kernel - use new td_ucred in numerous places
[dragonfly.git] / sys / vm / vm_swap.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1982, 1986, 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)vm_swap.c 8.5 (Berkeley) 2/17/94
34 * $FreeBSD: src/sys/vm/vm_swap.c,v 1.96.2.2 2001/10/14 18:46:47 iedowse Exp $
c34665ce 35 * $DragonFly: src/sys/vm/vm_swap.c,v 1.36 2007/07/20 17:21:54 dillon Exp $
984263bc
MD
36 */
37
38#include "opt_swap.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/sysproto.h>
43#include <sys/buf.h>
44#include <sys/proc.h>
895c1f85 45#include <sys/priv.h>
fad57d0e 46#include <sys/nlookup.h>
984263bc
MD
47#include <sys/dmap.h> /* XXX */
48#include <sys/vnode.h>
49#include <sys/fcntl.h>
50#include <sys/blist.h>
51#include <sys/kernel.h>
52#include <sys/lock.h>
53#include <sys/conf.h>
54#include <sys/stat.h>
cdd46d2e 55#include <sys/thread2.h>
984263bc
MD
56#include <vm/vm.h>
57#include <vm/vm_extern.h>
58#include <vm/swap_pager.h>
59#include <vm/vm_zone.h>
60
61/*
62 * Indirect driver for multi-controller paging.
63 */
64
65#ifndef NSWAPDEV
66#define NSWAPDEV 4
67#endif
68static struct swdevt should_be_malloced[NSWAPDEV];
460426e6 69struct swdevt *swdevt = should_be_malloced; /* exported to pstat/systat */
79634a66 70static swblk_t nswap; /* first block after the interleaved devs */
460426e6 71int nswdev = NSWAPDEV; /* exported to pstat/systat */
984263bc
MD
72int vm_swap_size;
73
1388df65 74static int swapdev_strategy (struct vop_strategy_args *ap);
984263bc
MD
75struct vnode *swapdev_vp;
76
77/*
78 * swapdev_strategy:
79 *
81b5c339 80 * vn_strategy() for swapdev_vp.
984263bc
MD
81 * Perform swap strategy interleave device selection.
82 *
10f3fee5 83 * The bp is expected to be locked and on call.
81b5c339
MD
84 *
85 * (struct vnode *a_vp, struct bio *b_bio)
984263bc
MD
86 */
87
88static int
81b5c339 89swapdev_strategy(struct vop_strategy_args *ap)
984263bc 90{
81b5c339
MD
91 struct bio *bio = ap->a_bio;
92 struct bio *nbio;
93 struct buf *bp = bio->bio_buf;
54078292 94 int sz, off, seg, index, blkno, nblkno;
5f910b2f 95 struct swdevt *sp;
984263bc 96 struct vnode *vp;
984263bc 97
81b5c339 98 vp = ap->a_vp;
984263bc 99 sz = howmany(bp->b_bcount, PAGE_SIZE);
54078292 100 blkno = (int)(bio->bio_offset >> PAGE_SHIFT);
984263bc
MD
101
102 /*
103 * Convert interleaved swap into per-device swap. Note that
104 * the block size is left in PAGE_SIZE'd chunks (for the newswap)
105 * here.
106 */
81b5c339 107 nbio = push_bio(bio);
984263bc 108 if (nswdev > 1) {
54078292 109 off = blkno % dmmax;
984263bc
MD
110 if (off + sz > dmmax) {
111 bp->b_error = EINVAL;
112 bp->b_flags |= B_ERROR;
81b5c339 113 biodone(bio);
984263bc
MD
114 return 0;
115 }
54078292 116 seg = blkno / dmmax;
984263bc
MD
117 index = seg % nswdev;
118 seg /= nswdev;
54078292 119 nbio->bio_offset = (off_t)(seg * dmmax + off) << PAGE_SHIFT;
984263bc
MD
120 } else {
121 index = 0;
54078292 122 nbio->bio_offset = bio->bio_offset;
984263bc 123 }
54078292 124 nblkno = (int)(nbio->bio_offset >> PAGE_SHIFT);
984263bc 125 sp = &swdevt[index];
54078292 126 if (nblkno + sz > sp->sw_nblks) {
984263bc
MD
127 bp->b_error = EINVAL;
128 bp->b_flags |= B_ERROR;
81b5c339
MD
129 /* I/O was never started on nbio, must biodone(bio) */
130 biodone(bio);
984263bc
MD
131 return 0;
132 }
984263bc
MD
133 if (sp->sw_vp == NULL) {
134 bp->b_error = ENODEV;
135 bp->b_flags |= B_ERROR;
81b5c339
MD
136 /* I/O was never started on nbio, must biodone(bio) */
137 biodone(bio);
984263bc
MD
138 return 0;
139 }
140
141 /*
81b5c339
MD
142 * Issue a strategy call on the appropriate swap vnode. Note that
143 * bp->b_vp is not modified. Strategy code is always supposed to
144 * use the passed vp.
145 *
c34665ce
MD
146 * We have to use vn_strategy() here even if we know we have a
147 * device in order to properly break up requests which exceed the
148 * device's DMA limits.
984263bc 149 */
81b5c339 150 vn_strategy(sp->sw_vp, nbio);
984263bc
MD
151 return 0;
152}
153
154/*
155 * Create a special vnode op vector for swapdev_vp - we only use
81b5c339 156 * vn_strategy(), everything else returns an error.
984263bc 157 */
66a1ddf5
MD
158static struct vop_ops swapdev_vnode_vops = {
159 .vop_default = vop_defaultop,
160 .vop_strategy = swapdev_strategy
984263bc 161};
66a1ddf5 162static struct vop_ops *swapdev_vnode_vops_p = &swapdev_vnode_vops;
984263bc 163
66a1ddf5 164VNODEOP_SET(swapdev_vnode_vops);
984263bc
MD
165
166/*
41c20dac
MD
167 * swapon_args(char *name)
168 *
984263bc
MD
169 * System call swapon(name) enables swapping on device name,
170 * which must be in the swdevsw. Return EBUSY
171 * if already swapping on this device.
3919ced0
MD
172 *
173 * MPALMOSTSAFE
984263bc 174 */
984263bc 175int
753fd850 176sys_swapon(struct swapon_args *uap)
984263bc 177{
dadab5e9 178 struct thread *td = curthread;
984263bc 179 struct vattr attr;
5f910b2f 180 struct vnode *vp;
fad57d0e 181 struct nlookupdata nd;
984263bc 182 int error;
dadab5e9 183 struct ucred *cred;
984263bc 184
9910d07b 185 cred = td->td_ucred;
dadab5e9 186
895c1f85 187 error = priv_check(td, PRIV_ROOT);
984263bc
MD
188 if (error)
189 return (error);
190
3919ced0 191 get_mplock();
fad57d0e
MD
192 vp = NULL;
193 error = nlookup_init(&nd, uap->name, UIO_USERSPACE, NLC_FOLLOW);
194 if (error == 0)
195 error = nlookup(&nd);
196 if (error == 0)
28623bf9 197 error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
fad57d0e 198 nlookup_done(&nd);
3919ced0
MD
199 if (error) {
200 rel_mplock();
984263bc 201 return (error);
3919ced0 202 }
984263bc 203
3919ced0 204 if (vn_isdisk(vp, &error)) {
e4c9c0c8 205 error = swaponvp(td, vp, 0);
3919ced0
MD
206 } else if (vp->v_type == VREG && vp->v_tag == VT_NFS &&
207 (error = VOP_GETATTR(vp, &attr)) == 0) {
984263bc
MD
208 /*
209 * Allow direct swapping to NFS regular files in the same
210 * way that nfs_mountroot() sets up diskless swapping.
211 */
e4c9c0c8 212 error = swaponvp(td, vp, attr.va_size / DEV_BSIZE);
984263bc 213 }
984263bc
MD
214 if (error)
215 vrele(vp);
3919ced0 216 rel_mplock();
984263bc
MD
217
218 return (error);
219}
220
221/*
222 * Swfree(index) frees the index'th portion of the swap map.
223 * Each of the nswdev devices provides 1/nswdev'th of the swap
224 * space, which is laid out with blocks of dmmax pages circularly
225 * among the devices.
226 *
227 * The new swap code uses page-sized blocks. The old swap code used
228 * DEV_BSIZE'd chunks.
229 *
230 * XXX locking when multiple swapon's run in parallel
231 */
232int
79634a66 233swaponvp(struct thread *td, struct vnode *vp, u_quad_t nblks)
984263bc 234{
79634a66 235 swblk_t aligned_nblks;
e0fc5693 236 int64_t dpsize;
e4c9c0c8 237 struct ucred *cred;
5f910b2f
RG
238 struct swdevt *sp;
239 swblk_t vsbase;
984263bc 240 swblk_t dvbase;
b13267a5 241 cdev_t dev;
e4c9c0c8 242 int index;
984263bc 243 int error;
e4c9c0c8 244 long blk;
dadab5e9 245
9910d07b 246 cred = td->td_ucred;
984263bc
MD
247
248 if (!swapdev_vp) {
66a1ddf5 249 error = getspecialvnode(VT_NON, NULL, &swapdev_vnode_vops_p,
3446c007 250 &swapdev_vp, 0, 0);
984263bc
MD
251 if (error)
252 panic("Cannot get vnode for swapdev");
253 swapdev_vp->v_type = VNON; /* Untyped */
5fd012e0 254 vx_unlock(swapdev_vp);
984263bc
MD
255 }
256
984263bc
MD
257 for (sp = swdevt, index = 0 ; index < nswdev; index++, sp++) {
258 if (sp->sw_vp == vp)
259 return EBUSY;
260 if (!sp->sw_vp)
261 goto found;
262
263 }
264 return EINVAL;
265 found:
ca466bae 266 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
87de5057 267 error = VOP_OPEN(vp, FREAD | FWRITE, cred, NULL);
a11aaa81 268 vn_unlock(vp);
984263bc
MD
269 if (error)
270 return (error);
271
e4c9c0c8
MD
272 /*
273 * v_rdev is not valid until after the VOP_OPEN() call. dev_psize()
274 * must be supported if a character device has been specified.
275 */
276 if (vp->v_type == VCHR)
277 dev = vp->v_rdev;
278 else
028066b1 279 dev = NULL;
e4c9c0c8 280
e0fc5693
MD
281 if (nblks == 0 && dev != NULL) {
282 dpsize = dev_dpsize(dev);
283 if (dpsize == -1) {
284 VOP_CLOSE(vp, FREAD | FWRITE);
285 return (ENXIO);
286 }
79634a66 287 nblks = (u_quad_t)dpsize;
984263bc
MD
288 }
289 if (nblks == 0) {
87de5057 290 VOP_CLOSE(vp, FREAD | FWRITE);
984263bc
MD
291 return (ENXIO);
292 }
293
294 /*
984263bc
MD
295 * nblks is in DEV_BSIZE'd chunks, convert to PAGE_SIZE'd chunks.
296 * First chop nblks off to page-align it, then convert.
297 *
298 * sw->sw_nblks is in page-sized chunks now too.
299 */
79634a66 300 nblks &= ~(u_quad_t)(ctodb(1) - 1);
984263bc
MD
301 nblks = dbtoc(nblks);
302
79634a66
MD
303 /*
304 * Post-conversion nblks must not be >= BLIST_MAXBLKS, and
305 * we impose a 4-swap-device limit so we have to divide it out
306 * further. Going beyond this will result in overflows in the
307 * blist code.
308 *
309 * Post-conversion nblks must fit within a (swblk_t), which
310 * this test also ensures.
311 */
312 if (nblks > BLIST_MAXBLKS / nswdev) {
313 kprintf("exceeded maximum of %d blocks per swap unit\n",
314 (int)BLIST_MAXBLKS / nswdev);
315 VOP_CLOSE(vp, FREAD | FWRITE);
316 return (ENXIO);
317 }
318
984263bc
MD
319 sp->sw_vp = vp;
320 sp->sw_dev = dev2udev(dev);
321 sp->sw_device = dev;
322 sp->sw_flags |= SW_FREED;
79634a66 323 sp->sw_nblks = (swblk_t)nblks;
984263bc
MD
324
325 /*
326 * nblks, nswap, and dmmax are PAGE_SIZE'd parameters now, not
327 * DEV_BSIZE'd. aligned_nblks is used to calculate the
328 * size of the swap bitmap, taking into account the stripe size.
329 */
79634a66 330 aligned_nblks = (swblk_t)((nblks + (dmmax - 1)) & ~(u_long)(dmmax - 1));
984263bc
MD
331
332 if (aligned_nblks * nswdev > nswap)
333 nswap = aligned_nblks * nswdev;
334
335 if (swapblist == NULL)
336 swapblist = blist_create(nswap);
337 else
338 blist_resize(&swapblist, nswap, 0);
339
340 for (dvbase = dmmax; dvbase < nblks; dvbase += dmmax) {
341 blk = min(nblks - dvbase, dmmax);
342 vsbase = index * dmmax + dvbase * nswdev;
343 blist_free(swapblist, vsbase, blk);
344 vm_swap_size += blk;
345 }
c84c24da 346 swap_pager_newswap();
984263bc
MD
347
348 return (0);
349}