geom_io.c revision 114511 
1/*-
2 * Copyright (c) 2002 Poul-Henning Kamp
3 * Copyright (c) 2002 Networks Associates Technology, Inc.
4 * All rights reserved.
5 *
6 * This software was developed for the FreeBSD Project by Poul-Henning Kamp
7 * and NAI Labs, the Security Research Division of Network Associates, Inc.
8 * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
9 * DARPA CHATS research program.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 *    notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 *    notice, this list of conditions and the following disclaimer in the
18 *    documentation and/or other materials provided with the distribution.
19 * 3. The names of the authors may not be used to endorse or promote
20 *    products derived from this software without specific prior written
21 *    permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * $FreeBSD: head/sys/geom/geom_io.c 114511 2003-05-02 06:42:59Z phk $
36 */
37
38
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/kernel.h>
42#include <sys/malloc.h>
43#include <sys/bio.h>
44
45#include <sys/errno.h>
46#include <geom/geom.h>
47#include <geom/geom_int.h>
48#include <sys/devicestat.h>
49
50static struct g_bioq g_bio_run_down;
51static struct g_bioq g_bio_run_up;
52static struct g_bioq g_bio_idle;
53
54static u_int pace;
55
56#include <machine/atomic.h>
57
58static void
59g_bioq_lock(struct g_bioq *bq)
60{
61
62	mtx_lock(&bq->bio_queue_lock);
63}
64
65static void
66g_bioq_unlock(struct g_bioq *bq)
67{
68
69	mtx_unlock(&bq->bio_queue_lock);
70}
71
72#if 0
73static void
74g_bioq_destroy(struct g_bioq *bq)
75{
76
77	mtx_destroy(&bq->bio_queue_lock);
78}
79#endif
80
81static void
82g_bioq_init(struct g_bioq *bq)
83{
84
85	TAILQ_INIT(&bq->bio_queue);
86	mtx_init(&bq->bio_queue_lock, "bio queue", NULL, MTX_DEF);
87}
88
89static struct bio *
90g_bioq_first(struct g_bioq *bq)
91{
92	struct bio *bp;
93
94	bp = TAILQ_FIRST(&bq->bio_queue);
95	if (bp != NULL) {
96		TAILQ_REMOVE(&bq->bio_queue, bp, bio_queue);
97		bq->bio_queue_length--;
98	}
99	return (bp);
100}
101
102static void
103g_bioq_enqueue_tail(struct bio *bp, struct g_bioq *rq)
104{
105
106	g_bioq_lock(rq);
107	TAILQ_INSERT_TAIL(&rq->bio_queue, bp, bio_queue);
108	rq->bio_queue_length++;
109	g_bioq_unlock(rq);
110}
111
112struct bio *
113g_new_bio(void)
114{
115	struct bio *bp;
116
117	g_bioq_lock(&g_bio_idle);
118	bp = g_bioq_first(&g_bio_idle);
119	g_bioq_unlock(&g_bio_idle);
120	if (bp == NULL)
121		bp = g_malloc(sizeof *bp, M_NOWAIT | M_ZERO);
122	/* g_trace(G_T_BIO, "g_new_bio() = %p", bp); */
123	return (bp);
124}
125
126void
127g_destroy_bio(struct bio *bp)
128{
129
130	/* g_trace(G_T_BIO, "g_destroy_bio(%p)", bp); */
131	bzero(bp, sizeof *bp);
132	g_bioq_enqueue_tail(bp, &g_bio_idle);
133}
134
135struct bio *
136g_clone_bio(struct bio *bp)
137{
138	struct bio *bp2;
139
140	bp2 = g_new_bio();
141	if (bp2 != NULL) {
142		bp2->bio_parent = bp;
143		bp2->bio_cmd = bp->bio_cmd;
144		bp2->bio_length = bp->bio_length;
145		bp2->bio_offset = bp->bio_offset;
146		bp2->bio_data = bp->bio_data;
147		bp2->bio_attribute = bp->bio_attribute;
148		bp->bio_children++;
149	}
150	/* g_trace(G_T_BIO, "g_clone_bio(%p) = %p", bp, bp2); */
151	return(bp2);
152}
153
154void
155g_io_init()
156{
157
158	g_bioq_init(&g_bio_run_down);
159	g_bioq_init(&g_bio_run_up);
160	g_bioq_init(&g_bio_idle);
161}
162
163int
164g_io_getattr(const char *attr, struct g_consumer *cp, int *len, void *ptr)
165{
166	struct bio *bp;
167	int error;
168
169	g_trace(G_T_BIO, "bio_getattr(%s)", attr);
170	bp = g_new_bio();
171	bp->bio_cmd = BIO_GETATTR;
172	bp->bio_done = NULL;
173	bp->bio_attribute = attr;
174	bp->bio_length = *len;
175	bp->bio_data = ptr;
176	g_io_request(bp, cp);
177	error = biowait(bp, "ggetattr");
178	*len = bp->bio_completed;
179	g_destroy_bio(bp);
180	return (error);
181}
182
183static int
184g_io_check(struct bio *bp)
185{
186	struct g_consumer *cp;
187	struct g_provider *pp;
188
189	cp = bp->bio_from;
190	pp = bp->bio_to;
191
192	/* Fail if access counters dont allow the operation */
193	switch(bp->bio_cmd) {
194	case BIO_READ:
195	case BIO_GETATTR:
196		if (cp->acr == 0)
197			return (EPERM);
198		break;
199	case BIO_WRITE:
200	case BIO_DELETE:
201		if (cp->acw == 0)
202			return (EPERM);
203		break;
204	default:
205		return (EPERM);
206	}
207	/* if provider is marked for error, don't disturb. */
208	if (pp->error)
209		return (pp->error);
210
211	switch(bp->bio_cmd) {
212	case BIO_READ:
213	case BIO_WRITE:
214	case BIO_DELETE:
215		/* Reject I/O not on sector boundary */
216		if (bp->bio_offset % pp->sectorsize)
217			return (EINVAL);
218		/* Reject I/O not integral sector long */
219		if (bp->bio_length % pp->sectorsize)
220			return (EINVAL);
221		/* Reject requests past the end of media. */
222		if (bp->bio_offset > pp->mediasize)
223			return (EIO);
224		break;
225	default:
226		break;
227	}
228	return (0);
229}
230
231void
232g_io_request(struct bio *bp, struct g_consumer *cp)
233{
234	struct g_provider *pp;
235
236	pp = cp->provider;
237	KASSERT(cp != NULL, ("NULL cp in g_io_request"));
238	KASSERT(bp != NULL, ("NULL bp in g_io_request"));
239	KASSERT(bp->bio_data != NULL, ("NULL bp->data in g_io_request"));
240	KASSERT(pp != NULL, ("consumer not attached in g_io_request"));
241
242	bp->bio_from = cp;
243	bp->bio_to = pp;
244	bp->bio_error = 0;
245	bp->bio_completed = 0;
246
247	if (g_collectstats) {
248		devstat_start_transaction_bio(cp->stat, bp);
249		devstat_start_transaction_bio(pp->stat, bp);
250	}
251	cp->nstart++;
252	pp->nstart++;
253
254	/* Pass it on down. */
255	g_trace(G_T_BIO, "bio_request(%p) from %p(%s) to %p(%s) cmd %d",
256	    bp, cp, cp->geom->name, pp, pp->name, bp->bio_cmd);
257	g_bioq_enqueue_tail(bp, &g_bio_run_down);
258	wakeup(&g_wait_down);
259}
260
261void
262g_io_deliver(struct bio *bp, int error)
263{
264	struct g_consumer *cp;
265	struct g_provider *pp;
266
267	cp = bp->bio_from;
268	pp = bp->bio_to;
269	KASSERT(bp != NULL, ("NULL bp in g_io_deliver"));
270	KASSERT(cp != NULL, ("NULL bio_from in g_io_deliver"));
271	KASSERT(cp->geom != NULL, ("NULL bio_from->geom in g_io_deliver"));
272	KASSERT(pp != NULL, ("NULL bio_to in g_io_deliver"));
273
274	g_trace(G_T_BIO,
275"g_io_deliver(%p) from %p(%s) to %p(%s) cmd %d error %d off %jd len %jd",
276	    bp, cp, cp->geom->name, pp, pp->name, bp->bio_cmd, error,
277	    (intmax_t)bp->bio_offset, (intmax_t)bp->bio_length);
278
279	bp->bio_bcount = bp->bio_length;
280	if (g_collectstats) {
281		bp->bio_resid = bp->bio_bcount - bp->bio_completed;
282		devstat_end_transaction_bio(cp->stat, bp);
283		devstat_end_transaction_bio(pp->stat, bp);
284	}
285	cp->nend++;
286	pp->nend++;
287
288	if (error == ENOMEM) {
289		printf("ENOMEM %p on %p(%s)\n", bp, pp, pp->name);
290		g_io_request(bp, cp);
291		pace++;
292		return;
293	}
294	bp->bio_error = error;
295	g_bioq_enqueue_tail(bp, &g_bio_run_up);
296	wakeup(&g_wait_up);
297}
298
299void
300g_io_schedule_down(struct thread *tp __unused)
301{
302	struct bio *bp;
303	off_t excess;
304	int error;
305	struct mtx mymutex;
306
307	bzero(&mymutex, sizeof mymutex);
308	mtx_init(&mymutex, "g_xdown", MTX_DEF, 0);
309
310	for(;;) {
311		g_bioq_lock(&g_bio_run_down);
312		bp = g_bioq_first(&g_bio_run_down);
313		if (bp == NULL) {
314			msleep(&g_wait_down, &g_bio_run_down.bio_queue_lock,
315			    PRIBIO | PDROP, "g_down", hz/10);
316			continue;
317		}
318		g_bioq_unlock(&g_bio_run_down);
319		if (pace > 0) {
320			msleep(&error, NULL, PRIBIO, "g_down", hz/10);
321			pace--;
322		}
323		error = g_io_check(bp);
324		if (error) {
325			g_io_deliver(bp, error);
326			continue;
327		}
328		switch (bp->bio_cmd) {
329		case BIO_READ:
330		case BIO_WRITE:
331		case BIO_DELETE:
332			/* Truncate requests to the end of providers media. */
333			excess = bp->bio_offset + bp->bio_length;
334			if (excess > bp->bio_to->mediasize) {
335				excess -= bp->bio_to->mediasize;
336				bp->bio_length -= excess;
337			}
338			/* Deliver zero length transfers right here. */
339			if (bp->bio_length == 0) {
340				g_io_deliver(bp, 0);
341				continue;
342			}
343			break;
344		default:
345			break;
346		}
347		mtx_lock(&mymutex);
348		bp->bio_to->geom->start(bp);
349		mtx_unlock(&mymutex);
350	}
351}
352
353void
354g_io_schedule_up(struct thread *tp __unused)
355{
356	struct bio *bp;
357	struct mtx mymutex;
358
359	bzero(&mymutex, sizeof mymutex);
360	mtx_init(&mymutex, "g_xup", MTX_DEF, 0);
361	for(;;) {
362		g_bioq_lock(&g_bio_run_up);
363		bp = g_bioq_first(&g_bio_run_up);
364		if (bp != NULL) {
365			g_bioq_unlock(&g_bio_run_up);
366			mtx_lock(&mymutex);
367			biodone(bp);
368			mtx_unlock(&mymutex);
369			continue;
370		}
371		msleep(&g_wait_up, &g_bio_run_up.bio_queue_lock,
372		    PRIBIO | PDROP, "g_up", hz/10);
373	}
374}
375
376void *
377g_read_data(struct g_consumer *cp, off_t offset, off_t length, int *error)
378{
379	struct bio *bp;
380	void *ptr;
381	int errorc;
382
383	bp = g_new_bio();
384	bp->bio_cmd = BIO_READ;
385	bp->bio_done = NULL;
386	bp->bio_offset = offset;
387	bp->bio_length = length;
388	ptr = g_malloc(length, M_WAITOK);
389	bp->bio_data = ptr;
390	g_io_request(bp, cp);
391	errorc = biowait(bp, "gread");
392	if (error != NULL)
393		*error = errorc;
394	g_destroy_bio(bp);
395	if (errorc) {
396		g_free(ptr);
397		ptr = NULL;
398	}
399	return (ptr);
400}
401
402int
403g_write_data(struct g_consumer *cp, off_t offset, void *ptr, off_t length)
404{
405	struct bio *bp;
406	int error;
407
408	bp = g_new_bio();
409	bp->bio_cmd = BIO_WRITE;
410	bp->bio_done = NULL;
411	bp->bio_offset = offset;
412	bp->bio_length = length;
413	bp->bio_data = ptr;
414	g_io_request(bp, cp);
415	error = biowait(bp, "gwrite");
416	g_destroy_bio(bp);
417	return (error);
418}
419