xref: /freebsd/sys/kern/subr_disk.c (revision 74bf4e164ba5851606a27d4feff27717452583e5)
1 /*
2  * ----------------------------------------------------------------------------
3  * "THE BEER-WARE LICENSE" (Revision 42):
4  * <phk@FreeBSD.ORG> wrote this file.  As long as you retain this notice you
5  * can do whatever you want with this stuff. If we meet some day, and you think
6  * this stuff is worth it, you can buy me a beer in return.   Poul-Henning Kamp
7  * ----------------------------------------------------------------------------
8  */
9 
10 #include <sys/cdefs.h>
11 __FBSDID("$FreeBSD$");
12 
13 #include "opt_geom.h"
14 
15 #include <sys/param.h>
16 #include <sys/systm.h>
17 #include <sys/bio.h>
18 #include <sys/conf.h>
19 #include <sys/disk.h>
20 #include <geom/geom_disk.h>
21 
22 /*-
23  * Disk error is the preface to plaintive error messages
24  * about failing disk transfers.  It prints messages of the form
25  * 	"hp0g: BLABLABLA cmd=read fsbn 12345 of 12344-12347"
26  * blkdone should be -1 if the position of the error is unknown.
27  * The message is printed with printf.
28  */
29 void
30 disk_err(struct bio *bp, const char *what, int blkdone, int nl)
31 {
32 	daddr_t sn;
33 
34 	if (bp->bio_dev != NULL)
35 		printf("%s: %s ", devtoname(bp->bio_dev), what);
36 	else if (bp->bio_disk != NULL)
37 		printf("%s%d: %s ",
38 		    bp->bio_disk->d_name, bp->bio_disk->d_unit, what);
39 	else
40 		printf("disk??: %s ", what);
41 	switch(bp->bio_cmd) {
42 	case BIO_READ:		printf("cmd=read "); break;
43 	case BIO_WRITE:		printf("cmd=write "); break;
44 	case BIO_DELETE:	printf("cmd=delete "); break;
45 	case BIO_GETATTR:	printf("cmd=getattr "); break;
46 	default:		printf("cmd=%x ", bp->bio_cmd); break;
47 	}
48 	sn = bp->bio_pblkno;
49 	if (bp->bio_bcount <= DEV_BSIZE) {
50 		printf("fsbn %jd%s", (intmax_t)sn, nl ? "\n" : "");
51 		return;
52 	}
53 	if (blkdone >= 0) {
54 		sn += blkdone;
55 		printf("fsbn %jd of ", (intmax_t)sn);
56 	}
57 	printf("%jd-%jd", (intmax_t)bp->bio_pblkno,
58 	    (intmax_t)(bp->bio_pblkno + (bp->bio_bcount - 1) / DEV_BSIZE));
59 	if (nl)
60 		printf("\n");
61 }
62 
63 /*
64  * BIO queue implementation
65  */
66 
67 void
68 bioq_init(struct bio_queue_head *head)
69 {
70 	TAILQ_INIT(&head->queue);
71 	head->last_offset = 0;
72 	head->insert_point = NULL;
73 	head->switch_point = NULL;
74 }
75 
76 void
77 bioq_remove(struct bio_queue_head *head, struct bio *bp)
78 {
79 	if (bp == head->switch_point)
80 		head->switch_point = TAILQ_NEXT(bp, bio_queue);
81 	if (bp == head->insert_point) {
82 		head->insert_point = TAILQ_PREV(bp, bio_queue, bio_queue);
83 		if (head->insert_point == NULL)
84 			head->last_offset = 0;
85 	} else if (bp == TAILQ_FIRST(&head->queue))
86 		head->last_offset = bp->bio_offset;
87 	TAILQ_REMOVE(&head->queue, bp, bio_queue);
88 	if (TAILQ_FIRST(&head->queue) == head->switch_point)
89 		head->switch_point = NULL;
90 }
91 
92 void
93 bioq_flush(struct bio_queue_head *head, struct devstat *stp, int error)
94 {
95 	struct bio *bp;
96 
97 	while ((bp = bioq_takefirst(head)) != NULL)
98 		biofinish(bp, stp, error);
99 }
100 
101 void
102 bioq_insert_tail(struct bio_queue_head *head, struct bio *bp)
103 {
104 
105 	TAILQ_INSERT_TAIL(&head->queue, bp, bio_queue);
106 }
107 
108 struct bio *
109 bioq_first(struct bio_queue_head *head)
110 {
111 
112 	return (TAILQ_FIRST(&head->queue));
113 }
114 
115 struct bio *
116 bioq_takefirst(struct bio_queue_head *head)
117 {
118 	struct bio *bp;
119 
120 	bp = TAILQ_FIRST(&head->queue);
121 	if (bp != NULL)
122 		bioq_remove(head, bp);
123 	return (bp);
124 }
125 
126 /*
127  * Seek sort for disks.
128  *
129  * The buf_queue keep two queues, sorted in ascending block order.  The first
130  * queue holds those requests which are positioned after the current block
131  * (in the first request); the second, which starts at queue->switch_point,
132  * holds requests which came in after their block number was passed.  Thus
133  * we implement a one way scan, retracting after reaching the end of the drive
134  * to the first request on the second queue, at which time it becomes the
135  * first queue.
136  *
137  * A one-way scan is natural because of the way UNIX read-ahead blocks are
138  * allocated.
139  */
140 
141 void
142 bioq_disksort(bioq, bp)
143 	struct bio_queue_head *bioq;
144 	struct bio *bp;
145 {
146 	struct bio *bq;
147 	struct bio *bn;
148 	struct bio *be;
149 
150 	be = TAILQ_LAST(&bioq->queue, bio_queue);
151 	/*
152 	 * If the queue is empty or we are an
153 	 * ordered transaction, then it's easy.
154 	 */
155 	if ((bq = bioq_first(bioq)) == NULL) {
156 		bioq_insert_tail(bioq, bp);
157 		return;
158 	} else if (bioq->insert_point != NULL) {
159 
160 		/*
161 		 * A certain portion of the list is
162 		 * "locked" to preserve ordering, so
163 		 * we can only insert after the insert
164 		 * point.
165 		 */
166 		bq = bioq->insert_point;
167 	} else {
168 
169 		/*
170 		 * If we lie before the last removed (currently active)
171 		 * request, and are not inserting ourselves into the
172 		 * "locked" portion of the list, then we must add ourselves
173 		 * to the second request list.
174 		 */
175 		if (bp->bio_offset < bioq->last_offset) {
176 
177 			bq = bioq->switch_point;
178 			/*
179 			 * If we are starting a new secondary list,
180 			 * then it's easy.
181 			 */
182 			if (bq == NULL) {
183 				bioq->switch_point = bp;
184 				bioq_insert_tail(bioq, bp);
185 				return;
186 			}
187 			/*
188 			 * If we lie ahead of the current switch point,
189 			 * insert us before the switch point and move
190 			 * the switch point.
191 			 */
192 			if (bp->bio_offset < bq->bio_offset) {
193 				bioq->switch_point = bp;
194 				TAILQ_INSERT_BEFORE(bq, bp, bio_queue);
195 				return;
196 			}
197 		} else {
198 			if (bioq->switch_point != NULL)
199 				be = TAILQ_PREV(bioq->switch_point,
200 						bio_queue, bio_queue);
201 			/*
202 			 * If we lie between last_offset and bq,
203 			 * insert before bq.
204 			 */
205 			if (bp->bio_offset < bq->bio_offset) {
206 				TAILQ_INSERT_BEFORE(bq, bp, bio_queue);
207 				return;
208 			}
209 		}
210 	}
211 
212 	/*
213 	 * Request is at/after our current position in the list.
214 	 * Optimize for sequential I/O by seeing if we go at the tail.
215 	 */
216 	if (bp->bio_offset > be->bio_offset) {
217 		TAILQ_INSERT_AFTER(&bioq->queue, be, bp, bio_queue);
218 		return;
219 	}
220 
221 	/* Otherwise, insertion sort */
222 	while ((bn = TAILQ_NEXT(bq, bio_queue)) != NULL) {
223 
224 		/*
225 		 * We want to go after the current request if it is the end
226 		 * of the first request list, or if the next request is a
227 		 * larger cylinder than our request.
228 		 */
229 		if (bn == bioq->switch_point
230 		 || bp->bio_offset < bn->bio_offset)
231 			break;
232 		bq = bn;
233 	}
234 	TAILQ_INSERT_AFTER(&bioq->queue, bq, bp, bio_queue);
235 }
236 
237 
238