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
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
|
/*-
* Copyright (c) 2008 Yahoo!, Inc.
* All rights reserved.
* Written by: John Baldwin <jhb@FreeBSD.org>
*
* 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. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/proc.h>
#include <sys/sglist.h>
#include <sys/uio.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <sys/ktr.h>
static MALLOC_DEFINE(M_SGLIST, "sglist", "scatter/gather lists");
/*
* Convenience macros to save the state of an sglist so it can be restored
* if an append attempt fails. Since sglist's only grow we only need to
* save the current count of segments and the length of the ending segment.
* Earlier segments will not be changed by an append, and the only change
* that can occur to the ending segment is that it can be extended.
*/
struct sgsave {
u_short sg_nseg;
size_t ss_len;
};
#define SGLIST_SAVE(sg, sgsave) do { \
(sgsave).sg_nseg = (sg)->sg_nseg; \
if ((sgsave).sg_nseg > 0) \
(sgsave).ss_len = (sg)->sg_segs[(sgsave).sg_nseg - 1].ss_len; \
else \
(sgsave).ss_len = 0; \
} while (0)
#define SGLIST_RESTORE(sg, sgsave) do { \
(sg)->sg_nseg = (sgsave).sg_nseg; \
if ((sgsave).sg_nseg > 0) \
(sg)->sg_segs[(sgsave).sg_nseg - 1].ss_len = (sgsave).ss_len; \
} while (0)
/*
* Append a single (paddr, len) to a sglist. sg is the list and ss is
* the current segment in the list. If we run out of segments then
* EFBIG will be returned.
*/
static __inline int
_sglist_append_range(struct sglist *sg, struct sglist_seg **ssp,
vm_paddr_t paddr, size_t len)
{
struct sglist_seg *ss;
ss = *ssp;
if (ss->ss_paddr + ss->ss_len == paddr)
ss->ss_len += len;
else {
if (sg->sg_nseg == sg->sg_maxseg)
return (EFBIG);
ss++;
ss->ss_paddr = paddr;
ss->ss_len = len;
sg->sg_nseg++;
*ssp = ss;
}
return (0);
}
/*
* Worker routine to append a virtual address range (either kernel or
* user) to a scatter/gather list.
*/
static __inline int
_sglist_append_buf(struct sglist *sg, void *buf, size_t len, pmap_t pmap,
size_t *donep)
{
struct sglist_seg *ss;
vm_offset_t vaddr, offset;
vm_paddr_t paddr;
size_t seglen;
int error;
if (donep)
*donep = 0;
if (len == 0)
return (0);
/* Do the first page. It may have an offset. */
vaddr = (vm_offset_t)buf;
offset = vaddr & PAGE_MASK;
if (pmap != NULL)
paddr = pmap_extract(pmap, vaddr);
else
paddr = pmap_kextract(vaddr);
seglen = MIN(len, PAGE_SIZE - offset);
if (sg->sg_nseg == 0) {
ss = sg->sg_segs;
ss->ss_paddr = paddr;
ss->ss_len = seglen;
sg->sg_nseg = 1;
} else {
ss = &sg->sg_segs[sg->sg_nseg - 1];
error = _sglist_append_range(sg, &ss, paddr, seglen);
if (error)
return (error);
}
vaddr += seglen;
len -= seglen;
if (donep)
*donep += seglen;
while (len > 0) {
seglen = MIN(len, PAGE_SIZE);
if (pmap != NULL)
paddr = pmap_extract(pmap, vaddr);
else
paddr = pmap_kextract(vaddr);
error = _sglist_append_range(sg, &ss, paddr, seglen);
if (error)
return (error);
vaddr += seglen;
len -= seglen;
if (donep)
*donep += seglen;
}
return (0);
}
/*
* Determine the number of scatter/gather list elements needed to
* describe a kernel virtual address range.
*/
int
sglist_count(void *buf, size_t len)
{
vm_offset_t vaddr, vendaddr;
vm_paddr_t lastaddr, paddr;
int nsegs;
if (len == 0)
return (0);
vaddr = trunc_page((vm_offset_t)buf);
vendaddr = (vm_offset_t)buf + len;
nsegs = 1;
lastaddr = pmap_kextract(vaddr);
vaddr += PAGE_SIZE;
while (vaddr < vendaddr) {
paddr = pmap_kextract(vaddr);
if (lastaddr + PAGE_SIZE != paddr)
nsegs++;
lastaddr = paddr;
vaddr += PAGE_SIZE;
}
return (nsegs);
}
/*
* Allocate a scatter/gather list along with 'nsegs' segments. The
* 'mflags' parameters are the same as passed to malloc(9). The caller
* should use sglist_free() to free this list.
*/
struct sglist *
sglist_alloc(int nsegs, int mflags)
{
struct sglist *sg;
sg = malloc(sizeof(struct sglist) + nsegs * sizeof(struct sglist_seg),
M_SGLIST, mflags);
if (sg == NULL)
return (NULL);
sglist_init(sg, nsegs, (struct sglist_seg *)(sg + 1));
return (sg);
}
/*
* Free a scatter/gather list allocated via sglist_allc().
*/
void
sglist_free(struct sglist *sg)
{
if (refcount_release(&sg->sg_refs))
free(sg, M_SGLIST);
}
/*
* Append the segments to describe a single kernel virtual address
* range to a scatter/gather list. If there are insufficient
* segments, then this fails with EFBIG.
*/
int
sglist_append(struct sglist *sg, void *buf, size_t len)
{
struct sgsave save;
int error;
if (sg->sg_maxseg == 0)
return (EINVAL);
SGLIST_SAVE(sg, save);
error = _sglist_append_buf(sg, buf, len, NULL, NULL);
if (error)
SGLIST_RESTORE(sg, save);
return (error);
}
/*
* Append a single physical address range to a scatter/gather list.
* If there are insufficient segments, then this fails with EFBIG.
*/
int
sglist_append_phys(struct sglist *sg, vm_paddr_t paddr, size_t len)
{
struct sglist_seg *ss;
struct sgsave save;
int error;
if (sg->sg_maxseg == 0)
return (EINVAL);
if (len == 0)
return (0);
if (sg->sg_nseg == 0) {
sg->sg_segs[0].ss_paddr = paddr;
sg->sg_segs[0].ss_len = len;
sg->sg_nseg = 1;
return (0);
}
ss = &sg->sg_segs[sg->sg_nseg - 1];
SGLIST_SAVE(sg, save);
error = _sglist_append_range(sg, &ss, paddr, len);
if (error)
SGLIST_RESTORE(sg, save);
return (error);
}
/*
* Append the segments that describe a single mbuf chain to a
* scatter/gather list. If there are insufficient segments, then this
* fails with EFBIG.
*/
int
sglist_append_mbuf(struct sglist *sg, struct mbuf *m0)
{
struct sgsave save;
struct mbuf *m;
int error;
if (sg->sg_maxseg == 0)
return (EINVAL);
error = 0;
SGLIST_SAVE(sg, save);
for (m = m0; m != NULL; m = m->m_next) {
if (m->m_len > 0) {
error = sglist_append(sg, m->m_data, m->m_len);
if (error) {
SGLIST_RESTORE(sg, save);
return (error);
}
}
}
return (0);
}
/*
* Append the segments that describe a single user address range to a
* scatter/gather list. If there are insufficient segments, then this
* fails with EFBIG.
*/
int
sglist_append_user(struct sglist *sg, void *buf, size_t len, struct thread *td)
{
struct sgsave save;
int error;
if (sg->sg_maxseg == 0)
return (EINVAL);
SGLIST_SAVE(sg, save);
error = _sglist_append_buf(sg, buf, len,
vmspace_pmap(td->td_proc->p_vmspace), NULL);
if (error)
SGLIST_RESTORE(sg, save);
return (error);
}
/*
* Append the segments that describe a single uio to a scatter/gather
* list. If there are insufficient segments, then this fails with
* EFBIG.
*/
int
sglist_append_uio(struct sglist *sg, struct uio *uio)
{
struct iovec *iov;
struct sgsave save;
size_t resid, minlen;
pmap_t pmap;
int error, i;
if (sg->sg_maxseg == 0)
return (EINVAL);
resid = uio->uio_resid;
iov = uio->uio_iov;
if (uio->uio_segflg == UIO_USERSPACE) {
KASSERT(uio->uio_td != NULL,
("sglist_append_uio: USERSPACE but no thread"));
pmap = vmspace_pmap(uio->uio_td->td_proc->p_vmspace);
} else
pmap = NULL;
error = 0;
SGLIST_SAVE(sg, save);
for (i = 0; i < uio->uio_iovcnt && resid != 0; i++) {
/*
* Now at the first iovec to load. Load each iovec
* until we have exhausted the residual count.
*/
minlen = MIN(resid, iov[i].iov_len);
if (minlen > 0) {
error = _sglist_append_buf(sg, iov[i].iov_base, minlen,
pmap, NULL);
if (error) {
SGLIST_RESTORE(sg, save);
return (error);
}
resid -= minlen;
}
}
return (0);
}
/*
* Append the segments that describe at most 'resid' bytes from a
* single uio to a scatter/gather list. If there are insufficient
* segments, then only the amount that fits is appended.
*/
int
sglist_consume_uio(struct sglist *sg, struct uio *uio, size_t resid)
{
struct iovec *iov;
size_t done;
pmap_t pmap;
int error, len;
if (sg->sg_maxseg == 0)
return (EINVAL);
if (uio->uio_segflg == UIO_USERSPACE) {
KASSERT(uio->uio_td != NULL,
("sglist_consume_uio: USERSPACE but no thread"));
pmap = vmspace_pmap(uio->uio_td->td_proc->p_vmspace);
} else
pmap = NULL;
error = 0;
while (resid > 0 && uio->uio_resid) {
iov = uio->uio_iov;
len = iov->iov_len;
if (len == 0) {
uio->uio_iov++;
uio->uio_iovcnt--;
continue;
}
if (len > resid)
len = resid;
/*
* Try to append this iovec. If we run out of room,
* then break out of the loop.
*/
error = _sglist_append_buf(sg, iov->iov_base, len, pmap, &done);
iov->iov_base = (char *)iov->iov_base + done;
iov->iov_len -= done;
uio->uio_resid -= done;
uio->uio_offset += done;
resid -= done;
if (error)
break;
}
return (0);
}
/*
* Allocate and populate a scatter/gather list to describe a single
* kernel virtual address range.
*/
struct sglist *
sglist_build(void *buf, size_t len, int mflags)
{
struct sglist *sg;
int nsegs;
if (len == 0)
return (NULL);
nsegs = sglist_count(buf, len);
sg = sglist_alloc(nsegs, mflags);
if (sg == NULL)
return (NULL);
if (sglist_append(sg, buf, len) != 0) {
sglist_free(sg);
return (NULL);
}
return (sg);
}
/*
* Clone a new copy of a scatter/gather list.
*/
struct sglist *
sglist_clone(struct sglist *sg, int mflags)
{
struct sglist *new;
if (sg == NULL)
return (NULL);
new = sglist_alloc(sg->sg_maxseg, mflags);
if (new == NULL)
return (NULL);
new->sg_nseg = sg->sg_nseg;
bcopy(sg->sg_segs, new->sg_segs, sizeof(struct sglist_seg) *
sg->sg_nseg);
return (new);
}
/*
* Calculate the total length of the segments described in a
* scatter/gather list.
*/
size_t
sglist_length(struct sglist *sg)
{
size_t space;
int i;
space = 0;
for (i = 0; i < sg->sg_nseg; i++)
space += sg->sg_segs[i].ss_len;
return (space);
}
/*
* Split a scatter/gather list into two lists. The scatter/gather
* entries for the first 'length' bytes of the 'original' list are
* stored in the '*head' list and are removed from 'original'.
*
* If '*head' is NULL, then a new list will be allocated using
* 'mflags'. If M_NOWAIT is specified and the allocation fails,
* ENOMEM will be returned.
*
* If '*head' is not NULL, it should point to an empty sglist. If it
* does not have enough room for the remaining space, then EFBIG will
* be returned. If '*head' is not empty, then EINVAL will be
* returned.
*
* If 'original' is shared (refcount > 1), then EDOOFUS will be
* returned.
*/
int
sglist_split(struct sglist *original, struct sglist **head, size_t length,
int mflags)
{
struct sglist *sg;
size_t space, split;
int count, i;
if (original->sg_refs > 1)
return (EDOOFUS);
/* Figure out how big of a sglist '*head' has to hold. */
count = 0;
space = 0;
split = 0;
for (i = 0; i < original->sg_nseg; i++) {
space += original->sg_segs[i].ss_len;
count++;
if (space >= length) {
/*
* If 'length' falls in the middle of a
* scatter/gather list entry, then 'split'
* holds how much of that entry will remain in
* 'original'.
*/
split = space - length;
break;
}
}
/* Nothing to do, so leave head empty. */
if (count == 0)
return (0);
if (*head == NULL) {
sg = sglist_alloc(count, mflags);
if (sg == NULL)
return (ENOMEM);
*head = sg;
} else {
sg = *head;
if (sg->sg_maxseg < count)
return (EFBIG);
if (sg->sg_nseg != 0)
return (EINVAL);
}
/* Copy 'count' entries to 'sg' from 'original'. */
bcopy(original->sg_segs, sg->sg_segs, count *
sizeof(struct sglist_seg));
sg->sg_nseg = count;
/*
* If we had to split a list entry, fixup the last entry in
* 'sg' and the new first entry in 'original'. We also
* decrement 'count' by 1 since we will only be removing
* 'count - 1' segments from 'original' now.
*/
if (split != 0) {
count--;
sg->sg_segs[count].ss_len -= split;
original->sg_segs[count].ss_paddr =
sg->sg_segs[count].ss_paddr + split;
original->sg_segs[count].ss_len = split;
}
/* Trim 'count' entries from the front of 'original'. */
original->sg_nseg -= count;
bcopy(original->sg_segs + count, original->sg_segs, count *
sizeof(struct sglist_seg));
return (0);
}
/*
* Append the scatter/gather list elements in 'second' to the
* scatter/gather list 'first'. If there is not enough space in
* 'first', EFBIG is returned.
*/
int
sglist_join(struct sglist *first, struct sglist *second)
{
struct sglist_seg *flast, *sfirst;
int append;
/* If 'second' is empty, there is nothing to do. */
if (second->sg_nseg == 0)
return (0);
/*
* If the first entry in 'second' can be appended to the last entry
* in 'first' then set append to '1'.
*/
append = 0;
flast = &first->sg_segs[first->sg_nseg - 1];
sfirst = &second->sg_segs[0];
if (first->sg_nseg != 0 &&
flast->ss_paddr + flast->ss_len == sfirst->ss_paddr)
append = 1;
/* Make sure 'first' has enough room. */
if (first->sg_nseg + second->sg_nseg - append > first->sg_maxseg)
return (EFBIG);
/* Merge last in 'first' and first in 'second' if needed. */
if (append)
flast->ss_len += sfirst->ss_len;
/* Append new segments from 'second' to 'first'. */
bcopy(first->sg_segs + first->sg_nseg, second->sg_segs + append,
(second->sg_nseg - append) * sizeof(struct sglist_seg));
first->sg_nseg += second->sg_nseg - append;
sglist_reset(second);
return (0);
}
/*
* Generate a new scatter/gather list from a range of an existing
* scatter/gather list. The 'offset' and 'length' parameters specify
* the logical range of the 'original' list to extract. If that range
* is not a subset of the length of 'original', then EINVAL is
* returned. The new scatter/gather list is stored in '*slice'.
*
* If '*slice' is NULL, then a new list will be allocated using
* 'mflags'. If M_NOWAIT is specified and the allocation fails,
* ENOMEM will be returned.
*
* If '*slice' is not NULL, it should point to an empty sglist. If it
* does not have enough room for the remaining space, then EFBIG will
* be returned. If '*slice' is not empty, then EINVAL will be
* returned.
*/
int
sglist_slice(struct sglist *original, struct sglist **slice, size_t offset,
size_t length, int mflags)
{
struct sglist *sg;
size_t space, end, foffs, loffs;
int count, i, fseg;
/* Nothing to do. */
if (length == 0)
return (0);
/* Figure out how many segments '*slice' needs to have. */
end = offset + length;
space = 0;
count = 0;
fseg = 0;
foffs = loffs = 0;
for (i = 0; i < original->sg_nseg; i++) {
space += original->sg_segs[i].ss_len;
if (space > offset) {
/*
* When we hit the first segment, store its index
* in 'fseg' and the offset into the first segment
* of 'offset' in 'foffs'.
*/
if (count == 0) {
fseg = i;
foffs = offset - (space -
original->sg_segs[i].ss_len);
CTR1(KTR_DEV, "sglist_slice: foffs = %08lx",
foffs);
}
count++;
/*
* When we hit the last segment, break out of
* the loop. Store the amount of extra space
* at the end of this segment in 'loffs'.
*/
if (space >= end) {
loffs = space - end;
CTR1(KTR_DEV, "sglist_slice: loffs = %08lx",
loffs);
break;
}
}
}
/* If we never hit 'end', then 'length' ran off the end, so fail. */
if (space < end)
return (EINVAL);
if (*slice == NULL) {
sg = sglist_alloc(count, mflags);
if (sg == NULL)
return (ENOMEM);
*slice = sg;
} else {
sg = *slice;
if (sg->sg_maxseg < count)
return (EFBIG);
if (sg->sg_nseg != 0)
return (EINVAL);
}
/*
* Copy over 'count' segments from 'original' starting at
* 'fseg' to 'sg'.
*/
bcopy(original->sg_segs + fseg, sg->sg_segs,
count * sizeof(struct sglist_seg));
sg->sg_nseg = count;
/* Fixup first and last segments if needed. */
if (foffs != 0) {
sg->sg_segs[0].ss_paddr += foffs;
sg->sg_segs[0].ss_len -= foffs;
CTR2(KTR_DEV, "sglist_slice seg[0]: %08lx:%08lx",
(long)sg->sg_segs[0].ss_paddr, sg->sg_segs[0].ss_len);
}
if (loffs != 0) {
sg->sg_segs[count - 1].ss_len -= loffs;
CTR2(KTR_DEV, "sglist_slice seg[%d]: len %08x", count - 1,
sg->sg_segs[count - 1].ss_len);
}
return (0);
}
|