + * Define EXT2_PREALLOCATE to preallocate data blocks for expanding files + */ +#define EXT2_PREALLOCATE +#define EXT2_DEFAULT_PREALLOC_BLOCKS 8 + +/* + * The second extended file system version + */ +#define EXT2FS_DATE "95/08/09" +#define EXT2FS_VERSION "0.5b" + +/* + * Debug code + */ +#ifdef EXT2FS_DEBUG +# define ext2_debug(f, a...) { \ + printf ("EXT2-fs DEBUG (%s, %d): %s:", \ + __FILE__, __LINE__, __FUNCTION__); \ + printf (f, ## a); \ + } +#else +# define ext2_debug(f, a...) /**/ +#endif + +/* + * Special inodes numbers + */ +#define EXT2_BAD_INO 1 /* Bad blocks inode */ +#define EXT2_ROOT_INO 2 /* Root inode */ +#define EXT2_ACL_IDX_INO 3 /* ACL inode */ +#define EXT2_ACL_DATA_INO 4 /* ACL inode */ +#define EXT2_BOOT_LOADER_INO 5 /* Boot loader inode */ +#define EXT2_UNDEL_DIR_INO 6 /* Undelete directory inode */ + +/* First non-reserved inode for old ext2 filesystems */ +#define EXT2_GOOD_OLD_FIRST_INO 11 + +/* + * The second extended file system magic number + */ +#define EXT2_SUPER_MAGIC 0xEF53 + +/* + * Maximal count of links to a file + */ +#define EXT2_LINK_MAX 32000 + +/* + * Note: under FreeBSD, the "user" versions of the following macros are + * used (and must be used) in most cases, because ((s)->u.ext2_sb.s_es is + * not accessible. This depends on __KERNEL__ not being defined for + * kernel builds under FreeBSD. + */ + +/* + * Macro-instructions used to manage several block sizes + */ +#define EXT2_MIN_BLOCK_SIZE 1024 +#define EXT2_MAX_BLOCK_SIZE 4096 +#define EXT2_MIN_BLOCK_LOG_SIZE 10 +#if defined(__KERNEL__) || (defined(__FreeBSD__) && defined(_KERNEL)) +# define EXT2_BLOCK_SIZE(s) ((s)->s_blocksize) +#else +# define EXT2_BLOCK_SIZE(s) (EXT2_MIN_BLOCK_SIZE << (s)->s_log_block_size) +#endif +#define EXT2_ACLE_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof (struct ext2_acl_entry)) +#define EXT2_ADDR_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof (__u32)) +#ifdef __KERNEL__ +# define EXT2_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) +#else +# define EXT2_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) +#endif +#ifdef notyet +#ifdef __KERNEL__ +#define EXT2_ADDR_PER_BLOCK_BITS(s) ((s)->u.ext2_sb.s_addr_per_block_bits) +#define EXT2_INODE_SIZE(s) ((s)->u.ext2_sb.s_inode_size) +#define EXT2_FIRST_INO(s) ((s)->u.ext2_sb.s_first_ino) +#else +#define EXT2_INODE_SIZE(s) (((s)->s_rev_level == EXT2_GOOD_OLD_REV) ? \ + EXT2_GOOD_OLD_INODE_SIZE : \ + (s)->s_inode_size) +#define EXT2_FIRST_INO(s) (((s)->s_rev_level == EXT2_GOOD_OLD_REV) ? \ + EXT2_GOOD_OLD_FIRST_INO : \ + (s)->s_first_ino) +#endif +#else /* !notyet */ +#define EXT2_INODES_PER_BLOCK(s) ((s)->s_inodes_per_block) +/* Should be sizeof(struct ext2_inode): */ +#define EXT2_INODE_SIZE 128 +#define EXT2_FIRST_INO 11 +#endif /* notyet */ + +/* + * Macro-instructions used to manage fragments + */ +#define EXT2_MIN_FRAG_SIZE 1024 +#define EXT2_MAX_FRAG_SIZE 4096 +#define EXT2_MIN_FRAG_LOG_SIZE 10 +#ifdef __KERNEL__ +# define EXT2_FRAG_SIZE(s) ((s)->u.ext2_sb.s_frag_size) +# define EXT2_FRAGS_PER_BLOCK(s) ((s)->u.ext2_sb.s_frags_per_block) +#else +# if defined(_KERNEL) && defined(__FreeBSD__) +# define EXT2_FRAG_SIZE(s) ((s)->s_frag_size) +# else +# define EXT2_FRAG_SIZE(s) (EXT2_MIN_FRAG_SIZE << (s)->s_log_frag_size) +# endif +# define EXT2_FRAGS_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / EXT2_FRAG_SIZE(s)) +#endif + +/* + * ACL structures + */ +struct ext2_acl_header /* Header of Access Control Lists */ +{ + __u32 aclh_size; + __u32 aclh_file_count; + __u32 aclh_acle_count; + __u32 aclh_first_acle; +}; + +struct ext2_acl_entry /* Access Control List Entry */ +{ + __u32 acle_size; + __u16 acle_perms; /* Access permissions */ + __u16 acle_type; /* Type of entry */ + __u16 acle_tag; /* User or group identity */ + __u16 acle_pad1; + __u32 acle_next; /* Pointer on next entry for the */ + /* same inode or on next free entry */ +}; + +/* + * Structure of a blocks group descriptor + */ +struct ext2_group_desc +{ + __u32 bg_block_bitmap; /* Blocks bitmap block */ + __u32 bg_inode_bitmap; /* Inodes bitmap block */ + __u32 bg_inode_table; /* Inodes table block */ + __u16 bg_free_blocks_count; /* Free blocks count */ + __u16 bg_free_inodes_count; /* Free inodes count */ + __u16 bg_used_dirs_count; /* Directories count */ + __u16 bg_pad; + __u32 bg_reserved[3]; +}; + +/* + * Macro-instructions used to manage group descriptors + */ +#ifdef __KERNEL__ +# define EXT2_BLOCKS_PER_GROUP(s) ((s)->u.ext2_sb.s_blocks_per_group) +# define EXT2_DESC_PER_BLOCK(s) ((s)->u.ext2_sb.s_desc_per_block) +# define EXT2_INODES_PER_GROUP(s) ((s)->u.ext2_sb.s_inodes_per_group) +# define EXT2_DESC_PER_BLOCK_BITS(s) ((s)->u.ext2_sb.s_desc_per_block_bits) +#else +# define EXT2_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) +# define EXT2_DESC_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof (struct ext2_group_desc)) +# define EXT2_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) +#endif + +/* + * Constants relative to the data blocks + */ +#define EXT2_NDIR_BLOCKS 12 +#define EXT2_IND_BLOCK EXT2_NDIR_BLOCKS +#define EXT2_DIND_BLOCK (EXT2_IND_BLOCK + 1) +#define EXT2_TIND_BLOCK (EXT2_DIND_BLOCK + 1) +#define EXT2_N_BLOCKS (EXT2_TIND_BLOCK + 1) +#define EXT2_MAXSYMLINKLEN (EXT2_N_BLOCKS * sizeof (__u32)) + +/* + * Inode flags + */ +#define EXT2_SECRM_FL 0x00000001 /* Secure deletion */ +#define EXT2_UNRM_FL 0x00000002 /* Undelete */ +#define EXT2_COMPR_FL 0x00000004 /* Compress file */ +#define EXT2_SYNC_FL 0x00000008 /* Synchronous updates */ +#define EXT2_IMMUTABLE_FL 0x00000010 /* Immutable file */ +#define EXT2_APPEND_FL 0x00000020 /* writes to file may only append */ +#define EXT2_NODUMP_FL 0x00000040 /* do not dump file */ +#define EXT2_NOATIME_FL 0x00000080 /* do not update atime */ +/* Reserved for compression usage... */ +#define EXT2_DIRTY_FL 0x00000100 +#define EXT2_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ +#define EXT2_NOCOMP_FL 0x00000400 /* Don't compress */ +#define EXT2_ECOMPR_FL 0x00000800 /* Compression error */ +/* End compression flags --- maybe not all used */ +#define EXT2_BTREE_FL 0x00001000 /* btree format dir */ +#define EXT2_RESERVED_FL 0x80000000 /* reserved for ext2 lib */ + +#define EXT2_FL_USER_VISIBLE 0x00001FFF /* User visible flags */ +#define EXT2_FL_USER_MODIFIABLE 0x000000FF /* User modifiable flags */ + +/* + * ioctl commands + */ +#define EXT2_IOC_GETFLAGS _IOR('f', 1, long) +#define EXT2_IOC_SETFLAGS _IOW('f', 2, long) +#define EXT2_IOC_GETVERSION _IOR('v', 1, long) +#define EXT2_IOC_SETVERSION _IOW('v', 2, long) + +/* + * Only declare `struct ext2_inode' if <ufs/ufs/inode.h> hasn't made things + * difficult by #defining i_mode and other struct members. The details of + * the struct are only needed in ext2_inode_cnv.c where the ext2fs on-disk + * inode is converted to a ufs in-core inode. + */ +#ifndef i_mode + +/* + * Structure of an inode on the disk + */ +struct ext2_inode { + __u16 i_mode; /* File mode */ + __u16 i_uid; /* Owner Uid */ + __u32 i_size; /* Size in bytes */ + __u32 i_atime; /* Access time */ + __u32 i_ctime; /* Creation time */ + __u32 i_mtime; /* Modification time */ + __u32 i_dtime; /* Deletion Time */ + __u16 i_gid; /* Group Id */ + __u16 i_links_count; /* Links count */ + __u32 i_blocks; /* Blocks count */ + __u32 i_flags; /* File flags */ + union { + struct { + __u32 l_i_reserved1; + } linux1; + struct { + __u32 h_i_translator; + } hurd1; + struct { + __u32 m_i_reserved1; + } masix1; + } osd1; /* OS dependent 1 */ + __u32 i_block[EXT2_N_BLOCKS];/* Pointers to blocks */ + __u32 i_generation; /* File version (for NFS) */ + __u32 i_file_acl; /* File ACL */ + __u32 i_dir_acl; /* Directory ACL */ + __u32 i_faddr; /* Fragment address */ + union { + struct { + __u8 l_i_frag; /* Fragment number */ + __u8 l_i_fsize; /* Fragment size */ + __u16 i_pad1; + __u32 l_i_reserved2[2]; + } linux2; + struct { + __u8 h_i_frag; /* Fragment number */ + __u8 h_i_fsize; /* Fragment size */ + __u16 h_i_mode_high; + __u16 h_i_uid_high; + __u16 h_i_gid_high; + __u32 h_i_author; + } hurd2; + struct { + __u8 m_i_frag; /* Fragment number */ + __u8 m_i_fsize; /* Fragment size */ + __u16 m_pad1; + __u32 m_i_reserved2[2]; + } masix2; + } osd2; /* OS dependent 2 */ +}; + +#define i_size_high i_dir_acl + +#if defined(__KERNEL__) || defined(__linux__) +#define i_reserved1 osd1.linux1.l_i_reserved1 +#define i_frag osd2.linux2.l_i_frag +#define i_fsize osd2.linux2.l_i_fsize +#define i_reserved2 osd2.linux2.l_i_reserved2 +#endif + +#ifdef __hurd__ +#define i_translator osd1.hurd1.h_i_translator +#define i_frag osd2.hurd2.h_i_frag; +#define i_fsize osd2.hurd2.h_i_fsize; +#define i_uid_high osd2.hurd2.h_i_uid_high +#define i_gid_high osd2.hurd2.h_i_gid_high +#define i_author osd2.hurd2.h_i_author +#endif + +#ifdef __masix__ +#define i_reserved1 osd1.masix1.m_i_reserved1 +#define i_frag osd2.masix2.m_i_frag +#define i_fsize osd2.masix2.m_i_fsize +#define i_reserved2 osd2.masix2.m_i_reserved2 +#endif + +#endif /* i_mode */ + +/* + * File system states + */ +#define EXT2_VALID_FS 0x0001 /* Unmounted cleanly */ +#define EXT2_ERROR_FS 0x0002 /* Errors detected */ + +/* + * Mount flags + */ +#define EXT2_MOUNT_CHECK_NORMAL 0x0001 /* Do some more checks */ +#define EXT2_MOUNT_CHECK_STRICT 0x0002 /* Do again more checks */ +#define EXT2_MOUNT_CHECK (EXT2_MOUNT_CHECK_NORMAL | \ + EXT2_MOUNT_CHECK_STRICT) +#define EXT2_MOUNT_GRPID 0x0004 /* Create files with directory's group */ +#define EXT2_MOUNT_DEBUG 0x0008 /* Some debugging messages */ +#define EXT2_MOUNT_ERRORS_CONT 0x0010 /* Continue on errors */ +#define EXT2_MOUNT_ERRORS_RO 0x0020 /* Remount fs ro on errors */ +#define EXT2_MOUNT_ERRORS_PANIC 0x0040 /* Panic on errors */ +#define EXT2_MOUNT_MINIX_DF 0x0080 /* Mimics the Minix statfs */ + +#define clear_opt(o, opt) o &= ~EXT2_MOUNT_##opt +#define set_opt(o, opt) o |= EXT2_MOUNT_##opt +#define test_opt(sb, opt) ((sb)->u.ext2_sb.s_mount_opt & \ + EXT2_MOUNT_##opt) +/* + * Maximal mount counts between two filesystem checks + */ +#define EXT2_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ +#define EXT2_DFL_CHECKINTERVAL 0 /* Don't use interval check */ + +/* + * Behaviour when detecting errors + */ +#define EXT2_ERRORS_CONTINUE 1 /* Continue execution */ +#define EXT2_ERRORS_RO 2 /* Remount fs read-only */ +#define EXT2_ERRORS_PANIC 3 /* Panic */ +#define EXT2_ERRORS_DEFAULT EXT2_ERRORS_CONTINUE + +/* + * Structure of the super block + */ +struct ext2_super_block { + __u32 s_inodes_count; /* Inodes count */ + __u32 s_blocks_count; /* Blocks count */ + __u32 s_r_blocks_count; /* Reserved blocks count */ + __u32 s_free_blocks_count; /* Free blocks count */ + __u32 s_free_inodes_count; /* Free inodes count */ + __u32 s_first_data_block; /* First Data Block */ + __u32 s_log_block_size; /* Block size */ + __s32 s_log_frag_size; /* Fragment size */ + __u32 s_blocks_per_group; /* # Blocks per group */ + __u32 s_frags_per_group; /* # Fragments per group */ + __u32 s_inodes_per_group; /* # Inodes per group */ + __u32 s_mtime; /* Mount time */ + __u32 s_wtime; /* Write time */ + __u16 s_mnt_count; /* Mount count */ + __s16 s_max_mnt_count; /* Maximal mount count */ + __u16 s_magic; /* Magic signature */ + __u16 s_state; /* File system state */ + __u16 s_errors; /* Behaviour when detecting errors */ + __u16 s_minor_rev_level; /* minor revision level */ + __u32 s_lastcheck; /* time of last check */ + __u32 s_checkinterval; /* max. time between checks */ + __u32 s_creator_os; /* OS */ + __u32 s_rev_level; /* Revision level */ + __u16 s_def_resuid; /* Default uid for reserved blocks */ + __u16 s_def_resgid; /* Default gid for reserved blocks */ + /* + * These fields are for EXT2_DYNAMIC_REV superblocks only. + * + * Note: the difference between the compatible feature set and + * the incompatible feature set is that if there is a bit set + * in the incompatible feature set that the kernel doesn't + * know about, it should refuse to mount the filesystem. + * + * e2fsck's requirements are more strict; if it doesn't know + * about a feature in either the compatible or incompatible + * feature set, it must abort and not try to meddle with + * things it doesn't understand... + */ + __u32 s_first_ino; /* First non-reserved inode */ + __u16 s_inode_size; /* size of inode structure */ + __u16 s_block_group_nr; /* block group # of this superblock */ + __u32 s_feature_compat; /* compatible feature set */ + __u32 s_feature_incompat; /* incompatible feature set */ + __u32 s_feature_ro_compat; /* readonly-compatible feature set */ + __u8 s_uuid[16]; /* 128-bit uuid for volume */ + char s_volume_name[16]; /* volume name */ + char s_last_mounted[64]; /* directory where last mounted */ + __u32 s_algorithm_usage_bitmap; /* For compression */ + /* + * Performance hints. Directory preallocation should only + * happen if the EXT2_COMPAT_PREALLOC flag is on. + */ + __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ + __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ + __u16 s_padding1; + __u32 s_reserved[204]; /* Padding to the end of the block */ +}; + +#ifdef __KERNEL__ +#define EXT2_SB(sb) (&((sb)->u.ext2_sb)) +#else +/* Assume that user mode programs are passing in an ext2fs superblock, not + * a kernel struct super_block. This will allow us to call the feature-test + * macros from user land. */ +#define EXT2_SB(sb) (sb) +#endif + +/* + * Codes for operating systems + */ +#define EXT2_OS_LINUX 0 +#define EXT2_OS_HURD 1 +#define EXT2_OS_MASIX 2 +#define EXT2_OS_FREEBSD 3 +#define EXT2_OS_LITES 4 + +/* + * Revision levels + */ +#define EXT2_GOOD_OLD_REV 0 /* The good old (original) format */ +#define EXT2_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ + +#define EXT2_CURRENT_REV EXT2_GOOD_OLD_REV +#define EXT2_MAX_SUPP_REV EXT2_DYNAMIC_REV + +#define EXT2_GOOD_OLD_INODE_SIZE 128 + +/* + * Feature set definitions + */ + +#define EXT2_HAS_COMPAT_FEATURE(sb,mask) \ + ( EXT2_SB(sb)->s_feature_compat & (mask) ) +#define EXT2_HAS_RO_COMPAT_FEATURE(sb,mask) \ + ( EXT2_SB(sb)->s_feature_ro_compat & (mask) ) +#define EXT2_HAS_INCOMPAT_FEATURE(sb,mask) \ + ( EXT2_SB(sb)->s_feature_incompat & (mask) ) + +#define EXT2_FEATURE_COMPAT_DIR_PREALLOC 0x0001 + +#define EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 +#define EXT2_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 +#define EXT2_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 + +#define EXT2_FEATURE_INCOMPAT_COMPRESSION 0x0001 +#define EXT2_FEATURE_INCOMPAT_FILETYPE 0x0002 + +#define EXT2_FEATURE_COMPAT_SUPP 0 +#define EXT2_FEATURE_INCOMPAT_SUPP EXT2_FEATURE_INCOMPAT_FILETYPE +#ifdef notyet +#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER| \ + EXT2_FEATURE_RO_COMPAT_LARGE_FILE| \ + EXT2_FEATURE_RO_COMPAT_BTREE_DIR) +#else +#define EXT2_FEATURE_RO_COMPAT_SUPP EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER +#endif + +/* + * Default values for user and/or group using reserved blocks + */ +#define EXT2_DEF_RESUID 0 +#define EXT2_DEF_RESGID 0 + +/* + * Structure of a directory entry + */ +#define EXT2_NAME_LEN 255 + +struct ext2_dir_entry { + __u32 inode; /* Inode number */ + __u16 rec_len; /* Directory entry length */ + __u16 name_len; /* Name length */ + char name[EXT2_NAME_LEN]; /* File name */ +}; + +/* + * The new version of the directory entry. Since EXT2 structures are + * stored in intel byte order, and the name_len field could never be + * bigger than 255 chars, it's safe to reclaim the extra byte for the + * file_type field. + */ +struct ext2_dir_entry_2 { + __u32 inode; /* Inode number */ + __u16 rec_len; /* Directory entry length */ + __u8 name_len; /* Name length */ + __u8 file_type; + char name[EXT2_NAME_LEN]; /* File name */ +}; + +/* + * Ext2 directory file types. Only the low 3 bits are used. The + * other bits are reserved for now. + */ +#define EXT2_FT_UNKNOWN 0 +#define EXT2_FT_REG_FILE 1 +#define EXT2_FT_DIR 2 +#define EXT2_FT_CHRDEV 3 +#define EXT2_FT_BLKDEV 4 +#define EXT2_FT_FIFO 5 +#define EXT2_FT_SOCK 6 +#define EXT2_FT_SYMLINK 7 + +#define EXT2_FT_MAX 8 + +/* + * EXT2_DIR_PAD defines the directory entries boundaries + * + * NOTE: It must be a multiple of 4 + */ +#define EXT2_DIR_PAD 4 +#define EXT2_DIR_ROUND (EXT2_DIR_PAD - 1) +#define EXT2_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT2_DIR_ROUND) & \ + ~EXT2_DIR_ROUND) + +#endif /* _LINUX_EXT2_FS_H */ diff --git a/sys/gnu/ext2fs/ext2_fs_sb.h b/sys/gnu/ext2fs/ext2_fs_sb.h new file mode 100644 index 000000000000..85f71bf39c7e --- /dev/null +++ b/sys/gnu/ext2fs/ext2_fs_sb.h @@ -0,0 +1,81 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * linux/include/linux/ext2_fs_sb.h + * + * Copyright (C) 1992, 1993, 1994, 1995 + * Remy Card (card@masi.ibp.fr) + * Laboratoire MASI - Institut Blaise Pascal + * Universite Pierre et Marie Curie (Paris VI) + * + * from + * + * linux/include/linux/minix_fs_sb.h + * + * Copyright (C) 1991, 1992 Linus Torvalds + */ + +#ifndef _LINUX_EXT2_FS_SB +#define _LINUX_EXT2_FS_SB + +/* + * The following is not needed anymore since the descriptors buffer + * heads are now dynamically allocated + */ +/* #define EXT2_MAX_GROUP_DESC 8 */ + +#define EXT2_MAX_GROUP_LOADED 8 + +#define buffer_head buf +#define MAXMNTLEN 512 + +/* + * second extended-fs super-block data in memory + */ +struct ext2_sb_info { + unsigned long s_frag_size; /* Size of a fragment in bytes */ + unsigned long s_frags_per_block;/* Number of fragments per block */ + unsigned long s_inodes_per_block;/* Number of inodes per block */ + unsigned long s_frags_per_group;/* Number of fragments in a group */ + unsigned long s_blocks_per_group;/* Number of blocks in a group */ + unsigned long s_inodes_per_group;/* Number of inodes in a group */ + unsigned long s_itb_per_group; /* Number of inode table blocks per group */ + unsigned long s_db_per_group; /* Number of descriptor blocks per group */ + unsigned long s_desc_per_block; /* Number of group descriptors per block */ + unsigned long s_groups_count; /* Number of groups in the fs */ + struct buffer_head * s_sbh; /* Buffer containing the super block */ + struct ext2_super_block * s_es; /* Pointer to the super block in the buffer */ + struct buffer_head ** s_group_desc; + unsigned short s_loaded_inode_bitmaps; + unsigned short s_loaded_block_bitmaps; + unsigned long s_inode_bitmap_number[EXT2_MAX_GROUP_LOADED]; + struct buffer_head * s_inode_bitmap[EXT2_MAX_GROUP_LOADED]; + unsigned long s_block_bitmap_number[EXT2_MAX_GROUP_LOADED]; + struct buffer_head * s_block_bitmap[EXT2_MAX_GROUP_LOADED]; + int s_rename_lock; + unsigned long s_mount_opt; + unsigned short s_resuid; + unsigned short s_resgid; + unsigned short s_mount_state; + /* + stuff that FFS keeps in its super block or that linux + has in its non-ext2 specific super block and which is + generally considered useful + */ + unsigned long s_blocksize; + unsigned long s_blocksize_bits; + unsigned int s_bshift; /* = log2(s_blocksize) */ + quad_t s_qbmask; /* = s_blocksize - 1 */ + unsigned int s_fsbtodb; /* shift to get disk block */ + char s_rd_only; /* read-only */ + char s_dirt; /* fs modified flag */ + char s_wasvalid; /* valid at mount time */ + + char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ +}; + +#endif /* _LINUX_EXT2_FS_SB */ diff --git a/sys/gnu/ext2fs/ext2_inode.c b/sys/gnu/ext2fs/ext2_inode.c new file mode 100644 index 000000000000..34a78fb47815 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_inode.c @@ -0,0 +1,494 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ext2_inode.c 8.5 (Berkeley) 12/30/93 + * $FreeBSD$ + */ + +#include "opt_quota.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/mount.h> +#include <sys/buf.h> +#include <sys/vnode.h> +#include <sys/malloc.h> + +#include <vm/vm.h> +#include <vm/vm_extern.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufsmount.h> +#include <ufs/ufs/ufs_extern.h> + +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> +#include <gnu/ext2fs/ext2_extern.h> + +static int ext2_indirtrunc __P((struct inode *, daddr_t, daddr_t, daddr_t, int, + long *)); + +int +ext2_init(struct vfsconf *vfsp) +{ + return (ufs_init(vfsp)); +} + +/* + * Update the access, modified, and inode change times as specified by the + * IN_ACCESS, IN_UPDATE, and IN_CHANGE flags respectively. Write the inode + * to disk if the IN_MODIFIED flag is set (it may be set initially, or by + * the timestamp update). The IN_LAZYMOD flag is set to force a write + * later if not now. If we write now, then clear both IN_MODIFIED and + * IN_LAZYMOD to reflect the presumably successful write, and if waitfor is + * set, then wait for the write to complete. + */ +int +ext2_update(vp, waitfor) + struct vnode *vp; + int waitfor; +{ + register struct ext2_sb_info *fs; + struct buf *bp; + struct inode *ip; + int error; + + ufs_itimes(vp); + ip = VTOI(vp); + if ((ip->i_flag & IN_MODIFIED) == 0) + return (0); + ip->i_flag &= ~(IN_LAZYMOD | IN_MODIFIED); + if (vp->v_mount->mnt_flag & MNT_RDONLY) + return (0); + fs = ip->i_e2fs; + if ((error = bread(ip->i_devvp, + fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), + (int)fs->s_blocksize, NOCRED, &bp)) != 0) { + brelse(bp); + return (error); + } + ext2_di2ei( &ip->i_din, (struct ext2_inode *) ((char *)bp->b_data + EXT2_INODE_SIZE * + ino_to_fsbo(fs, ip->i_number))); +/* + if (waitfor && (vp->v_mount->mnt_flag & MNT_ASYNC) == 0) + return (bwrite(bp)); + else { +*/ + bdwrite(bp); + return (0); +/* + } +*/ +} + +#define SINGLE 0 /* index of single indirect block */ +#define DOUBLE 1 /* index of double indirect block */ +#define TRIPLE 2 /* index of triple indirect block */ +/* + * Truncate the inode oip to at most length size, freeing the + * disk blocks. + */ +int +ext2_truncate(vp, length, flags, cred, p) + struct vnode *vp; + off_t length; + int flags; + struct ucred *cred; + struct proc *p; +{ + register struct vnode *ovp = vp; + register daddr_t lastblock; + register struct inode *oip; + daddr_t bn, lbn, lastiblock[NIADDR], indir_lbn[NIADDR]; + daddr_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; + register struct ext2_sb_info *fs; + struct buf *bp; + int offset, size, level; + long count, nblocks, blocksreleased = 0; + register int i; + int aflags, error, allerror; + off_t osize; +/* +printf("ext2_truncate called %d to %d\n", VTOI(ovp)->i_number, length); +*/ /* + * negative file sizes will totally break the code below and + * are not meaningful anyways. + */ + if (length < 0) + return EFBIG; + + oip = VTOI(ovp); + if (ovp->v_type == VLNK && + oip->i_size < ovp->v_mount->mnt_maxsymlinklen) { +#if DIAGNOSTIC + if (length != 0) + panic("ext2_truncate: partial truncate of symlink"); +#endif + bzero((char *)&oip->i_shortlink, (u_int)oip->i_size); + oip->i_size = 0; + oip->i_flag |= IN_CHANGE | IN_UPDATE; + return (UFS_UPDATE(ovp, 1)); + } + if (oip->i_size == length) { + oip->i_flag |= IN_CHANGE | IN_UPDATE; + return (UFS_UPDATE(ovp, 0)); + } +#if QUOTA + if ((error = getinoquota(oip)) != 0) + return (error); +#endif + fs = oip->i_e2fs; + osize = oip->i_size; + ext2_discard_prealloc(oip); + /* + * Lengthen the size of the file. We must ensure that the + * last byte of the file is allocated. Since the smallest + * value of oszie is 0, length will be at least 1. + */ + if (osize < length) { + offset = blkoff(fs, length - 1); + lbn = lblkno(fs, length - 1); + aflags = B_CLRBUF; + if (flags & IO_SYNC) + aflags |= B_SYNC; + vnode_pager_setsize(ovp, length); + if ((error = ext2_balloc(oip, lbn, offset + 1, cred, &bp, + aflags)) != 0) + return (error); + oip->i_size = length; + if (aflags & IO_SYNC) + bwrite(bp); + else + bawrite(bp); + oip->i_flag |= IN_CHANGE | IN_UPDATE; + return (UFS_UPDATE(ovp, 1)); + } + /* + * Shorten the size of the file. If the file is not being + * truncated to a block boundry, the contents of the + * partial block following the end of the file must be + * zero'ed in case it ever become accessable again because + * of subsequent file growth. + */ + /* I don't understand the comment above */ + offset = blkoff(fs, length); + if (offset == 0) { + oip->i_size = length; + } else { + lbn = lblkno(fs, length); + aflags = B_CLRBUF; + if (flags & IO_SYNC) + aflags |= B_SYNC; + if ((error = ext2_balloc(oip, lbn, offset, cred, &bp, + aflags)) != 0) + return (error); + oip->i_size = length; + size = blksize(fs, oip, lbn); + bzero((char *)bp->b_data + offset, (u_int)(size - offset)); + allocbuf(bp, size); + if (aflags & IO_SYNC) + bwrite(bp); + else + bawrite(bp); + } + /* + * Calculate index into inode's block list of + * last direct and indirect blocks (if any) + * which we want to keep. Lastblock is -1 when + * the file is truncated to 0. + */ + lastblock = lblkno(fs, length + fs->s_blocksize - 1) - 1; + lastiblock[SINGLE] = lastblock - NDADDR; + lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); + lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); + nblocks = btodb(fs->s_blocksize); + /* + * Update file and block pointers on disk before we start freeing + * blocks. If we crash before free'ing blocks below, the blocks + * will be returned to the free list. lastiblock values are also + * normalized to -1 for calls to ext2_indirtrunc below. + */ + bcopy((caddr_t)&oip->i_db[0], (caddr_t)oldblks, sizeof oldblks); + for (level = TRIPLE; level >= SINGLE; level--) + if (lastiblock[level] < 0) { + oip->i_ib[level] = 0; + lastiblock[level] = -1; + } + for (i = NDADDR - 1; i > lastblock; i--) + oip->i_db[i] = 0; + oip->i_flag |= IN_CHANGE | IN_UPDATE; + allerror = UFS_UPDATE(ovp, 1); + + /* + * Having written the new inode to disk, save its new configuration + * and put back the old block pointers long enough to process them. + * Note that we save the new block configuration so we can check it + * when we are done. + */ + bcopy((caddr_t)&oip->i_db[0], (caddr_t)newblks, sizeof newblks); + bcopy((caddr_t)oldblks, (caddr_t)&oip->i_db[0], sizeof oldblks); + oip->i_size = osize; + error = vtruncbuf(ovp, cred, p, length, (int)fs->s_blocksize); + if (error && (allerror == 0)) + allerror = error; + + /* + * Indirect blocks first. + */ + indir_lbn[SINGLE] = -NDADDR; + indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) - 1; + indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; + for (level = TRIPLE; level >= SINGLE; level--) { + bn = oip->i_ib[level]; + if (bn != 0) { + error = ext2_indirtrunc(oip, indir_lbn[level], + fsbtodb(fs, bn), lastiblock[level], level, &count); + if (error) + allerror = error; + blocksreleased += count; + if (lastiblock[level] < 0) { + oip->i_ib[level] = 0; + ext2_blkfree(oip, bn, fs->s_frag_size); + blocksreleased += nblocks; + } + } + if (lastiblock[level] >= 0) + goto done; + } + + /* + * All whole direct blocks or frags. + */ + for (i = NDADDR - 1; i > lastblock; i--) { + register long bsize; + + bn = oip->i_db[i]; + if (bn == 0) + continue; + oip->i_db[i] = 0; + bsize = blksize(fs, oip, i); + ext2_blkfree(oip, bn, bsize); + blocksreleased += btodb(bsize); + } + if (lastblock < 0) + goto done; + + /* + * Finally, look for a change in size of the + * last direct block; release any frags. + */ + bn = oip->i_db[lastblock]; + if (bn != 0) { + long oldspace, newspace; + + /* + * Calculate amount of space we're giving + * back as old block size minus new block size. + */ + oldspace = blksize(fs, oip, lastblock); + oip->i_size = length; + newspace = blksize(fs, oip, lastblock); + if (newspace == 0) + panic("itrunc: newspace"); + if (oldspace - newspace > 0) { + /* + * Block number of space to be free'd is + * the old block # plus the number of frags + * required for the storage we're keeping. + */ + bn += numfrags(fs, newspace); + ext2_blkfree(oip, bn, oldspace - newspace); + blocksreleased += btodb(oldspace - newspace); + } + } +done: +#if DIAGNOSTIC + for (level = SINGLE; level <= TRIPLE; level++) + if (newblks[NDADDR + level] != oip->i_ib[level]) + panic("itrunc1"); + for (i = 0; i < NDADDR; i++) + if (newblks[i] != oip->i_db[i]) + panic("itrunc2"); + if (length == 0 && (!TAILQ_EMPTY(&ovp->v_dirtyblkhd) || + !TAILQ_EMPTY(&ovp->v_cleanblkhd))) + panic("itrunc3"); +#endif /* DIAGNOSTIC */ + /* + * Put back the real size. + */ + oip->i_size = length; + oip->i_blocks -= blocksreleased; + if (oip->i_blocks < 0) /* sanity */ + oip->i_blocks = 0; + oip->i_flag |= IN_CHANGE; + vnode_pager_setsize(ovp, length); +#if QUOTA + (void) chkdq(oip, -blocksreleased, NOCRED, 0); +#endif + return (allerror); +} + +/* + * Release blocks associated with the inode ip and stored in the indirect + * block bn. Blocks are free'd in LIFO order up to (but not including) + * lastbn. If level is greater than SINGLE, the block is an indirect block + * and recursive calls to indirtrunc must be used to cleanse other indirect + * blocks. + * + * NB: triple indirect blocks are untested. + */ + +static int +ext2_indirtrunc(ip, lbn, dbn, lastbn, level, countp) + register struct inode *ip; + daddr_t lbn, lastbn; + daddr_t dbn; + int level; + long *countp; +{ + register int i; + struct buf *bp; + register struct ext2_sb_info *fs = ip->i_e2fs; + register daddr_t *bap; + struct vnode *vp; + daddr_t *copy, nb, nlbn, last; + long blkcount, factor; + int nblocks, blocksreleased = 0; + int error = 0, allerror = 0; + + /* + * Calculate index in current block of last + * block to be kept. -1 indicates the entire + * block so we need not calculate the index. + */ + factor = 1; + for (i = SINGLE; i < level; i++) + factor *= NINDIR(fs); + last = lastbn; + if (lastbn > 0) + last /= factor; + nblocks = btodb(fs->s_blocksize); + /* + * Get buffer of block pointers, zero those entries corresponding + * to blocks to be free'd, and update on disk copy first. Since + * double(triple) indirect before single(double) indirect, calls + * to bmap on these blocks will fail. However, we already have + * the on disk address, so we have to set the b_blkno field + * explicitly instead of letting bread do everything for us. + */ + vp = ITOV(ip); + bp = getblk(vp, lbn, (int)fs->s_blocksize, 0, 0); + if (bp->b_flags & (B_DONE | B_DELWRI)) { + } else { + bp->b_flags |= B_READ; + if (bp->b_bcount > bp->b_bufsize) + panic("ext2_indirtrunc: bad buffer size"); + bp->b_blkno = dbn; + vfs_busy_pages(bp, 0); + VOP_STRATEGY(vp, bp); + error = biowait(bp); + } + if (error) { + brelse(bp); + *countp = 0; + return (error); + } + + bap = (daddr_t *)bp->b_data; + MALLOC(copy, daddr_t *, fs->s_blocksize, M_TEMP, M_WAITOK); + bcopy((caddr_t)bap, (caddr_t)copy, (u_int)fs->s_blocksize); + bzero((caddr_t)&bap[last + 1], + (u_int)(NINDIR(fs) - (last + 1)) * sizeof (daddr_t)); + if (last == -1) + bp->b_flags |= B_INVAL; + error = bwrite(bp); + if (error) + allerror = error; + bap = copy; + + /* + * Recursively free totally unused blocks. + */ + for (i = NINDIR(fs) - 1, nlbn = lbn + 1 - i * factor; i > last; + i--, nlbn += factor) { + nb = bap[i]; + if (nb == 0) + continue; + if (level > SINGLE) { + if ((error = ext2_indirtrunc(ip, nlbn, + fsbtodb(fs, nb), (daddr_t)-1, level - 1, &blkcount)) != 0) + allerror = error; + blocksreleased += blkcount; + } + ext2_blkfree(ip, nb, fs->s_blocksize); + blocksreleased += nblocks; + } + + /* + * Recursively free last partial block. + */ + if (level > SINGLE && lastbn >= 0) { + last = lastbn % factor; + nb = bap[i]; + if (nb != 0) { + if ((error = ext2_indirtrunc(ip, nlbn, fsbtodb(fs, nb), + last, level - 1, &blkcount)) != 0) + allerror = error; + blocksreleased += blkcount; + } + } + FREE(copy, M_TEMP); + *countp = blocksreleased; + return (allerror); +} + +/* + * discard preallocated blocks + */ +int +ext2_inactive(ap) + struct vop_inactive_args /* { + struct vnode *a_vp; + } */ *ap; +{ + ext2_discard_prealloc(VTOI(ap->a_vp)); + return ufs_inactive(ap); +} + diff --git a/sys/gnu/ext2fs/ext2_inode_cnv.c b/sys/gnu/ext2fs/ext2_inode_cnv.c new file mode 100644 index 000000000000..3e8e0ff73445 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_inode_cnv.c @@ -0,0 +1,160 @@ +/* + * Copyright (c) 1995 The University of Utah and + * the Computer Systems Laboratory at the University of Utah (CSL). + * All rights reserved. + * + * Permission to use, copy, modify and distribute this software is hereby + * granted provided that (1) source code retains these copyright, permission, + * and disclaimer notices, and (2) redistributions including binaries + * reproduce the notices in supporting documentation, and (3) all advertising + * materials mentioning features or use of this software display the following + * acknowledgement: ``This product includes software developed by the + * Computer Systems Laboratory at the University of Utah.'' + * + * THE UNIVERSITY OF UTAH AND CSL ALLOW FREE USE OF THIS SOFTWARE IN ITS "AS + * IS" CONDITION. THE UNIVERSITY OF UTAH AND CSL DISCLAIM ANY LIABILITY OF + * ANY KIND FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. + * + * CSL requests users of this software to return to csl-dist@cs.utah.edu any + * improvements that they make and grant CSL redistribution rights. + * + * Utah $Hdr$ + * $FreeBSD$ + */ + +/* + * routines to convert on disk ext2 inodes in dinodes and back + */ +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/lock.h> +#include <sys/stat.h> +#include <sys/vnode.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> + +/* + * Undo the definitions in <ufs/ufs/inode.h> that would destroy the include + * of <gnu/ext2fs/ext2_fs.h>. + */ +#undef i_atime +#undef i_blocks +#undef i_ctime +#undef i_db +#undef i_flags +#undef i_gen +#undef i_gid +#undef i_ib +#undef i_mode +#undef i_mtime +#undef i_nlink +#undef i_rdev +#undef i_shortlink +#undef i_size +#undef i_uid + +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_extern.h> + +void +ext2_print_dinode( di ) + struct dinode *di; +{ + int i; + printf( /* "Inode: %5d" */ + " Type: %10s Mode: 0x%o Flags: 0x%x Version: %d\n", + "n/a", di->di_mode, di->di_flags, di->di_gen); + printf( "User: %5lu Group: %5lu Size: %lu\n", + (unsigned long)di->di_uid, (unsigned long)di->di_gid, + (unsigned long)di->di_size); + printf( "Links: %3d Blockcount: %d\n", + di->di_nlink, di->di_blocks); + printf( "ctime: 0x%x", di->di_ctime); + printf( "atime: 0x%x", di->di_atime); + printf( "mtime: 0x%x", di->di_mtime); + printf( "BLOCKS: "); + for(i=0; i < (di->di_blocks <= 24 ? ((di->di_blocks+1)/2): 12); i++) + printf("%d ", di->di_db[i]); + printf("\n"); +} + +void +ext2_print_inode( in ) + struct inode *in; +{ + printf( "Inode: %5d", in->i_number); + ext2_print_dinode(&in->i_din); +} + +/* + * raw ext2 inode to dinode + */ +void +ext2_ei2di(ei, di) + struct ext2_inode *ei; + struct dinode *di; +{ + int i; + + di->di_nlink = ei->i_links_count; + /* Godmar thinks - if the link count is zero, then the inode is + unused - according to ext2 standards. Ufs marks this fact + by setting i_mode to zero - why ? + I can see that this might lead to problems in an undelete. + */ + di->di_mode = ei->i_links_count ? ei->i_mode : 0; + di->di_size = ei->i_size; + di->di_atime = ei->i_atime; + di->di_mtime = ei->i_mtime; + di->di_ctime = ei->i_ctime; + di->di_flags = 0; + di->di_flags |= (ei->i_flags & EXT2_APPEND_FL) ? APPEND : 0; + di->di_flags |= (ei->i_flags & EXT2_IMMUTABLE_FL) ? IMMUTABLE : 0; + di->di_blocks = ei->i_blocks; + di->di_gen = ei->i_generation; + di->di_uid = ei->i_uid; + di->di_gid = ei->i_gid; + /* XXX use memcpy */ + for(i = 0; i < NDADDR; i++) + di->di_db[i] = ei->i_block[i]; + for(i = 0; i < NIADDR; i++) + di->di_ib[i] = ei->i_block[EXT2_NDIR_BLOCKS + i]; +} + +/* + * dinode to raw ext2 inode + */ +void +ext2_di2ei(di, ei) + struct dinode *di; + struct ext2_inode *ei; +{ + int i; + + ei->i_mode = di->di_mode; + ei->i_links_count = di->di_nlink; + /* + Godmar thinks: if dtime is nonzero, ext2 says this inode + has been deleted, this would correspond to a zero link count + */ + ei->i_dtime = ei->i_links_count ? 0 : di->di_mtime; + ei->i_size = di->di_size; + ei->i_atime = di->di_atime; + ei->i_mtime = di->di_mtime; + ei->i_ctime = di->di_ctime; + ei->i_flags = di->di_flags; + ei->i_flags = 0; + ei->i_flags |= (di->di_flags & APPEND) ? EXT2_APPEND_FL: 0; + ei->i_flags |= (di->di_flags & IMMUTABLE) + ? EXT2_IMMUTABLE_FL: 0; + ei->i_blocks = di->di_blocks; + ei->i_generation = di->di_gen; + ei->i_uid = di->di_uid; + ei->i_gid = di->di_gid; + /* XXX use memcpy */ + for(i = 0; i < NDADDR; i++) + ei->i_block[i] = di->di_db[i]; + for(i = 0; i < NIADDR; i++) + ei->i_block[EXT2_NDIR_BLOCKS + i] = di->di_ib[i]; +} diff --git a/sys/gnu/ext2fs/ext2_linux_balloc.c b/sys/gnu/ext2fs/ext2_linux_balloc.c new file mode 100644 index 000000000000..fba300e41ae0 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_linux_balloc.c @@ -0,0 +1,610 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + * + * $FreeBSD$ + */ +/* + * linux/fs/ext2/balloc.c + * + * Copyright (C) 1992, 1993, 1994, 1995 + * Remy Card (card@masi.ibp.fr) + * Laboratoire MASI - Institut Blaise Pascal + * Universite Pierre et Marie Curie (Paris VI) + * + * Enhanced block allocation by Stephen Tweedie (sct@dcs.ed.ac.uk), 1993 + */ + +/* + * The free blocks are managed by bitmaps. A file system contains several + * blocks groups. Each group contains 1 bitmap block for blocks, 1 bitmap + * block for inodes, N blocks for the inode table and data blocks. + * + * The file system contains group descriptors which are located after the + * super block. Each descriptor contains the number of the bitmap block and + * the free blocks count in the block. The descriptors are loaded in memory + * when a file system is mounted (see ext2_read_super). + */ + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/buf.h> +#include <sys/proc.h> +#include <sys/mount.h> +#include <sys/vnode.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/ufsmount.h> +#include <gnu/ext2fs/ext2_extern.h> +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> + +#ifdef __i386__ +#include <gnu/ext2fs/i386-bitops.h> +#elif defined (__alpha__) +#include <gnu/ext2fs/alpha-bitops.h> +#else +#error Provide an bitops.h file, please ! +#endif + +#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) + +/* got rid of get_group_desc since it can already be found in + * ext2_linux_ialloc.c + */ + +static void read_block_bitmap (struct mount * mp, + unsigned int block_group, + unsigned long bitmap_nr) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct ext2_group_desc * gdp; + struct buffer_head * bh; + int error; + + gdp = get_group_desc (mp, block_group, NULL); + if ((error = bread (VFSTOUFS(mp)->um_devvp, + fsbtodb(sb, gdp->bg_block_bitmap),sb->s_blocksize, NOCRED, &bh)) != 0) + panic ( "read_block_bitmap: " + "Cannot read block bitmap - " + "block_group = %d, block_bitmap = %lu", + block_group, (unsigned long) gdp->bg_block_bitmap); + sb->s_block_bitmap_number[bitmap_nr] = block_group; + sb->s_block_bitmap[bitmap_nr] = bh; + LCK_BUF(bh) +} + +/* + * load_block_bitmap loads the block bitmap for a blocks group + * + * It maintains a cache for the last bitmaps loaded. This cache is managed + * with a LRU algorithm. + * + * Notes: + * 1/ There is one cache per mounted file system. + * 2/ If the file system contains less than EXT2_MAX_GROUP_LOADED groups, + * this function reads the bitmap without maintaining a LRU cache. + */ +static int load__block_bitmap (struct mount * mp, + unsigned int block_group) +{ + int i, j; + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + unsigned long block_bitmap_number; + struct buffer_head * block_bitmap; + + if (block_group >= sb->s_groups_count) + panic ( "load_block_bitmap: " + "block_group >= groups_count - " + "block_group = %d, groups_count = %lu", + block_group, sb->s_groups_count); + + if (sb->s_groups_count <= EXT2_MAX_GROUP_LOADED) { + if (sb->s_block_bitmap[block_group]) { + if (sb->s_block_bitmap_number[block_group] != + block_group) + panic ( "load_block_bitmap: " + "block_group != block_bitmap_number"); + else + return block_group; + } else { + read_block_bitmap (mp, block_group, block_group); + return block_group; + } + } + + for (i = 0; i < sb->s_loaded_block_bitmaps && + sb->s_block_bitmap_number[i] != block_group; i++) + ; + if (i < sb->s_loaded_block_bitmaps && + sb->s_block_bitmap_number[i] == block_group) { + block_bitmap_number = sb->s_block_bitmap_number[i]; + block_bitmap = sb->s_block_bitmap[i]; + for (j = i; j > 0; j--) { + sb->s_block_bitmap_number[j] = + sb->s_block_bitmap_number[j - 1]; + sb->s_block_bitmap[j] = + sb->s_block_bitmap[j - 1]; + } + sb->s_block_bitmap_number[0] = block_bitmap_number; + sb->s_block_bitmap[0] = block_bitmap; + } else { + if (sb->s_loaded_block_bitmaps < EXT2_MAX_GROUP_LOADED) + sb->s_loaded_block_bitmaps++; + else + ULCK_BUF(sb->s_block_bitmap[EXT2_MAX_GROUP_LOADED - 1]) + + for (j = sb->s_loaded_block_bitmaps - 1; j > 0; j--) { + sb->s_block_bitmap_number[j] = + sb->s_block_bitmap_number[j - 1]; + sb->s_block_bitmap[j] = + sb->s_block_bitmap[j - 1]; + } + read_block_bitmap (mp, block_group, 0); + } + return 0; +} + +static __inline int load_block_bitmap (struct mount * mp, + unsigned int block_group) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + if (sb->s_loaded_block_bitmaps > 0 && + sb->s_block_bitmap_number[0] == block_group) + return 0; + + if (sb->s_groups_count <= EXT2_MAX_GROUP_LOADED && + sb->s_block_bitmap_number[block_group] == block_group && + sb->s_block_bitmap[block_group]) + return block_group; + + return load__block_bitmap (mp, block_group); +} + +void ext2_free_blocks (struct mount * mp, unsigned long block, + unsigned long count) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct buffer_head * bh; + struct buffer_head * bh2; + unsigned long block_group; + unsigned long bit; + unsigned long i; + int bitmap_nr; + struct ext2_group_desc * gdp; + struct ext2_super_block * es = sb->s_es; + + if (!sb) { + printf ("ext2_free_blocks: nonexistent device"); + return; + } + lock_super (VFSTOUFS(mp)->um_devvp); + if (block < es->s_first_data_block || + (block + count) > es->s_blocks_count) { + printf ( "ext2_free_blocks: " + "Freeing blocks not in datazone - " + "block = %lu, count = %lu", block, count); + unlock_super (VFSTOUFS(mp)->um_devvp); + return; + } + + ext2_debug ("freeing blocks %lu to %lu\n", block, block+count-1); + + block_group = (block - es->s_first_data_block) / + EXT2_BLOCKS_PER_GROUP(sb); + bit = (block - es->s_first_data_block) % EXT2_BLOCKS_PER_GROUP(sb); + if (bit + count > EXT2_BLOCKS_PER_GROUP(sb)) + panic ( "ext2_free_blocks: " + "Freeing blocks across group boundary - " + "Block = %lu, count = %lu", + block, count); + bitmap_nr = load_block_bitmap (mp, block_group); + bh = sb->s_block_bitmap[bitmap_nr]; + gdp = get_group_desc (mp, block_group, &bh2); + + if (/* test_opt (sb, CHECK_STRICT) && assume always strict ! */ + (in_range (gdp->bg_block_bitmap, block, count) || + in_range (gdp->bg_inode_bitmap, block, count) || + in_range (block, gdp->bg_inode_table, + sb->s_itb_per_group) || + in_range (block + count - 1, gdp->bg_inode_table, + sb->s_itb_per_group))) + panic ( "ext2_free_blocks: " + "Freeing blocks in system zones - " + "Block = %lu, count = %lu", + block, count); + + for (i = 0; i < count; i++) { + if (!clear_bit (bit + i, bh->b_data)) + printf ("ext2_free_blocks: " + "bit already cleared for block %lu", + block); + else { + gdp->bg_free_blocks_count++; + es->s_free_blocks_count++; + } + } + + mark_buffer_dirty(bh2); + mark_buffer_dirty(bh); +/**** + if (sb->s_flags & MS_SYNCHRONOUS) { + ll_rw_block (WRITE, 1, &bh); + wait_on_buffer (bh); + } +****/ + sb->s_dirt = 1; + unlock_super (VFSTOUFS(mp)->um_devvp); + return; +} + +/* + * ext2_new_block uses a goal block to assist allocation. If the goal is + * free, or there is a free block within 32 blocks of the goal, that block + * is allocated. Otherwise a forward search is made for a free block; within + * each block group the search first looks for an entire free byte in the block + * bitmap, and then for any free bit if that fails. + */ +int ext2_new_block (struct mount * mp, unsigned long goal, + u_int32_t * prealloc_count, + u_int32_t * prealloc_block) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct buffer_head * bh; + struct buffer_head * bh2; + char * p, * r; + int i, j, k, tmp; + int bitmap_nr; + struct ext2_group_desc * gdp; + struct ext2_super_block * es = sb->s_es; + +#ifdef EXT2FS_DEBUG + static int goal_hits = 0, goal_attempts = 0; +#endif + if (!sb) { + printf ("ext2_new_block: nonexistent device"); + return 0; + } + lock_super (VFSTOUFS(mp)->um_devvp); + + ext2_debug ("goal=%lu.\n", goal); + +repeat: + /* + * First, test whether the goal block is free. + */ + if (goal < es->s_first_data_block || goal >= es->s_blocks_count) + goal = es->s_first_data_block; + i = (goal - es->s_first_data_block) / EXT2_BLOCKS_PER_GROUP(sb); + gdp = get_group_desc (mp, i, &bh2); + if (gdp->bg_free_blocks_count > 0) { + j = ((goal - es->s_first_data_block) % EXT2_BLOCKS_PER_GROUP(sb)); +#ifdef EXT2FS_DEBUG + if (j) + goal_attempts++; +#endif + bitmap_nr = load_block_bitmap (mp, i); + bh = sb->s_block_bitmap[bitmap_nr]; + + ext2_debug ("goal is at %d:%d.\n", i, j); + + if (!test_bit(j, bh->b_data)) { +#ifdef EXT2FS_DEBUG + goal_hits++; + ext2_debug ("goal bit allocated.\n"); +#endif + goto got_block; + } + if (j) { + /* + * The goal was occupied; search forward for a free + * block within the next XX blocks. + * + * end_goal is more or less random, but it has to be + * less than EXT2_BLOCKS_PER_GROUP. Aligning up to the + * next 64-bit boundary is simple.. + */ + int end_goal = (j + 63) & ~63; + j = find_next_zero_bit(bh->b_data, end_goal, j); + if (j < end_goal) + goto got_block; + } + + ext2_debug ("Bit not found near goal\n"); + + /* + * There has been no free block found in the near vicinity + * of the goal: do a search forward through the block groups, + * searching in each group first for an entire free byte in + * the bitmap and then for any free bit. + * + * Search first in the remainder of the current group; then, + * cyclicly search through the rest of the groups. + */ + p = ((char *) bh->b_data) + (j >> 3); + r = memscan(p, 0, (EXT2_BLOCKS_PER_GROUP(sb) - j + 7) >> 3); + k = (r - ((char *) bh->b_data)) << 3; + if (k < EXT2_BLOCKS_PER_GROUP(sb)) { + j = k; + goto search_back; + } + k = find_next_zero_bit ((unsigned long *) bh->b_data, + EXT2_BLOCKS_PER_GROUP(sb), + j); + if (k < EXT2_BLOCKS_PER_GROUP(sb)) { + j = k; + goto got_block; + } + } + + ext2_debug ("Bit not found in block group %d.\n", i); + + /* + * Now search the rest of the groups. We assume that + * i and gdp correctly point to the last group visited. + */ + for (k = 0; k < sb->s_groups_count; k++) { + i++; + if (i >= sb->s_groups_count) + i = 0; + gdp = get_group_desc (mp, i, &bh2); + if (gdp->bg_free_blocks_count > 0) + break; + } + if (k >= sb->s_groups_count) { + unlock_super (VFSTOUFS(mp)->um_devvp); + return 0; + } + bitmap_nr = load_block_bitmap (mp, i); + bh = sb->s_block_bitmap[bitmap_nr]; + r = memscan(bh->b_data, 0, EXT2_BLOCKS_PER_GROUP(sb) >> 3); + j = (r - bh->b_data) << 3; + + if (j < EXT2_BLOCKS_PER_GROUP(sb)) + goto search_back; + else + j = find_first_zero_bit ((unsigned long *) bh->b_data, + EXT2_BLOCKS_PER_GROUP(sb)); + if (j >= EXT2_BLOCKS_PER_GROUP(sb)) { + printf ( "ext2_new_block: " + "Free blocks count corrupted for block group %d", i); + unlock_super (VFSTOUFS(mp)->um_devvp); + return 0; + } + +search_back: + /* + * We have succeeded in finding a free byte in the block + * bitmap. Now search backwards up to 7 bits to find the + * start of this group of free blocks. + */ + for (k = 0; k < 7 && j > 0 && !test_bit (j - 1, bh->b_data); k++, j--); + +got_block: + + ext2_debug ("using block group %d(%d)\n", i, gdp->bg_free_blocks_count); + + tmp = j + i * EXT2_BLOCKS_PER_GROUP(sb) + es->s_first_data_block; + + if (/* test_opt (sb, CHECK_STRICT) && we are always strict. */ + (tmp == gdp->bg_block_bitmap || + tmp == gdp->bg_inode_bitmap || + in_range (tmp, gdp->bg_inode_table, sb->s_itb_per_group))) + panic ( "ext2_new_block: " + "Allocating block in system zone - " + "%dth block = %u in group %u", j, tmp, i); + + if (set_bit (j, bh->b_data)) { + printf ( "ext2_new_block: " + "bit already set for block %d", j); + goto repeat; + } + + ext2_debug ("found bit %d\n", j); + + /* + * Do block preallocation now if required. + */ +#ifdef EXT2_PREALLOCATE + if (prealloc_block) { + *prealloc_count = 0; + *prealloc_block = tmp + 1; + for (k = 1; + k < 8 && (j + k) < EXT2_BLOCKS_PER_GROUP(sb); k++) { + if (set_bit (j + k, bh->b_data)) + break; + (*prealloc_count)++; + } + gdp->bg_free_blocks_count -= *prealloc_count; + es->s_free_blocks_count -= *prealloc_count; + ext2_debug ("Preallocated a further %lu bits.\n", + *prealloc_count); + } +#endif + + j = tmp; + + mark_buffer_dirty(bh); +/**** + if (sb->s_flags & MS_SYNCHRONOUS) { + ll_rw_block (WRITE, 1, &bh); + wait_on_buffer (bh); + } +****/ + if (j >= es->s_blocks_count) { + printf ( "ext2_new_block: " + "block >= blocks count - " + "block_group = %d, block=%d", i, j); + unlock_super (VFSTOUFS(mp)->um_devvp); + return 0; + } + + ext2_debug ("allocating block %d. " + "Goal hits %d of %d.\n", j, goal_hits, goal_attempts); + + gdp->bg_free_blocks_count--; + mark_buffer_dirty(bh2); + es->s_free_blocks_count--; + sb->s_dirt = 1; + unlock_super (VFSTOUFS(mp)->um_devvp); + return j; +} + +#ifdef unused +static unsigned long ext2_count_free_blocks (struct mount * mp) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; +#ifdef EXT2FS_DEBUG + struct ext2_super_block * es; + unsigned long desc_count, bitmap_count, x; + int bitmap_nr; + struct ext2_group_desc * gdp; + int i; + + lock_super (VFSTOUFS(mp)->um_devvp); + es = sb->s_es; + desc_count = 0; + bitmap_count = 0; + gdp = NULL; + for (i = 0; i < sb->s_groups_count; i++) { + gdp = get_group_desc (mp, i, NULL); + desc_count += gdp->bg_free_blocks_count; + bitmap_nr = load_block_bitmap (mp, i); + x = ext2_count_free (sb->s_block_bitmap[bitmap_nr], + sb->s_blocksize); + ext2_debug ("group %d: stored = %d, counted = %lu\n", + i, gdp->bg_free_blocks_count, x); + bitmap_count += x; + } + ext2_debug( "stored = %lu, computed = %lu, %lu\n", + es->s_free_blocks_count, desc_count, bitmap_count); + unlock_super (VFSTOUFS(mp)->um_devvp); + return bitmap_count; +#else + return sb->s_es->s_free_blocks_count; +#endif +} +#endif /* unused */ + +static __inline int block_in_use (unsigned long block, + struct ext2_sb_info * sb, + unsigned char * map) +{ + return test_bit ((block - sb->s_es->s_first_data_block) % + EXT2_BLOCKS_PER_GROUP(sb), map); +} + +static int test_root(int a, int b) +{ + if (a == 0) + return 1; + while (1) { + if (a == 1) + return 1; + if (a % b) + return 0; + a = a / b; + } +} + +int ext2_group_sparse(int group) +{ + return (test_root(group, 3) || test_root(group, 5) || + test_root(group, 7)); +} + +#ifdef unused +static void ext2_check_blocks_bitmap (struct mount * mp) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct buffer_head * bh; + struct ext2_super_block * es; + unsigned long desc_count, bitmap_count, x; + unsigned long desc_blocks; + int bitmap_nr; + struct ext2_group_desc * gdp; + int i, j; + + lock_super (VFSTOUFS(mp)->um_devvp); + es = sb->s_es; + desc_count = 0; + bitmap_count = 0; + gdp = NULL; + desc_blocks = (sb->s_groups_count + EXT2_DESC_PER_BLOCK(sb) - 1) / + EXT2_DESC_PER_BLOCK(sb); + for (i = 0; i < sb->s_groups_count; i++) { + gdp = get_group_desc (mp, i, NULL); + desc_count += gdp->bg_free_blocks_count; + bitmap_nr = load_block_bitmap (mp, i); + bh = sb->s_block_bitmap[bitmap_nr]; + + if (!(es->s_feature_ro_compat & + EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER) || + ext2_group_sparse(i)) { + if (!test_bit (0, bh->b_data)) + printf ("ext2_check_blocks_bitmap: " + "Superblock in group %d " + "is marked free", i); + + for (j = 0; j < desc_blocks; j++) + if (!test_bit (j + 1, bh->b_data)) + printf ("ext2_check_blocks_bitmap: " + "Descriptor block #%d in group " + "%d is marked free", j, i); + } + + if (!block_in_use (gdp->bg_block_bitmap, sb, bh->b_data)) + printf ("ext2_check_blocks_bitmap: " + "Block bitmap for group %d is marked free", + i); + + if (!block_in_use (gdp->bg_inode_bitmap, sb, bh->b_data)) + printf ("ext2_check_blocks_bitmap: " + "Inode bitmap for group %d is marked free", + i); + + for (j = 0; j < sb->s_itb_per_group; j++) + if (!block_in_use (gdp->bg_inode_table + j, sb, bh->b_data)) + printf ("ext2_check_blocks_bitmap: " + "Block #%d of the inode table in " + "group %d is marked free", j, i); + + x = ext2_count_free (bh, sb->s_blocksize); + if (gdp->bg_free_blocks_count != x) + printf ("ext2_check_blocks_bitmap: " + "Wrong free blocks count for group %d, " + "stored = %d, counted = %lu", i, + gdp->bg_free_blocks_count, x); + bitmap_count += x; + } + if (es->s_free_blocks_count != bitmap_count) + printf ("ext2_check_blocks_bitmap: " + "Wrong free blocks count in super block, " + "stored = %lu, counted = %lu", + (unsigned long) es->s_free_blocks_count, bitmap_count); + unlock_super (VFSTOUFS(mp)->um_devvp); +} +#endif /* unused */ + +/* + * this function is taken from + * linux/fs/ext2/bitmap.c + */ + +static int nibblemap[] = {4, 3, 3, 2, 3, 2, 2, 1, 3, 2, 2, 1, 2, 1, 1, 0}; + +unsigned long ext2_count_free (struct buffer_head * map, unsigned int numchars) +{ + unsigned int i; + unsigned long sum = 0; + + if (!map) + return (0); + for (i = 0; i < numchars; i++) + sum += nibblemap[map->b_data[i] & 0xf] + + nibblemap[(map->b_data[i] >> 4) & 0xf]; + return (sum); +} diff --git a/sys/gnu/ext2fs/ext2_linux_ialloc.c b/sys/gnu/ext2fs/ext2_linux_ialloc.c new file mode 100644 index 000000000000..8f2d33f608ec --- /dev/null +++ b/sys/gnu/ext2fs/ext2_linux_ialloc.c @@ -0,0 +1,515 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + * + * $FreeBSD$ + */ +/* + * linux/fs/ext2/ialloc.c + * + * Copyright (C) 1992, 1993, 1994, 1995 + * Remy Card (card@masi.ibp.fr) + * Laboratoire MASI - Institut Blaise Pascal + * Universite Pierre et Marie Curie (Paris VI) + * + * BSD ufs-inspired inode and directory allocation by + * Stephen Tweedie (sct@dcs.ed.ac.uk), 1993 + */ + +/* + * The free inodes are managed by bitmaps. A file system contains several + * blocks groups. Each group contains 1 bitmap block for blocks, 1 bitmap + * block for inodes, N blocks for the inode table and data blocks. + * + * The file system contains group descriptors which are located after the + * super block. Each descriptor contains the number of the bitmap block and + * the free blocks count in the block. The descriptors are loaded in memory + * when a file system is mounted (see ext2_read_super). + */ + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/buf.h> +#include <sys/proc.h> +#include <sys/mount.h> +#include <sys/vnode.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufsmount.h> +#include <gnu/ext2fs/ext2_extern.h> +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> +#include <sys/stat.h> + +#ifdef __i386__ +#include <gnu/ext2fs/i386-bitops.h> +#elif defined(__alpha__) +#include <gnu/ext2fs/alpha-bitops.h> +#else +#error please provide bit operation functions +#endif + +/* this is supposed to mark a buffer dirty on ready for delayed writing + */ +void mark_buffer_dirty(struct buf *bh) +{ + int s; + + s = splbio(); + bh->b_flags |= B_DIRTY; + splx(s); +} + +struct ext2_group_desc * get_group_desc (struct mount * mp, + unsigned int block_group, + struct buffer_head ** bh) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + unsigned long group_desc; + unsigned long desc; + struct ext2_group_desc * gdp; + + if (block_group >= sb->s_groups_count) + panic ("get_group_desc: " + "block_group >= groups_count - " + "block_group = %d, groups_count = %lu", + block_group, sb->s_groups_count); + + group_desc = block_group / EXT2_DESC_PER_BLOCK(sb); + desc = block_group % EXT2_DESC_PER_BLOCK(sb); + if (!sb->s_group_desc[group_desc]) + panic ( "get_group_desc:" + "Group descriptor not loaded - " + "block_group = %d, group_desc = %lu, desc = %lu", + block_group, group_desc, desc); + gdp = (struct ext2_group_desc *) + sb->s_group_desc[group_desc]->b_data; + if (bh) + *bh = sb->s_group_desc[group_desc]; + return gdp + desc; +} + +static void read_inode_bitmap (struct mount * mp, + unsigned long block_group, + unsigned int bitmap_nr) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct ext2_group_desc * gdp; + struct buffer_head * bh; + int error; + + gdp = get_group_desc (mp, block_group, NULL); + if ((error = bread (VFSTOUFS(mp)->um_devvp, + fsbtodb(sb, gdp->bg_inode_bitmap), + sb->s_blocksize, + NOCRED, &bh)) != 0) + panic ( "read_inode_bitmap:" + "Cannot read inode bitmap - " + "block_group = %lu, inode_bitmap = %lu", + block_group, (unsigned long) gdp->bg_inode_bitmap); + sb->s_inode_bitmap_number[bitmap_nr] = block_group; + sb->s_inode_bitmap[bitmap_nr] = bh; + LCK_BUF(bh) +} + +/* + * load_inode_bitmap loads the inode bitmap for a blocks group + * + * It maintains a cache for the last bitmaps loaded. This cache is managed + * with a LRU algorithm. + * + * Notes: + * 1/ There is one cache per mounted file system. + * 2/ If the file system contains less than EXT2_MAX_GROUP_LOADED groups, + * this function reads the bitmap without maintaining a LRU cache. + */ +static int load_inode_bitmap (struct mount * mp, + unsigned int block_group) +{ + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + int i, j; + unsigned long inode_bitmap_number; + struct buffer_head * inode_bitmap; + + if (block_group >= sb->s_groups_count) + panic ("load_inode_bitmap:" + "block_group >= groups_count - " + "block_group = %d, groups_count = %lu", + block_group, sb->s_groups_count); + if (sb->s_loaded_inode_bitmaps > 0 && + sb->s_inode_bitmap_number[0] == block_group) + return 0; + if (sb->s_groups_count <= EXT2_MAX_GROUP_LOADED) { + if (sb->s_inode_bitmap[block_group]) { + if (sb->s_inode_bitmap_number[block_group] != + block_group) + panic ( "load_inode_bitmap:" + "block_group != inode_bitmap_number"); + else + return block_group; + } else { + read_inode_bitmap (mp, block_group, block_group); + return block_group; + } + } + + for (i = 0; i < sb->s_loaded_inode_bitmaps && + sb->s_inode_bitmap_number[i] != block_group; + i++) + ; + if (i < sb->s_loaded_inode_bitmaps && + sb->s_inode_bitmap_number[i] == block_group) { + inode_bitmap_number = sb->s_inode_bitmap_number[i]; + inode_bitmap = sb->s_inode_bitmap[i]; + for (j = i; j > 0; j--) { + sb->s_inode_bitmap_number[j] = + sb->s_inode_bitmap_number[j - 1]; + sb->s_inode_bitmap[j] = + sb->s_inode_bitmap[j - 1]; + } + sb->s_inode_bitmap_number[0] = inode_bitmap_number; + sb->s_inode_bitmap[0] = inode_bitmap; + } else { + if (sb->s_loaded_inode_bitmaps < EXT2_MAX_GROUP_LOADED) + sb->s_loaded_inode_bitmaps++; + else + ULCK_BUF(sb->s_inode_bitmap[EXT2_MAX_GROUP_LOADED - 1]) + for (j = sb->s_loaded_inode_bitmaps - 1; j > 0; j--) { + sb->s_inode_bitmap_number[j] = + sb->s_inode_bitmap_number[j - 1]; + sb->s_inode_bitmap[j] = + sb->s_inode_bitmap[j - 1]; + } + read_inode_bitmap (mp, block_group, 0); + } + return 0; +} + + +void ext2_free_inode (struct inode * inode) +{ + struct ext2_sb_info * sb; + struct buffer_head * bh; + struct buffer_head * bh2; + unsigned long block_group; + unsigned long bit; + int bitmap_nr; + struct ext2_group_desc * gdp; + struct ext2_super_block * es; + + if (!inode) + return; + + if (inode->i_nlink) { + printf ("ext2_free_inode: inode has nlink=%d\n", + inode->i_nlink); + return; + } + + ext2_debug ("freeing inode %lu\n", inode->i_number); + + sb = inode->i_e2fs; + lock_super (DEVVP(inode)); + if (inode->i_number < EXT2_FIRST_INO || + inode->i_number > sb->s_es->s_inodes_count) { + printf ("free_inode reserved inode or nonexistent inode"); + unlock_super (DEVVP(inode)); + return; + } + es = sb->s_es; + block_group = (inode->i_number - 1) / EXT2_INODES_PER_GROUP(sb); + bit = (inode->i_number - 1) % EXT2_INODES_PER_GROUP(sb); + bitmap_nr = load_inode_bitmap (ITOV(inode)->v_mount, block_group); + bh = sb->s_inode_bitmap[bitmap_nr]; + if (!clear_bit (bit, bh->b_data)) + printf ( "ext2_free_inode:" + "bit already cleared for inode %lu", + (unsigned long)inode->i_number); + else { + gdp = get_group_desc (ITOV(inode)->v_mount, block_group, &bh2); + gdp->bg_free_inodes_count++; + if (S_ISDIR(inode->i_mode)) + gdp->bg_used_dirs_count--; + mark_buffer_dirty(bh2); + es->s_free_inodes_count++; + } + mark_buffer_dirty(bh); +/*** XXX + if (sb->s_flags & MS_SYNCHRONOUS) { + ll_rw_block (WRITE, 1, &bh); + wait_on_buffer (bh); + } +***/ + sb->s_dirt = 1; + unlock_super (DEVVP(inode)); +} + +#if linux +/* + * This function increments the inode version number + * + * This may be used one day by the NFS server + */ +static void inc_inode_version (struct inode * inode, + struct ext2_group_desc *gdp, + int mode) +{ + unsigned long inode_block; + struct buffer_head * bh; + struct ext2_inode * raw_inode; + + inode_block = gdp->bg_inode_table + (((inode->i_number - 1) % + EXT2_INODES_PER_GROUP(inode->i_sb)) / + EXT2_INODES_PER_BLOCK(inode->i_sb)); + bh = bread (inode->i_sb->s_dev, inode_block, inode->i_sb->s_blocksize); + if (!bh) { + printf ("inc_inode_version Cannot load inode table block - " + "inode=%lu, inode_block=%lu\n", + inode->i_number, inode_block); + inode->u.ext2_i.i_version = 1; + return; + } + raw_inode = ((struct ext2_inode *) bh->b_data) + + (((inode->i_number - 1) % + EXT2_INODES_PER_GROUP(inode->i_sb)) % + EXT2_INODES_PER_BLOCK(inode->i_sb)); + raw_inode->i_version++; + inode->u.ext2_i.i_version = raw_inode->i_version; + bdwrite (bh); +} + +#endif /* linux */ + +/* + * There are two policies for allocating an inode. If the new inode is + * a directory, then a forward search is made for a block group with both + * free space and a low directory-to-inode ratio; if that fails, then of + * the groups with above-average free space, that group with the fewest + * directories already is chosen. + * + * For other inodes, search forward from the parent directory\'s block + * group to find a free inode. + */ +/* + * this functino has been reduced to the actual 'find the inode number' part + */ +ino_t ext2_new_inode (const struct inode * dir, int mode) +{ + struct ext2_sb_info * sb; + struct buffer_head * bh; + struct buffer_head * bh2; + int i, j, avefreei; + int bitmap_nr; + struct ext2_group_desc * gdp; + struct ext2_group_desc * tmp; + struct ext2_super_block * es; + + if (!dir) + return 0; + sb = dir->i_e2fs; + + lock_super (DEVVP(dir)); + es = sb->s_es; +repeat: + gdp = NULL; i=0; + + if (S_ISDIR(mode)) { + avefreei = es->s_free_inodes_count / + sb->s_groups_count; +/* I am not yet convinced that this next bit is necessary. + i = dir->u.ext2_i.i_block_group; + for (j = 0; j < sb->u.ext2_sb.s_groups_count; j++) { + tmp = get_group_desc (sb, i, &bh2); + if ((tmp->bg_used_dirs_count << 8) < + tmp->bg_free_inodes_count) { + gdp = tmp; + break; + } + else + i = ++i % sb->u.ext2_sb.s_groups_count; + } +*/ + if (!gdp) { + for (j = 0; j < sb->s_groups_count; j++) { + tmp = get_group_desc(ITOV(dir)->v_mount,j,&bh2); + if (tmp->bg_free_inodes_count && + tmp->bg_free_inodes_count >= avefreei) { + if (!gdp || + (tmp->bg_free_blocks_count > + gdp->bg_free_blocks_count)) { + i = j; + gdp = tmp; + } + } + } + } + } + else + { + /* + * Try to place the inode in its parent directory + */ + i = dir->i_block_group; + tmp = get_group_desc (ITOV(dir)->v_mount, i, &bh2); + if (tmp->bg_free_inodes_count) + gdp = tmp; + else + { + /* + * Use a quadratic hash to find a group with a + * free inode + */ + for (j = 1; j < sb->s_groups_count; j <<= 1) { + i += j; + if (i >= sb->s_groups_count) + i -= sb->s_groups_count; + tmp = get_group_desc(ITOV(dir)->v_mount,i,&bh2); + if (tmp->bg_free_inodes_count) { + gdp = tmp; + break; + } + } + } + if (!gdp) { + /* + * That failed: try linear search for a free inode + */ + i = dir->i_block_group + 1; + for (j = 2; j < sb->s_groups_count; j++) { + if (++i >= sb->s_groups_count) + i = 0; + tmp = get_group_desc(ITOV(dir)->v_mount,i,&bh2); + if (tmp->bg_free_inodes_count) { + gdp = tmp; + break; + } + } + } + } + + if (!gdp) { + unlock_super (DEVVP(dir)); + return 0; + } + bitmap_nr = load_inode_bitmap (ITOV(dir)->v_mount, i); + bh = sb->s_inode_bitmap[bitmap_nr]; + if ((j = find_first_zero_bit ((unsigned long *) bh->b_data, + EXT2_INODES_PER_GROUP(sb))) < + EXT2_INODES_PER_GROUP(sb)) { + if (set_bit (j, bh->b_data)) { + printf ( "ext2_new_inode:" + "bit already set for inode %d", j); + goto repeat; + } +/* Linux now does the following: + mark_buffer_dirty(bh); + if (sb->s_flags & MS_SYNCHRONOUS) { + ll_rw_block (WRITE, 1, &bh); + wait_on_buffer (bh); + } +*/ + mark_buffer_dirty(bh); + } else { + if (gdp->bg_free_inodes_count != 0) { + printf ( "ext2_new_inode:" + "Free inodes count corrupted in group %d", + i); + unlock_super (DEVVP(dir)); + return 0; + } + goto repeat; + } + j += i * EXT2_INODES_PER_GROUP(sb) + 1; + if (j < EXT2_FIRST_INO || j > es->s_inodes_count) { + printf ( "ext2_new_inode:" + "reserved inode or inode > inodes count - " + "block_group = %d,inode=%d", i, j); + unlock_super (DEVVP(dir)); + return 0; + } + gdp->bg_free_inodes_count--; + if (S_ISDIR(mode)) + gdp->bg_used_dirs_count++; + mark_buffer_dirty(bh2); + es->s_free_inodes_count--; + /* mark_buffer_dirty(sb->u.ext2_sb.s_sbh, 1); */ + sb->s_dirt = 1; + unlock_super (DEVVP(dir)); + return j; +} + +#ifdef unused +static unsigned long ext2_count_free_inodes (struct mount * mp) +{ +#ifdef EXT2FS_DEBUG + struct ext2_sb_info *sb = VFSTOUFS(mp)->um_e2fs; + struct ext2_super_block * es; + unsigned long desc_count, bitmap_count, x; + int bitmap_nr; + struct ext2_group_desc * gdp; + int i; + + lock_super (VFSTOUFS(mp)->um_devvp); + es = sb->s_es; + desc_count = 0; + bitmap_count = 0; + gdp = NULL; + for (i = 0; i < sb->s_groups_count; i++) { + gdp = get_group_desc (mp, i, NULL); + desc_count += gdp->bg_free_inodes_count; + bitmap_nr = load_inode_bitmap (mp, i); + x = ext2_count_free (sb->s_inode_bitmap[bitmap_nr], + EXT2_INODES_PER_GROUP(sb) / 8); + ext2_debug ("group %d: stored = %d, counted = %lu\n", + i, gdp->bg_free_inodes_count, x); + bitmap_count += x; + } + ext2_debug("stored = %lu, computed = %lu, %lu\n", + es->s_free_inodes_count, desc_count, bitmap_count); + unlock_super (VFSTOUFS(mp)->um_devvp); + return desc_count; +#else + return VFSTOUFS(mp)->um_e2fsb->s_free_inodes_count; +#endif +} +#endif /* unused */ + +#ifdef LATER +void ext2_check_inodes_bitmap (struct mount * mp) +{ + struct ext2_super_block * es; + unsigned long desc_count, bitmap_count, x; + int bitmap_nr; + struct ext2_group_desc * gdp; + int i; + + lock_super (sb); + es = sb->u.ext2_sb.s_es; + desc_count = 0; + bitmap_count = 0; + gdp = NULL; + for (i = 0; i < sb->u.ext2_sb.s_groups_count; i++) { + gdp = get_group_desc (sb, i, NULL); + desc_count += gdp->bg_free_inodes_count; + bitmap_nr = load_inode_bitmap (sb, i); + x = ext2_count_free (sb->u.ext2_sb.s_inode_bitmap[bitmap_nr], + EXT2_INODES_PER_GROUP(sb) / 8); + if (gdp->bg_free_inodes_count != x) + printf ( "ext2_check_inodes_bitmap:" + "Wrong free inodes count in group %d, " + "stored = %d, counted = %lu", i, + gdp->bg_free_inodes_count, x); + bitmap_count += x; + } + if (es->s_free_inodes_count != bitmap_count) + printf ( "ext2_check_inodes_bitmap:" + "Wrong free inodes count in super block, " + "stored = %lu, counted = %lu", + (unsigned long) es->s_free_inodes_count, bitmap_count); + unlock_super (sb); +} +#endif diff --git a/sys/gnu/ext2fs/ext2_lookup.c b/sys/gnu/ext2fs/ext2_lookup.c new file mode 100644 index 000000000000..628f6b066fc8 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_lookup.c @@ -0,0 +1,1078 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + * + * $FreeBSD$ + */ +/* + * Copyright (c) 1989, 1993 + * The Regents of the University of California. All rights reserved. + * (c) UNIX System Laboratories, Inc. + * All or some portions of this file are derived from material licensed + * to the University of California by American Telephone and Telegraph + * Co. or Unix System Laboratories, Inc. and are reproduced herein with + * the permission of UNIX System Laboratories, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ufs_lookup.c 8.6 (Berkeley) 4/1/94 + */ + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/namei.h> +#include <sys/buf.h> +#include <sys/mount.h> +#include <sys/vnode.h> +#include <sys/malloc.h> +#include <sys/dirent.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/dir.h> +#include <ufs/ufs/ufsmount.h> +#include <ufs/ufs/ufs_extern.h> + +#include <gnu/ext2fs/ext2_extern.h> +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> + +/* + DIRBLKSIZE in ffs is DEV_BSIZE (in most cases 512) + while it is the native blocksize in ext2fs - thus, a #define + is no longer appropriate +*/ +#undef DIRBLKSIZ + +extern int dirchk; + +static u_char ext2_ft_to_dt[] = { + DT_UNKNOWN, /* EXT2_FT_UNKNOWN */ + DT_REG, /* EXT2_FT_REG_FILE */ + DT_DIR, /* EXT2_FT_DIR */ + DT_CHR, /* EXT2_FT_CHRDEV */ + DT_BLK, /* EXT2_FT_BLKDEV */ + DT_FIFO, /* EXT2_FT_FIFO */ + DT_SOCK, /* EXT2_FT_SOCK */ + DT_LNK, /* EXT2_FT_SYMLINK */ +}; +#define FTTODT(ft) \ + ((ft) > sizeof(ext2_ft_to_dt) / sizeof(ext2_ft_to_dt[0]) ? \ + DT_UNKNOWN : ext2_ft_to_dt[(ft)]) + +static u_char dt_to_ext2_ft[] = { + EXT2_FT_UNKNOWN, /* DT_UNKNOWN */ + EXT2_FT_FIFO, /* DT_FIFO */ + EXT2_FT_CHRDEV, /* DT_CHR */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_DIR, /* DT_DIR */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_BLKDEV, /* DT_BLK */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_REG_FILE, /* DT_REG */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_SYMLINK, /* DT_LNK */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_SOCK, /* DT_SOCK */ + EXT2_FT_UNKNOWN, /* unused */ + EXT2_FT_UNKNOWN, /* DT_WHT */ +}; +#define DTTOFT(dt) \ + ((dt) > sizeof(dt_to_ext2_ft) / sizeof(dt_to_ext2_ft[0]) ? \ + EXT2_FT_UNKNOWN : dt_to_ext2_ft[(dt)]) + +static int ext2_dirbadentry __P((struct vnode *dp, + struct ext2_dir_entry_2 *de, + int entryoffsetinblock)); + +/* + * Vnode op for reading directories. + * + * The routine below assumes that the on-disk format of a directory + * is the same as that defined by <sys/dirent.h>. If the on-disk + * format changes, then it will be necessary to do a conversion + * from the on-disk format that read returns to the format defined + * by <sys/dirent.h>. + */ +/* + * this is exactly what we do here - the problem is that the conversion + * will blow up some entries by four bytes, so it can't be done in place. + * This is too bad. Right now the conversion is done entry by entry, the + * converted entry is sent via uiomove. + * + * XXX allocate a buffer, convert as many entries as possible, then send + * the whole buffer to uiomove + */ +int +ext2_readdir(ap) + struct vop_readdir_args /* { + struct vnode *a_vp; + struct uio *a_uio; + struct ucred *a_cred; + } */ *ap; +{ + register struct uio *uio = ap->a_uio; + int count, error; + + struct ext2_dir_entry_2 *edp, *dp; + int ncookies; + struct dirent dstdp; + struct uio auio; + struct iovec aiov; + caddr_t dirbuf; + int DIRBLKSIZ = VTOI(ap->a_vp)->i_e2fs->s_blocksize; + int readcnt; + off_t startoffset = uio->uio_offset; + + count = uio->uio_resid; + /* + * Avoid complications for partial directory entries by adjusting + * the i/o to end at a block boundary. Don't give up (like ufs + * does) if the initial adjustment gives a negative count, since + * many callers don't supply a large enough buffer. The correct + * size is a little larger than DIRBLKSIZ to allow for expansion + * of directory entries, but some callers just use 512. + */ + count -= (uio->uio_offset + count) & (DIRBLKSIZ -1); + if (count <= 0) + count += DIRBLKSIZ; + +#ifdef EXT2FS_DEBUG + printf("ext2_readdir: uio_offset = %lld, uio_resid = %d, count = %d\n", + uio->uio_offset, uio->uio_resid, count); +#endif + + auio = *uio; + auio.uio_iov = &aiov; + auio.uio_iovcnt = 1; + auio.uio_resid = count; + auio.uio_segflg = UIO_SYSSPACE; + aiov.iov_len = count; + MALLOC(dirbuf, caddr_t, count, M_TEMP, M_WAITOK); + aiov.iov_base = dirbuf; + error = VOP_READ(ap->a_vp, &auio, 0, ap->a_cred); + if (error == 0) { + readcnt = count - auio.uio_resid; + edp = (struct ext2_dir_entry_2 *)&dirbuf[readcnt]; + ncookies = 0; + bzero(&dstdp, offsetof(struct dirent, d_name)); + for (dp = (struct ext2_dir_entry_2 *)dirbuf; + !error && uio->uio_resid > 0 && dp < edp; ) { + /*- + * "New" ext2fs directory entries differ in 3 ways + * from ufs on-disk ones: + * - the name is not necessarily NUL-terminated. + * - the file type field always exists and always + * follows the name length field. + * - the file type is encoded in a different way. + * + * "Old" ext2fs directory entries need no special + * conversions, since they binary compatible with + * "new" entries having a file type of 0 (i.e., + * EXT2_FT_UNKNOWN). Splitting the old name length + * field didn't make a mess like it did in ufs, + * because ext2fs uses a machine-dependent disk + * layout. + */ + dstdp.d_fileno = dp->inode; + dstdp.d_type = FTTODT(dp->file_type); + dstdp.d_namlen = dp->name_len; + dstdp.d_reclen = GENERIC_DIRSIZ(&dstdp); + bcopy(dp->name, dstdp.d_name, dstdp.d_namlen); + bzero(dstdp.d_name + dstdp.d_namlen, + dstdp.d_reclen - offsetof(struct dirent, d_name) - + dstdp.d_namlen); + + if (dp->rec_len > 0) { + if(dstdp.d_reclen <= uio->uio_resid) { + /* advance dp */ + dp = (struct ext2_dir_entry_2 *) + ((char *)dp + dp->rec_len); + error = + uiomove((caddr_t)&dstdp, + dstdp.d_reclen, uio); + if (!error) + ncookies++; + } else + break; + } else { + error = EIO; + break; + } + } + /* we need to correct uio_offset */ + uio->uio_offset = startoffset + (caddr_t)dp - dirbuf; + + if (!error && ap->a_ncookies != NULL) { + u_long *cookiep, *cookies, *ecookies; + off_t off; + + if (uio->uio_segflg != UIO_SYSSPACE || uio->uio_iovcnt != 1) + panic("ext2fs_readdir: unexpected uio from NFS server"); + MALLOC(cookies, u_long *, ncookies * sizeof(u_long), M_TEMP, + M_WAITOK); + off = startoffset; + for (dp = (struct ext2_dir_entry_2 *)dirbuf, + cookiep = cookies, ecookies = cookies + ncookies; + cookiep < ecookies; + dp = (struct ext2_dir_entry_2 *)((caddr_t) dp + dp->rec_len)) { + off += dp->rec_len; + *cookiep++ = (u_long) off; + } + *ap->a_ncookies = ncookies; + *ap->a_cookies = cookies; + } + } + FREE(dirbuf, M_TEMP); + if (ap->a_eofflag) + *ap->a_eofflag = VTOI(ap->a_vp)->i_size <= uio->uio_offset; + return (error); +} + +/* + * Convert a component of a pathname into a pointer to a locked inode. + * This is a very central and rather complicated routine. + * If the file system is not maintained in a strict tree hierarchy, + * this can result in a deadlock situation (see comments in code below). + * + * The cnp->cn_nameiop argument is LOOKUP, CREATE, RENAME, or DELETE depending + * on whether the name is to be looked up, created, renamed, or deleted. + * When CREATE, RENAME, or DELETE is specified, information usable in + * creating, renaming, or deleting a directory entry may be calculated. + * If flag has LOCKPARENT or'ed into it and the target of the pathname + * exists, lookup returns both the target and its parent directory locked. + * When creating or renaming and LOCKPARENT is specified, the target may + * not be ".". When deleting and LOCKPARENT is specified, the target may + * be "."., but the caller must check to ensure it does an vrele and vput + * instead of two vputs. + * + * Overall outline of ufs_lookup: + * + * search for name in directory, to found or notfound + * notfound: + * if creating, return locked directory, leaving info on available slots + * else return error + * found: + * if at end of path and deleting, return information to allow delete + * if at end of path and rewriting (RENAME and LOCKPARENT), lock target + * inode and return info to allow rewrite + * if not at end, add name to cache; if at end and neither creating + * nor deleting, add name to cache + */ +int +ext2_lookup(ap) + struct vop_cachedlookup_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + } */ *ap; +{ + register struct vnode *vdp; /* vnode for directory being searched */ + register struct inode *dp; /* inode for directory being searched */ + struct buf *bp; /* a buffer of directory entries */ + register struct ext2_dir_entry_2 *ep; /* the current directory entry */ + int entryoffsetinblock; /* offset of ep in bp's buffer */ + enum {NONE, COMPACT, FOUND} slotstatus; + doff_t slotoffset; /* offset of area with free space */ + int slotsize; /* size of area at slotoffset */ + int slotfreespace; /* amount of space free in slot */ + int slotneeded; /* size of the entry we're seeking */ + int numdirpasses; /* strategy for directory search */ + doff_t endsearch; /* offset to end directory search */ + doff_t prevoff; /* prev entry dp->i_offset */ + struct vnode *pdp; /* saved dp during symlink work */ + struct vnode *tdp; /* returned by VFS_VGET */ + doff_t enduseful; /* pointer past last used dir slot */ + u_long bmask; /* block offset mask */ + int lockparent; /* 1 => lockparent flag is set */ + int wantparent; /* 1 => wantparent or lockparent flag */ + int namlen, error; + struct vnode **vpp = ap->a_vpp; + struct componentname *cnp = ap->a_cnp; + struct ucred *cred = cnp->cn_cred; + int flags = cnp->cn_flags; + int nameiop = cnp->cn_nameiop; + struct proc *p = cnp->cn_proc; + + int DIRBLKSIZ = VTOI(ap->a_dvp)->i_e2fs->s_blocksize; + + bp = NULL; + slotoffset = -1; + *vpp = NULL; + vdp = ap->a_dvp; + dp = VTOI(vdp); + lockparent = flags & LOCKPARENT; + wantparent = flags & (LOCKPARENT|WANTPARENT); + + /* + * We now have a segment name to search for, and a directory to search. + */ + + /* + * Suppress search for slots unless creating + * file and at end of pathname, in which case + * we watch for a place to put the new file in + * case it doesn't already exist. + */ + slotstatus = FOUND; + slotfreespace = slotsize = slotneeded = 0; + if ((nameiop == CREATE || nameiop == RENAME) && + (flags & ISLASTCN)) { + slotstatus = NONE; + slotneeded = EXT2_DIR_REC_LEN(cnp->cn_namelen); + /* was + slotneeded = (sizeof(struct direct) - MAXNAMLEN + + cnp->cn_namelen + 3) &~ 3; */ + } + + /* + * If there is cached information on a previous search of + * this directory, pick up where we last left off. + * We cache only lookups as these are the most common + * and have the greatest payoff. Caching CREATE has little + * benefit as it usually must search the entire directory + * to determine that the entry does not exist. Caching the + * location of the last DELETE or RENAME has not reduced + * profiling time and hence has been removed in the interest + * of simplicity. + */ + bmask = VFSTOUFS(vdp->v_mount)->um_mountp->mnt_stat.f_iosize - 1; + if (nameiop != LOOKUP || dp->i_diroff == 0 || + dp->i_diroff > dp->i_size) { + entryoffsetinblock = 0; + dp->i_offset = 0; + numdirpasses = 1; + } else { + dp->i_offset = dp->i_diroff; + if ((entryoffsetinblock = dp->i_offset & bmask) && + (error = UFS_BLKATOFF(vdp, (off_t)dp->i_offset, NULL, &bp))) + return (error); + numdirpasses = 2; + nchstats.ncs_2passes++; + } + prevoff = dp->i_offset; + endsearch = roundup(dp->i_size, DIRBLKSIZ); + enduseful = 0; + +searchloop: + while (dp->i_offset < endsearch) { + /* + * If necessary, get the next directory block. + */ + if ((dp->i_offset & bmask) == 0) { + if (bp != NULL) + brelse(bp); + if ((error = + UFS_BLKATOFF(vdp, (off_t)dp->i_offset, NULL, &bp)) != 0) + return (error); + entryoffsetinblock = 0; + } + /* + * If still looking for a slot, and at a DIRBLKSIZE + * boundary, have to start looking for free space again. + */ + if (slotstatus == NONE && + (entryoffsetinblock & (DIRBLKSIZ - 1)) == 0) { + slotoffset = -1; + slotfreespace = 0; + } + /* + * Get pointer to next entry. + * Full validation checks are slow, so we only check + * enough to insure forward progress through the + * directory. Complete checks can be run by patching + * "dirchk" to be true. + */ + ep = (struct ext2_dir_entry_2 *) + ((char *)bp->b_data + entryoffsetinblock); + if (ep->rec_len == 0 || + (dirchk && ext2_dirbadentry(vdp, ep, entryoffsetinblock))) { + int i; + ufs_dirbad(dp, dp->i_offset, "mangled entry"); + i = DIRBLKSIZ - (entryoffsetinblock & (DIRBLKSIZ - 1)); + dp->i_offset += i; + entryoffsetinblock += i; + continue; + } + + /* + * If an appropriate sized slot has not yet been found, + * check to see if one is available. Also accumulate space + * in the current block so that we can determine if + * compaction is viable. + */ + if (slotstatus != FOUND) { + int size = ep->rec_len; + + if (ep->inode != 0) + size -= EXT2_DIR_REC_LEN(ep->name_len); + if (size > 0) { + if (size >= slotneeded) { + slotstatus = FOUND; + slotoffset = dp->i_offset; + slotsize = ep->rec_len; + } else if (slotstatus == NONE) { + slotfreespace += size; + if (slotoffset == -1) + slotoffset = dp->i_offset; + if (slotfreespace >= slotneeded) { + slotstatus = COMPACT; + slotsize = dp->i_offset + + ep->rec_len - slotoffset; + } + } + } + } + + /* + * Check for a name match. + */ + if (ep->inode) { + namlen = ep->name_len; + if (namlen == cnp->cn_namelen && + !bcmp(cnp->cn_nameptr, ep->name, + (unsigned)namlen)) { + /* + * Save directory entry's inode number and + * reclen in ndp->ni_ufs area, and release + * directory buffer. + */ + dp->i_ino = ep->inode; + dp->i_reclen = ep->rec_len; + brelse(bp); + goto found; + } + } + prevoff = dp->i_offset; + dp->i_offset += ep->rec_len; + entryoffsetinblock += ep->rec_len; + if (ep->inode) + enduseful = dp->i_offset; + } +/* notfound: */ + /* + * If we started in the middle of the directory and failed + * to find our target, we must check the beginning as well. + */ + if (numdirpasses == 2) { + numdirpasses--; + dp->i_offset = 0; + endsearch = dp->i_diroff; + goto searchloop; + } + if (bp != NULL) + brelse(bp); + /* + * If creating, and at end of pathname and current + * directory has not been removed, then can consider + * allowing file to be created. + */ + if ((nameiop == CREATE || nameiop == RENAME) && + (flags & ISLASTCN) && dp->i_nlink != 0) { + /* + * Access for write is interpreted as allowing + * creation of files in the directory. + */ + if ((error = VOP_ACCESS(vdp, VWRITE, cred, cnp->cn_proc)) != 0) + return (error); + /* + * Return an indication of where the new directory + * entry should be put. If we didn't find a slot, + * then set dp->i_count to 0 indicating + * that the new slot belongs at the end of the + * directory. If we found a slot, then the new entry + * can be put in the range from dp->i_offset to + * dp->i_offset + dp->i_count. + */ + if (slotstatus == NONE) { + dp->i_offset = roundup(dp->i_size, DIRBLKSIZ); + dp->i_count = 0; + enduseful = dp->i_offset; + } else { + dp->i_offset = slotoffset; + dp->i_count = slotsize; + if (enduseful < slotoffset + slotsize) + enduseful = slotoffset + slotsize; + } + dp->i_endoff = roundup(enduseful, DIRBLKSIZ); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + /* + * We return with the directory locked, so that + * the parameters we set up above will still be + * valid if we actually decide to do a direnter(). + * We return ni_vp == NULL to indicate that the entry + * does not currently exist; we leave a pointer to + * the (locked) directory inode in ndp->ni_dvp. + * The pathname buffer is saved so that the name + * can be obtained later. + * + * NB - if the directory is unlocked, then this + * information cannot be used. + */ + cnp->cn_flags |= SAVENAME; + if (!lockparent) + VOP_UNLOCK(vdp, 0, p); + return (EJUSTRETURN); + } + /* + * Insert name into cache (as non-existent) if appropriate. + */ + if ((cnp->cn_flags & MAKEENTRY) && nameiop != CREATE) + cache_enter(vdp, *vpp, cnp); + return (ENOENT); + +found: + if (numdirpasses == 2) + nchstats.ncs_pass2++; + /* + * Check that directory length properly reflects presence + * of this entry. + */ + if (entryoffsetinblock + EXT2_DIR_REC_LEN(ep->name_len) + > dp->i_size) { + ufs_dirbad(dp, dp->i_offset, "i_size too small"); + dp->i_size = entryoffsetinblock+EXT2_DIR_REC_LEN(ep->name_len); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + } + + /* + * Found component in pathname. + * If the final component of path name, save information + * in the cache as to where the entry was found. + */ + if ((flags & ISLASTCN) && nameiop == LOOKUP) + dp->i_diroff = dp->i_offset &~ (DIRBLKSIZ - 1); + + /* + * If deleting, and at end of pathname, return + * parameters which can be used to remove file. + * If the wantparent flag isn't set, we return only + * the directory (in ndp->ni_dvp), otherwise we go + * on and lock the inode, being careful with ".". + */ + if (nameiop == DELETE && (flags & ISLASTCN)) { + /* + * Write access to directory required to delete files. + */ + if ((error = VOP_ACCESS(vdp, VWRITE, cred, cnp->cn_proc)) != 0) + return (error); + /* + * Return pointer to current entry in dp->i_offset, + * and distance past previous entry (if there + * is a previous entry in this block) in dp->i_count. + * Save directory inode pointer in ndp->ni_dvp for dirremove(). + */ + if ((dp->i_offset & (DIRBLKSIZ - 1)) == 0) + dp->i_count = 0; + else + dp->i_count = dp->i_offset - prevoff; + if (dp->i_number == dp->i_ino) { + VREF(vdp); + *vpp = vdp; + return (0); + } + if ((error = VFS_VGET(vdp->v_mount, dp->i_ino, &tdp)) != 0) + return (error); + /* + * If directory is "sticky", then user must own + * the directory, or the file in it, else she + * may not delete it (unless she's root). This + * implements append-only directories. + */ + if ((dp->i_mode & ISVTX) && + cred->cr_uid != 0 && + cred->cr_uid != dp->i_uid && + VTOI(tdp)->i_uid != cred->cr_uid) { + vput(tdp); + return (EPERM); + } + *vpp = tdp; + if (!lockparent) + VOP_UNLOCK(vdp, 0, p); + return (0); + } + + /* + * If rewriting (RENAME), return the inode and the + * information required to rewrite the present directory + * Must get inode of directory entry to verify it's a + * regular file, or empty directory. + */ + if (nameiop == RENAME && wantparent && + (flags & ISLASTCN)) { + if ((error = VOP_ACCESS(vdp, VWRITE, cred, cnp->cn_proc)) != 0) + return (error); + /* + * Careful about locking second inode. + * This can only occur if the target is ".". + */ + if (dp->i_number == dp->i_ino) + return (EISDIR); + if ((error = VFS_VGET(vdp->v_mount, dp->i_ino, &tdp)) != 0) + return (error); + *vpp = tdp; + cnp->cn_flags |= SAVENAME; + if (!lockparent) + VOP_UNLOCK(vdp, 0, p); + return (0); + } + + /* + * Step through the translation in the name. We do not `vput' the + * directory because we may need it again if a symbolic link + * is relative to the current directory. Instead we save it + * unlocked as "pdp". We must get the target inode before unlocking + * the directory to insure that the inode will not be removed + * before we get it. We prevent deadlock by always fetching + * inodes from the root, moving down the directory tree. Thus + * when following backward pointers ".." we must unlock the + * parent directory before getting the requested directory. + * There is a potential race condition here if both the current + * and parent directories are removed before the VFS_VGET for the + * inode associated with ".." returns. We hope that this occurs + * infrequently since we cannot avoid this race condition without + * implementing a sophisticated deadlock detection algorithm. + * Note also that this simple deadlock detection scheme will not + * work if the file system has any hard links other than ".." + * that point backwards in the directory structure. + */ + pdp = vdp; + if (flags & ISDOTDOT) { + VOP_UNLOCK(pdp, 0, p); /* race to get the inode */ + if ((error = VFS_VGET(vdp->v_mount, dp->i_ino, &tdp)) != 0) { + vn_lock(pdp, LK_EXCLUSIVE | LK_RETRY, p); + return (error); + } + if (lockparent && (flags & ISLASTCN) && + (error = vn_lock(pdp, LK_EXCLUSIVE, p))) { + vput(tdp); + return (error); + } + *vpp = tdp; + } else if (dp->i_number == dp->i_ino) { + VREF(vdp); /* we want ourself, ie "." */ + *vpp = vdp; + } else { + if ((error = VFS_VGET(vdp->v_mount, dp->i_ino, &tdp)) != 0) + return (error); + if (!lockparent || !(flags & ISLASTCN)) + VOP_UNLOCK(pdp, 0, p); + *vpp = tdp; + } + + /* + * Insert name into cache if appropriate. + */ + if (cnp->cn_flags & MAKEENTRY) + cache_enter(vdp, *vpp, cnp); + return (0); +} + +/* + * Do consistency checking on a directory entry: + * record length must be multiple of 4 + * entry must fit in rest of its DIRBLKSIZ block + * record must be large enough to contain entry + * name is not longer than MAXNAMLEN + * name must be as long as advertised, and null terminated + */ +/* + * changed so that it confirms to ext2_check_dir_entry + */ +static int +ext2_dirbadentry(dp, de, entryoffsetinblock) + struct vnode *dp; + register struct ext2_dir_entry_2 *de; + int entryoffsetinblock; +{ + int DIRBLKSIZ = VTOI(dp)->i_e2fs->s_blocksize; + + char * error_msg = NULL; + + if (de->rec_len < EXT2_DIR_REC_LEN(1)) + error_msg = "rec_len is smaller than minimal"; + else if (de->rec_len % 4 != 0) + error_msg = "rec_len % 4 != 0"; + else if (de->rec_len < EXT2_DIR_REC_LEN(de->name_len)) + error_msg = "reclen is too small for name_len"; + else if (entryoffsetinblock + de->rec_len > DIRBLKSIZ) + error_msg = "directory entry across blocks"; + /* else LATER + if (de->inode > dir->i_sb->u.ext2_sb.s_es->s_inodes_count) + error_msg = "inode out of bounds"; + */ + + if (error_msg != NULL) { + printf("bad directory entry: %s\n", error_msg); + printf("offset=%d, inode=%lu, rec_len=%u, name_len=%u\n", + entryoffsetinblock, (unsigned long)de->inode, + de->rec_len, de->name_len); + } + return error_msg == NULL ? 0 : 1; +} + +/* + * Write a directory entry after a call to namei, using the parameters + * that it left in nameidata. The argument ip is the inode which the new + * directory entry will refer to. Dvp is a pointer to the directory to + * be written, which was left locked by namei. Remaining parameters + * (dp->i_offset, dp->i_count) indicate how the space for the new + * entry is to be obtained. + */ +int +ext2_direnter(ip, dvp, cnp) + struct inode *ip; + struct vnode *dvp; + register struct componentname *cnp; +{ + register struct ext2_dir_entry_2 *ep, *nep; + register struct inode *dp; + struct buf *bp; + struct ext2_dir_entry_2 newdir; + struct iovec aiov; + struct uio auio; + u_int dsize; + int error, loc, newentrysize, spacefree; + char *dirbuf; + int DIRBLKSIZ = ip->i_e2fs->s_blocksize; + + +#if DIAGNOSTIC + if ((cnp->cn_flags & SAVENAME) == 0) + panic("direnter: missing name"); +#endif + dp = VTOI(dvp); + newdir.inode = ip->i_number; + newdir.name_len = cnp->cn_namelen; + if (EXT2_HAS_INCOMPAT_FEATURE(ip->i_e2fs->s_es, + EXT2_FEATURE_INCOMPAT_FILETYPE)) + newdir.file_type = DTTOFT(IFTODT(ip->i_mode)); + else + newdir.file_type = EXT2_FT_UNKNOWN; + bcopy(cnp->cn_nameptr, newdir.name, (unsigned)cnp->cn_namelen + 1); + newentrysize = EXT2_DIR_REC_LEN(newdir.name_len); + if (dp->i_count == 0) { + /* + * If dp->i_count is 0, then namei could find no + * space in the directory. Here, dp->i_offset will + * be on a directory block boundary and we will write the + * new entry into a fresh block. + */ + if (dp->i_offset & (DIRBLKSIZ - 1)) + panic("ext2_direnter: newblk"); + auio.uio_offset = dp->i_offset; + newdir.rec_len = DIRBLKSIZ; + auio.uio_resid = newentrysize; + aiov.iov_len = newentrysize; + aiov.iov_base = (caddr_t)&newdir; + auio.uio_iov = &aiov; + auio.uio_iovcnt = 1; + auio.uio_rw = UIO_WRITE; + auio.uio_segflg = UIO_SYSSPACE; + auio.uio_procp = (struct proc *)0; + error = VOP_WRITE(dvp, &auio, IO_SYNC, cnp->cn_cred); + if (DIRBLKSIZ > + VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_bsize) + /* XXX should grow with balloc() */ + panic("ext2_direnter: frag size"); + else if (!error) { + dp->i_size = roundup(dp->i_size, DIRBLKSIZ); + dp->i_flag |= IN_CHANGE; + } + return (error); + } + + /* + * If dp->i_count is non-zero, then namei found space + * for the new entry in the range dp->i_offset to + * dp->i_offset + dp->i_count in the directory. + * To use this space, we may have to compact the entries located + * there, by copying them together towards the beginning of the + * block, leaving the free space in one usable chunk at the end. + */ + + /* + * Increase size of directory if entry eats into new space. + * This should never push the size past a new multiple of + * DIRBLKSIZE. + * + * N.B. - THIS IS AN ARTIFACT OF 4.2 AND SHOULD NEVER HAPPEN. + */ + if (dp->i_offset + dp->i_count > dp->i_size) + dp->i_size = dp->i_offset + dp->i_count; + /* + * Get the block containing the space for the new directory entry. + */ + if ((error = UFS_BLKATOFF(dvp, (off_t)dp->i_offset, &dirbuf, &bp)) != 0) + return (error); + /* + * Find space for the new entry. In the simple case, the entry at + * offset base will have the space. If it does not, then namei + * arranged that compacting the region dp->i_offset to + * dp->i_offset + dp->i_count would yield the + * space. + */ + ep = (struct ext2_dir_entry_2 *)dirbuf; + dsize = EXT2_DIR_REC_LEN(ep->name_len); + spacefree = ep->rec_len - dsize; + for (loc = ep->rec_len; loc < dp->i_count; ) { + nep = (struct ext2_dir_entry_2 *)(dirbuf + loc); + if (ep->inode) { + /* trim the existing slot */ + ep->rec_len = dsize; + ep = (struct ext2_dir_entry_2 *)((char *)ep + dsize); + } else { + /* overwrite; nothing there; header is ours */ + spacefree += dsize; + } + dsize = EXT2_DIR_REC_LEN(nep->name_len); + spacefree += nep->rec_len - dsize; + loc += nep->rec_len; + bcopy((caddr_t)nep, (caddr_t)ep, dsize); + } + /* + * Update the pointer fields in the previous entry (if any), + * copy in the new entry, and write out the block. + */ + if (ep->inode == 0) { + if (spacefree + dsize < newentrysize) + panic("ext2_direnter: compact1"); + newdir.rec_len = spacefree + dsize; + } else { + if (spacefree < newentrysize) + panic("ext2_direnter: compact2"); + newdir.rec_len = spacefree; + ep->rec_len = dsize; + ep = (struct ext2_dir_entry_2 *)((char *)ep + dsize); + } + bcopy((caddr_t)&newdir, (caddr_t)ep, (u_int)newentrysize); + error = VOP_BWRITE(bp->b_vp, bp); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + if (!error && dp->i_endoff && dp->i_endoff < dp->i_size) + error = UFS_TRUNCATE(dvp, (off_t)dp->i_endoff, IO_SYNC, + cnp->cn_cred, cnp->cn_proc); + return (error); +} + +/* + * Remove a directory entry after a call to namei, using + * the parameters which it left in nameidata. The entry + * dp->i_offset contains the offset into the directory of the + * entry to be eliminated. The dp->i_count field contains the + * size of the previous record in the directory. If this + * is 0, the first entry is being deleted, so we need only + * zero the inode number to mark the entry as free. If the + * entry is not the first in the directory, we must reclaim + * the space of the now empty record by adding the record size + * to the size of the previous entry. + */ +int +ext2_dirremove(dvp, cnp) + struct vnode *dvp; + struct componentname *cnp; +{ + register struct inode *dp; + struct ext2_dir_entry_2 *ep; + struct buf *bp; + int error; + + dp = VTOI(dvp); + if (dp->i_count == 0) { + /* + * First entry in block: set d_ino to zero. + */ + if ((error = + UFS_BLKATOFF(dvp, (off_t)dp->i_offset, (char **)&ep, &bp)) != 0) + return (error); + ep->inode = 0; + error = VOP_BWRITE(bp->b_vp, bp); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + return (error); + } + /* + * Collapse new free space into previous entry. + */ + if ((error = UFS_BLKATOFF(dvp, (off_t)(dp->i_offset - dp->i_count), + (char **)&ep, &bp)) != 0) + return (error); + ep->rec_len += dp->i_reclen; + error = VOP_BWRITE(bp->b_vp, bp); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + return (error); +} + +/* + * Rewrite an existing directory entry to point at the inode + * supplied. The parameters describing the directory entry are + * set up by a call to namei. + */ +int +ext2_dirrewrite(dp, ip, cnp) + struct inode *dp, *ip; + struct componentname *cnp; +{ + struct buf *bp; + struct ext2_dir_entry_2 *ep; + struct vnode *vdp = ITOV(dp); + int error; + + if ((error = UFS_BLKATOFF(vdp, (off_t)dp->i_offset, (char **)&ep, &bp)) != 0) + return (error); + ep->inode = ip->i_number; + if (EXT2_HAS_INCOMPAT_FEATURE(ip->i_e2fs->s_es, + EXT2_FEATURE_INCOMPAT_FILETYPE)) + ep->file_type = DTTOFT(IFTODT(ip->i_mode)); + else + ep->file_type = EXT2_FT_UNKNOWN; + error = VOP_BWRITE(bp->b_vp, bp); + dp->i_flag |= IN_CHANGE | IN_UPDATE; + return (error); +} + +/* + * Check if a directory is empty or not. + * Inode supplied must be locked. + * + * Using a struct dirtemplate here is not precisely + * what we want, but better than using a struct direct. + * + * NB: does not handle corrupted directories. + */ +int +ext2_dirempty(ip, parentino, cred) + register struct inode *ip; + ino_t parentino; + struct ucred *cred; +{ + register off_t off; + struct dirtemplate dbuf; + register struct ext2_dir_entry_2 *dp = (struct ext2_dir_entry_2 *)&dbuf; + int error, count, namlen; + +#define MINDIRSIZ (sizeof (struct dirtemplate) / 2) + + for (off = 0; off < ip->i_size; off += dp->rec_len) { + error = vn_rdwr(UIO_READ, ITOV(ip), (caddr_t)dp, MINDIRSIZ, off, + UIO_SYSSPACE, IO_NODELOCKED, cred, &count, (struct proc *)0); + /* + * Since we read MINDIRSIZ, residual must + * be 0 unless we're at end of file. + */ + if (error || count != 0) + return (0); + /* avoid infinite loops */ + if (dp->rec_len == 0) + return (0); + /* skip empty entries */ + if (dp->inode == 0) + continue; + /* accept only "." and ".." */ + namlen = dp->name_len; + if (namlen > 2) + return (0); + if (dp->name[0] != '.') + return (0); + /* + * At this point namlen must be 1 or 2. + * 1 implies ".", 2 implies ".." if second + * char is also "." + */ + if (namlen == 1) + continue; + if (dp->name[1] == '.' && dp->inode == parentino) + continue; + return (0); + } + return (1); +} + +/* + * Check if source directory is in the path of the target directory. + * Target is supplied locked, source is unlocked. + * The target is always vput before returning. + */ +int +ext2_checkpath(source, target, cred) + struct inode *source, *target; + struct ucred *cred; +{ + struct vnode *vp; + int error, rootino, namlen; + struct dirtemplate dirbuf; + + vp = ITOV(target); + if (target->i_number == source->i_number) { + error = EEXIST; + goto out; + } + rootino = ROOTINO; + error = 0; + if (target->i_number == rootino) + goto out; + + for (;;) { + if (vp->v_type != VDIR) { + error = ENOTDIR; + break; + } + error = vn_rdwr(UIO_READ, vp, (caddr_t)&dirbuf, + sizeof (struct dirtemplate), (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED, cred, (int *)0, (struct proc *)0); + if (error != 0) + break; + namlen = dirbuf.dotdot_type; /* like ufs little-endian */ + if (namlen != 2 || + dirbuf.dotdot_name[0] != '.' || + dirbuf.dotdot_name[1] != '.') { + error = ENOTDIR; + break; + } + if (dirbuf.dotdot_ino == source->i_number) { + error = EINVAL; + break; + } + if (dirbuf.dotdot_ino == rootino) + break; + vput(vp); + if ((error = VFS_VGET(vp->v_mount, dirbuf.dotdot_ino, &vp)) != 0) { + vp = NULL; + break; + } + } + +out: + if (error == ENOTDIR) + printf("checkpath: .. not a directory\n"); + if (vp != NULL) + vput(vp); + return (error); +} + diff --git a/sys/gnu/ext2fs/ext2_readwrite.c b/sys/gnu/ext2fs/ext2_readwrite.c new file mode 100644 index 000000000000..380f4899c258 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_readwrite.c @@ -0,0 +1,308 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ufs_readwrite.c 8.7 (Berkeley) 1/21/94 + * $FreeBSD$ + */ + +#define BLKSIZE(a, b, c) blksize(a, b, c) +#define FS struct ext2_sb_info +#define I_FS i_e2fs +#define READ ext2_read +#define READ_S "ext2_read" +#define WRITE ext2_write +#define WRITE_S "ext2_write" + +/* + * Vnode op for reading. + */ +/* ARGSUSED */ +static int +READ(ap) + struct vop_read_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; +{ + register struct vnode *vp; + register struct inode *ip; + register struct uio *uio; + register FS *fs; + struct buf *bp; + daddr_t lbn, nextlbn; + off_t bytesinfile; + long size, xfersize, blkoffset; + int error, orig_resid; + int seqcount = ap->a_ioflag >> 16; + u_short mode; + + vp = ap->a_vp; + ip = VTOI(vp); + mode = ip->i_mode; + uio = ap->a_uio; + +#ifdef DIAGNOSTIC + if (uio->uio_rw != UIO_READ) + panic("%s: mode", READ_S); + + if (vp->v_type == VLNK) { + if ((int)ip->i_size < vp->v_mount->mnt_maxsymlinklen) + panic("%s: short symlink", READ_S); + } else if (vp->v_type != VREG && vp->v_type != VDIR) + panic("%s: type %d", READ_S, vp->v_type); +#endif + fs = ip->I_FS; +#if 0 + if ((u_quad_t)uio->uio_offset > fs->fs_maxfilesize) + return (EFBIG); +#endif + + orig_resid = uio->uio_resid; + for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) { + if ((bytesinfile = ip->i_size - uio->uio_offset) <= 0) + break; + lbn = lblkno(fs, uio->uio_offset); + nextlbn = lbn + 1; + size = BLKSIZE(fs, ip, lbn); + blkoffset = blkoff(fs, uio->uio_offset); + + xfersize = fs->s_frag_size - blkoffset; + if (uio->uio_resid < xfersize) + xfersize = uio->uio_resid; + if (bytesinfile < xfersize) + xfersize = bytesinfile; + + if (lblktosize(fs, nextlbn) >= ip->i_size) + error = bread(vp, lbn, size, NOCRED, &bp); + else if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0) + error = cluster_read(vp, + ip->i_size, lbn, size, NOCRED, + uio->uio_resid, (ap->a_ioflag >> 16), &bp); + else if (seqcount > 1) { + int nextsize = BLKSIZE(fs, ip, nextlbn); + error = breadn(vp, lbn, + size, &nextlbn, &nextsize, 1, NOCRED, &bp); + } else + error = bread(vp, lbn, size, NOCRED, &bp); + if (error) { + brelse(bp); + bp = NULL; + break; + } + + /* + * We should only get non-zero b_resid when an I/O error + * has occurred, which should cause us to break above. + * However, if the short read did not cause an error, + * then we want to ensure that we do not uiomove bad + * or uninitialized data. + */ + size -= bp->b_resid; + if (size < xfersize) { + if (size == 0) + break; + xfersize = size; + } + error = + uiomove((char *)bp->b_data + blkoffset, (int)xfersize, uio); + if (error) + break; + + bqrelse(bp); + } + if (bp != NULL) + bqrelse(bp); + if (orig_resid > 0 && (error == 0 || uio->uio_resid != orig_resid) && + (vp->v_mount->mnt_flag & MNT_NOATIME) == 0) + ip->i_flag |= IN_ACCESS; + return (error); +} + +/* + * Vnode op for writing. + */ +static int +WRITE(ap) + struct vop_write_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; +{ + register struct vnode *vp; + register struct uio *uio; + register struct inode *ip; + register FS *fs; + struct buf *bp; + struct proc *p; + daddr_t lbn; + off_t osize; + int seqcount; + int blkoffset, error, flags, ioflag, resid, size, xfersize; + + ioflag = ap->a_ioflag; + seqcount = ap->a_ioflag >> 16; + uio = ap->a_uio; + vp = ap->a_vp; + ip = VTOI(vp); + +#ifdef DIAGNOSTIC + if (uio->uio_rw != UIO_WRITE) + panic("%s: mode", WRITE_S); +#endif + + switch (vp->v_type) { + case VREG: + if (ioflag & IO_APPEND) + uio->uio_offset = ip->i_size; + if ((ip->i_flags & APPEND) && uio->uio_offset != ip->i_size) + return (EPERM); + /* FALLTHROUGH */ + case VLNK: + break; + case VDIR: + if ((ioflag & IO_SYNC) == 0) + panic("%s: nonsync dir write", WRITE_S); + break; + default: + panic("%s: type", WRITE_S); + } + + fs = ip->I_FS; +#if 0 + if (uio->uio_offset < 0 || + (u_quad_t)uio->uio_offset + uio->uio_resid > fs->fs_maxfilesize) + return (EFBIG); +#endif + /* + * Maybe this should be above the vnode op call, but so long as + * file servers have no limits, I don't think it matters. + */ + p = uio->uio_procp; + if (vp->v_type == VREG && p && + uio->uio_offset + uio->uio_resid > + p->p_rlimit[RLIMIT_FSIZE].rlim_cur) { + psignal(p, SIGXFSZ); + return (EFBIG); + } + + resid = uio->uio_resid; + osize = ip->i_size; + flags = ioflag & IO_SYNC ? B_SYNC : 0; + + for (error = 0; uio->uio_resid > 0;) { + lbn = lblkno(fs, uio->uio_offset); + blkoffset = blkoff(fs, uio->uio_offset); + xfersize = fs->s_frag_size - blkoffset; + if (uio->uio_resid < xfersize) + xfersize = uio->uio_resid; + + if (uio->uio_offset + xfersize > ip->i_size) + vnode_pager_setsize(vp, uio->uio_offset + xfersize); + + /* + * Avoid a data-consistency race between write() and mmap() + * by ensuring that newly allocated blocks are zerod. The + * race can occur even in the case where the write covers + * the entire block. + */ + flags |= B_CLRBUF; +#if 0 + if (fs->s_frag_size > xfersize) + flags |= B_CLRBUF; + else + flags &= ~B_CLRBUF; +#endif + + error = ext2_balloc(ip, + lbn, blkoffset + xfersize, ap->a_cred, &bp, flags); + if (error) + break; + + if (uio->uio_offset + xfersize > ip->i_size) { + ip->i_size = uio->uio_offset + xfersize; + } + + size = BLKSIZE(fs, ip, lbn) - bp->b_resid; + if (size < xfersize) + xfersize = size; + + error = + uiomove((char *)bp->b_data + blkoffset, (int)xfersize, uio); + if ((ioflag & IO_VMIO) && + (LIST_FIRST(&bp->b_dep) == NULL)) /* in ext2fs? */ + bp->b_flags |= B_RELBUF; + + if (ioflag & IO_SYNC) { + (void)bwrite(bp); + } else if (xfersize + blkoffset == fs->s_frag_size) { + if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERW) == 0) { + bp->b_flags |= B_CLUSTEROK; + cluster_write(bp, ip->i_size, seqcount); + } else { + bawrite(bp); + } + } else { + bp->b_flags |= B_CLUSTEROK; + bdwrite(bp); + } + if (error || xfersize == 0) + break; + ip->i_flag |= IN_CHANGE | IN_UPDATE; + } + /* + * If we successfully wrote any data, and we are not the superuser + * we clear the setuid and setgid bits as a precaution against + * tampering. + */ + if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0) + ip->i_mode &= ~(ISUID | ISGID); + if (error) { + if (ioflag & IO_UNIT) { + (void)UFS_TRUNCATE(vp, osize, + ioflag & IO_SYNC, ap->a_cred, uio->uio_procp); + uio->uio_offset -= resid - uio->uio_resid; + uio->uio_resid = resid; + } + } else if (resid > uio->uio_resid && (ioflag & IO_SYNC)) + error = UFS_UPDATE(vp, 1); + return (error); +} diff --git a/sys/gnu/ext2fs/ext2_subr.c b/sys/gnu/ext2fs/ext2_subr.c new file mode 100644 index 000000000000..6a87b78d47e7 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_subr.c @@ -0,0 +1,128 @@ +/* + * modified for Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ext2_subr.c 8.2 (Berkeley) 9/21/93 + * $FreeBSD$ + */ + +#include <sys/param.h> +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> + +#include <sys/lock.h> +#include <sys/systm.h> +#include <sys/ucred.h> +#include <sys/vnode.h> +#include <gnu/ext2fs/ext2_extern.h> +#include <sys/buf.h> +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> + +#include "opt_ddb.h" + +#ifdef DDB +void ext2_checkoverlap __P((struct buf *, struct inode *)); +#endif + +/* + * Return buffer with the contents of block "offset" from the beginning of + * directory "ip". If "res" is non-zero, fill it in with a pointer to the + * remaining space in the directory. + */ +int +ext2_blkatoff(vp, offset, res, bpp) + struct vnode *vp; + off_t offset; + char **res; + struct buf **bpp; +{ + struct inode *ip; + register struct ext2_sb_info *fs; + struct buf *bp; + daddr_t lbn; + int bsize, error; + + ip = VTOI(vp); + fs = ip->i_e2fs; + lbn = lblkno(fs, offset); + bsize = blksize(fs, ip, lbn); + + *bpp = NULL; + if ((error = bread(vp, lbn, bsize, NOCRED, &bp)) != 0) { + brelse(bp); + return (error); + } + if (res) + *res = (char *)bp->b_data + blkoff(fs, offset); + *bpp = bp; + return (0); +} + +#ifdef DDB +void +ext2_checkoverlap(bp, ip) + struct buf *bp; + struct inode *ip; +{ + register struct buf *ebp, *ep; + register daddr_t start, last; + struct vnode *vp; + + ebp = &buf[nbuf]; + start = bp->b_blkno; + last = start + btodb(bp->b_bcount) - 1; + for (ep = buf; ep < ebp; ep++) { + if (ep == bp || (ep->b_flags & B_INVAL) || + ep->b_vp == NULLVP) + continue; + if (VOP_BMAP(ep->b_vp, (daddr_t)0, &vp, (daddr_t)0, NULL, NULL)) + continue; + if (vp != ip->i_devvp) + continue; + /* look for overlap */ + if (ep->b_bcount == 0 || ep->b_blkno > last || + ep->b_blkno + btodb(ep->b_bcount) <= start) + continue; + vprint("Disk overlap", vp); + (void)printf("\tstart %d, end %d overlap start %d, end %d\n", + start, last, ep->b_blkno, + ep->b_blkno + btodb(ep->b_bcount) - 1); + panic("Disk buffer overlap"); + } +} +#endif /* DDB */ diff --git a/sys/gnu/ext2fs/ext2_vfsops.c b/sys/gnu/ext2fs/ext2_vfsops.c new file mode 100644 index 000000000000..a61afc032c3b --- /dev/null +++ b/sys/gnu/ext2fs/ext2_vfsops.c @@ -0,0 +1,1207 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1989, 1991, 1993, 1994 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ffs_vfsops.c 8.8 (Berkeley) 4/18/94 + * $FreeBSD$ + */ + +#include "opt_quota.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/namei.h> +#include <sys/proc.h> +#include <sys/kernel.h> +#include <sys/vnode.h> +#include <sys/mount.h> +#include <sys/buf.h> +#include <sys/conf.h> +#include <sys/fcntl.h> +#include <sys/disklabel.h> +#include <sys/malloc.h> +#include <sys/stat.h> + +#include <ufs/ufs/quota.h> +#include <ufs/ufs/ufsmount.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufs_extern.h> + +#include <vm/vm_zone.h> + +#include <gnu/ext2fs/fs.h> +#include <gnu/ext2fs/ext2_extern.h> +#include <gnu/ext2fs/ext2_fs.h> +#include <gnu/ext2fs/ext2_fs_sb.h> + +static int ext2_fhtovp __P((struct mount *, struct fid *, struct vnode **)); +static int ext2_flushfiles __P((struct mount *mp, int flags, struct proc *p)); +static int ext2_mount __P((struct mount *, + char *, caddr_t, struct nameidata *, struct proc *)); +static int ext2_mountfs __P((struct vnode *, struct mount *, struct proc *)); +static int ext2_reload __P((struct mount *mountp, struct ucred *cred, + struct proc *p)); +static int ext2_sbupdate __P((struct ufsmount *, int)); +static int ext2_statfs __P((struct mount *, struct statfs *, struct proc *)); +static int ext2_sync __P((struct mount *, int, struct ucred *, struct proc *)); +static int ext2_unmount __P((struct mount *, int, struct proc *)); +static int ext2_vget __P((struct mount *, ino_t, struct vnode **)); +static int ext2_vptofh __P((struct vnode *, struct fid *)); + +static MALLOC_DEFINE(M_EXT2NODE, "EXT2 node", "EXT2 vnode private part"); + +static struct vfsops ext2fs_vfsops = { + ext2_mount, + ufs_start, /* empty function */ + ext2_unmount, + ufs_root, /* root inode via vget */ + ufs_quotactl, /* does operations associated with quotas */ + ext2_statfs, + ext2_sync, + ext2_vget, + ext2_fhtovp, + ufs_check_export, + ext2_vptofh, + ext2_init, + vfs_stduninit, + vfs_stdextattrctl, +}; + +VFS_SET(ext2fs_vfsops, ext2fs, 0); +#define bsd_malloc malloc +#define bsd_free free + +static int ext2fs_inode_hash_lock; + +static int ext2_check_sb_compat __P((struct ext2_super_block *es, + dev_t dev, int ronly)); +static int compute_sb_data __P((struct vnode * devvp, + struct ext2_super_block * es, + struct ext2_sb_info * fs)); + +#ifdef notyet +static int ext2_mountroot __P((void)); + +/* + * Called by main() when ext2fs is going to be mounted as root. + * + * Name is updated by mount(8) after booting. + */ +#define ROOTNAME "root_device" + +static int +ext2_mountroot() +{ + register struct ext2_sb_info *fs; + register struct mount *mp; + struct proc *p = curproc; + struct ufsmount *ump; + u_int size; + int error; + + if ((error = bdevvp(rootdev, &rootvp))) { + printf("ext2_mountroot: can't find rootvp\n"); + return (error); + } + mp = bsd_malloc((u_long)sizeof(struct mount), M_MOUNT, M_WAITOK); + bzero((char *)mp, (u_long)sizeof(struct mount)); + TAILQ_INIT(&mp->mnt_nvnodelist); + TAILQ_INIT(&mp->mnt_reservedvnlist); + mp->mnt_op = &ext2fs_vfsops; + mp->mnt_flag = MNT_RDONLY; + if (error = ext2_mountfs(rootvp, mp, p)) { + bsd_free(mp, M_MOUNT); + return (error); + } + if (error = vfs_lock(mp)) { + (void)ext2_unmount(mp, 0, p); + bsd_free(mp, M_MOUNT); + return (error); + } + TAILQ_INSERT_HEAD(&mountlist, mp, mnt_list); + mp->mnt_flag |= MNT_ROOTFS; + mp->mnt_vnodecovered = NULLVP; + ump = VFSTOUFS(mp); + fs = ump->um_e2fs; + bzero(fs->fs_fsmnt, sizeof(fs->fs_fsmnt)); + fs->fs_fsmnt[0] = '/'; + bcopy((caddr_t)fs->fs_fsmnt, (caddr_t)mp->mnt_stat.f_mntonname, + MNAMELEN); + (void) copystr(ROOTNAME, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, + &size); + bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size); + (void)ext2_statfs(mp, &mp->mnt_stat, p); + vfs_unlock(mp); + inittodr(fs->s_es->s_wtime); /* this helps to set the time */ + return (0); +} +#endif + +/* + * VFS Operations. + * + * mount system call + */ +static int +ext2_mount(mp, path, data, ndp, p) + register struct mount *mp; + char *path; + caddr_t data; /* this is actually a (struct ufs_args *) */ + struct nameidata *ndp; + struct proc *p; +{ + struct vnode *devvp; + struct ufs_args args; + struct ufsmount *ump = 0; + register struct ext2_sb_info *fs; + size_t size; + int error, flags; + mode_t accessmode; + + if ((error = copyin(data, (caddr_t)&args, sizeof (struct ufs_args))) != 0) + return (error); + /* + * If updating, check whether changing from read-only to + * read/write; if there is no device name, that's all we do. + */ + if (mp->mnt_flag & MNT_UPDATE) { + ump = VFSTOUFS(mp); + fs = ump->um_e2fs; + error = 0; + if (fs->s_rd_only == 0 && (mp->mnt_flag & MNT_RDONLY)) { + flags = WRITECLOSE; + if (mp->mnt_flag & MNT_FORCE) + flags |= FORCECLOSE; + if (vfs_busy(mp, LK_NOWAIT, 0, p)) + return (EBUSY); + error = ext2_flushfiles(mp, flags, p); + vfs_unbusy(mp, p); + if (!error && fs->s_wasvalid) { + fs->s_es->s_state |= EXT2_VALID_FS; + ext2_sbupdate(ump, MNT_WAIT); + } + fs->s_rd_only = 1; + } + if (!error && (mp->mnt_flag & MNT_RELOAD)) + error = ext2_reload(mp, ndp->ni_cnd.cn_cred, p); + if (error) + return (error); + devvp = ump->um_devvp; + if (ext2_check_sb_compat(fs->s_es, devvp->v_rdev, + (mp->mnt_kern_flag & MNTK_WANTRDWR) == 0) != 0) + return (EPERM); + if (fs->s_rd_only && (mp->mnt_kern_flag & MNTK_WANTRDWR)) { + /* + * If upgrade to read-write by non-root, then verify + * that user has necessary permissions on the device. + */ + if (p->p_ucred->cr_uid != 0) { + vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY, p); + if ((error = VOP_ACCESS(devvp, VREAD | VWRITE, + p->p_ucred, p)) != 0) { + VOP_UNLOCK(devvp, 0, p); + return (error); + } + VOP_UNLOCK(devvp, 0, p); + } + + if ((fs->s_es->s_state & EXT2_VALID_FS) == 0 || + (fs->s_es->s_state & EXT2_ERROR_FS)) { + if (mp->mnt_flag & MNT_FORCE) { + printf( +"WARNING: %s was not properly dismounted\n", + fs->fs_fsmnt); + } else { + printf( +"WARNING: R/W mount of %s denied. Filesystem is not clean - run fsck\n", + fs->fs_fsmnt); + return (EPERM); + } + } + fs->s_es->s_state &= ~EXT2_VALID_FS; + ext2_sbupdate(ump, MNT_WAIT); + fs->s_rd_only = 0; + } + if (args.fspec == 0) { + /* + * Process export requests. + */ + return (vfs_export(mp, &ump->um_export, &args.export)); + } + } + /* + * Not an update, or updating the name: look up the name + * and verify that it refers to a sensible block device. + */ + NDINIT(ndp, LOOKUP, FOLLOW, UIO_USERSPACE, args.fspec, p); + if ((error = namei(ndp)) != 0) + return (error); + NDFREE(ndp, NDF_ONLY_PNBUF); + devvp = ndp->ni_vp; + + if (!vn_isdisk(devvp, &error)) { + vrele(devvp); + return (error); + } + + /* + * If mount by non-root, then verify that user has necessary + * permissions on the device. + */ + if (p->p_ucred->cr_uid != 0) { + accessmode = VREAD; + if ((mp->mnt_flag & MNT_RDONLY) == 0) + accessmode |= VWRITE; + vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY, p); + if ((error = VOP_ACCESS(devvp, accessmode, p->p_ucred, p)) != 0) { + vput(devvp); + return (error); + } + VOP_UNLOCK(devvp, 0, p); + } + + if ((mp->mnt_flag & MNT_UPDATE) == 0) { + error = ext2_mountfs(devvp, mp, p); + } else { + if (devvp != ump->um_devvp) + error = EINVAL; /* needs translation */ + else + vrele(devvp); + } + if (error) { + vrele(devvp); + return (error); + } + ump = VFSTOUFS(mp); + fs = ump->um_e2fs; + (void) copyinstr(path, fs->fs_fsmnt, sizeof(fs->fs_fsmnt) - 1, &size); + bzero(fs->fs_fsmnt + size, sizeof(fs->fs_fsmnt) - size); + bcopy((caddr_t)fs->fs_fsmnt, (caddr_t)mp->mnt_stat.f_mntonname, + MNAMELEN); + (void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, + &size); + bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size); + (void)ext2_statfs(mp, &mp->mnt_stat, p); + return (0); +} + +/* + * checks that the data in the descriptor blocks make sense + * this is taken from ext2/super.c + */ +static int ext2_check_descriptors (struct ext2_sb_info * sb) +{ + int i; + int desc_block = 0; + unsigned long block = sb->s_es->s_first_data_block; + struct ext2_group_desc * gdp = NULL; + + /* ext2_debug ("Checking group descriptors"); */ + + for (i = 0; i < sb->s_groups_count; i++) + { + /* examine next descriptor block */ + if ((i % EXT2_DESC_PER_BLOCK(sb)) == 0) + gdp = (struct ext2_group_desc *) + sb->s_group_desc[desc_block++]->b_data; + if (gdp->bg_block_bitmap < block || + gdp->bg_block_bitmap >= block + EXT2_BLOCKS_PER_GROUP(sb)) + { + printf ("ext2_check_descriptors: " + "Block bitmap for group %d" + " not in group (block %lu)!\n", + i, (unsigned long) gdp->bg_block_bitmap); + return 0; + } + if (gdp->bg_inode_bitmap < block || + gdp->bg_inode_bitmap >= block + EXT2_BLOCKS_PER_GROUP(sb)) + { + printf ("ext2_check_descriptors: " + "Inode bitmap for group %d" + " not in group (block %lu)!\n", + i, (unsigned long) gdp->bg_inode_bitmap); + return 0; + } + if (gdp->bg_inode_table < block || + gdp->bg_inode_table + sb->s_itb_per_group >= + block + EXT2_BLOCKS_PER_GROUP(sb)) + { + printf ("ext2_check_descriptors: " + "Inode table for group %d" + " not in group (block %lu)!\n", + i, (unsigned long) gdp->bg_inode_table); + return 0; + } + block += EXT2_BLOCKS_PER_GROUP(sb); + gdp++; + } + return 1; +} + +static int +ext2_check_sb_compat(es, dev, ronly) + struct ext2_super_block *es; + dev_t dev; + int ronly; +{ + + if (es->s_magic != EXT2_SUPER_MAGIC) { + printf("ext2fs: %s: wrong magic number %#x (expected %#x)\n", + devtoname(dev), es->s_magic, EXT2_SUPER_MAGIC); + return (1); + } + if (es->s_rev_level > EXT2_GOOD_OLD_REV) { + if (es->s_feature_incompat & ~EXT2_FEATURE_INCOMPAT_SUPP) { + printf( +"WARNING: mount of %s denied due to unsupported optional features\n", + devtoname(dev)); + return (1); + } + if (!ronly && + (es->s_feature_ro_compat & ~EXT2_FEATURE_RO_COMPAT_SUPP)) { + printf( +"WARNING: R/W mount of %s denied due to unsupported optional features\n", + devtoname(dev)); + return (1); + } + } + return (0); +} + +/* + * this computes the fields of the ext2_sb_info structure from the + * data in the ext2_super_block structure read in + */ +static int compute_sb_data(devvp, es, fs) + struct vnode * devvp; + struct ext2_super_block * es; + struct ext2_sb_info * fs; +{ + int db_count, error; + int i, j; + int logic_sb_block = 1; /* XXX for now */ + +#if 1 +#define V(v) +#else +#define V(v) printf(#v"= %d\n", fs->v); +#endif + + fs->s_blocksize = EXT2_MIN_BLOCK_SIZE << es->s_log_block_size; + V(s_blocksize) + fs->s_bshift = EXT2_MIN_BLOCK_LOG_SIZE + es->s_log_block_size; + V(s_bshift) + fs->s_fsbtodb = es->s_log_block_size + 1; + V(s_fsbtodb) + fs->s_qbmask = fs->s_blocksize - 1; + V(s_bmask) + fs->s_blocksize_bits = EXT2_BLOCK_SIZE_BITS(es); + V(s_blocksize_bits) + fs->s_frag_size = EXT2_MIN_FRAG_SIZE << es->s_log_frag_size; + V(s_frag_size) + if (fs->s_frag_size) + fs->s_frags_per_block = fs->s_blocksize / fs->s_frag_size; + V(s_frags_per_block) + fs->s_blocks_per_group = es->s_blocks_per_group; + V(s_blocks_per_group) + fs->s_frags_per_group = es->s_frags_per_group; + V(s_frags_per_group) + fs->s_inodes_per_group = es->s_inodes_per_group; + V(s_inodes_per_group) + fs->s_inodes_per_block = fs->s_blocksize / EXT2_INODE_SIZE; + V(s_inodes_per_block) + fs->s_itb_per_group = fs->s_inodes_per_group /fs->s_inodes_per_block; + V(s_itb_per_group) + fs->s_desc_per_block = fs->s_blocksize / sizeof (struct ext2_group_desc); + V(s_desc_per_block) + /* s_resuid / s_resgid ? */ + fs->s_groups_count = (es->s_blocks_count - + es->s_first_data_block + + EXT2_BLOCKS_PER_GROUP(fs) - 1) / + EXT2_BLOCKS_PER_GROUP(fs); + V(s_groups_count) + db_count = (fs->s_groups_count + EXT2_DESC_PER_BLOCK(fs) - 1) / + EXT2_DESC_PER_BLOCK(fs); + fs->s_db_per_group = db_count; + V(s_db_per_group) + + fs->s_group_desc = bsd_malloc(db_count * sizeof (struct buf *), + M_UFSMNT, M_WAITOK); + + /* adjust logic_sb_block */ + if(fs->s_blocksize > SBSIZE) + /* Godmar thinks: if the blocksize is greater than 1024, then + the superblock is logically part of block zero. + */ + logic_sb_block = 0; + + for (i = 0; i < db_count; i++) { + error = bread(devvp , fsbtodb(fs, logic_sb_block + i + 1), + fs->s_blocksize, NOCRED, &fs->s_group_desc[i]); + if(error) { + for (j = 0; j < i; j++) + brelse(fs->s_group_desc[j]); + bsd_free(fs->s_group_desc, M_UFSMNT); + printf("EXT2-fs: unable to read group descriptors (%d)\n", error); + return EIO; + } + /* Set the B_LOCKED flag on the buffer, then brelse() it */ + LCK_BUF(fs->s_group_desc[i]) + } + if(!ext2_check_descriptors(fs)) { + for (j = 0; j < db_count; j++) + ULCK_BUF(fs->s_group_desc[j]) + bsd_free(fs->s_group_desc, M_UFSMNT); + printf("EXT2-fs: (ext2_check_descriptors failure) " + "unable to read group descriptors\n"); + return EIO; + } + + for (i = 0; i < EXT2_MAX_GROUP_LOADED; i++) { + fs->s_inode_bitmap_number[i] = 0; + fs->s_inode_bitmap[i] = NULL; + fs->s_block_bitmap_number[i] = 0; + fs->s_block_bitmap[i] = NULL; + } + fs->s_loaded_inode_bitmaps = 0; + fs->s_loaded_block_bitmaps = 0; + return 0; +} + +/* + * Reload all incore data for a filesystem (used after running fsck on + * the root filesystem and finding things to fix). The filesystem must + * be mounted read-only. + * + * Things to do to update the mount: + * 1) invalidate all cached meta-data. + * 2) re-read superblock from disk. + * 3) re-read summary information from disk. + * 4) invalidate all inactive vnodes. + * 5) invalidate all cached file data. + * 6) re-read inode data for all active vnodes. + */ +static int +ext2_reload(mountp, cred, p) + register struct mount *mountp; + struct ucred *cred; + struct proc *p; +{ + register struct vnode *vp, *nvp, *devvp; + struct inode *ip; + struct buf *bp; + struct ext2_super_block * es; + struct ext2_sb_info *fs; + int error; + + if ((mountp->mnt_flag & MNT_RDONLY) == 0) + return (EINVAL); + /* + * Step 1: invalidate all cached meta-data. + */ + devvp = VFSTOUFS(mountp)->um_devvp; + if (vinvalbuf(devvp, 0, cred, p, 0, 0)) + panic("ext2_reload: dirty1"); + /* + * Step 2: re-read superblock from disk. + * constants have been adjusted for ext2 + */ + if ((error = bread(devvp, SBLOCK, SBSIZE, NOCRED, &bp)) != 0) + return (error); + es = (struct ext2_super_block *)bp->b_data; + if (ext2_check_sb_compat(es, devvp->v_rdev, 0) != 0) { + brelse(bp); + return (EIO); /* XXX needs translation */ + } + fs = VFSTOUFS(mountp)->um_e2fs; + bcopy(bp->b_data, fs->s_es, sizeof(struct ext2_super_block)); + + if((error = compute_sb_data(devvp, es, fs)) != 0) { + brelse(bp); + return error; + } +#ifdef UNKLAR + if (fs->fs_sbsize < SBSIZE) + bp->b_flags |= B_INVAL; +#endif + brelse(bp); + +loop: + simple_lock(&mntvnode_slock); + for (vp = TAILQ_FIRST(&mountp->mnt_nvnodelist); vp != NULL; vp = nvp) { + if (vp->v_mount != mountp) { + simple_unlock(&mntvnode_slock); + goto loop; + } + nvp = TAILQ_NEXT(vp, v_nmntvnodes); + /* + * Step 4: invalidate all inactive vnodes. + */ + if (vrecycle(vp, &mntvnode_slock, p)) + goto loop; + /* + * Step 5: invalidate all cached file data. + */ + simple_lock(&vp->v_interlock); + simple_unlock(&mntvnode_slock); + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, p)) { + goto loop; + } + if (vinvalbuf(vp, 0, cred, p, 0, 0)) + panic("ext2_reload: dirty2"); + /* + * Step 6: re-read inode data for all active vnodes. + */ + ip = VTOI(vp); + error = + bread(devvp, fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), + (int)fs->s_blocksize, NOCRED, &bp); + if (error) { + vput(vp); + return (error); + } + ext2_ei2di((struct ext2_inode *) ((char *)bp->b_data + + EXT2_INODE_SIZE * ino_to_fsbo(fs, ip->i_number)), + &ip->i_din); + brelse(bp); + vput(vp); + simple_lock(&mntvnode_slock); + } + simple_unlock(&mntvnode_slock); + return (0); +} + +/* + * Common code for mount and mountroot + */ +static int +ext2_mountfs(devvp, mp, p) + register struct vnode *devvp; + struct mount *mp; + struct proc *p; +{ + register struct ufsmount *ump; + struct buf *bp; + register struct ext2_sb_info *fs; + struct ext2_super_block * es; + dev_t dev = devvp->v_rdev; + struct partinfo dpart; + int havepart = 0; + int error, i, size; + int ronly; + + /* + * Disallow multiple mounts of the same device. + * Disallow mounting of a device that is currently in use + * (except for root, which might share swap device for miniroot). + * Flush out any old buffers remaining from a previous use. + */ + if ((error = vfs_mountedon(devvp)) != 0) + return (error); + if (vcount(devvp) > 1 && devvp != rootvp) + return (EBUSY); + if ((error = vinvalbuf(devvp, V_SAVE, p->p_ucred, p, 0, 0)) != 0) + return (error); +#ifdef READONLY +/* turn on this to force it to be read-only */ + mp->mnt_flag |= MNT_RDONLY; +#endif + + ronly = (mp->mnt_flag & MNT_RDONLY) != 0; + vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY, p); + error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, FSCRED, p); + VOP_UNLOCK(devvp, 0, p); + if (error) + return (error); + if (devvp->v_rdev->si_iosize_max != 0) + mp->mnt_iosize_max = devvp->v_rdev->si_iosize_max; + if (mp->mnt_iosize_max > MAXPHYS) + mp->mnt_iosize_max = MAXPHYS; + if (VOP_IOCTL(devvp, DIOCGPART, (caddr_t)&dpart, FREAD, NOCRED, p) != 0) + size = DEV_BSIZE; + else { + havepart = 1; + size = dpart.disklab->d_secsize; + } + + bp = NULL; + ump = NULL; + if ((error = bread(devvp, SBLOCK, SBSIZE, NOCRED, &bp)) != 0) + goto out; + es = (struct ext2_super_block *)bp->b_data; + if (ext2_check_sb_compat(es, dev, ronly) != 0) { + error = EINVAL; /* XXX needs translation */ + goto out; + } + if ((es->s_state & EXT2_VALID_FS) == 0 || + (es->s_state & EXT2_ERROR_FS)) { + if (ronly || (mp->mnt_flag & MNT_FORCE)) { + printf( +"WARNING: Filesystem was not properly dismounted\n"); + } else { + printf( +"WARNING: R/W mount denied. Filesystem is not clean - run fsck\n"); + error = EPERM; + goto out; + } + } + ump = bsd_malloc(sizeof *ump, M_UFSMNT, M_WAITOK); + bzero((caddr_t)ump, sizeof *ump); + ump->um_malloctype = M_EXT2NODE; + ump->um_blkatoff = ext2_blkatoff; + ump->um_truncate = ext2_truncate; + ump->um_update = ext2_update; + ump->um_valloc = ext2_valloc; + ump->um_vfree = ext2_vfree; + /* I don't know whether this is the right strategy. Note that + we dynamically allocate both a ext2_sb_info and a ext2_super_block + while Linux keeps the super block in a locked buffer + */ + ump->um_e2fs = bsd_malloc(sizeof(struct ext2_sb_info), + M_UFSMNT, M_WAITOK); + ump->um_e2fs->s_es = bsd_malloc(sizeof(struct ext2_super_block), + M_UFSMNT, M_WAITOK); + bcopy(es, ump->um_e2fs->s_es, (u_int)sizeof(struct ext2_super_block)); + if ((error = compute_sb_data(devvp, ump->um_e2fs->s_es, ump->um_e2fs))) + goto out; + /* + * We don't free the group descriptors allocated by compute_sb_data() + * until ext2_unmount(). This is OK since the mount will succeed. + */ + brelse(bp); + bp = NULL; + fs = ump->um_e2fs; + fs->s_rd_only = ronly; /* ronly is set according to mnt_flags */ + /* if the fs is not mounted read-only, make sure the super block is + always written back on a sync() + */ + fs->s_wasvalid = fs->s_es->s_state & EXT2_VALID_FS ? 1 : 0; + if (ronly == 0) { + fs->s_dirt = 1; /* mark it modified */ + fs->s_es->s_state &= ~EXT2_VALID_FS; /* set fs invalid */ + } + mp->mnt_data = (qaddr_t)ump; + mp->mnt_stat.f_fsid.val[0] = dev2udev(dev); + mp->mnt_stat.f_fsid.val[1] = mp->mnt_vfc->vfc_typenum; + mp->mnt_maxsymlinklen = EXT2_MAXSYMLINKLEN; + mp->mnt_flag |= MNT_LOCAL; + ump->um_mountp = mp; + ump->um_dev = dev; + ump->um_devvp = devvp; + /* setting those two parameters allows us to use + ufs_bmap w/o changse ! + */ + ump->um_nindir = EXT2_ADDR_PER_BLOCK(fs); + ump->um_bptrtodb = fs->s_es->s_log_block_size + 1; + ump->um_seqinc = EXT2_FRAGS_PER_BLOCK(fs); + for (i = 0; i < MAXQUOTAS; i++) + ump->um_quotas[i] = NULLVP; + devvp->v_specmountpoint = mp; + if (ronly == 0) + ext2_sbupdate(ump, MNT_WAIT); + return (0); +out: + if (bp) + brelse(bp); + (void)VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, NOCRED, p); + if (ump) { + bsd_free(ump->um_e2fs->s_es, M_UFSMNT); + bsd_free(ump->um_e2fs, M_UFSMNT); + bsd_free(ump, M_UFSMNT); + mp->mnt_data = (qaddr_t)0; + } + return (error); +} + +/* + * unmount system call + */ +static int +ext2_unmount(mp, mntflags, p) + struct mount *mp; + int mntflags; + struct proc *p; +{ + register struct ufsmount *ump; + register struct ext2_sb_info *fs; + int error, flags, ronly, i; + + flags = 0; + if (mntflags & MNT_FORCE) { + if (mp->mnt_flag & MNT_ROOTFS) + return (EINVAL); + flags |= FORCECLOSE; + } + if ((error = ext2_flushfiles(mp, flags, p)) != 0) + return (error); + ump = VFSTOUFS(mp); + fs = ump->um_e2fs; + ronly = fs->s_rd_only; + if (ronly == 0) { + if (fs->s_wasvalid) + fs->s_es->s_state |= EXT2_VALID_FS; + ext2_sbupdate(ump, MNT_WAIT); + } + + /* release buffers containing group descriptors */ + for(i = 0; i < fs->s_db_per_group; i++) + ULCK_BUF(fs->s_group_desc[i]) + bsd_free(fs->s_group_desc, M_UFSMNT); + + /* release cached inode/block bitmaps */ + for (i = 0; i < EXT2_MAX_GROUP_LOADED; i++) + if (fs->s_inode_bitmap[i]) + ULCK_BUF(fs->s_inode_bitmap[i]) + + for (i = 0; i < EXT2_MAX_GROUP_LOADED; i++) + if (fs->s_block_bitmap[i]) + ULCK_BUF(fs->s_block_bitmap[i]) + + ump->um_devvp->v_specmountpoint = NULL; + error = VOP_CLOSE(ump->um_devvp, ronly ? FREAD : FREAD|FWRITE, + NOCRED, p); + vrele(ump->um_devvp); + bsd_free(fs->s_es, M_UFSMNT); + bsd_free(fs, M_UFSMNT); + bsd_free(ump, M_UFSMNT); + mp->mnt_data = (qaddr_t)0; + mp->mnt_flag &= ~MNT_LOCAL; + return (error); +} + +/* + * Flush out all the files in a filesystem. + */ +static int +ext2_flushfiles(mp, flags, p) + register struct mount *mp; + int flags; + struct proc *p; +{ + register struct ufsmount *ump; + int error; +#if QUOTA + int i; +#endif + + ump = VFSTOUFS(mp); +#if QUOTA + if (mp->mnt_flag & MNT_QUOTA) { + if ((error = vflush(mp, 0, SKIPSYSTEM|flags)) != 0) + return (error); + for (i = 0; i < MAXQUOTAS; i++) { + if (ump->um_quotas[i] == NULLVP) + continue; + quotaoff(p, mp, i); + } + /* + * Here we fall through to vflush again to ensure + * that we have gotten rid of all the system vnodes. + */ + } +#endif + error = vflush(mp, 0, flags); + return (error); +} + +/* + * Get file system statistics. + * taken from ext2/super.c ext2_statfs + */ +static int +ext2_statfs(mp, sbp, p) + struct mount *mp; + register struct statfs *sbp; + struct proc *p; +{ + unsigned long overhead; + register struct ufsmount *ump; + register struct ext2_sb_info *fs; + register struct ext2_super_block *es; + int i, nsb; + + ump = VFSTOUFS(mp); + fs = ump->um_e2fs; + es = fs->s_es; + + if (es->s_magic != EXT2_SUPER_MAGIC) + panic("ext2_statfs - magic number spoiled"); + + /* + * Compute the overhead (FS structures) + */ + if (es->s_feature_ro_compat & EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER) { + nsb = 0; + for (i = 0 ; i < fs->s_groups_count; i++) + if (ext2_group_sparse(i)) + nsb++; + } else + nsb = fs->s_groups_count; + overhead = es->s_first_data_block + + /* Superblocks and block group descriptors: */ + nsb * (1 + fs->s_db_per_group) + + /* Inode bitmap, block bitmap, and inode table: */ + fs->s_groups_count * (1 + 1 + fs->s_itb_per_group); + + sbp->f_bsize = EXT2_FRAG_SIZE(fs); + sbp->f_iosize = EXT2_BLOCK_SIZE(fs); + sbp->f_blocks = es->s_blocks_count - overhead; + sbp->f_bfree = es->s_free_blocks_count; + sbp->f_bavail = sbp->f_bfree - es->s_r_blocks_count; + sbp->f_files = es->s_inodes_count; + sbp->f_ffree = es->s_free_inodes_count; + if (sbp != &mp->mnt_stat) { + sbp->f_type = mp->mnt_vfc->vfc_typenum; + bcopy((caddr_t)mp->mnt_stat.f_mntonname, + (caddr_t)&sbp->f_mntonname[0], MNAMELEN); + bcopy((caddr_t)mp->mnt_stat.f_mntfromname, + (caddr_t)&sbp->f_mntfromname[0], MNAMELEN); + } + return (0); +} + +/* + * Go through the disk queues to initiate sandbagged IO; + * go through the inodes to write those that have been modified; + * initiate the writing of the super block if it has been modified. + * + * Note: we are always called with the filesystem marked `MPBUSY'. + */ +static int +ext2_sync(mp, waitfor, cred, p) + struct mount *mp; + int waitfor; + struct ucred *cred; + struct proc *p; +{ + struct vnode *nvp, *vp; + struct inode *ip; + struct ufsmount *ump = VFSTOUFS(mp); + struct ext2_sb_info *fs; + int error, allerror = 0; + + fs = ump->um_e2fs; + if (fs->s_dirt != 0 && fs->s_rd_only != 0) { /* XXX */ + printf("fs = %s\n", fs->fs_fsmnt); + panic("ext2_sync: rofs mod"); + } + /* + * Write back each (modified) inode. + */ + simple_lock(&mntvnode_slock); +loop: + for (vp = TAILQ_FIRST(&mp->mnt_nvnodelist); vp != NULL; vp = nvp) { + /* + * If the vnode that we are about to sync is no longer + * associated with this mount point, start over. + */ + if (vp->v_mount != mp) + goto loop; + simple_lock(&vp->v_interlock); + nvp = TAILQ_NEXT(vp, v_nmntvnodes); + ip = VTOI(vp); + if (vp->v_type == VNON || + ((ip->i_flag & + (IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) == 0 && + (TAILQ_EMPTY(&vp->v_dirtyblkhd) || waitfor == MNT_LAZY))) { + simple_unlock(&vp->v_interlock); + continue; + } + simple_unlock(&mntvnode_slock); + error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, p); + if (error) { + simple_lock(&mntvnode_slock); + if (error == ENOENT) + goto loop; + continue; + } + if ((error = VOP_FSYNC(vp, cred, waitfor, p)) != 0) + allerror = error; + VOP_UNLOCK(vp, 0, p); + vrele(vp); + simple_lock(&mntvnode_slock); + } + simple_unlock(&mntvnode_slock); + /* + * Force stale file system control information to be flushed. + */ + if (waitfor != MNT_LAZY) { + vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY, p); + if ((error = VOP_FSYNC(ump->um_devvp, cred, waitfor, p)) != 0) + allerror = error; + VOP_UNLOCK(ump->um_devvp, 0, p); + } +#if QUOTA + qsync(mp); +#endif + /* + * Write back modified superblock. + */ + if (fs->s_dirt != 0) { + fs->s_dirt = 0; + fs->s_es->s_wtime = time_second; + if ((error = ext2_sbupdate(ump, waitfor)) != 0) + allerror = error; + } + return (allerror); +} + +/* + * Look up a EXT2FS dinode number to find its incore vnode, otherwise read it + * in from disk. If it is in core, wait for the lock bit to clear, then + * return the inode locked. Detection and handling of mount points must be + * done by the calling routine. + */ +static int +ext2_vget(mp, ino, vpp) + struct mount *mp; + ino_t ino; + struct vnode **vpp; +{ + register struct ext2_sb_info *fs; + register struct inode *ip; + struct ufsmount *ump; + struct buf *bp; + struct vnode *vp; + dev_t dev; + int i, error; + int used_blocks; + + ump = VFSTOUFS(mp); + dev = ump->um_dev; +restart: + if ((*vpp = ufs_ihashget(dev, ino)) != NULL) + return (0); + + /* + * Lock out the creation of new entries in the FFS hash table in + * case getnewvnode() or MALLOC() blocks, otherwise a duplicate + * may occur! + */ + if (ext2fs_inode_hash_lock) { + while (ext2fs_inode_hash_lock) { + ext2fs_inode_hash_lock = -1; + tsleep(&ext2fs_inode_hash_lock, PVM, "e2vget", 0); + } + goto restart; + } + ext2fs_inode_hash_lock = 1; + + /* + * If this MALLOC() is performed after the getnewvnode() + * it might block, leaving a vnode with a NULL v_data to be + * found by ext2_sync() if a sync happens to fire right then, + * which will cause a panic because ext2_sync() blindly + * dereferences vp->v_data (as well it should). + */ + MALLOC(ip, struct inode *, sizeof(struct inode), M_EXT2NODE, M_WAITOK); + + /* Allocate a new vnode/inode. */ + if ((error = getnewvnode(VT_UFS, mp, ext2_vnodeop_p, &vp)) != 0) { + if (ext2fs_inode_hash_lock < 0) + wakeup(&ext2fs_inode_hash_lock); + ext2fs_inode_hash_lock = 0; + *vpp = NULL; + FREE(ip, M_EXT2NODE); + return (error); + } + bzero((caddr_t)ip, sizeof(struct inode)); + lockinit(&ip->i_lock, PINOD, "ext2in", 0, 0); + vp->v_data = ip; + ip->i_vnode = vp; + ip->i_e2fs = fs = ump->um_e2fs; + ip->i_dev = dev; + ip->i_number = ino; +#if QUOTA + for (i = 0; i < MAXQUOTAS; i++) + ip->i_dquot[i] = NODQUOT; +#endif + /* + * Put it onto its hash chain and lock it so that other requests for + * this inode will block if they arrive while we are sleeping waiting + * for old data structures to be purged or for the contents of the + * disk portion of this inode to be read. + */ + ufs_ihashins(ip); + + if (ext2fs_inode_hash_lock < 0) + wakeup(&ext2fs_inode_hash_lock); + ext2fs_inode_hash_lock = 0; + + /* Read in the disk contents for the inode, copy into the inode. */ +#if 0 +printf("ext2_vget(%d) dbn= %d ", ino, fsbtodb(fs, ino_to_fsba(fs, ino))); +#endif + if ((error = bread(ump->um_devvp, fsbtodb(fs, ino_to_fsba(fs, ino)), + (int)fs->s_blocksize, NOCRED, &bp)) != 0) { + /* + * The inode does not contain anything useful, so it would + * be misleading to leave it on its hash chain. With mode + * still zero, it will be unlinked and returned to the free + * list by vput(). + */ + vput(vp); + brelse(bp); + *vpp = NULL; + return (error); + } + /* convert ext2 inode to dinode */ + ext2_ei2di((struct ext2_inode *) ((char *)bp->b_data + EXT2_INODE_SIZE * + ino_to_fsbo(fs, ino)), &ip->i_din); + ip->i_block_group = ino_to_cg(fs, ino); + ip->i_next_alloc_block = 0; + ip->i_next_alloc_goal = 0; + ip->i_prealloc_count = 0; + ip->i_prealloc_block = 0; + /* now we want to make sure that block pointers for unused + blocks are zeroed out - ext2_balloc depends on this + although for regular files and directories only + */ + if(S_ISDIR(ip->i_mode) || S_ISREG(ip->i_mode)) { + used_blocks = (ip->i_size+fs->s_blocksize-1) / fs->s_blocksize; + for(i = used_blocks; i < EXT2_NDIR_BLOCKS; i++) + ip->i_db[i] = 0; + } +/* + ext2_print_inode(ip); +*/ + brelse(bp); + + /* + * Initialize the vnode from the inode, check for aliases. + * Note that the underlying vnode may have changed. + */ + if ((error = ufs_vinit(mp, ext2_specop_p, ext2_fifoop_p, &vp)) != 0) { + vput(vp); + *vpp = NULL; + return (error); + } + /* + * Finish inode initialization now that aliasing has been resolved. + */ + ip->i_devvp = ump->um_devvp; + VREF(ip->i_devvp); + /* + * Set up a generation number for this inode if it does not + * already have one. This should only happen on old filesystems. + */ + if (ip->i_gen == 0) { + ip->i_gen = random() / 2 + 1; + if ((vp->v_mount->mnt_flag & MNT_RDONLY) == 0) + ip->i_flag |= IN_MODIFIED; + } + *vpp = vp; + return (0); +} + +/* + * File handle to vnode + * + * Have to be really careful about stale file handles: + * - check that the inode number is valid + * - call ext2_vget() to get the locked inode + * - check for an unallocated inode (i_mode == 0) + * - check that the given client host has export rights and return + * those rights via. exflagsp and credanonp + */ +static int +ext2_fhtovp(mp, fhp, vpp) + register struct mount *mp; + struct fid *fhp; + struct vnode **vpp; +{ + register struct ufid *ufhp; + struct ext2_sb_info *fs; + + ufhp = (struct ufid *)fhp; + fs = VFSTOUFS(mp)->um_e2fs; + if (ufhp->ufid_ino < ROOTINO || + ufhp->ufid_ino >= fs->s_groups_count * fs->s_es->s_inodes_per_group) + return (ESTALE); + return (ufs_fhtovp(mp, ufhp, vpp)); +} + +/* + * Vnode pointer to File handle + */ +/* ARGSUSED */ +static int +ext2_vptofh(vp, fhp) + struct vnode *vp; + struct fid *fhp; +{ + register struct inode *ip; + register struct ufid *ufhp; + + ip = VTOI(vp); + ufhp = (struct ufid *)fhp; + ufhp->ufid_len = sizeof(struct ufid); + ufhp->ufid_ino = ip->i_number; + ufhp->ufid_gen = ip->i_gen; + return (0); +} + +/* + * Write a superblock and associated information back to disk. + */ +static int +ext2_sbupdate(mp, waitfor) + struct ufsmount *mp; + int waitfor; +{ + register struct ext2_sb_info *fs = mp->um_e2fs; + register struct ext2_super_block *es = fs->s_es; + register struct buf *bp; + int error = 0; +/* +printf("\nupdating superblock, waitfor=%s\n", waitfor == MNT_WAIT ? "yes":"no"); +*/ + bp = getblk(mp->um_devvp, SBLOCK, SBSIZE, 0, 0); + bcopy((caddr_t)es, bp->b_data, (u_int)sizeof(struct ext2_super_block)); + if (waitfor == MNT_WAIT) + error = bwrite(bp); + else + bawrite(bp); + + /* + * The buffers for group descriptors, inode bitmaps and block bitmaps + * are not busy at this point and are (hopefully) written by the + * usual sync mechanism. No need to write them here + */ + + return (error); +} diff --git a/sys/gnu/ext2fs/ext2_vnops.c b/sys/gnu/ext2fs/ext2_vnops.c new file mode 100644 index 000000000000..8076fd888c77 --- /dev/null +++ b/sys/gnu/ext2fs/ext2_vnops.c @@ -0,0 +1,1215 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * (c) UNIX System Laboratories, Inc. + * All or some portions of this file are derived from material licensed + * to the University of California by American Telephone and Telegraph + * Co. or Unix System Laboratories, Inc. and are reproduced herein with + * the permission of UNIX System Laboratories, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ufs_vnops.c 8.27 (Berkeley) 5/27/95 + * @(#)ext2_vnops.c 8.7 (Berkeley) 2/3/94 + * $FreeBSD$ + */ + +#include "opt_quota.h" +#include "opt_suiddir.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/resourcevar.h> +#include <sys/kernel.h> +#include <sys/stat.h> +#include <sys/buf.h> +#include <sys/proc.h> +#include <sys/mount.h> +#include <sys/time.h> +#include <sys/vnode.h> +#include <sys/namei.h> + +#include <vm/vm.h> +#include <vm/vm_extern.h> +#include <vm/vm_zone.h> +#include <vm/vnode_pager.h> + +#include <sys/signalvar.h> +#include <ufs/ufs/dir.h> +#include <ufs/ufs/quota.h> +#include <ufs/ufs/inode.h> +#include <ufs/ufs/ufsmount.h> +#include <ufs/ufs/ufs_extern.h> + +#include <gnu/ext2fs/ext2_fs_sb.h> +#include <gnu/ext2fs/fs.h> +#include <gnu/ext2fs/ext2_extern.h> +#include <gnu/ext2fs/ext2_fs.h> + +static int ext2_makeinode __P((int mode, struct vnode *, struct vnode **, struct componentname *)); + +static int ext2_fsync __P((struct vop_fsync_args *)); +static int ext2_read __P((struct vop_read_args *)); +static int ext2_write __P((struct vop_write_args *)); +static int ext2_remove __P((struct vop_remove_args *)); +static int ext2_link __P((struct vop_link_args *)); +static int ext2_rename __P((struct vop_rename_args *)); +static int ext2_mkdir __P((struct vop_mkdir_args *)); +static int ext2_rmdir __P((struct vop_rmdir_args *)); +static int ext2_create __P((struct vop_create_args *)); +static int ext2_mknod __P((struct vop_mknod_args *)); +static int ext2_symlink __P((struct vop_symlink_args *)); +static int ext2_getpages __P((struct vop_getpages_args *)); +static int ext2_putpages __P((struct vop_putpages_args *)); + +/* Global vfs data structures for ufs. */ +vop_t **ext2_vnodeop_p; +static struct vnodeopv_entry_desc ext2_vnodeop_entries[] = { + { &vop_default_desc, (vop_t *) ufs_vnoperate }, + { &vop_cachedlookup_desc, (vop_t *) ext2_lookup }, + { &vop_fsync_desc, (vop_t *) ext2_fsync }, + { &vop_inactive_desc, (vop_t *) ext2_inactive }, + { &vop_lookup_desc, (vop_t *) vfs_cache_lookup }, + { &vop_read_desc, (vop_t *) ext2_read }, + { &vop_readdir_desc, (vop_t *) ext2_readdir }, + { &vop_reallocblks_desc, (vop_t *) ext2_reallocblks }, + { &vop_write_desc, (vop_t *) ext2_write }, + { &vop_remove_desc, (vop_t *) ext2_remove }, + { &vop_link_desc, (vop_t *) ext2_link }, + { &vop_rename_desc, (vop_t *) ext2_rename }, + { &vop_mkdir_desc, (vop_t *) ext2_mkdir }, + { &vop_rmdir_desc, (vop_t *) ext2_rmdir }, + { &vop_create_desc, (vop_t *) ext2_create }, + { &vop_mknod_desc, (vop_t *) ext2_mknod }, + { &vop_symlink_desc, (vop_t *) ext2_symlink }, + { &vop_getpages_desc, (vop_t *) ext2_getpages }, + { &vop_putpages_desc, (vop_t *) ext2_putpages }, + { NULL, NULL } +}; +static struct vnodeopv_desc ext2fs_vnodeop_opv_desc = + { &ext2_vnodeop_p, ext2_vnodeop_entries }; + +vop_t **ext2_specop_p; +static struct vnodeopv_entry_desc ext2_specop_entries[] = { + { &vop_default_desc, (vop_t *) ufs_vnoperatespec }, + { &vop_fsync_desc, (vop_t *) ext2_fsync }, + { &vop_inactive_desc, (vop_t *) ext2_inactive }, + { NULL, NULL } +}; +static struct vnodeopv_desc ext2fs_specop_opv_desc = + { &ext2_specop_p, ext2_specop_entries }; + +vop_t **ext2_fifoop_p; +static struct vnodeopv_entry_desc ext2_fifoop_entries[] = { + { &vop_default_desc, (vop_t *) ufs_vnoperatefifo }, + { &vop_fsync_desc, (vop_t *) ext2_fsync }, + { &vop_inactive_desc, (vop_t *) ext2_inactive }, + { NULL, NULL } +}; +static struct vnodeopv_desc ext2fs_fifoop_opv_desc = + { &ext2_fifoop_p, ext2_fifoop_entries }; + + VNODEOP_SET(ext2fs_vnodeop_opv_desc); + VNODEOP_SET(ext2fs_specop_opv_desc); + VNODEOP_SET(ext2fs_fifoop_opv_desc); + +#include <gnu/ext2fs/ext2_readwrite.c> + +/* + * A virgin directory (no blushing please). + * Note that the type and namlen fields are reversed relative to ufs. + * Also, we don't use `struct odirtemplate', since it would just cause + * endianness problems. + */ +static struct dirtemplate mastertemplate = { + 0, 12, 1, EXT2_FT_DIR, ".", + 0, DIRBLKSIZ - 12, 2, EXT2_FT_DIR, ".." +}; +static struct dirtemplate omastertemplate = { + 0, 12, 1, EXT2_FT_UNKNOWN, ".", + 0, DIRBLKSIZ - 12, 2, EXT2_FT_UNKNOWN, ".." +}; + +/* + * Create a regular file + */ +static int +ext2_create(ap) + struct vop_create_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; +{ + int error; + + error = + ext2_makeinode(MAKEIMODE(ap->a_vap->va_type, ap->a_vap->va_mode), + ap->a_dvp, ap->a_vpp, ap->a_cnp); + if (error) + return (error); + return (0); +} + +/* + * Synch an open file. + */ +/* ARGSUSED */ +static int +ext2_fsync(ap) + struct vop_fsync_args /* { + struct vnode *a_vp; + struct ucred *a_cred; + int a_waitfor; + struct proc *a_p; + } */ *ap; +{ + register struct vnode *vp = ap->a_vp; + register struct buf *bp; + struct buf *nbp; + int s; + + /* + * XXX why is all this fs specific? + */ + + /* + * Flush all dirty buffers associated with a vnode. + */ + ext2_discard_prealloc(VTOI(vp)); + +loop: + s = splbio(); + for (bp = TAILQ_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) { + nbp = TAILQ_NEXT(bp, b_vnbufs); + if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT)) + continue; + if ((bp->b_flags & B_DELWRI) == 0) + panic("ext2_fsync: not dirty"); + bremfree(bp); + splx(s); + /* + * Wait for I/O associated with indirect blocks to complete, + * since there is no way to quickly wait for them below. + */ + if (bp->b_vp == vp || ap->a_waitfor == MNT_NOWAIT) + (void) bawrite(bp); + else + (void) bwrite(bp); + goto loop; + } + if (ap->a_waitfor == MNT_WAIT) { + while (vp->v_numoutput) { + vp->v_flag |= VBWAIT; + tsleep(&vp->v_numoutput, PRIBIO + 1, "e2fsyn", 0); + } +#if DIAGNOSTIC + if (!TAILQ_EMPTY(&vp->v_dirtyblkhd)) { + vprint("ext2_fsync: dirty", vp); + goto loop; + } +#endif + } + splx(s); + return (UFS_UPDATE(ap->a_vp, ap->a_waitfor == MNT_WAIT)); +} + +/* + * Mknod vnode call + */ +/* ARGSUSED */ +static int +ext2_mknod(ap) + struct vop_mknod_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; +{ + struct vattr *vap = ap->a_vap; + struct vnode **vpp = ap->a_vpp; + struct inode *ip; + ino_t ino; + int error; + + error = ext2_makeinode(MAKEIMODE(vap->va_type, vap->va_mode), + ap->a_dvp, vpp, ap->a_cnp); + if (error) + return (error); + ip = VTOI(*vpp); + ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE; + if (vap->va_rdev != VNOVAL) { + /* + * Want to be able to use this to make badblock + * inodes, so don't truncate the dev number. + */ + ip->i_rdev = vap->va_rdev; + } + /* + * Remove inode, then reload it through VFS_VGET so it is + * checked to see if it is an alias of an existing entry in + * the inode cache. + */ + vput(*vpp); + (*vpp)->v_type = VNON; + ino = ip->i_number; /* Save this before vgone() invalidates ip. */ + vgone(*vpp); + error = VFS_VGET(ap->a_dvp->v_mount, ino, vpp); + if (error) { + *vpp = NULL; + return (error); + } + return (0); +} + +static int +ext2_remove(ap) + struct vop_remove_args /* { + struct vnode *a_dvp; + struct vnode *a_vp; + struct componentname *a_cnp; + } */ *ap; +{ + struct inode *ip; + struct vnode *vp = ap->a_vp; + struct vnode *dvp = ap->a_dvp; + int error; + + ip = VTOI(vp); + if ((ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND)) || + (VTOI(dvp)->i_flags & APPEND)) { + error = EPERM; + goto out; + } + error = ext2_dirremove(dvp, ap->a_cnp); + if (error == 0) { + ip->i_nlink--; + ip->i_flag |= IN_CHANGE; + } +out: + return (error); +} + +/* + * link vnode call + */ +static int +ext2_link(ap) + struct vop_link_args /* { + struct vnode *a_tdvp; + struct vnode *a_vp; + struct componentname *a_cnp; + } */ *ap; +{ + struct vnode *vp = ap->a_vp; + struct vnode *tdvp = ap->a_tdvp; + struct componentname *cnp = ap->a_cnp; + struct proc *p = cnp->cn_proc; + struct inode *ip; + int error; + +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_link: no name"); +#endif + if (tdvp->v_mount != vp->v_mount) { + error = EXDEV; + goto out2; + } + if (tdvp != vp && (error = vn_lock(vp, LK_EXCLUSIVE, p))) { + goto out2; + } + ip = VTOI(vp); + if ((nlink_t)ip->i_nlink >= LINK_MAX) { + error = EMLINK; + goto out1; + } + if (ip->i_flags & (IMMUTABLE | APPEND)) { + error = EPERM; + goto out1; + } + ip->i_nlink++; + ip->i_flag |= IN_CHANGE; + error = UFS_UPDATE(vp, 1); + if (!error) + error = ext2_direnter(ip, tdvp, cnp); + if (error) { + ip->i_nlink--; + ip->i_flag |= IN_CHANGE; + } +out1: + if (tdvp != vp) + VOP_UNLOCK(vp, 0, p); +out2: + return (error); +} + +/* + * Rename system call. + * See comments in sys/ufs/ufs/ufs_vnops.c + */ +static int +ext2_rename(ap) + struct vop_rename_args /* { + struct vnode *a_fdvp; + struct vnode *a_fvp; + struct componentname *a_fcnp; + struct vnode *a_tdvp; + struct vnode *a_tvp; + struct componentname *a_tcnp; + } */ *ap; +{ + struct vnode *tvp = ap->a_tvp; + register struct vnode *tdvp = ap->a_tdvp; + struct vnode *fvp = ap->a_fvp; + struct vnode *fdvp = ap->a_fdvp; + struct componentname *tcnp = ap->a_tcnp; + struct componentname *fcnp = ap->a_fcnp; + struct proc *p = fcnp->cn_proc; + struct inode *ip, *xp, *dp; + struct dirtemplate dirbuf; + int doingdirectory = 0, oldparent = 0, newparent = 0; + int error = 0; + u_char namlen; + +#ifdef DIAGNOSTIC + if ((tcnp->cn_flags & HASBUF) == 0 || + (fcnp->cn_flags & HASBUF) == 0) + panic("ufs_rename: no name"); +#endif + /* + * Check for cross-device rename. + */ + if ((fvp->v_mount != tdvp->v_mount) || + (tvp && (fvp->v_mount != tvp->v_mount))) { + error = EXDEV; +abortit: + if (tdvp == tvp) + vrele(tdvp); + else + vput(tdvp); + if (tvp) + vput(tvp); + vrele(fdvp); + vrele(fvp); + return (error); + } + + if (tvp && ((VTOI(tvp)->i_flags & (NOUNLINK | IMMUTABLE | APPEND)) || + (VTOI(tdvp)->i_flags & APPEND))) { + error = EPERM; + goto abortit; + } + + /* + * Check if just deleting a link name or if we've lost a race. + * If another process completes the same rename after we've looked + * up the source and have blocked looking up the target, then the + * source and target inodes may be identical now although the + * names were never linked. + */ + if (fvp == tvp) { + if (fvp->v_type == VDIR) { + /* + * Linked directories are impossible, so we must + * have lost the race. Pretend that the rename + * completed before the lookup. + */ +#ifdef UFS_RENAME_DEBUG + printf("ufs_rename: fvp == tvp for directories\n"); +#endif + error = ENOENT; + goto abortit; + } + + /* Release destination completely. */ + vput(tdvp); + vput(tvp); + + /* + * Delete source. There is another race now that everything + * is unlocked, but this doesn't cause any new complications. + * Relookup() may find a file that is unrelated to the + * original one, or it may fail. Too bad. + */ + vrele(fdvp); + vrele(fvp); + fcnp->cn_flags &= ~MODMASK; + fcnp->cn_flags |= LOCKPARENT | LOCKLEAF; + fcnp->cn_nameiop = DELETE; + VREF(fdvp); + error = relookup(fdvp, &fvp, fcnp); + if (error == 0) + vrele(fdvp); + if (fvp == NULL) { +#ifdef UFS_RENAME_DEBUG + printf("ufs_rename: from name disappeared\n"); +#endif + return (ENOENT); + } + error = VOP_REMOVE(fdvp, fvp, fcnp); + if (fdvp == fvp) + vrele(fdvp); + else + vput(fdvp); + if (fvp != NULLVP) + vput(fvp); + return (error); + } + if ((error = vn_lock(fvp, LK_EXCLUSIVE, p)) != 0) + goto abortit; + dp = VTOI(fdvp); + ip = VTOI(fvp); + if (ip->i_nlink >= LINK_MAX) { + VOP_UNLOCK(fvp, 0, p); + error = EMLINK; + goto abortit; + } + if ((ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND)) + || (dp->i_flags & APPEND)) { + VOP_UNLOCK(fvp, 0, p); + error = EPERM; + goto abortit; + } + if ((ip->i_mode & IFMT) == IFDIR) { + /* + * Avoid ".", "..", and aliases of "." for obvious reasons. + */ + if ((fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.') || + dp == ip || (fcnp->cn_flags | tcnp->cn_flags) & ISDOTDOT || + (ip->i_flag & IN_RENAME)) { + VOP_UNLOCK(fvp, 0, p); + error = EINVAL; + goto abortit; + } + ip->i_flag |= IN_RENAME; + oldparent = dp->i_number; + doingdirectory++; + } + vrele(fdvp); + + /* + * When the target exists, both the directory + * and target vnodes are returned locked. + */ + dp = VTOI(tdvp); + xp = NULL; + if (tvp) + xp = VTOI(tvp); + + /* + * 1) Bump link count while we're moving stuff + * around. If we crash somewhere before + * completing our work, the link count + * may be wrong, but correctable. + */ + ip->i_nlink++; + ip->i_flag |= IN_CHANGE; + if ((error = UFS_UPDATE(fvp, 1)) != 0) { + VOP_UNLOCK(fvp, 0, p); + goto bad; + } + + /* + * If ".." must be changed (ie the directory gets a new + * parent) then the source directory must not be in the + * directory heirarchy above the target, as this would + * orphan everything below the source directory. Also + * the user must have write permission in the source so + * as to be able to change "..". We must repeat the call + * to namei, as the parent directory is unlocked by the + * call to checkpath(). + */ + error = VOP_ACCESS(fvp, VWRITE, tcnp->cn_cred, tcnp->cn_proc); + VOP_UNLOCK(fvp, 0, p); + if (oldparent != dp->i_number) + newparent = dp->i_number; + if (doingdirectory && newparent) { + if (error) /* write access check above */ + goto bad; + if (xp != NULL) + vput(tvp); + error = ext2_checkpath(ip, dp, tcnp->cn_cred); + if (error) + goto out; + VREF(tdvp); + error = relookup(tdvp, &tvp, tcnp); + if (error) + goto out; + vrele(tdvp); + dp = VTOI(tdvp); + xp = NULL; + if (tvp) + xp = VTOI(tvp); + } + /* + * 2) If target doesn't exist, link the target + * to the source and unlink the source. + * Otherwise, rewrite the target directory + * entry to reference the source inode and + * expunge the original entry's existence. + */ + if (xp == NULL) { + if (dp->i_dev != ip->i_dev) + panic("ufs_rename: EXDEV"); + /* + * Account for ".." in new directory. + * When source and destination have the same + * parent we don't fool with the link count. + */ + if (doingdirectory && newparent) { + if ((nlink_t)dp->i_nlink >= LINK_MAX) { + error = EMLINK; + goto bad; + } + dp->i_nlink++; + dp->i_flag |= IN_CHANGE; + error = UFS_UPDATE(tdvp, 1); + if (error) + goto bad; + } + error = ext2_direnter(ip, tdvp, tcnp); + if (error) { + if (doingdirectory && newparent) { + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + (void)UFS_UPDATE(tdvp, 1); + } + goto bad; + } + vput(tdvp); + } else { + if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev) + panic("ufs_rename: EXDEV"); + /* + * Short circuit rename(foo, foo). + */ + if (xp->i_number == ip->i_number) + panic("ufs_rename: same file"); + /* + * If the parent directory is "sticky", then the user must + * own the parent directory, or the destination of the rename, + * otherwise the destination may not be changed (except by + * root). This implements append-only directories. + */ + if ((dp->i_mode & S_ISTXT) && tcnp->cn_cred->cr_uid != 0 && + tcnp->cn_cred->cr_uid != dp->i_uid && + xp->i_uid != tcnp->cn_cred->cr_uid) { + error = EPERM; + goto bad; + } + /* + * Target must be empty if a directory and have no links + * to it. Also, ensure source and target are compatible + * (both directories, or both not directories). + */ + if ((xp->i_mode&IFMT) == IFDIR) { + if (! ext2_dirempty(xp, dp->i_number, tcnp->cn_cred) || + xp->i_nlink > 2) { + error = ENOTEMPTY; + goto bad; + } + if (!doingdirectory) { + error = ENOTDIR; + goto bad; + } + cache_purge(tdvp); + } else if (doingdirectory) { + error = EISDIR; + goto bad; + } + error = ext2_dirrewrite(dp, ip, tcnp); + if (error) + goto bad; + /* + * If the target directory is in the same + * directory as the source directory, + * decrement the link count on the parent + * of the target directory. + */ + if (doingdirectory && !newparent) { + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + } + vput(tdvp); + /* + * Adjust the link count of the target to + * reflect the dirrewrite above. If this is + * a directory it is empty and there are + * no links to it, so we can squash the inode and + * any space associated with it. We disallowed + * renaming over top of a directory with links to + * it above, as the remaining link would point to + * a directory without "." or ".." entries. + */ + xp->i_nlink--; + if (doingdirectory) { + if (--xp->i_nlink != 0) + panic("ufs_rename: linked directory"); + error = UFS_TRUNCATE(tvp, (off_t)0, IO_SYNC, + tcnp->cn_cred, tcnp->cn_proc); + } + xp->i_flag |= IN_CHANGE; + vput(tvp); + xp = NULL; + } + + /* + * 3) Unlink the source. + */ + fcnp->cn_flags &= ~MODMASK; + fcnp->cn_flags |= LOCKPARENT | LOCKLEAF; + VREF(fdvp); + error = relookup(fdvp, &fvp, fcnp); + if (error == 0) + vrele(fdvp); + if (fvp != NULL) { + xp = VTOI(fvp); + dp = VTOI(fdvp); + } else { + /* + * From name has disappeared. + */ + if (doingdirectory) + panic("ufs_rename: lost dir entry"); + vrele(ap->a_fvp); + return (0); + } + /* + * Ensure that the directory entry still exists and has not + * changed while the new name has been entered. If the source is + * a file then the entry may have been unlinked or renamed. In + * either case there is no further work to be done. If the source + * is a directory then it cannot have been rmdir'ed; its link + * count of three would cause a rmdir to fail with ENOTEMPTY. + * The IN_RENAME flag ensures that it cannot be moved by another + * rename. + */ + if (xp != ip) { + if (doingdirectory) + panic("ufs_rename: lost dir entry"); + } else { + /* + * If the source is a directory with a + * new parent, the link count of the old + * parent directory must be decremented + * and ".." set to point to the new parent. + */ + if (doingdirectory && newparent) { + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + error = vn_rdwr(UIO_READ, fvp, (caddr_t)&dirbuf, + sizeof (struct dirtemplate), (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, + tcnp->cn_cred, (int *)0, (struct proc *)0); + if (error == 0) { + /* Like ufs little-endian: */ + namlen = dirbuf.dotdot_type; + if (namlen != 2 || + dirbuf.dotdot_name[0] != '.' || + dirbuf.dotdot_name[1] != '.') { + ufs_dirbad(xp, (doff_t)12, + "rename: mangled dir"); + } else { + dirbuf.dotdot_ino = newparent; + (void) vn_rdwr(UIO_WRITE, fvp, + (caddr_t)&dirbuf, + sizeof (struct dirtemplate), + (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED|IO_SYNC, + tcnp->cn_cred, (int *)0, + (struct proc *)0); + cache_purge(fdvp); + } + } + } + error = ext2_dirremove(fdvp, fcnp); + if (!error) { + xp->i_nlink--; + xp->i_flag |= IN_CHANGE; + } + xp->i_flag &= ~IN_RENAME; + } + if (dp) + vput(fdvp); + if (xp) + vput(fvp); + vrele(ap->a_fvp); + return (error); + +bad: + if (xp) + vput(ITOV(xp)); + vput(ITOV(dp)); +out: + if (doingdirectory) + ip->i_flag &= ~IN_RENAME; + if (vn_lock(fvp, LK_EXCLUSIVE, p) == 0) { + ip->i_nlink--; + ip->i_flag |= IN_CHANGE; + ip->i_flag &= ~IN_RENAME; + vput(fvp); + } else + vrele(fvp); + return (error); +} + +/* + * Mkdir system call + */ +static int +ext2_mkdir(ap) + struct vop_mkdir_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; +{ + register struct vnode *dvp = ap->a_dvp; + register struct vattr *vap = ap->a_vap; + register struct componentname *cnp = ap->a_cnp; + register struct inode *ip, *dp; + struct vnode *tvp; + struct dirtemplate dirtemplate, *dtp; + int error, dmode; + +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_mkdir: no name"); +#endif + dp = VTOI(dvp); + if ((nlink_t)dp->i_nlink >= LINK_MAX) { + error = EMLINK; + goto out; + } + dmode = vap->va_mode & 0777; + dmode |= IFDIR; + /* + * Must simulate part of ext2_makeinode here to acquire the inode, + * but not have it entered in the parent directory. The entry is + * made later after writing "." and ".." entries. + */ + error = UFS_VALLOC(dvp, dmode, cnp->cn_cred, &tvp); + if (error) + goto out; + ip = VTOI(tvp); + ip->i_gid = dp->i_gid; +#ifdef SUIDDIR + { +#ifdef QUOTA + struct ucred ucred, *ucp; + ucp = cnp->cn_cred; +#endif I + /* + * if we are hacking owners here, (only do this where told to) + * and we are not giving it TOO root, (would subvert quotas) + * then go ahead and give it to the other user. + * The new directory also inherits the SUID bit. + * If user's UID and dir UID are the same, + * 'give it away' so that the SUID is still forced on. + */ + if ( (dvp->v_mount->mnt_flag & MNT_SUIDDIR) && + (dp->i_mode & ISUID) && dp->i_uid) { + dmode |= ISUID; + ip->i_uid = dp->i_uid; +#ifdef QUOTA + if (dp->i_uid != cnp->cn_cred->cr_uid) { + /* + * make sure the correct user gets charged + * for the space. + * Make a dummy credential for the victim. + * XXX This seems to never be accessed out of + * our context so a stack variable is ok. + */ + ucred.cr_ref = 1; + ucred.cr_uid = ip->i_uid; + ucred.cr_ngroups = 1; + ucred.cr_groups[0] = dp->i_gid; + ucp = &ucred; + } +#endif I + } else { + ip->i_uid = cnp->cn_cred->cr_uid; + } +#ifdef QUOTA + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, ucp, 0))) { + UFS_VFREE(tvp, ip->i_number, dmode); + vput(tvp); + return (error); + } +#endif + } +#else + ip->i_uid = cnp->cn_cred->cr_uid; +#ifdef QUOTA + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, cnp->cn_cred, 0))) { + UFS_VFREE(tvp, ip->i_number, dmode); + vput(tvp); + return (error); + } +#endif +#endif + ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE; + ip->i_mode = dmode; + tvp->v_type = VDIR; /* Rest init'd in getnewvnode(). */ + ip->i_nlink = 2; + if (cnp->cn_flags & ISWHITEOUT) + ip->i_flags |= UF_OPAQUE; + error = UFS_UPDATE(tvp, 1); + + /* + * Bump link count in parent directory + * to reflect work done below. Should + * be done before reference is created + * so reparation is possible if we crash. + */ + dp->i_nlink++; + dp->i_flag |= IN_CHANGE; + error = UFS_UPDATE(dvp, 1); + if (error) + goto bad; + + /* Initialize directory with "." and ".." from static template. */ + if (EXT2_HAS_INCOMPAT_FEATURE(ip->i_e2fs->s_es, + EXT2_FEATURE_INCOMPAT_FILETYPE)) + dtp = &mastertemplate; + else + dtp = &omastertemplate; + dirtemplate = *dtp; + dirtemplate.dot_ino = ip->i_number; + dirtemplate.dotdot_ino = dp->i_number; + /* note that in ext2 DIRBLKSIZ == blocksize, not DEV_BSIZE + * so let's just redefine it - for this function only + */ +#undef DIRBLKSIZ +#define DIRBLKSIZ VTOI(dvp)->i_e2fs->s_blocksize + dirtemplate.dotdot_reclen = DIRBLKSIZ - 12; + error = vn_rdwr(UIO_WRITE, tvp, (caddr_t)&dirtemplate, + sizeof (dirtemplate), (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED|IO_SYNC, cnp->cn_cred, (int *)0, (struct proc *)0); + if (error) { + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + goto bad; + } + if (DIRBLKSIZ > VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_bsize) + panic("ufs_mkdir: blksize"); /* XXX should grow with balloc() */ + else { + ip->i_size = DIRBLKSIZ; + ip->i_flag |= IN_CHANGE; + } + + /* Directory set up, now install its entry in the parent directory. */ + error = ext2_direnter(ip, dvp, cnp); + if (error) { + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + } +bad: + /* + * No need to do an explicit VOP_TRUNCATE here, vrele will do this + * for us because we set the link count to 0. + */ + if (error) { + ip->i_nlink = 0; + ip->i_flag |= IN_CHANGE; + vput(tvp); + } else + *ap->a_vpp = tvp; +out: + return (error); +#undef DIRBLKSIZ +#define DIRBLKSIZ DEV_BSIZE +} + +/* + * Rmdir system call. + */ +static int +ext2_rmdir(ap) + struct vop_rmdir_args /* { + struct vnode *a_dvp; + struct vnode *a_vp; + struct componentname *a_cnp; + } */ *ap; +{ + struct vnode *vp = ap->a_vp; + struct vnode *dvp = ap->a_dvp; + struct componentname *cnp = ap->a_cnp; + struct proc *p = cnp->cn_proc; + struct inode *ip, *dp; + int error; + + ip = VTOI(vp); + dp = VTOI(dvp); + + /* + * Verify the directory is empty (and valid). + * (Rmdir ".." won't be valid since + * ".." will contain a reference to + * the current directory and thus be + * non-empty.) + */ + error = 0; + if (ip->i_nlink != 2 || !ext2_dirempty(ip, dp->i_number, cnp->cn_cred)) { + error = ENOTEMPTY; + goto out; + } + if ((dp->i_flags & APPEND) + || (ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND))) { + error = EPERM; + goto out; + } + /* + * Delete reference to directory before purging + * inode. If we crash in between, the directory + * will be reattached to lost+found, + */ + error = ext2_dirremove(dvp, cnp); + if (error) + goto out; + dp->i_nlink--; + dp->i_flag |= IN_CHANGE; + cache_purge(dvp); + VOP_UNLOCK(dvp, 0, p); + /* + * Truncate inode. The only stuff left + * in the directory is "." and "..". The + * "." reference is inconsequential since + * we're quashing it. The ".." reference + * has already been adjusted above. We've + * removed the "." reference and the reference + * in the parent directory, but there may be + * other hard links so decrement by 2 and + * worry about them later. + */ + ip->i_nlink -= 2; + error = UFS_TRUNCATE(vp, (off_t)0, IO_SYNC, cnp->cn_cred, p); + cache_purge(ITOV(ip)); + vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY, p); +out: + return (error); +} + +/* + * symlink -- make a symbolic link + */ +static int +ext2_symlink(ap) + struct vop_symlink_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + char *a_target; + } */ *ap; +{ + register struct vnode *vp, **vpp = ap->a_vpp; + register struct inode *ip; + int len, error; + + error = ext2_makeinode(IFLNK | ap->a_vap->va_mode, ap->a_dvp, + vpp, ap->a_cnp); + if (error) + return (error); + vp = *vpp; + len = strlen(ap->a_target); + if (len < vp->v_mount->mnt_maxsymlinklen) { + ip = VTOI(vp); + bcopy(ap->a_target, (char *)ip->i_shortlink, len); + ip->i_size = len; + ip->i_flag |= IN_CHANGE | IN_UPDATE; + } else + error = vn_rdwr(UIO_WRITE, vp, ap->a_target, len, (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, ap->a_cnp->cn_cred, (int *)0, + (struct proc *)0); + if (error) + vput(vp); + return (error); +} + +/* + * Allocate a new inode. + */ +static int +ext2_makeinode(mode, dvp, vpp, cnp) + int mode; + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; +{ + register struct inode *ip, *pdir; + struct vnode *tvp; + int error; + + pdir = VTOI(dvp); +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ext2_makeinode: no name"); +#endif + *vpp = NULL; + if ((mode & IFMT) == 0) + mode |= IFREG; + + error = UFS_VALLOC(dvp, mode, cnp->cn_cred, &tvp); + if (error) { + return (error); + } + ip = VTOI(tvp); + ip->i_gid = pdir->i_gid; +#ifdef SUIDDIR + { +#ifdef QUOTA + struct ucred ucred, *ucp; + ucp = cnp->cn_cred; +#endif I + /* + * if we are + * not the owner of the directory, + * and we are hacking owners here, (only do this where told to) + * and we are not giving it TOO root, (would subvert quotas) + * then go ahead and give it to the other user. + * Note that this drops off the execute bits for security. + */ + if ( (dvp->v_mount->mnt_flag & MNT_SUIDDIR) && + (pdir->i_mode & ISUID) && + (pdir->i_uid != cnp->cn_cred->cr_uid) && pdir->i_uid) { + ip->i_uid = pdir->i_uid; + mode &= ~07111; +#ifdef QUOTA + /* + * make sure the correct user gets charged + * for the space. + * Quickly knock up a dummy credential for the victim. + * XXX This seems to never be accessed out of our + * context so a stack variable is ok. + */ + ucred.cr_ref = 1; + ucred.cr_uid = ip->i_uid; + ucred.cr_ngroups = 1; + ucred.cr_groups[0] = pdir->i_gid; + ucp = &ucred; +#endif I + } else { + ip->i_uid = cnp->cn_cred->cr_uid; + } + +#ifdef QUOTA + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, ucp, 0))) { + UFS_VFREE(tvp, ip->i_number, mode); + vput(tvp); + return (error); + } +#endif + } +#else + ip->i_uid = cnp->cn_cred->cr_uid; +#ifdef QUOTA + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, cnp->cn_cred, 0))) { + UFS_VFREE(tvp, ip->i_number, mode); + vput(tvp); + return (error); + } +#endif +#endif + ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE; + ip->i_mode = mode; + tvp->v_type = IFTOVT(mode); /* Rest init'd in getnewvnode(). */ + ip->i_nlink = 1; + if ((ip->i_mode & ISGID) && !groupmember(ip->i_gid, cnp->cn_cred) && + suser_xxx(cnp->cn_cred, 0, PRISON_ROOT)) + ip->i_mode &= ~ISGID; + + if (cnp->cn_flags & ISWHITEOUT) + ip->i_flags |= UF_OPAQUE; + + /* + * Make sure inode goes to disk before directory entry. + */ + error = UFS_UPDATE(tvp, 1); + if (error) + goto bad; + error = ext2_direnter(ip, dvp, cnp); + if (error) + goto bad; + + *vpp = tvp; + return (0); + +bad: + /* + * Write error occurred trying to update the inode + * or the directory so must deallocate the inode. + */ + ip->i_nlink = 0; + ip->i_flag |= IN_CHANGE; + vput(tvp); + return (error); +} + +/* + * get page routine + * + * XXX By default, wimp out... note that a_offset is ignored (and always + * XXX has been). + */ +static int +ext2_getpages(ap) + struct vop_getpages_args *ap; +{ + return (vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count, + ap->a_reqpage)); +} + +/* + * put page routine + * + * XXX By default, wimp out... note that a_offset is ignored (and always + * XXX has been). + */ +static int +ext2_putpages(ap) + struct vop_putpages_args *ap; +{ + return (vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count, + ap->a_sync, ap->a_rtvals)); +} diff --git a/sys/gnu/ext2fs/fs.h b/sys/gnu/ext2fs/fs.h new file mode 100644 index 000000000000..8bedb05f330c --- /dev/null +++ b/sys/gnu/ext2fs/fs.h @@ -0,0 +1,184 @@ +/* + * modified for EXT2FS support in Lites 1.1 + * + * Aug 1995, Godmar Back (gback@cs.utah.edu) + * University of Utah, Department of Computer Science + */ +/* + * Copyright (c) 1982, 1986, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)fs.h 8.7 (Berkeley) 4/19/94 + * $FreeBSD$ + */ + +/* + * Each disk drive contains some number of file systems. + * A file system consists of a number of cylinder groups. + * Each cylinder group has inodes and data. + * + * A file system is described by its super-block, which in turn + * describes the cylinder groups. The super-block is critical + * data and is replicated in each cylinder group to protect against + * catastrophic loss. This is done at `newfs' time and the critical + * super-block data does not change, so the copies need not be + * referenced further unless disaster strikes. + * + * The first boot and super blocks are given in absolute disk addresses. + * The byte-offset forms are preferred, as they don't imply a sector size. + */ +#define BBSIZE 1024 +#define SBSIZE 1024 +#define BBOFF ((off_t)(0)) +#define SBOFF ((off_t)(BBOFF + BBSIZE)) +#define BBLOCK ((daddr_t)(0)) +#define SBLOCK ((daddr_t)(BBLOCK + BBSIZE / DEV_BSIZE)) + +/* + * The path name on which the file system is mounted is maintained + * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in + * the super block for this name. + */ +#define MAXMNTLEN 512 + +/* + * Macros for access to superblock array structures + */ + +/* + * Convert cylinder group to base address of its global summary info. + */ +#define fs_cs(fs, cgindx) (((struct ext2_group_desc *) \ + (fs->s_group_desc[cgindx / EXT2_DESC_PER_BLOCK(fs)]->b_data)) \ + [cgindx % EXT2_DESC_PER_BLOCK(fs)]) + +/* + * Turn file system block numbers into disk block addresses. + * This maps file system blocks to device size blocks. + */ +#define fsbtodb(fs, b) ((b) << ((fs)->s_fsbtodb)) +#define dbtofsb(fs, b) ((b) >> ((fs)->s_fsbtodb)) + +/* get group containing inode */ +#define ino_to_cg(fs, x) (((x) - 1) / EXT2_INODES_PER_GROUP(fs)) + +/* get block containing inode from its number x */ +#define ino_to_fsba(fs, x) fs_cs(fs, ino_to_cg(fs, x)).bg_inode_table + \ + (((x)-1) % EXT2_INODES_PER_GROUP(fs))/EXT2_INODES_PER_BLOCK(fs) + +/* get offset for inode in block */ +#define ino_to_fsbo(fs, x) ((x-1) % EXT2_INODES_PER_BLOCK(fs)) + +/* + * Give cylinder group number for a file system block. + * Give cylinder group block number for a file system block. + */ +#define dtog(fs, d) (((d) - fs->s_es->s_first_data_block) / \ + EXT2_BLOCKS_PER_GROUP(fs)) +#define dtogd(fs, d) (((d) - fs->s_es->s_first_data_block) % \ + EXT2_BLOCKS_PER_GROUP(fs)) + +/* + * The following macros optimize certain frequently calculated + * quantities by using shifts and masks in place of divisions + * modulos and multiplications. + */ +#define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \ + ((loc) & (fs)->s_qbmask) + +#define lblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \ + ((blk) << (fs->s_bshift)) + +#define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \ + ((loc) >> (fs->s_bshift)) + +/* no fragments -> logical block number equal # of frags */ +#define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \ + ((loc) >> (fs->s_bshift)) + +#define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \ + roundup(size, fs->s_frag_size) + /* was (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask) */ + +/* + * Determining the size of a file block in the file system. + * easy w/o fragments + */ +#define blksize(fs, ip, lbn) ((fs)->s_frag_size) + +/* + * INOPB is the number of inodes in a secondary storage block. + */ +#define INOPB(fs) EXT2_INODES_PER_BLOCK(fs) + +/* + * NINDIR is the number of indirects in a file system block. + */ +#define NINDIR(fs) (EXT2_ADDR_PER_BLOCK(fs)) + +extern int inside[], around[]; +extern u_char *fragtbl[]; + +/* a few remarks about superblock locking/unlocking + * Linux provides special routines for doing so + * I haven't figured out yet what BSD does + * I think I'll try a VOP_LOCK/VOP_UNLOCK on the device vnode + */ +#define DEVVP(inode) (VFSTOUFS(ITOV(inode)->v_mount)->um_devvp) +#define lock_super(devvp) vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY, curproc) +#define unlock_super(devvp) VOP_UNLOCK(devvp, 0, curproc) + +/* + * To lock a buffer, set the B_LOCKED flag and then brelse() it. To unlock, + * reset the B_LOCKED flag and brelse() the buffer back on the LRU list + */ +#define LCK_BUF(bp) { \ + int s; \ + s = splbio(); \ + (bp)->b_flags |= B_LOCKED; \ + splx(s); \ + brelse(bp); \ +} + +#define ULCK_BUF(bp) { \ + long flags; \ + int s; \ + s = splbio(); \ + flags = (bp)->b_flags; \ + (bp)->b_flags &= ~(B_DIRTY | B_LOCKED); \ + BUF_LOCK(bp, LK_EXCLUSIVE); \ + bremfree(bp); \ + splx(s); \ + if (flags & B_DIRTY) \ + bwrite(bp); \ + else \ + brelse(bp); \ +} diff --git a/sys/gnu/ext2fs/i386-bitops.h b/sys/gnu/ext2fs/i386-bitops.h new file mode 100644 index 000000000000..595e870526b3 --- /dev/null +++ b/sys/gnu/ext2fs/i386-bitops.h @@ -0,0 +1,162 @@ +/* $FreeBSD$ */ +/* + * this is mixture of i386/bitops.h and asm/string.h + * taken from the Linux source tree + * + * XXX replace with Mach routines or reprogram in C + */ +#ifndef _SYS_GNU_EXT2FS_I386_BITOPS_H_ +#define _SYS_GNU_EXT2FS_I386_BITOPS_H_ + +/* + * Copyright 1992, Linus Torvalds. + */ + +/* + * These have to be done with inline assembly: that way the bit-setting + * is guaranteed to be atomic. All bit operations return 0 if the bit + * was cleared before the operation and != 0 if it was not. + * + * bit 0 is the LSB of addr; bit 32 is the LSB of (addr+1). + */ + +/* + * Some hacks to defeat gcc over-optimizations.. + */ +struct __dummy { unsigned long a[100]; }; +#define ADDR (*(struct __dummy *) addr) + +static __inline__ int set_bit(int nr, void * addr) +{ + int oldbit; + + __asm__ __volatile__("btsl %2,%1\n\tsbbl %0,%0" + :"=r" (oldbit),"=m" (ADDR) + :"ir" (nr)); + return oldbit; +} + +static __inline__ int clear_bit(int nr, void * addr) +{ + int oldbit; + + __asm__ __volatile__("btrl %2,%1\n\tsbbl %0,%0" + :"=r" (oldbit),"=m" (ADDR) + :"ir" (nr)); + return oldbit; +} + +static __inline__ int change_bit(int nr, void * addr) +{ + int oldbit; + + __asm__ __volatile__("btcl %2,%1\n\tsbbl %0,%0" + :"=r" (oldbit),"=m" (ADDR) + :"ir" (nr)); + return oldbit; +} + +/* + * This routine doesn't need to be atomic, but it's faster to code it + * this way. + */ +static __inline__ int test_bit(int nr, void * addr) +{ + int oldbit; + + __asm__ __volatile__("btl %2,%1\n\tsbbl %0,%0" + :"=r" (oldbit) + :"m" (ADDR),"ir" (nr)); + return oldbit; +} + +/* + * Find-bit routines.. + */ +static __inline__ int find_first_zero_bit(void * addr, unsigned size) +{ + int res; + int _count = (size + 31) >> 5; + + if (!size) + return 0; + __asm__(" \n\ + cld \n\ + movl $-1,%%eax \n\ + xorl %%edx,%%edx \n\ + repe; scasl \n\ + je 1f \n\ + xorl -4(%%edi),%%eax \n\ + subl $4,%%edi \n\ + bsfl %%eax,%%edx \n\ +1: subl %%ebx,%%edi \n\ + shll $3,%%edi \n\ + addl %%edi,%%edx" + : "=c" (_count), "=D" (addr), "=d" (res) + : "0" (_count), "1" (addr), "b" (addr) + : "ax"); + return res; +} + +static __inline__ int find_next_zero_bit (void * addr, int size, int offset) +{ + unsigned long * p = ((unsigned long *) addr) + (offset >> 5); + int set = 0, bit = offset & 31, res; + + if (bit) { + /* + * Look for zero in first byte + */ + __asm__(" \n\ + bsfl %1,%0 \n\ + jne 1f \n\ + movl $32, %0 \n\ +1: " + : "=r" (set) + : "r" (~(*p >> bit))); + if (set < (32 - bit)) + return set + offset; + set = 32 - bit; + p++; + } + /* + * No zero yet, search remaining full bytes for a zero + */ + res = find_first_zero_bit (p, size - 32 * (p - (unsigned long *) addr)); + return (offset + set + res); +} + +/* + * ffz = Find First Zero in word. Undefined if no zero exists, + * so code should check against ~0UL first.. + */ +static __inline__ unsigned long ffz(unsigned long word) +{ + __asm__("bsfl %1,%0" + :"=r" (word) + :"r" (~word)); + return word; +} + +/* + * memscan() taken from linux asm/string.h + */ +/* + * find the first occurrence of byte 'c', or 1 past the area if none + */ +static __inline__ char * memscan(void * addr, unsigned char c, int size) +{ + if (!size) + return addr; + __asm__(" \n\ + cld \n\ + repnz; scasb \n\ + jnz 1f \n\ + dec %%edi \n\ +1: " + : "=D" (addr), "=c" (size) + : "0" (addr), "1" (size), "a" (c)); + return addr; +} + +#endif /* !_SYS_GNU_EXT2FS_I386_BITOPS_H_ */ |
