1 | /* $NetBSD: ufs_inode.c,v 1.97 2016/10/28 20:38:12 jdolecek Exp $ */ |
2 | |
3 | /* |
4 | * Copyright (c) 1991, 1993 |
5 | * The Regents of the University of California. All rights reserved. |
6 | * (c) UNIX System Laboratories, Inc. |
7 | * All or some portions of this file are derived from material licensed |
8 | * to the University of California by American Telephone and Telegraph |
9 | * Co. or Unix System Laboratories, Inc. and are reproduced herein with |
10 | * the permission of UNIX System Laboratories, Inc. |
11 | * |
12 | * Redistribution and use in source and binary forms, with or without |
13 | * modification, are permitted provided that the following conditions |
14 | * are met: |
15 | * 1. Redistributions of source code must retain the above copyright |
16 | * notice, this list of conditions and the following disclaimer. |
17 | * 2. Redistributions in binary form must reproduce the above copyright |
18 | * notice, this list of conditions and the following disclaimer in the |
19 | * documentation and/or other materials provided with the distribution. |
20 | * 3. Neither the name of the University nor the names of its contributors |
21 | * may be used to endorse or promote products derived from this software |
22 | * without specific prior written permission. |
23 | * |
24 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
25 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
26 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
27 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
28 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
29 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
30 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
31 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
32 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
33 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
34 | * SUCH DAMAGE. |
35 | * |
36 | * @(#)ufs_inode.c 8.9 (Berkeley) 5/14/95 |
37 | */ |
38 | |
39 | #include <sys/cdefs.h> |
40 | __KERNEL_RCSID(0, "$NetBSD: ufs_inode.c,v 1.97 2016/10/28 20:38:12 jdolecek Exp $" ); |
41 | |
42 | #if defined(_KERNEL_OPT) |
43 | #include "opt_ffs.h" |
44 | #include "opt_quota.h" |
45 | #include "opt_wapbl.h" |
46 | #endif |
47 | |
48 | #include <sys/param.h> |
49 | #include <sys/systm.h> |
50 | #include <sys/proc.h> |
51 | #include <sys/vnode.h> |
52 | #include <sys/mount.h> |
53 | #include <sys/kernel.h> |
54 | #include <sys/namei.h> |
55 | #include <sys/kauth.h> |
56 | #include <sys/wapbl.h> |
57 | #include <sys/fstrans.h> |
58 | #include <sys/kmem.h> |
59 | |
60 | #include <ufs/ufs/inode.h> |
61 | #include <ufs/ufs/ufsmount.h> |
62 | #include <ufs/ufs/ufs_extern.h> |
63 | #include <ufs/ufs/ufs_wapbl.h> |
64 | #ifdef UFS_DIRHASH |
65 | #include <ufs/ufs/dirhash.h> |
66 | #endif |
67 | #ifdef UFS_EXTATTR |
68 | #include <ufs/ufs/extattr.h> |
69 | #endif |
70 | |
71 | #include <uvm/uvm.h> |
72 | |
73 | extern int prtactive; |
74 | |
75 | /* |
76 | * Last reference to an inode. If necessary, write or delete it. |
77 | */ |
78 | int |
79 | ufs_inactive(void *v) |
80 | { |
81 | struct vop_inactive_args /* { |
82 | struct vnode *a_vp; |
83 | struct bool *a_recycle; |
84 | } */ *ap = v; |
85 | struct vnode *vp = ap->a_vp; |
86 | struct inode *ip = VTOI(vp); |
87 | struct mount *mp = vp->v_mount; |
88 | mode_t mode; |
89 | int allerror = 0, error; |
90 | bool wapbl_locked = false; |
91 | |
92 | UFS_WAPBL_JUNLOCK_ASSERT(mp); |
93 | |
94 | fstrans_start(mp, FSTRANS_LAZY); |
95 | |
96 | /* |
97 | * Ignore inodes related to stale file handles. |
98 | */ |
99 | if (ip->i_mode == 0) |
100 | goto out; |
101 | |
102 | if (ip->i_nlink <= 0 && (mp->mnt_flag & MNT_RDONLY) == 0) { |
103 | #ifdef UFS_EXTATTR |
104 | ufs_extattr_vnode_inactive(vp, curlwp); |
105 | #endif |
106 | |
107 | /* |
108 | * All file blocks must be freed before we can let the vnode |
109 | * be reclaimed, so can't postpone full truncating any further. |
110 | */ |
111 | if (ip->i_size != 0) { |
112 | allerror = ufs_truncate_retry(vp, 0, NOCRED); |
113 | if (allerror) |
114 | goto out; |
115 | } |
116 | |
117 | #if defined(QUOTA) || defined(QUOTA2) |
118 | error = UFS_WAPBL_BEGIN(mp); |
119 | if (error) { |
120 | allerror = error; |
121 | } else { |
122 | wapbl_locked = true; |
123 | (void)chkiq(ip, -1, NOCRED, 0); |
124 | } |
125 | #endif |
126 | DIP_ASSIGN(ip, rdev, 0); |
127 | mode = ip->i_mode; |
128 | ip->i_mode = 0; |
129 | ip->i_omode = mode; |
130 | DIP_ASSIGN(ip, mode, 0); |
131 | ip->i_flag |= IN_CHANGE | IN_UPDATE; |
132 | /* |
133 | * Defer final inode free and update to ufs_reclaim(). |
134 | */ |
135 | } |
136 | |
137 | if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) { |
138 | if (! wapbl_locked) { |
139 | error = UFS_WAPBL_BEGIN(mp); |
140 | if (error) { |
141 | allerror = error; |
142 | goto out; |
143 | } |
144 | wapbl_locked = true; |
145 | } |
146 | UFS_UPDATE(vp, NULL, NULL, 0); |
147 | } |
148 | out: |
149 | if (wapbl_locked) |
150 | UFS_WAPBL_END(mp); |
151 | /* |
152 | * If we are done with the inode, reclaim it |
153 | * so that it can be reused immediately. |
154 | */ |
155 | *ap->a_recycle = (ip->i_mode == 0); |
156 | VOP_UNLOCK(vp); |
157 | fstrans_done(mp); |
158 | return (allerror); |
159 | } |
160 | |
161 | /* |
162 | * Reclaim an inode so that it can be used for other purposes. |
163 | */ |
164 | int |
165 | ufs_reclaim(struct vnode *vp) |
166 | { |
167 | struct inode *ip = VTOI(vp); |
168 | |
169 | if (prtactive && vp->v_usecount > 1) |
170 | vprint("ufs_reclaim: pushing active" , vp); |
171 | |
172 | if (!UFS_WAPBL_BEGIN(vp->v_mount)) { |
173 | UFS_UPDATE(vp, NULL, NULL, UPDATE_CLOSE); |
174 | UFS_WAPBL_END(vp->v_mount); |
175 | } |
176 | UFS_UPDATE(vp, NULL, NULL, UPDATE_CLOSE); |
177 | |
178 | if (ip->i_devvp) { |
179 | vrele(ip->i_devvp); |
180 | ip->i_devvp = 0; |
181 | } |
182 | #if defined(QUOTA) || defined(QUOTA2) |
183 | ufsquota_free(ip); |
184 | #endif |
185 | #ifdef UFS_DIRHASH |
186 | if (ip->i_dirhash != NULL) |
187 | ufsdirhash_free(ip); |
188 | #endif |
189 | return (0); |
190 | } |
191 | |
192 | /* |
193 | * allocate a range of blocks in a file. |
194 | * after this function returns, any page entirely contained within the range |
195 | * will map to invalid data and thus must be overwritten before it is made |
196 | * accessible to others. |
197 | */ |
198 | |
199 | int |
200 | ufs_balloc_range(struct vnode *vp, off_t off, off_t len, kauth_cred_t cred, |
201 | int flags) |
202 | { |
203 | off_t neweof; /* file size after the operation */ |
204 | off_t neweob; /* offset next to the last block after the operation */ |
205 | off_t pagestart; /* starting offset of range covered by pgs */ |
206 | off_t eob; /* offset next to allocated blocks */ |
207 | struct uvm_object *uobj; |
208 | int i, delta, error, npages; |
209 | int bshift = vp->v_mount->mnt_fs_bshift; |
210 | int bsize = 1 << bshift; |
211 | int ppb = MAX(bsize >> PAGE_SHIFT, 1); |
212 | struct vm_page **pgs; |
213 | size_t pgssize; |
214 | UVMHIST_FUNC("ufs_balloc_range" ); UVMHIST_CALLED(ubchist); |
215 | UVMHIST_LOG(ubchist, "vp %p off 0x%x len 0x%x u_size 0x%x" , |
216 | vp, off, len, vp->v_size); |
217 | |
218 | neweof = MAX(vp->v_size, off + len); |
219 | GOP_SIZE(vp, neweof, &neweob, 0); |
220 | |
221 | error = 0; |
222 | uobj = &vp->v_uobj; |
223 | |
224 | /* |
225 | * read or create pages covering the range of the allocation and |
226 | * keep them locked until the new block is allocated, so there |
227 | * will be no window where the old contents of the new block are |
228 | * visible to racing threads. |
229 | */ |
230 | |
231 | pagestart = trunc_page(off) & ~(bsize - 1); |
232 | npages = MIN(ppb, (round_page(neweob) - pagestart) >> PAGE_SHIFT); |
233 | pgssize = npages * sizeof(struct vm_page *); |
234 | pgs = kmem_zalloc(pgssize, KM_SLEEP); |
235 | |
236 | /* |
237 | * adjust off to be block-aligned. |
238 | */ |
239 | |
240 | delta = off & (bsize - 1); |
241 | off -= delta; |
242 | len += delta; |
243 | |
244 | genfs_node_wrlock(vp); |
245 | mutex_enter(uobj->vmobjlock); |
246 | error = VOP_GETPAGES(vp, pagestart, pgs, &npages, 0, |
247 | VM_PROT_WRITE, 0, PGO_SYNCIO | PGO_PASTEOF | PGO_NOBLOCKALLOC | |
248 | PGO_NOTIMESTAMP | PGO_GLOCKHELD); |
249 | if (error) { |
250 | genfs_node_unlock(vp); |
251 | goto out; |
252 | } |
253 | |
254 | /* |
255 | * now allocate the range. |
256 | */ |
257 | |
258 | error = GOP_ALLOC(vp, off, len, flags, cred); |
259 | genfs_node_unlock(vp); |
260 | |
261 | /* |
262 | * if the allocation succeeded, clear PG_CLEAN on all the pages |
263 | * and clear PG_RDONLY on any pages that are now fully backed |
264 | * by disk blocks. if the allocation failed, we do not invalidate |
265 | * the pages since they might have already existed and been dirty, |
266 | * in which case we need to keep them around. if we created the pages, |
267 | * they will be clean and read-only, and leaving such pages |
268 | * in the cache won't cause any problems. |
269 | */ |
270 | |
271 | GOP_SIZE(vp, off + len, &eob, 0); |
272 | mutex_enter(uobj->vmobjlock); |
273 | mutex_enter(&uvm_pageqlock); |
274 | for (i = 0; i < npages; i++) { |
275 | KASSERT((pgs[i]->flags & PG_RELEASED) == 0); |
276 | if (!error) { |
277 | if (off <= pagestart + (i << PAGE_SHIFT) && |
278 | pagestart + ((i + 1) << PAGE_SHIFT) <= eob) { |
279 | pgs[i]->flags &= ~PG_RDONLY; |
280 | } |
281 | pgs[i]->flags &= ~PG_CLEAN; |
282 | } |
283 | uvm_pageactivate(pgs[i]); |
284 | } |
285 | mutex_exit(&uvm_pageqlock); |
286 | uvm_page_unbusy(pgs, npages); |
287 | mutex_exit(uobj->vmobjlock); |
288 | |
289 | out: |
290 | kmem_free(pgs, pgssize); |
291 | return error; |
292 | } |
293 | |
294 | int |
295 | ufs_truncate_retry(struct vnode *vp, uint64_t newsize, kauth_cred_t cred) |
296 | { |
297 | struct inode *ip = VTOI(vp); |
298 | struct mount *mp = vp->v_mount; |
299 | int error = 0; |
300 | |
301 | UFS_WAPBL_JUNLOCK_ASSERT(mp); |
302 | |
303 | /* |
304 | * Truncate might temporarily fail, loop until done. |
305 | */ |
306 | while (ip->i_size != newsize) { |
307 | error = UFS_WAPBL_BEGIN(mp); |
308 | if (error) |
309 | goto out; |
310 | |
311 | error = UFS_TRUNCATE(vp, newsize, 0, cred); |
312 | UFS_WAPBL_END(mp); |
313 | |
314 | if (error != 0 && error != EAGAIN) |
315 | goto out; |
316 | } |
317 | |
318 | out: |
319 | return error; |
320 | } |
321 | |