summaryrefslogtreecommitdiff
path: root/sys/kern/kern_physio.c
blob: 9d891a757b3268d65403e8d225480b40beea60e5 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
/*	$OpenBSD: kern_physio.c,v 1.12 2001/06/26 06:27:39 aaron Exp $	*/
/*	$NetBSD: kern_physio.c,v 1.28 1997/05/19 10:43:28 pk Exp $	*/

/*-
 * Copyright (c) 1994 Christopher G. Demetriou
 * Copyright (c) 1982, 1986, 1990, 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.
 *
 *	@(#)kern_physio.c	8.1 (Berkeley) 6/10/93
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/proc.h>
#include <sys/malloc.h>

#include <vm/vm.h>

#if defined(UVM)
#include <uvm/uvm_extern.h>
#endif

/*
 * The routines implemented in this file are described in:
 *	Leffler, et al.: The Design and Implementation of the 4.3BSD
 *	    UNIX Operating System (Addison Welley, 1989)
 * on pages 231-233.
 *
 * The routines "getphysbuf" and "putphysbuf" steal and return a swap
 * buffer.  Leffler, et al., says that swap buffers are used to do the
 * I/O, so raw I/O requests don't have to be single-threaded.
 */

struct buf *getphysbuf __P((void));
void putphysbuf __P((struct buf *bp));

/*
 * Do "physical I/O" on behalf of a user.  "Physical I/O" is I/O directly
 * from the raw device to user buffers, and bypasses the buffer cache.
 *
 * Comments in brackets are from Leffler, et al.'s pseudo-code implementation.
 */
int
physio(strategy, bp, dev, flags, minphys, uio)
	void (*strategy) __P((struct buf *));
	struct buf *bp;
	dev_t dev;
	int flags;
	void (*minphys) __P((struct buf *));
	struct uio *uio;
{
	struct iovec *iovp;
	struct proc *p = curproc;
	int error, done, i, nobuf, s, todo;

	error = 0;
	flags &= B_READ | B_WRITE;

	/*
	 * [check user read/write access to the data buffer]
	 *
	 * Check each iov one by one.  Note that we know if we're reading or
	 * writing, so we ignore the uio's rw parameter.  Also note that if
	 * we're doing a read, that's a *write* to user-space.
	 */
	if (uio->uio_segflg == UIO_USERSPACE)
		for (i = 0; i < uio->uio_iovcnt; i++)
#if defined(UVM) /* XXXCDC: map not locked, rethink */
			/* XXX - obsolete now that vslock can error? */
			if (!uvm_useracc(uio->uio_iov[i].iov_base,
			    uio->uio_iov[i].iov_len,
			    (flags == B_READ) ? B_WRITE : B_READ))
				return (EFAULT);
#else
			if (!useracc(uio->uio_iov[i].iov_base,
			    uio->uio_iov[i].iov_len,
			    (flags == B_READ) ? B_WRITE : B_READ))
				return (EFAULT);
#endif

	/* Make sure we have a buffer, creating one if necessary. */
	if ((nobuf = (bp == NULL)) != 0)
		bp = getphysbuf();

	/* [raise the processor priority level to splbio;] */
	s = splbio();

	/* [while the buffer is marked busy] */
	while (bp->b_flags & B_BUSY) {
		/* [mark the buffer wanted] */
		bp->b_flags |= B_WANTED;
		/* [wait until the buffer is available] */
		tsleep((caddr_t)bp, PRIBIO+1, "physbuf", 0);
	}

	/* Mark it busy, so nobody else will use it. */
	bp->b_flags |= B_BUSY;

	/* [lower the priority level] */
	splx(s);

	/* [set up the fixed part of the buffer for a transfer] */
	bp->b_dev = dev;
	bp->b_error = 0;
	bp->b_proc = p;
	LIST_INIT(&bp->b_dep);

	/*
	 * [while there are data to transfer and no I/O error]
	 * Note that I/O errors are handled with a 'goto' at the bottom
	 * of the 'while' loop.
	 */
	for (i = 0; i < uio->uio_iovcnt; i++) {
		iovp = &uio->uio_iov[i];
		while (iovp->iov_len > 0) {
			/*
			 * [mark the buffer busy for physical I/O]
			 * (i.e. set B_PHYS (because it's an I/O to user
			 * memory, and B_RAW, because B_RAW is to be
			 * "Set by physio for raw transfers.", in addition
			 * to the "busy" and read/write flag.)
			 */
			bp->b_flags = B_BUSY | B_PHYS | B_RAW | flags;

			/* [set up the buffer for a maximum-sized transfer] */
			bp->b_blkno = btodb(uio->uio_offset);
			bp->b_bcount = iovp->iov_len;
			bp->b_data = iovp->iov_base;

			/*
			 * [call minphys to bound the tranfer size]
			 * and remember the amount of data to transfer,
			 * for later comparison.
			 */
			(*minphys)(bp);
			todo = bp->b_bcount;
#ifdef DIAGNOSTIC
			if (todo < 0)
				panic("todo < 0; minphys broken");
			if (todo > MAXPHYS)
				panic("todo > MAXPHYS; minphys broken");
#endif

			/*
			 * [lock the part of the user address space involved
			 *    in the transfer]
			 * Beware vmapbuf(); it clobbers b_data and
			 * saves it in b_saveaddr.  However, vunmapbuf()
			 * restores it.
			 */
			PHOLD(p);
#if defined(UVM)
			if (uvm_vslock(p, bp->b_data, todo, (flags & B_READ) ?
			    VM_PROT_READ | VM_PROT_WRITE : VM_PROT_READ) !=
			    KERN_SUCCESS) {
				bp->b_flags |= B_ERROR;
				bp->b_error = EFAULT;
				goto after_unlock;
			}
#else
			vslock(bp->b_data, todo);
#endif
			vmapbuf(bp, todo);

			/* [call strategy to start the transfer] */
			(*strategy)(bp);

			/*
			 * Note that the raise/wait/lower/get error
			 * steps below would be done by biowait(), but
			 * we want to unlock the address space before
			 * we lower the priority.
			 *
			 * [raise the priority level to splbio]
			 */
			s = splbio();

			/* [wait for the transfer to complete] */
			while ((bp->b_flags & B_DONE) == 0)
				tsleep((caddr_t) bp, PRIBIO + 1, "physio", 0);

			/* Mark it busy again, so nobody else will use it. */
			bp->b_flags |= B_BUSY;

			/* [lower the priority level] */
			splx(s);

			/*
			 * [unlock the part of the address space previously
			 *    locked]
			 */
			vunmapbuf(bp, todo);
#if defined(UVM)
			uvm_vsunlock(p, bp->b_data, todo);
after_unlock:
#else
			vsunlock(bp->b_data, todo);
#endif
			PRELE(p);

			/* remember error value (save a splbio/splx pair) */
			if (bp->b_flags & B_ERROR)
				error = (bp->b_error ? bp->b_error : EIO);

			/*
			 * [deduct the transfer size from the total number
			 *    of data to transfer]
			 */
			done = bp->b_bcount - bp->b_resid;
#ifdef DIAGNOSTIC
			if (done < 0)
				panic("done < 0; strategy broken");
			if (done > todo)
				panic("done > todo; strategy broken");
#endif
			iovp->iov_len -= done;
			iovp->iov_base = (caddr_t)iovp->iov_base + done;
			uio->uio_offset += done;
			uio->uio_resid -= done;

			/*
			 * Now, check for an error.
			 * Also, handle weird end-of-disk semantics.
			 */
			if (error || done < todo)
				goto done;
		}
	}

done:
	/*
	 * [clean up the state of the buffer]
	 * Remember if somebody wants it, so we can wake them up below.
	 * Also, if we had to steal it, give it back.
	 */
	s = splbio();
	bp->b_flags &= ~(B_BUSY | B_PHYS | B_RAW);
	if (nobuf)
		putphysbuf(bp);
	else {
		/*
		 * [if another process is waiting for the raw I/O buffer,
		 *    wake up processes waiting to do physical I/O;
		 */
		if (bp->b_flags & B_WANTED) {
			bp->b_flags &= ~B_WANTED;
			wakeup(bp);
		}
	}
	splx(s);

	return (error);
}

/*
 * Get a swap buffer structure, for use in physical I/O.
 * Mostly taken from /sys/vm/swap_pager.c, except that it no longer
 * records buffer list-empty conditions, and sleeps at PRIBIO + 1,
 * rather than PSWP + 1 (and on a different wchan).
 */
struct buf *
getphysbuf()
{
	struct buf *bp;
#if !defined(UVM)
	int s;

	s = splbio();
	while (bswlist.b_actf == NULL) {
		bswlist.b_flags |= B_WANTED;
		tsleep((caddr_t)&bswlist, PRIBIO + 1, "getphys", 0);
	}
	bp = bswlist.b_actf;
	bswlist.b_actf = bp->b_actf;
	splx(s);
#else

	bp = malloc(sizeof(*bp), M_TEMP, M_WAITOK);
	bzero(bp, sizeof(*bp));

	/* XXXCDC: are the following two lines necessary? */
	bp->b_rcred = bp->b_wcred = NOCRED;
	bp->b_vnbufs.le_next = NOLIST;
#endif
	return (bp);
}

/*
 * Get rid of a swap buffer structure which has been used in physical I/O.
 * Mostly taken from /sys/vm/swap_pager.c, except that it now uses
 * wakeup() rather than the VM-internal thread_wakeup(), and that the caller
 * must mask disk interrupts, rather than putphysbuf() itself.
 */
void
putphysbuf(bp)
	struct buf *bp;
{
#if !defined(UVM)
	bp->b_actf = bswlist.b_actf;
	bswlist.b_actf = bp;
	if (bp->b_vp)
		brelvp(bp);
	if (bswlist.b_flags & B_WANTED) {
		bswlist.b_flags &= ~B_WANTED;
		wakeup(&bswlist);
	}
#else
	/* XXXCDC: is this necesary? */
	if (bp->b_vp)
		brelvp(bp);

	if (bp->b_flags & B_WANTED)
		panic("putphysbuf: private buf B_WANTED");
	free(bp, M_TEMP);
#endif
}

/*
 * Leffler, et al., says on p. 231:
 * "The minphys() routine is called by physio() to adjust the
 * size of each I/O transfer before the latter is passed to
 * the strategy routine..."
 *
 * so, just adjust the buffer's count accounting to MAXPHYS here,
 * and return the new count;
 */
void
minphys(bp)
	struct buf *bp;
{

	if (bp->b_bcount > MAXPHYS)
		bp->b_bcount = MAXPHYS;
}