?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fs: create and use seq_show_option for escaping
[linux/fpc-iii.git] / fs / nfs / flexfilelayout / flexfilelayout.c
blobb3289d701eea21623f4081fee1b2807e3e2f4b3a
1 /*
2 * Module for pnfs flexfile layout driver.
3 *
4 * Copyright (c) 2014, Primary Data, Inc. All rights reserved.
5 *
6 * Tao Peng <bergwolf@primarydata.com>
7 */
8
9 #include <linux/nfs_fs.h>
10 #include <linux/nfs_page.h>
11 #include <linux/module.h>
12
13 #include <linux/sunrpc/metrics.h>
14
15 #include "flexfilelayout.h"
16 #include "../nfs4session.h"
17 #include "../nfs4idmap.h"
18 #include "../internal.h"
19 #include "../delegation.h"
20 #include "../nfs4trace.h"
21 #include "../iostat.h"
22 #include "../nfs.h"
23 #include "../nfs42.h"
24
25 #define NFSDBG_FACILITY NFSDBG_PNFS_LD
26
27 #define FF_LAYOUT_POLL_RETRY_MAX (15*HZ)
28
29 static struct pnfs_layout_hdr *
30 ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
31 {
32 struct nfs4_flexfile_layout *ffl;
33
34 ffl = kzalloc(sizeof(*ffl), gfp_flags);
35 if (ffl) {
36 INIT_LIST_HEAD(&ffl->error_list);
37 return &ffl->generic_hdr;
38 } else
39 return NULL;
40 }
41
42 static void
43 ff_layout_free_layout_hdr(struct pnfs_layout_hdr *lo)
44 {
45 struct nfs4_ff_layout_ds_err *err, *n;
46
47 list_for_each_entry_safe(err, n, &FF_LAYOUT_FROM_HDR(lo)->error_list,
48 list) {
49 list_del(&err->list);
50 kfree(err);
51 }
52 kfree(FF_LAYOUT_FROM_HDR(lo));
53 }
54
55 static int decode_stateid(struct xdr_stream *xdr, nfs4_stateid *stateid)
56 {
57 __be32 *p;
58
59 p = xdr_inline_decode(xdr, NFS4_STATEID_SIZE);
60 if (unlikely(p == NULL))
61 return -ENOBUFS;
62 memcpy(stateid, p, NFS4_STATEID_SIZE);
63 dprintk("%s: stateid id= [%x%x%x%x]\n", __func__,
64 p[0], p[1], p[2], p[3]);
65 return 0;
66 }
67
68 static int decode_deviceid(struct xdr_stream *xdr, struct nfs4_deviceid *devid)
69 {
70 __be32 *p;
71
72 p = xdr_inline_decode(xdr, NFS4_DEVICEID4_SIZE);
73 if (unlikely(!p))
74 return -ENOBUFS;
75 memcpy(devid, p, NFS4_DEVICEID4_SIZE);
76 nfs4_print_deviceid(devid);
77 return 0;
78 }
79
80 static int decode_nfs_fh(struct xdr_stream *xdr, struct nfs_fh *fh)
81 {
82 __be32 *p;
83
84 p = xdr_inline_decode(xdr, 4);
85 if (unlikely(!p))
86 return -ENOBUFS;
87 fh->size = be32_to_cpup(p++);
88 if (fh->size > sizeof(struct nfs_fh)) {
89 printk(KERN_ERR "NFS flexfiles: Too big fh received %d\n",
90 fh->size);
91 return -EOVERFLOW;
92 }
93 /* fh.data */
94 p = xdr_inline_decode(xdr, fh->size);
95 if (unlikely(!p))
96 return -ENOBUFS;
97 memcpy(&fh->data, p, fh->size);
98 dprintk("%s: fh len %d\n", __func__, fh->size);
99
100 return 0;
101 }
102
103 /*
104 * Currently only stringified uids and gids are accepted.
105 * I.e., kerberos is not supported to the DSes, so no pricipals.
106 *
107 * That means that one common function will suffice, but when
108 * principals are added, this should be split to accomodate
109 * calls to both nfs_map_name_to_uid() and nfs_map_group_to_gid().
110 */
111 static int
112 decode_name(struct xdr_stream *xdr, u32 *id)
113 {
114 __be32 *p;
115 int len;
116
117 /* opaque_length(4)*/
118 p = xdr_inline_decode(xdr, 4);
119 if (unlikely(!p))
120 return -ENOBUFS;
121 len = be32_to_cpup(p++);
122 if (len < 0)
123 return -EINVAL;
124
125 dprintk("%s: len %u\n", __func__, len);
126
127 /* opaque body */
128 p = xdr_inline_decode(xdr, len);
129 if (unlikely(!p))
130 return -ENOBUFS;
131
132 if (!nfs_map_string_to_numeric((char *)p, len, id))
133 return -EINVAL;
134
135 return 0;
136 }
137
138 static void ff_layout_free_mirror_array(struct nfs4_ff_layout_segment *fls)
139 {
140 int i;
141
142 if (fls->mirror_array) {
143 for (i = 0; i < fls->mirror_array_cnt; i++) {
144 /* normally mirror_ds is freed in
145 * .free_deviceid_node but we still do it here
146 * for .alloc_lseg error path */
147 if (fls->mirror_array[i]) {
148 kfree(fls->mirror_array[i]->fh_versions);
149 nfs4_ff_layout_put_deviceid(fls->mirror_array[i]->mirror_ds);
150 kfree(fls->mirror_array[i]);
151 }
152 }
153 kfree(fls->mirror_array);
154 fls->mirror_array = NULL;
155 }
156 }
157
158 static int ff_layout_check_layout(struct nfs4_layoutget_res *lgr)
159 {
160 int ret = 0;
161
162 dprintk("--> %s\n", __func__);
163
164 /* FIXME: remove this check when layout segment support is added */
165 if (lgr->range.offset != 0 ||
166 lgr->range.length != NFS4_MAX_UINT64) {
167 dprintk("%s Only whole file layouts supported. Use MDS i/o\n",
168 __func__);
169 ret = -EINVAL;
170 }
171
172 dprintk("--> %s returns %d\n", __func__, ret);
173 return ret;
174 }
175
176 static void _ff_layout_free_lseg(struct nfs4_ff_layout_segment *fls)
177 {
178 if (fls) {
179 ff_layout_free_mirror_array(fls);
180 kfree(fls);
181 }
182 }
183
184 static void ff_layout_sort_mirrors(struct nfs4_ff_layout_segment *fls)
185 {
186 int i, j;
187
188 for (i = 0; i < fls->mirror_array_cnt - 1; i++) {
189 for (j = i + 1; j < fls->mirror_array_cnt; j++)
190 if (fls->mirror_array[i]->efficiency <
191 fls->mirror_array[j]->efficiency)
192 swap(fls->mirror_array[i],
193 fls->mirror_array[j]);
194 }
195 }
196
197 static struct pnfs_layout_segment *
198 ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
199 struct nfs4_layoutget_res *lgr,
200 gfp_t gfp_flags)
201 {
202 struct pnfs_layout_segment *ret;
203 struct nfs4_ff_layout_segment *fls = NULL;
204 struct xdr_stream stream;
205 struct xdr_buf buf;
206 struct page *scratch;
207 u64 stripe_unit;
208 u32 mirror_array_cnt;
209 __be32 *p;
210 int i, rc;
211
212 dprintk("--> %s\n", __func__);
213 scratch = alloc_page(gfp_flags);
214 if (!scratch)
215 return ERR_PTR(-ENOMEM);
216
217 xdr_init_decode_pages(&stream, &buf, lgr->layoutp->pages,
218 lgr->layoutp->len);
219 xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
220
221 /* stripe unit and mirror_array_cnt */
222 rc = -EIO;
223 p = xdr_inline_decode(&stream, 8 + 4);
224 if (!p)
225 goto out_err_free;
226
227 p = xdr_decode_hyper(p, &stripe_unit);
228 mirror_array_cnt = be32_to_cpup(p++);
229 dprintk("%s: stripe_unit=%llu mirror_array_cnt=%u\n", __func__,
230 stripe_unit, mirror_array_cnt);
231
232 if (mirror_array_cnt > NFS4_FLEXFILE_LAYOUT_MAX_MIRROR_CNT ||
233 mirror_array_cnt == 0)
234 goto out_err_free;
235
236 rc = -ENOMEM;
237 fls = kzalloc(sizeof(*fls), gfp_flags);
238 if (!fls)
239 goto out_err_free;
240
241 fls->mirror_array_cnt = mirror_array_cnt;
242 fls->stripe_unit = stripe_unit;
243 fls->mirror_array = kcalloc(fls->mirror_array_cnt,
244 sizeof(fls->mirror_array[0]), gfp_flags);
245 if (fls->mirror_array == NULL)
246 goto out_err_free;
247
248 for (i = 0; i < fls->mirror_array_cnt; i++) {
249 struct nfs4_deviceid devid;
250 struct nfs4_deviceid_node *idnode;
251 u32 ds_count;
252 u32 fh_count;
253 int j;
254
255 rc = -EIO;
256 p = xdr_inline_decode(&stream, 4);
257 if (!p)
258 goto out_err_free;
259 ds_count = be32_to_cpup(p);
260
261 /* FIXME: allow for striping? */
262 if (ds_count != 1)
263 goto out_err_free;
264
265 fls->mirror_array[i] =
266 kzalloc(sizeof(struct nfs4_ff_layout_mirror),
267 gfp_flags);
268 if (fls->mirror_array[i] == NULL) {
269 rc = -ENOMEM;
270 goto out_err_free;
271 }
272
273 spin_lock_init(&fls->mirror_array[i]->lock);
274 fls->mirror_array[i]->ds_count = ds_count;
275 fls->mirror_array[i]->lseg = &fls->generic_hdr;
276
277 /* deviceid */
278 rc = decode_deviceid(&stream, &devid);
279 if (rc)
280 goto out_err_free;
281
282 idnode = nfs4_find_get_deviceid(NFS_SERVER(lh->plh_inode),
283 &devid, lh->plh_lc_cred,
284 gfp_flags);
285 /*
286 * upon success, mirror_ds is allocated by previous
287 * getdeviceinfo, or newly by .alloc_deviceid_node
288 * nfs4_find_get_deviceid failure is indeed getdeviceinfo falure
289 */
290 if (idnode)
291 fls->mirror_array[i]->mirror_ds =
292 FF_LAYOUT_MIRROR_DS(idnode);
293 else
294 goto out_err_free;
295
296 /* efficiency */
297 rc = -EIO;
298 p = xdr_inline_decode(&stream, 4);
299 if (!p)
300 goto out_err_free;
301 fls->mirror_array[i]->efficiency = be32_to_cpup(p);
302
303 /* stateid */
304 rc = decode_stateid(&stream, &fls->mirror_array[i]->stateid);
305 if (rc)
306 goto out_err_free;
307
308 /* fh */
309 p = xdr_inline_decode(&stream, 4);
310 if (!p)
311 goto out_err_free;
312 fh_count = be32_to_cpup(p);
313
314 fls->mirror_array[i]->fh_versions =
315 kzalloc(fh_count * sizeof(struct nfs_fh),
316 gfp_flags);
317 if (fls->mirror_array[i]->fh_versions == NULL) {
318 rc = -ENOMEM;
319 goto out_err_free;
320 }
321
322 for (j = 0; j < fh_count; j++) {
323 rc = decode_nfs_fh(&stream,
324 &fls->mirror_array[i]->fh_versions[j]);
325 if (rc)
326 goto out_err_free;
327 }
328
329 fls->mirror_array[i]->fh_versions_cnt = fh_count;
330
331 /* user */
332 rc = decode_name(&stream, &fls->mirror_array[i]->uid);
333 if (rc)
334 goto out_err_free;
335
336 /* group */
337 rc = decode_name(&stream, &fls->mirror_array[i]->gid);
338 if (rc)
339 goto out_err_free;
340
341 dprintk("%s: uid %d gid %d\n", __func__,
342 fls->mirror_array[i]->uid,
343 fls->mirror_array[i]->gid);
344 }
345
346 p = xdr_inline_decode(&stream, 4);
347 if (p)
348 fls->flags = be32_to_cpup(p);
349
350 ff_layout_sort_mirrors(fls);
351 rc = ff_layout_check_layout(lgr);
352 if (rc)
353 goto out_err_free;
354
355 ret = &fls->generic_hdr;
356 dprintk("<-- %s (success)\n", __func__);
357 out_free_page:
358 __free_page(scratch);
359 return ret;
360 out_err_free:
361 _ff_layout_free_lseg(fls);
362 ret = ERR_PTR(rc);
363 dprintk("<-- %s (%d)\n", __func__, rc);
364 goto out_free_page;
365 }
366
367 static bool ff_layout_has_rw_segments(struct pnfs_layout_hdr *layout)
368 {
369 struct pnfs_layout_segment *lseg;
370
371 list_for_each_entry(lseg, &layout->plh_segs, pls_list)
372 if (lseg->pls_range.iomode == IOMODE_RW)
373 return true;
374
375 return false;
376 }
377
378 static void
379 ff_layout_free_lseg(struct pnfs_layout_segment *lseg)
380 {
381 struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg);
382 int i;
383
384 dprintk("--> %s\n", __func__);
385
386 for (i = 0; i < fls->mirror_array_cnt; i++) {
387 if (fls->mirror_array[i]) {
388 nfs4_ff_layout_put_deviceid(fls->mirror_array[i]->mirror_ds);
389 fls->mirror_array[i]->mirror_ds = NULL;
390 if (fls->mirror_array[i]->cred) {
391 put_rpccred(fls->mirror_array[i]->cred);
392 fls->mirror_array[i]->cred = NULL;
393 }
394 }
395 }
396
397 if (lseg->pls_range.iomode == IOMODE_RW) {
398 struct nfs4_flexfile_layout *ffl;
399 struct inode *inode;
400
401 ffl = FF_LAYOUT_FROM_HDR(lseg->pls_layout);
402 inode = ffl->generic_hdr.plh_inode;
403 spin_lock(&inode->i_lock);
404 if (!ff_layout_has_rw_segments(lseg->pls_layout)) {
405 ffl->commit_info.nbuckets = 0;
406 kfree(ffl->commit_info.buckets);
407 ffl->commit_info.buckets = NULL;
408 }
409 spin_unlock(&inode->i_lock);
410 }
411 _ff_layout_free_lseg(fls);
412 }
413
414 /* Return 1 until we have multiple lsegs support */
415 static int
416 ff_layout_get_lseg_count(struct nfs4_ff_layout_segment *fls)
417 {
418 return 1;
419 }
420
421 static void
422 nfs4_ff_start_busy_timer(struct nfs4_ff_busy_timer *timer)
423 {
424 /* first IO request? */
425 if (atomic_inc_return(&timer->n_ops) == 1) {
426 timer->start_time = ktime_get();
427 }
428 }
429
430 static ktime_t
431 nfs4_ff_end_busy_timer(struct nfs4_ff_busy_timer *timer)
432 {
433 ktime_t start, now;
434
435 if (atomic_dec_return(&timer->n_ops) < 0)
436 WARN_ON_ONCE(1);
437
438 now = ktime_get();
439 start = timer->start_time;
440 timer->start_time = now;
441 return ktime_sub(now, start);
442 }
443
444 static ktime_t
445 nfs4_ff_layout_calc_completion_time(struct rpc_task *task)
446 {
447 return ktime_sub(ktime_get(), task->tk_start);
448 }
449
450 static bool
451 nfs4_ff_layoutstat_start_io(struct nfs4_ff_layout_mirror *mirror,
452 struct nfs4_ff_layoutstat *layoutstat)
453 {
454 static const ktime_t notime = {0};
455 ktime_t now = ktime_get();
456
457 nfs4_ff_start_busy_timer(&layoutstat->busy_timer);
458 if (ktime_equal(mirror->start_time, notime))
459 mirror->start_time = now;
460 if (ktime_equal(mirror->last_report_time, notime))
461 mirror->last_report_time = now;
462 if (ktime_to_ms(ktime_sub(now, mirror->last_report_time)) >=
463 FF_LAYOUTSTATS_REPORT_INTERVAL) {
464 mirror->last_report_time = now;
465 return true;
466 }
467
468 return false;
469 }
470
471 static void
472 nfs4_ff_layout_stat_io_update_requested(struct nfs4_ff_layoutstat *layoutstat,
473 __u64 requested)
474 {
475 struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
476
477 iostat->ops_requested++;
478 iostat->bytes_requested += requested;
479 }
480
481 static void
482 nfs4_ff_layout_stat_io_update_completed(struct nfs4_ff_layoutstat *layoutstat,
483 __u64 requested,
484 __u64 completed,
485 ktime_t time_completed)
486 {
487 struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
488 ktime_t timer;
489
490 iostat->ops_completed++;
491 iostat->bytes_completed += completed;
492 iostat->bytes_not_delivered += requested - completed;
493
494 timer = nfs4_ff_end_busy_timer(&layoutstat->busy_timer);
495 iostat->total_busy_time =
496 ktime_add(iostat->total_busy_time, timer);
497 iostat->aggregate_completion_time =
498 ktime_add(iostat->aggregate_completion_time, time_completed);
499 }
500
501 static void
502 nfs4_ff_layout_stat_io_start_read(struct nfs4_ff_layout_mirror *mirror,
503 __u64 requested)
504 {
505 bool report;
506
507 spin_lock(&mirror->lock);
508 report = nfs4_ff_layoutstat_start_io(mirror, &mirror->read_stat);
509 nfs4_ff_layout_stat_io_update_requested(&mirror->read_stat, requested);
510 spin_unlock(&mirror->lock);
511
512 if (report)
513 pnfs_report_layoutstat(mirror->lseg->pls_layout->plh_inode);
514 }
515
516 static void
517 nfs4_ff_layout_stat_io_end_read(struct rpc_task *task,
518 struct nfs4_ff_layout_mirror *mirror,
519 __u64 requested,
520 __u64 completed)
521 {
522 spin_lock(&mirror->lock);
523 nfs4_ff_layout_stat_io_update_completed(&mirror->read_stat,
524 requested, completed,
525 nfs4_ff_layout_calc_completion_time(task));
526 spin_unlock(&mirror->lock);
527 }
528
529 static void
530 nfs4_ff_layout_stat_io_start_write(struct nfs4_ff_layout_mirror *mirror,
531 __u64 requested)
532 {
533 bool report;
534
535 spin_lock(&mirror->lock);
536 report = nfs4_ff_layoutstat_start_io(mirror , &mirror->write_stat);
537 nfs4_ff_layout_stat_io_update_requested(&mirror->write_stat, requested);
538 spin_unlock(&mirror->lock);
539
540 if (report)
541 pnfs_report_layoutstat(mirror->lseg->pls_layout->plh_inode);
542 }
543
544 static void
545 nfs4_ff_layout_stat_io_end_write(struct rpc_task *task,
546 struct nfs4_ff_layout_mirror *mirror,
547 __u64 requested,
548 __u64 completed,
549 enum nfs3_stable_how committed)
550 {
551 if (committed == NFS_UNSTABLE)
552 requested = completed = 0;
553
554 spin_lock(&mirror->lock);
555 nfs4_ff_layout_stat_io_update_completed(&mirror->write_stat,
556 requested, completed,
557 nfs4_ff_layout_calc_completion_time(task));
558 spin_unlock(&mirror->lock);
559 }
560
561 static int
562 ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
563 struct nfs_commit_info *cinfo,
564 gfp_t gfp_flags)
565 {
566 struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg);
567 struct pnfs_commit_bucket *buckets;
568 int size;
569
570 if (cinfo->ds->nbuckets != 0) {
571 /* This assumes there is only one RW lseg per file.
572 * To support multiple lseg per file, we need to
573 * change struct pnfs_commit_bucket to allow dynamic
574 * increasing nbuckets.
575 */
576 return 0;
577 }
578
579 size = ff_layout_get_lseg_count(fls) * FF_LAYOUT_MIRROR_COUNT(lseg);
580
581 buckets = kcalloc(size, sizeof(struct pnfs_commit_bucket),
582 gfp_flags);
583 if (!buckets)
584 return -ENOMEM;
585 else {
586 int i;
587
588 spin_lock(cinfo->lock);
589 if (cinfo->ds->nbuckets != 0)
590 kfree(buckets);
591 else {
592 cinfo->ds->buckets = buckets;
593 cinfo->ds->nbuckets = size;
594 for (i = 0; i < size; i++) {
595 INIT_LIST_HEAD(&buckets[i].written);
596 INIT_LIST_HEAD(&buckets[i].committing);
597 /* mark direct verifier as unset */
598 buckets[i].direct_verf.committed =
599 NFS_INVALID_STABLE_HOW;
600 }
601 }
602 spin_unlock(cinfo->lock);
603 return 0;
604 }
605 }
606
607 static struct nfs4_pnfs_ds *
608 ff_layout_choose_best_ds_for_read(struct nfs_pageio_descriptor *pgio,
609 int *best_idx)
610 {
611 struct nfs4_ff_layout_segment *fls;
612 struct nfs4_pnfs_ds *ds;
613 int idx;
614
615 fls = FF_LAYOUT_LSEG(pgio->pg_lseg);
616 /* mirrors are sorted by efficiency */
617 for (idx = 0; idx < fls->mirror_array_cnt; idx++) {
618 ds = nfs4_ff_layout_prepare_ds(pgio->pg_lseg, idx, false);
619 if (ds) {
620 *best_idx = idx;
621 return ds;
622 }
623 }
624
625 return NULL;
626 }
627
628 static void
629 ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
630 struct nfs_page *req)
631 {
632 struct nfs_pgio_mirror *pgm;
633 struct nfs4_ff_layout_mirror *mirror;
634 struct nfs4_pnfs_ds *ds;
635 int ds_idx;
636
637 /* Use full layout for now */
638 if (!pgio->pg_lseg)
639 pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
640 req->wb_context,
641 0,
642 NFS4_MAX_UINT64,
643 IOMODE_READ,
644 GFP_KERNEL);
645 /* If no lseg, fall back to read through mds */
646 if (pgio->pg_lseg == NULL)
647 goto out_mds;
648
649 ds = ff_layout_choose_best_ds_for_read(pgio, &ds_idx);
650 if (!ds)
651 goto out_mds;
652 mirror = FF_LAYOUT_COMP(pgio->pg_lseg, ds_idx);
653
654 pgio->pg_mirror_idx = ds_idx;
655
656 /* read always uses only one mirror - idx 0 for pgio layer */
657 pgm = &pgio->pg_mirrors[0];
658 pgm->pg_bsize = mirror->mirror_ds->ds_versions[0].rsize;
659
660 return;
661 out_mds:
662 pnfs_put_lseg(pgio->pg_lseg);
663 pgio->pg_lseg = NULL;
664 nfs_pageio_reset_read_mds(pgio);
665 }
666
667 static void
668 ff_layout_pg_init_write(struct nfs_pageio_descriptor *pgio,
669 struct nfs_page *req)
670 {
671 struct nfs4_ff_layout_mirror *mirror;
672 struct nfs_pgio_mirror *pgm;
673 struct nfs_commit_info cinfo;
674 struct nfs4_pnfs_ds *ds;
675 int i;
676 int status;
677
678 if (!pgio->pg_lseg)
679 pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
680 req->wb_context,
681 0,
682 NFS4_MAX_UINT64,
683 IOMODE_RW,
684 GFP_NOFS);
685 /* If no lseg, fall back to write through mds */
686 if (pgio->pg_lseg == NULL)
687 goto out_mds;
688
689 nfs_init_cinfo(&cinfo, pgio->pg_inode, pgio->pg_dreq);
690 status = ff_layout_alloc_commit_info(pgio->pg_lseg, &cinfo, GFP_NOFS);
691 if (status < 0)
692 goto out_mds;
693
694 /* Use a direct mapping of ds_idx to pgio mirror_idx */
695 if (WARN_ON_ONCE(pgio->pg_mirror_count !=
696 FF_LAYOUT_MIRROR_COUNT(pgio->pg_lseg)))
697 goto out_mds;
698
699 for (i = 0; i < pgio->pg_mirror_count; i++) {
700 ds = nfs4_ff_layout_prepare_ds(pgio->pg_lseg, i, true);
701 if (!ds)
702 goto out_mds;
703 pgm = &pgio->pg_mirrors[i];
704 mirror = FF_LAYOUT_COMP(pgio->pg_lseg, i);
705 pgm->pg_bsize = mirror->mirror_ds->ds_versions[0].wsize;
706 }
707
708 return;
709
710 out_mds:
711 pnfs_put_lseg(pgio->pg_lseg);
712 pgio->pg_lseg = NULL;
713 nfs_pageio_reset_write_mds(pgio);
714 }
715
716 static unsigned int
717 ff_layout_pg_get_mirror_count_write(struct nfs_pageio_descriptor *pgio,
718 struct nfs_page *req)
719 {
720 if (!pgio->pg_lseg)
721 pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
722 req->wb_context,
723 0,
724 NFS4_MAX_UINT64,
725 IOMODE_RW,
726 GFP_NOFS);
727 if (pgio->pg_lseg)
728 return FF_LAYOUT_MIRROR_COUNT(pgio->pg_lseg);
729
730 /* no lseg means that pnfs is not in use, so no mirroring here */
731 pnfs_put_lseg(pgio->pg_lseg);
732 pgio->pg_lseg = NULL;
733 nfs_pageio_reset_write_mds(pgio);
734 return 1;
735 }
736
737 static const struct nfs_pageio_ops ff_layout_pg_read_ops = {
738 .pg_init = ff_layout_pg_init_read,
739 .pg_test = pnfs_generic_pg_test,
740 .pg_doio = pnfs_generic_pg_readpages,
741 .pg_cleanup = pnfs_generic_pg_cleanup,
742 };
743
744 static const struct nfs_pageio_ops ff_layout_pg_write_ops = {
745 .pg_init = ff_layout_pg_init_write,
746 .pg_test = pnfs_generic_pg_test,
747 .pg_doio = pnfs_generic_pg_writepages,
748 .pg_get_mirror_count = ff_layout_pg_get_mirror_count_write,
749 .pg_cleanup = pnfs_generic_pg_cleanup,
750 };
751
752 static void ff_layout_reset_write(struct nfs_pgio_header *hdr, bool retry_pnfs)
753 {
754 struct rpc_task *task = &hdr->task;
755
756 pnfs_layoutcommit_inode(hdr->inode, false);
757
758 if (retry_pnfs) {
759 dprintk("%s Reset task %5u for i/o through pNFS "
760 "(req %s/%llu, %u bytes @ offset %llu)\n", __func__,
761 hdr->task.tk_pid,
762 hdr->inode->i_sb->s_id,
763 (unsigned long long)NFS_FILEID(hdr->inode),
764 hdr->args.count,
765 (unsigned long long)hdr->args.offset);
766
767 if (!hdr->dreq) {
768 struct nfs_open_context *ctx;
769
770 ctx = nfs_list_entry(hdr->pages.next)->wb_context;
771 set_bit(NFS_CONTEXT_RESEND_WRITES, &ctx->flags);
772 hdr->completion_ops->error_cleanup(&hdr->pages);
773 } else {
774 nfs_direct_set_resched_writes(hdr->dreq);
775 /* fake unstable write to let common nfs resend pages */
776 hdr->verf.committed = NFS_UNSTABLE;
777 hdr->good_bytes = hdr->args.count;
778 }
779 return;
780 }
781
782 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
783 dprintk("%s Reset task %5u for i/o through MDS "
784 "(req %s/%llu, %u bytes @ offset %llu)\n", __func__,
785 hdr->task.tk_pid,
786 hdr->inode->i_sb->s_id,
787 (unsigned long long)NFS_FILEID(hdr->inode),
788 hdr->args.count,
789 (unsigned long long)hdr->args.offset);
790
791 task->tk_status = pnfs_write_done_resend_to_mds(hdr);
792 }
793 }
794
795 static void ff_layout_reset_read(struct nfs_pgio_header *hdr)
796 {
797 struct rpc_task *task = &hdr->task;
798
799 pnfs_layoutcommit_inode(hdr->inode, false);
800
801 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
802 dprintk("%s Reset task %5u for i/o through MDS "
803 "(req %s/%llu, %u bytes @ offset %llu)\n", __func__,
804 hdr->task.tk_pid,
805 hdr->inode->i_sb->s_id,
806 (unsigned long long)NFS_FILEID(hdr->inode),
807 hdr->args.count,
808 (unsigned long long)hdr->args.offset);
809
810 task->tk_status = pnfs_read_done_resend_to_mds(hdr);
811 }
812 }
813
814 static int ff_layout_async_handle_error_v4(struct rpc_task *task,
815 struct nfs4_state *state,
816 struct nfs_client *clp,
817 struct pnfs_layout_segment *lseg,
818 int idx)
819 {
820 struct pnfs_layout_hdr *lo = lseg->pls_layout;
821 struct inode *inode = lo->plh_inode;
822 struct nfs_server *mds_server = NFS_SERVER(inode);
823
824 struct nfs4_deviceid_node *devid = FF_LAYOUT_DEVID_NODE(lseg, idx);
825 struct nfs_client *mds_client = mds_server->nfs_client;
826 struct nfs4_slot_table *tbl = &clp->cl_session->fc_slot_table;
827
828 if (task->tk_status >= 0)
829 return 0;
830
831 switch (task->tk_status) {
832 /* MDS state errors */
833 case -NFS4ERR_DELEG_REVOKED:
834 case -NFS4ERR_ADMIN_REVOKED:
835 case -NFS4ERR_BAD_STATEID:
836 if (state == NULL)
837 break;
838 nfs_remove_bad_delegation(state->inode);
839 case -NFS4ERR_OPENMODE:
840 if (state == NULL)
841 break;
842 if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
843 goto out_bad_stateid;
844 goto wait_on_recovery;
845 case -NFS4ERR_EXPIRED:
846 if (state != NULL) {
847 if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
848 goto out_bad_stateid;
849 }
850 nfs4_schedule_lease_recovery(mds_client);
851 goto wait_on_recovery;
852 /* DS session errors */
853 case -NFS4ERR_BADSESSION:
854 case -NFS4ERR_BADSLOT:
855 case -NFS4ERR_BAD_HIGH_SLOT:
856 case -NFS4ERR_DEADSESSION:
857 case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
858 case -NFS4ERR_SEQ_FALSE_RETRY:
859 case -NFS4ERR_SEQ_MISORDERED:
860 dprintk("%s ERROR %d, Reset session. Exchangeid "
861 "flags 0x%x\n", __func__, task->tk_status,
862 clp->cl_exchange_flags);
863 nfs4_schedule_session_recovery(clp->cl_session, task->tk_status);
864 break;
865 case -NFS4ERR_DELAY:
866 case -NFS4ERR_GRACE:
867 rpc_delay(task, FF_LAYOUT_POLL_RETRY_MAX);
868 break;
869 case -NFS4ERR_RETRY_UNCACHED_REP:
870 break;
871 /* Invalidate Layout errors */
872 case -NFS4ERR_PNFS_NO_LAYOUT:
873 case -ESTALE: /* mapped NFS4ERR_STALE */
874 case -EBADHANDLE: /* mapped NFS4ERR_BADHANDLE */
875 case -EISDIR: /* mapped NFS4ERR_ISDIR */
876 case -NFS4ERR_FHEXPIRED:
877 case -NFS4ERR_WRONG_TYPE:
878 dprintk("%s Invalid layout error %d\n", __func__,
879 task->tk_status);
880 /*
881 * Destroy layout so new i/o will get a new layout.
882 * Layout will not be destroyed until all current lseg
883 * references are put. Mark layout as invalid to resend failed
884 * i/o and all i/o waiting on the slot table to the MDS until
885 * layout is destroyed and a new valid layout is obtained.
886 */
887 pnfs_destroy_layout(NFS_I(inode));
888 rpc_wake_up(&tbl->slot_tbl_waitq);
889 goto reset;
890 /* RPC connection errors */
891 case -ECONNREFUSED:
892 case -EHOSTDOWN:
893 case -EHOSTUNREACH:
894 case -ENETUNREACH:
895 case -EIO:
896 case -ETIMEDOUT:
897 case -EPIPE:
898 dprintk("%s DS connection error %d\n", __func__,
899 task->tk_status);
900 nfs4_mark_deviceid_unavailable(devid);
901 rpc_wake_up(&tbl->slot_tbl_waitq);
902 /* fall through */
903 default:
904 if (ff_layout_has_available_ds(lseg))
905 return -NFS4ERR_RESET_TO_PNFS;
906 reset:
907 dprintk("%s Retry through MDS. Error %d\n", __func__,
908 task->tk_status);
909 return -NFS4ERR_RESET_TO_MDS;
910 }
911 out:
912 task->tk_status = 0;
913 return -EAGAIN;
914 out_bad_stateid:
915 task->tk_status = -EIO;
916 return 0;
917 wait_on_recovery:
918 rpc_sleep_on(&mds_client->cl_rpcwaitq, task, NULL);
919 if (test_bit(NFS4CLNT_MANAGER_RUNNING, &mds_client->cl_state) == 0)
920 rpc_wake_up_queued_task(&mds_client->cl_rpcwaitq, task);
921 goto out;
922 }
923
924 /* Retry all errors through either pNFS or MDS except for -EJUKEBOX */
925 static int ff_layout_async_handle_error_v3(struct rpc_task *task,
926 struct pnfs_layout_segment *lseg,
927 int idx)
928 {
929 struct nfs4_deviceid_node *devid = FF_LAYOUT_DEVID_NODE(lseg, idx);
930
931 if (task->tk_status >= 0)
932 return 0;
933
934 if (task->tk_status != -EJUKEBOX) {
935 dprintk("%s DS connection error %d\n", __func__,
936 task->tk_status);
937 nfs4_mark_deviceid_unavailable(devid);
938 if (ff_layout_has_available_ds(lseg))
939 return -NFS4ERR_RESET_TO_PNFS;
940 else
941 return -NFS4ERR_RESET_TO_MDS;
942 }
943
944 if (task->tk_status == -EJUKEBOX)
945 nfs_inc_stats(lseg->pls_layout->plh_inode, NFSIOS_DELAY);
946 task->tk_status = 0;
947 rpc_restart_call(task);
948 rpc_delay(task, NFS_JUKEBOX_RETRY_TIME);
949 return -EAGAIN;
950 }
951
952 static int ff_layout_async_handle_error(struct rpc_task *task,
953 struct nfs4_state *state,
954 struct nfs_client *clp,
955 struct pnfs_layout_segment *lseg,
956 int idx)
957 {
958 int vers = clp->cl_nfs_mod->rpc_vers->number;
959
960 switch (vers) {
961 case 3:
962 return ff_layout_async_handle_error_v3(task, lseg, idx);
963 case 4:
964 return ff_layout_async_handle_error_v4(task, state, clp,
965 lseg, idx);
966 default:
967 /* should never happen */
968 WARN_ON_ONCE(1);
969 return 0;
970 }
971 }
972
973 static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
974 int idx, u64 offset, u64 length,
975 u32 status, int opnum)
976 {
977 struct nfs4_ff_layout_mirror *mirror;
978 int err;
979
980 mirror = FF_LAYOUT_COMP(lseg, idx);
981 err = ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
982 mirror, offset, length, status, opnum,
983 GFP_NOIO);
984 dprintk("%s: err %d op %d status %u\n", __func__, err, opnum, status);
985 }
986
987 /* NFS_PROTO call done callback routines */
988
989 static int ff_layout_read_done_cb(struct rpc_task *task,
990 struct nfs_pgio_header *hdr)
991 {
992 struct inode *inode;
993 int err;
994
995 trace_nfs4_pnfs_read(hdr, task->tk_status);
996 if (task->tk_status == -ETIMEDOUT && !hdr->res.op_status)
997 hdr->res.op_status = NFS4ERR_NXIO;
998 if (task->tk_status < 0 && hdr->res.op_status)
999 ff_layout_io_track_ds_error(hdr->lseg, hdr->pgio_mirror_idx,
1000 hdr->args.offset, hdr->args.count,
1001 hdr->res.op_status, OP_READ);
1002 err = ff_layout_async_handle_error(task, hdr->args.context->state,
1003 hdr->ds_clp, hdr->lseg,
1004 hdr->pgio_mirror_idx);
1005
1006 switch (err) {
1007 case -NFS4ERR_RESET_TO_PNFS:
1008 set_bit(NFS_LAYOUT_RETURN_BEFORE_CLOSE,
1009 &hdr->lseg->pls_layout->plh_flags);
1010 pnfs_read_resend_pnfs(hdr);
1011 return task->tk_status;
1012 case -NFS4ERR_RESET_TO_MDS:
1013 inode = hdr->lseg->pls_layout->plh_inode;
1014 pnfs_error_mark_layout_for_return(inode, hdr->lseg);
1015 ff_layout_reset_read(hdr);
1016 return task->tk_status;
1017 case -EAGAIN:
1018 rpc_restart_call_prepare(task);
1019 return -EAGAIN;
1020 }
1021
1022 return 0;
1023 }
1024
1025 static bool
1026 ff_layout_need_layoutcommit(struct pnfs_layout_segment *lseg)
1027 {
1028 return !(FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_LAYOUTCOMMIT);
1029 }
1030
1031 /*
1032 * We reference the rpc_cred of the first WRITE that triggers the need for
1033 * a LAYOUTCOMMIT, and use it to send the layoutcommit compound.
1034 * rfc5661 is not clear about which credential should be used.
1035 *
1036 * Flexlayout client should treat DS replied FILE_SYNC as DATA_SYNC, so
1037 * to follow http://www.rfc-editor.org/errata_search.php?rfc=5661&eid=2751
1038 * we always send layoutcommit after DS writes.
1039 */
1040 static void
1041 ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
1042 {
1043 if (!ff_layout_need_layoutcommit(hdr->lseg))
1044 return;
1045
1046 pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
1047 hdr->mds_offset + hdr->res.count);
1048 dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
1049 (unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
1050 }
1051
1052 static bool
1053 ff_layout_reset_to_mds(struct pnfs_layout_segment *lseg, int idx)
1054 {
1055 /* No mirroring for now */
1056 struct nfs4_deviceid_node *node = FF_LAYOUT_DEVID_NODE(lseg, idx);
1057
1058 return ff_layout_test_devid_unavailable(node);
1059 }
1060
1061 static int ff_layout_read_prepare_common(struct rpc_task *task,
1062 struct nfs_pgio_header *hdr)
1063 {
1064 nfs4_ff_layout_stat_io_start_read(
1065 FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
1066 hdr->args.count);
1067
1068 if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
1069 rpc_exit(task, -EIO);
1070 return -EIO;
1071 }
1072 if (ff_layout_reset_to_mds(hdr->lseg, hdr->pgio_mirror_idx)) {
1073 dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid);
1074 if (ff_layout_has_available_ds(hdr->lseg))
1075 pnfs_read_resend_pnfs(hdr);
1076 else
1077 ff_layout_reset_read(hdr);
1078 rpc_exit(task, 0);
1079 return -EAGAIN;
1080 }
1081 hdr->pgio_done_cb = ff_layout_read_done_cb;
1082
1083 return 0;
1084 }
1085
1086 /*
1087 * Call ops for the async read/write cases
1088 * In the case of dense layouts, the offset needs to be reset to its
1089 * original value.
1090 */
1091 static void ff_layout_read_prepare_v3(struct rpc_task *task, void *data)
1092 {
1093 struct nfs_pgio_header *hdr = data;
1094
1095 if (ff_layout_read_prepare_common(task, hdr))
1096 return;
1097
1098 rpc_call_start(task);
1099 }
1100
1101 static int ff_layout_setup_sequence(struct nfs_client *ds_clp,
1102 struct nfs4_sequence_args *args,
1103 struct nfs4_sequence_res *res,
1104 struct rpc_task *task)
1105 {
1106 if (ds_clp->cl_session)
1107 return nfs41_setup_sequence(ds_clp->cl_session,
1108 args,
1109 res,
1110 task);
1111 return nfs40_setup_sequence(ds_clp->cl_slot_tbl,
1112 args,
1113 res,
1114 task);
1115 }
1116
1117 static void ff_layout_read_prepare_v4(struct rpc_task *task, void *data)
1118 {
1119 struct nfs_pgio_header *hdr = data;
1120
1121 if (ff_layout_setup_sequence(hdr->ds_clp,
1122 &hdr->args.seq_args,
1123 &hdr->res.seq_res,
1124 task))
1125 return;
1126
1127 if (ff_layout_read_prepare_common(task, hdr))
1128 return;
1129
1130 if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
1131 hdr->args.lock_context, FMODE_READ) == -EIO)
1132 rpc_exit(task, -EIO); /* lost lock, terminate I/O */
1133 }
1134
1135 static void ff_layout_read_call_done(struct rpc_task *task, void *data)
1136 {
1137 struct nfs_pgio_header *hdr = data;
1138
1139 dprintk("--> %s task->tk_status %d\n", __func__, task->tk_status);
1140
1141 nfs4_ff_layout_stat_io_end_read(task,
1142 FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
1143 hdr->args.count, hdr->res.count);
1144
1145 if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
1146 task->tk_status == 0) {
1147 nfs4_sequence_done(task, &hdr->res.seq_res);
1148 return;
1149 }
1150
1151 /* Note this may cause RPC to be resent */
1152 hdr->mds_ops->rpc_call_done(task, hdr);
1153 }
1154
1155 static void ff_layout_read_count_stats(struct rpc_task *task, void *data)
1156 {
1157 struct nfs_pgio_header *hdr = data;
1158
1159 rpc_count_iostats_metrics(task,
1160 &NFS_CLIENT(hdr->inode)->cl_metrics[NFSPROC4_CLNT_READ]);
1161 }
1162
1163 static int ff_layout_write_done_cb(struct rpc_task *task,
1164 struct nfs_pgio_header *hdr)
1165 {
1166 struct inode *inode;
1167 int err;
1168
1169 trace_nfs4_pnfs_write(hdr, task->tk_status);
1170 if (task->tk_status == -ETIMEDOUT && !hdr->res.op_status)
1171 hdr->res.op_status = NFS4ERR_NXIO;
1172 if (task->tk_status < 0 && hdr->res.op_status)
1173 ff_layout_io_track_ds_error(hdr->lseg, hdr->pgio_mirror_idx,
1174 hdr->args.offset, hdr->args.count,
1175 hdr->res.op_status, OP_WRITE);
1176 err = ff_layout_async_handle_error(task, hdr->args.context->state,
1177 hdr->ds_clp, hdr->lseg,
1178 hdr->pgio_mirror_idx);
1179
1180 switch (err) {
1181 case -NFS4ERR_RESET_TO_PNFS:
1182 case -NFS4ERR_RESET_TO_MDS:
1183 inode = hdr->lseg->pls_layout->plh_inode;
1184 pnfs_error_mark_layout_for_return(inode, hdr->lseg);
1185 if (err == -NFS4ERR_RESET_TO_PNFS) {
1186 pnfs_set_retry_layoutget(hdr->lseg->pls_layout);
1187 ff_layout_reset_write(hdr, true);
1188 } else {
1189 pnfs_clear_retry_layoutget(hdr->lseg->pls_layout);
1190 ff_layout_reset_write(hdr, false);
1191 }
1192 return task->tk_status;
1193 case -EAGAIN:
1194 rpc_restart_call_prepare(task);
1195 return -EAGAIN;
1196 }
1197
1198 if (hdr->res.verf->committed == NFS_FILE_SYNC ||
1199 hdr->res.verf->committed == NFS_DATA_SYNC)
1200 ff_layout_set_layoutcommit(hdr);
1201
1202 return 0;
1203 }
1204
1205 static int ff_layout_commit_done_cb(struct rpc_task *task,
1206 struct nfs_commit_data *data)
1207 {
1208 struct inode *inode;
1209 int err;
1210
1211 trace_nfs4_pnfs_commit_ds(data, task->tk_status);
1212 if (task->tk_status == -ETIMEDOUT && !data->res.op_status)
1213 data->res.op_status = NFS4ERR_NXIO;
1214 if (task->tk_status < 0 && data->res.op_status)
1215 ff_layout_io_track_ds_error(data->lseg, data->ds_commit_index,
1216 data->args.offset, data->args.count,
1217 data->res.op_status, OP_COMMIT);
1218 err = ff_layout_async_handle_error(task, NULL, data->ds_clp,
1219 data->lseg, data->ds_commit_index);
1220
1221 switch (err) {
1222 case -NFS4ERR_RESET_TO_PNFS:
1223 case -NFS4ERR_RESET_TO_MDS:
1224 inode = data->lseg->pls_layout->plh_inode;
1225 pnfs_error_mark_layout_for_return(inode, data->lseg);
1226 if (err == -NFS4ERR_RESET_TO_PNFS)
1227 pnfs_set_retry_layoutget(data->lseg->pls_layout);
1228 else
1229 pnfs_clear_retry_layoutget(data->lseg->pls_layout);
1230 pnfs_generic_prepare_to_resend_writes(data);
1231 return -EAGAIN;
1232 case -EAGAIN:
1233 rpc_restart_call_prepare(task);
1234 return -EAGAIN;
1235 }
1236
1237 if (data->verf.committed == NFS_UNSTABLE
1238 && ff_layout_need_layoutcommit(data->lseg))
1239 pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
1240
1241 return 0;
1242 }
1243
1244 static int ff_layout_write_prepare_common(struct rpc_task *task,
1245 struct nfs_pgio_header *hdr)
1246 {
1247 nfs4_ff_layout_stat_io_start_write(
1248 FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
1249 hdr->args.count);
1250
1251 if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
1252 rpc_exit(task, -EIO);
1253 return -EIO;
1254 }
1255
1256 if (ff_layout_reset_to_mds(hdr->lseg, hdr->pgio_mirror_idx)) {
1257 bool retry_pnfs;
1258
1259 retry_pnfs = ff_layout_has_available_ds(hdr->lseg);
1260 dprintk("%s task %u reset io to %s\n", __func__,
1261 task->tk_pid, retry_pnfs ? "pNFS" : "MDS");
1262 ff_layout_reset_write(hdr, retry_pnfs);
1263 rpc_exit(task, 0);
1264 return -EAGAIN;
1265 }
1266
1267 return 0;
1268 }
1269
1270 static void ff_layout_write_prepare_v3(struct rpc_task *task, void *data)
1271 {
1272 struct nfs_pgio_header *hdr = data;
1273
1274 if (ff_layout_write_prepare_common(task, hdr))
1275 return;
1276
1277 rpc_call_start(task);
1278 }
1279
1280 static void ff_layout_write_prepare_v4(struct rpc_task *task, void *data)
1281 {
1282 struct nfs_pgio_header *hdr = data;
1283
1284 if (ff_layout_setup_sequence(hdr->ds_clp,
1285 &hdr->args.seq_args,
1286 &hdr->res.seq_res,
1287 task))
1288 return;
1289
1290 if (ff_layout_write_prepare_common(task, hdr))
1291 return;
1292
1293 if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
1294 hdr->args.lock_context, FMODE_WRITE) == -EIO)
1295 rpc_exit(task, -EIO); /* lost lock, terminate I/O */
1296 }
1297
1298 static void ff_layout_write_call_done(struct rpc_task *task, void *data)
1299 {
1300 struct nfs_pgio_header *hdr = data;
1301
1302 nfs4_ff_layout_stat_io_end_write(task,
1303 FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
1304 hdr->args.count, hdr->res.count,
1305 hdr->res.verf->committed);
1306
1307 if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
1308 task->tk_status == 0) {
1309 nfs4_sequence_done(task, &hdr->res.seq_res);
1310 return;
1311 }
1312
1313 /* Note this may cause RPC to be resent */
1314 hdr->mds_ops->rpc_call_done(task, hdr);
1315 }
1316
1317 static void ff_layout_write_count_stats(struct rpc_task *task, void *data)
1318 {
1319 struct nfs_pgio_header *hdr = data;
1320
1321 rpc_count_iostats_metrics(task,
1322 &NFS_CLIENT(hdr->inode)->cl_metrics[NFSPROC4_CLNT_WRITE]);
1323 }
1324
1325 static void ff_layout_commit_prepare_common(struct rpc_task *task,
1326 struct nfs_commit_data *cdata)
1327 {
1328 nfs4_ff_layout_stat_io_start_write(
1329 FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
1330 0);
1331 }
1332
1333 static void ff_layout_commit_prepare_v3(struct rpc_task *task, void *data)
1334 {
1335 ff_layout_commit_prepare_common(task, data);
1336 rpc_call_start(task);
1337 }
1338
1339 static void ff_layout_commit_prepare_v4(struct rpc_task *task, void *data)
1340 {
1341 struct nfs_commit_data *wdata = data;
1342
1343 if (ff_layout_setup_sequence(wdata->ds_clp,
1344 &wdata->args.seq_args,
1345 &wdata->res.seq_res,
1346 task))
1347 return;
1348 ff_layout_commit_prepare_common(task, data);
1349 }
1350
1351 static void ff_layout_commit_done(struct rpc_task *task, void *data)
1352 {
1353 struct nfs_commit_data *cdata = data;
1354 struct nfs_page *req;
1355 __u64 count = 0;
1356
1357 if (task->tk_status == 0) {
1358 list_for_each_entry(req, &cdata->pages, wb_list)
1359 count += req->wb_bytes;
1360 }
1361
1362 nfs4_ff_layout_stat_io_end_write(task,
1363 FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
1364 count, count, NFS_FILE_SYNC);
1365
1366 pnfs_generic_write_commit_done(task, data);
1367 }
1368
1369 static void ff_layout_commit_count_stats(struct rpc_task *task, void *data)
1370 {
1371 struct nfs_commit_data *cdata = data;
1372
1373 rpc_count_iostats_metrics(task,
1374 &NFS_CLIENT(cdata->inode)->cl_metrics[NFSPROC4_CLNT_COMMIT]);
1375 }
1376
1377 static const struct rpc_call_ops ff_layout_read_call_ops_v3 = {
1378 .rpc_call_prepare = ff_layout_read_prepare_v3,
1379 .rpc_call_done = ff_layout_read_call_done,
1380 .rpc_count_stats = ff_layout_read_count_stats,
1381 .rpc_release = pnfs_generic_rw_release,
1382 };
1383
1384 static const struct rpc_call_ops ff_layout_read_call_ops_v4 = {
1385 .rpc_call_prepare = ff_layout_read_prepare_v4,
1386 .rpc_call_done = ff_layout_read_call_done,
1387 .rpc_count_stats = ff_layout_read_count_stats,
1388 .rpc_release = pnfs_generic_rw_release,
1389 };
1390
1391 static const struct rpc_call_ops ff_layout_write_call_ops_v3 = {
1392 .rpc_call_prepare = ff_layout_write_prepare_v3,
1393 .rpc_call_done = ff_layout_write_call_done,
1394 .rpc_count_stats = ff_layout_write_count_stats,
1395 .rpc_release = pnfs_generic_rw_release,
1396 };
1397
1398 static const struct rpc_call_ops ff_layout_write_call_ops_v4 = {
1399 .rpc_call_prepare = ff_layout_write_prepare_v4,
1400 .rpc_call_done = ff_layout_write_call_done,
1401 .rpc_count_stats = ff_layout_write_count_stats,
1402 .rpc_release = pnfs_generic_rw_release,
1403 };
1404
1405 static const struct rpc_call_ops ff_layout_commit_call_ops_v3 = {
1406 .rpc_call_prepare = ff_layout_commit_prepare_v3,
1407 .rpc_call_done = ff_layout_commit_done,
1408 .rpc_count_stats = ff_layout_commit_count_stats,
1409 .rpc_release = pnfs_generic_commit_release,
1410 };
1411
1412 static const struct rpc_call_ops ff_layout_commit_call_ops_v4 = {
1413 .rpc_call_prepare = ff_layout_commit_prepare_v4,
1414 .rpc_call_done = ff_layout_commit_done,
1415 .rpc_count_stats = ff_layout_commit_count_stats,
1416 .rpc_release = pnfs_generic_commit_release,
1417 };
1418
1419 static enum pnfs_try_status
1420 ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
1421 {
1422 struct pnfs_layout_segment *lseg = hdr->lseg;
1423 struct nfs4_pnfs_ds *ds;
1424 struct rpc_clnt *ds_clnt;
1425 struct rpc_cred *ds_cred;
1426 loff_t offset = hdr->args.offset;
1427 u32 idx = hdr->pgio_mirror_idx;
1428 int vers;
1429 struct nfs_fh *fh;
1430
1431 dprintk("--> %s ino %lu pgbase %u req %Zu@%llu\n",
1432 __func__, hdr->inode->i_ino,
1433 hdr->args.pgbase, (size_t)hdr->args.count, offset);
1434
1435 ds = nfs4_ff_layout_prepare_ds(lseg, idx, false);
1436 if (!ds)
1437 goto out_failed;
1438
1439 ds_clnt = nfs4_ff_find_or_create_ds_client(lseg, idx, ds->ds_clp,
1440 hdr->inode);
1441 if (IS_ERR(ds_clnt))
1442 goto out_failed;
1443
1444 ds_cred = ff_layout_get_ds_cred(lseg, idx, hdr->cred);
1445 if (IS_ERR(ds_cred))
1446 goto out_failed;
1447
1448 vers = nfs4_ff_layout_ds_version(lseg, idx);
1449
1450 dprintk("%s USE DS: %s cl_count %d vers %d\n", __func__,
1451 ds->ds_remotestr, atomic_read(&ds->ds_clp->cl_count), vers);
1452
1453 atomic_inc(&ds->ds_clp->cl_count);
1454 hdr->ds_clp = ds->ds_clp;
1455 fh = nfs4_ff_layout_select_ds_fh(lseg, idx);
1456 if (fh)
1457 hdr->args.fh = fh;
1458 /*
1459 * Note that if we ever decide to split across DSes,
1460 * then we may need to handle dense-like offsets.
1461 */
1462 hdr->args.offset = offset;
1463 hdr->mds_offset = offset;
1464
1465 /* Perform an asynchronous read to ds */
1466 nfs_initiate_pgio(ds_clnt, hdr, ds_cred, ds->ds_clp->rpc_ops,
1467 vers == 3 ? &ff_layout_read_call_ops_v3 :
1468 &ff_layout_read_call_ops_v4,
1469 0, RPC_TASK_SOFTCONN);
1470
1471 return PNFS_ATTEMPTED;
1472
1473 out_failed:
1474 if (ff_layout_has_available_ds(lseg))
1475 return PNFS_TRY_AGAIN;
1476 return PNFS_NOT_ATTEMPTED;
1477 }
1478
1479 /* Perform async writes. */
1480 static enum pnfs_try_status
1481 ff_layout_write_pagelist(struct nfs_pgio_header *hdr, int sync)
1482 {
1483 struct pnfs_layout_segment *lseg = hdr->lseg;
1484 struct nfs4_pnfs_ds *ds;
1485 struct rpc_clnt *ds_clnt;
1486 struct rpc_cred *ds_cred;
1487 loff_t offset = hdr->args.offset;
1488 int vers;
1489 struct nfs_fh *fh;
1490 int idx = hdr->pgio_mirror_idx;
1491
1492 ds = nfs4_ff_layout_prepare_ds(lseg, idx, true);
1493 if (!ds)
1494 return PNFS_NOT_ATTEMPTED;
1495
1496 ds_clnt = nfs4_ff_find_or_create_ds_client(lseg, idx, ds->ds_clp,
1497 hdr->inode);
1498 if (IS_ERR(ds_clnt))
1499 return PNFS_NOT_ATTEMPTED;
1500
1501 ds_cred = ff_layout_get_ds_cred(lseg, idx, hdr->cred);
1502 if (IS_ERR(ds_cred))
1503 return PNFS_NOT_ATTEMPTED;
1504
1505 vers = nfs4_ff_layout_ds_version(lseg, idx);
1506
1507 dprintk("%s ino %lu sync %d req %Zu@%llu DS: %s cl_count %d vers %d\n",
1508 __func__, hdr->inode->i_ino, sync, (size_t) hdr->args.count,
1509 offset, ds->ds_remotestr, atomic_read(&ds->ds_clp->cl_count),
1510 vers);
1511
1512 hdr->pgio_done_cb = ff_layout_write_done_cb;
1513 atomic_inc(&ds->ds_clp->cl_count);
1514 hdr->ds_clp = ds->ds_clp;
1515 hdr->ds_commit_idx = idx;
1516 fh = nfs4_ff_layout_select_ds_fh(lseg, idx);
1517 if (fh)
1518 hdr->args.fh = fh;
1519
1520 /*
1521 * Note that if we ever decide to split across DSes,
1522 * then we may need to handle dense-like offsets.
1523 */
1524 hdr->args.offset = offset;
1525
1526 /* Perform an asynchronous write */
1527 nfs_initiate_pgio(ds_clnt, hdr, ds_cred, ds->ds_clp->rpc_ops,
1528 vers == 3 ? &ff_layout_write_call_ops_v3 :
1529 &ff_layout_write_call_ops_v4,
1530 sync, RPC_TASK_SOFTCONN);
1531 return PNFS_ATTEMPTED;
1532 }
1533
1534 static u32 calc_ds_index_from_commit(struct pnfs_layout_segment *lseg, u32 i)
1535 {
1536 return i;
1537 }
1538
1539 static struct nfs_fh *
1540 select_ds_fh_from_commit(struct pnfs_layout_segment *lseg, u32 i)
1541 {
1542 struct nfs4_ff_layout_segment *flseg = FF_LAYOUT_LSEG(lseg);
1543
1544 /* FIXME: Assume that there is only one NFS version available
1545 * for the DS.
1546 */
1547 return &flseg->mirror_array[i]->fh_versions[0];
1548 }
1549
1550 static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
1551 {
1552 struct pnfs_layout_segment *lseg = data->lseg;
1553 struct nfs4_pnfs_ds *ds;
1554 struct rpc_clnt *ds_clnt;
1555 struct rpc_cred *ds_cred;
1556 u32 idx;
1557 int vers;
1558 struct nfs_fh *fh;
1559
1560 idx = calc_ds_index_from_commit(lseg, data->ds_commit_index);
1561 ds = nfs4_ff_layout_prepare_ds(lseg, idx, true);
1562 if (!ds)
1563 goto out_err;
1564
1565 ds_clnt = nfs4_ff_find_or_create_ds_client(lseg, idx, ds->ds_clp,
1566 data->inode);
1567 if (IS_ERR(ds_clnt))
1568 goto out_err;
1569
1570 ds_cred = ff_layout_get_ds_cred(lseg, idx, data->cred);
1571 if (IS_ERR(ds_cred))
1572 goto out_err;
1573
1574 vers = nfs4_ff_layout_ds_version(lseg, idx);
1575
1576 dprintk("%s ino %lu, how %d cl_count %d vers %d\n", __func__,
1577 data->inode->i_ino, how, atomic_read(&ds->ds_clp->cl_count),
1578 vers);
1579 data->commit_done_cb = ff_layout_commit_done_cb;
1580 data->cred = ds_cred;
1581 atomic_inc(&ds->ds_clp->cl_count);
1582 data->ds_clp = ds->ds_clp;
1583 fh = select_ds_fh_from_commit(lseg, data->ds_commit_index);
1584 if (fh)
1585 data->args.fh = fh;
1586
1587 return nfs_initiate_commit(ds_clnt, data, ds->ds_clp->rpc_ops,
1588 vers == 3 ? &ff_layout_commit_call_ops_v3 :
1589 &ff_layout_commit_call_ops_v4,
1590 how, RPC_TASK_SOFTCONN);
1591 out_err:
1592 pnfs_generic_prepare_to_resend_writes(data);
1593 pnfs_generic_commit_release(data);
1594 return -EAGAIN;
1595 }
1596
1597 static int
1598 ff_layout_commit_pagelist(struct inode *inode, struct list_head *mds_pages,
1599 int how, struct nfs_commit_info *cinfo)
1600 {
1601 return pnfs_generic_commit_pagelist(inode, mds_pages, how, cinfo,
1602 ff_layout_initiate_commit);
1603 }
1604
1605 static struct pnfs_ds_commit_info *
1606 ff_layout_get_ds_info(struct inode *inode)
1607 {
1608 struct pnfs_layout_hdr *layout = NFS_I(inode)->layout;
1609
1610 if (layout == NULL)
1611 return NULL;
1612
1613 return &FF_LAYOUT_FROM_HDR(layout)->commit_info;
1614 }
1615
1616 static void
1617 ff_layout_free_deviceid_node(struct nfs4_deviceid_node *d)
1618 {
1619 nfs4_ff_layout_free_deviceid(container_of(d, struct nfs4_ff_layout_ds,
1620 id_node));
1621 }
1622
1623 static int ff_layout_encode_ioerr(struct nfs4_flexfile_layout *flo,
1624 struct xdr_stream *xdr,
1625 const struct nfs4_layoutreturn_args *args)
1626 {
1627 struct pnfs_layout_hdr *hdr = &flo->generic_hdr;
1628 __be32 *start;
1629 int count = 0, ret = 0;
1630
1631 start = xdr_reserve_space(xdr, 4);
1632 if (unlikely(!start))
1633 return -E2BIG;
1634
1635 /* This assume we always return _ALL_ layouts */
1636 spin_lock(&hdr->plh_inode->i_lock);
1637 ret = ff_layout_encode_ds_ioerr(flo, xdr, &count, &args->range);
1638 spin_unlock(&hdr->plh_inode->i_lock);
1639
1640 *start = cpu_to_be32(count);
1641
1642 return ret;
1643 }
1644
1645 /* report nothing for now */
1646 static void ff_layout_encode_iostats(struct nfs4_flexfile_layout *flo,
1647 struct xdr_stream *xdr,
1648 const struct nfs4_layoutreturn_args *args)
1649 {
1650 __be32 *p;
1651
1652 p = xdr_reserve_space(xdr, 4);
1653 if (likely(p))
1654 *p = cpu_to_be32(0);
1655 }
1656
1657 static struct nfs4_deviceid_node *
1658 ff_layout_alloc_deviceid_node(struct nfs_server *server,
1659 struct pnfs_device *pdev, gfp_t gfp_flags)
1660 {
1661 struct nfs4_ff_layout_ds *dsaddr;
1662
1663 dsaddr = nfs4_ff_alloc_deviceid_node(server, pdev, gfp_flags);
1664 if (!dsaddr)
1665 return NULL;
1666 return &dsaddr->id_node;
1667 }
1668
1669 static void
1670 ff_layout_encode_layoutreturn(struct pnfs_layout_hdr *lo,
1671 struct xdr_stream *xdr,
1672 const struct nfs4_layoutreturn_args *args)
1673 {
1674 struct nfs4_flexfile_layout *flo = FF_LAYOUT_FROM_HDR(lo);
1675 __be32 *start;
1676
1677 dprintk("%s: Begin\n", __func__);
1678 start = xdr_reserve_space(xdr, 4);
1679 BUG_ON(!start);
1680
1681 if (ff_layout_encode_ioerr(flo, xdr, args))
1682 goto out;
1683
1684 ff_layout_encode_iostats(flo, xdr, args);
1685 out:
1686 *start = cpu_to_be32((xdr->p - start - 1) * 4);
1687 dprintk("%s: Return\n", __func__);
1688 }
1689
1690 static int
1691 ff_layout_ntop4(const struct sockaddr *sap, char *buf, const size_t buflen)
1692 {
1693 const struct sockaddr_in *sin = (struct sockaddr_in *)sap;
1694
1695 return snprintf(buf, buflen, "%pI4", &sin->sin_addr);
1696 }
1697
1698 static size_t
1699 ff_layout_ntop6_noscopeid(const struct sockaddr *sap, char *buf,
1700 const int buflen)
1701 {
1702 const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap;
1703 const struct in6_addr *addr = &sin6->sin6_addr;
1704
1705 /*
1706 * RFC 4291, Section 2.2.2
1707 *
1708 * Shorthanded ANY address
1709 */
1710 if (ipv6_addr_any(addr))
1711 return snprintf(buf, buflen, "::");
1712
1713 /*
1714 * RFC 4291, Section 2.2.2
1715 *
1716 * Shorthanded loopback address
1717 */
1718 if (ipv6_addr_loopback(addr))
1719 return snprintf(buf, buflen, "::1");
1720
1721 /*
1722 * RFC 4291, Section 2.2.3
1723 *
1724 * Special presentation address format for mapped v4
1725 * addresses.
1726 */
1727 if (ipv6_addr_v4mapped(addr))
1728 return snprintf(buf, buflen, "::ffff:%pI4",
1729 &addr->s6_addr32[3]);
1730
1731 /*
1732 * RFC 4291, Section 2.2.1
1733 */
1734 return snprintf(buf, buflen, "%pI6c", addr);
1735 }
1736
1737 /* Derived from rpc_sockaddr2uaddr */
1738 static void
1739 ff_layout_encode_netaddr(struct xdr_stream *xdr, struct nfs4_pnfs_ds_addr *da)
1740 {
1741 struct sockaddr *sap = (struct sockaddr *)&da->da_addr;
1742 char portbuf[RPCBIND_MAXUADDRPLEN];
1743 char addrbuf[RPCBIND_MAXUADDRLEN];
1744 char *netid;
1745 unsigned short port;
1746 int len, netid_len;
1747 __be32 *p;
1748
1749 switch (sap->sa_family) {
1750 case AF_INET:
1751 if (ff_layout_ntop4(sap, addrbuf, sizeof(addrbuf)) == 0)
1752 return;
1753 port = ntohs(((struct sockaddr_in *)sap)->sin_port);
1754 netid = "tcp";
1755 netid_len = 3;
1756 break;
1757 case AF_INET6:
1758 if (ff_layout_ntop6_noscopeid(sap, addrbuf, sizeof(addrbuf)) == 0)
1759 return;
1760 port = ntohs(((struct sockaddr_in6 *)sap)->sin6_port);
1761 netid = "tcp6";
1762 netid_len = 4;
1763 break;
1764 default:
1765 /* we only support tcp and tcp6 */
1766 WARN_ON_ONCE(1);
1767 return;
1768 }
1769
1770 snprintf(portbuf, sizeof(portbuf), ".%u.%u", port >> 8, port & 0xff);
1771 len = strlcat(addrbuf, portbuf, sizeof(addrbuf));
1772
1773 p = xdr_reserve_space(xdr, 4 + netid_len);
1774 xdr_encode_opaque(p, netid, netid_len);
1775
1776 p = xdr_reserve_space(xdr, 4 + len);
1777 xdr_encode_opaque(p, addrbuf, len);
1778 }
1779
1780 static void
1781 ff_layout_encode_nfstime(struct xdr_stream *xdr,
1782 ktime_t t)
1783 {
1784 struct timespec64 ts;
1785 __be32 *p;
1786
1787 p = xdr_reserve_space(xdr, 12);
1788 ts = ktime_to_timespec64(t);
1789 p = xdr_encode_hyper(p, ts.tv_sec);
1790 *p++ = cpu_to_be32(ts.tv_nsec);
1791 }
1792
1793 static void
1794 ff_layout_encode_io_latency(struct xdr_stream *xdr,
1795 struct nfs4_ff_io_stat *stat)
1796 {
1797 __be32 *p;
1798
1799 p = xdr_reserve_space(xdr, 5 * 8);
1800 p = xdr_encode_hyper(p, stat->ops_requested);
1801 p = xdr_encode_hyper(p, stat->bytes_requested);
1802 p = xdr_encode_hyper(p, stat->ops_completed);
1803 p = xdr_encode_hyper(p, stat->bytes_completed);
1804 p = xdr_encode_hyper(p, stat->bytes_not_delivered);
1805 ff_layout_encode_nfstime(xdr, stat->total_busy_time);
1806 ff_layout_encode_nfstime(xdr, stat->aggregate_completion_time);
1807 }
1808
1809 static void
1810 ff_layout_encode_layoutstats(struct xdr_stream *xdr,
1811 struct nfs42_layoutstat_args *args,
1812 struct nfs42_layoutstat_devinfo *devinfo)
1813 {
1814 struct nfs4_ff_layout_mirror *mirror = devinfo->layout_private;
1815 struct nfs4_pnfs_ds_addr *da;
1816 struct nfs4_pnfs_ds *ds = mirror->mirror_ds->ds;
1817 struct nfs_fh *fh = &mirror->fh_versions[0];
1818 __be32 *p, *start;
1819
1820 da = list_first_entry(&ds->ds_addrs, struct nfs4_pnfs_ds_addr, da_node);
1821 dprintk("%s: DS %s: encoding address %s\n",
1822 __func__, ds->ds_remotestr, da->da_remotestr);
1823 /* layoutupdate length */
1824 start = xdr_reserve_space(xdr, 4);
1825 /* netaddr4 */
1826 ff_layout_encode_netaddr(xdr, da);
1827 /* nfs_fh4 */
1828 p = xdr_reserve_space(xdr, 4 + fh->size);
1829 xdr_encode_opaque(p, fh->data, fh->size);
1830 /* ff_io_latency4 read */
1831 spin_lock(&mirror->lock);
1832 ff_layout_encode_io_latency(xdr, &mirror->read_stat.io_stat);
1833 /* ff_io_latency4 write */
1834 ff_layout_encode_io_latency(xdr, &mirror->write_stat.io_stat);
1835 spin_unlock(&mirror->lock);
1836 /* nfstime4 */
1837 ff_layout_encode_nfstime(xdr, ktime_sub(ktime_get(), mirror->start_time));
1838 /* bool */
1839 p = xdr_reserve_space(xdr, 4);
1840 *p = cpu_to_be32(false);
1841
1842 *start = cpu_to_be32((xdr->p - start - 1) * 4);
1843 }
1844
1845 static bool
1846 ff_layout_mirror_prepare_stats(struct nfs42_layoutstat_args *args,
1847 struct pnfs_layout_segment *pls,
1848 int *dev_count, int dev_limit)
1849 {
1850 struct nfs4_ff_layout_mirror *mirror;
1851 struct nfs4_deviceid_node *dev;
1852 struct nfs42_layoutstat_devinfo *devinfo;
1853 int i;
1854
1855 for (i = 0; i < FF_LAYOUT_MIRROR_COUNT(pls); i++) {
1856 if (*dev_count >= dev_limit)
1857 break;
1858 mirror = FF_LAYOUT_COMP(pls, i);
1859 if (!mirror || !mirror->mirror_ds)
1860 continue;
1861 dev = FF_LAYOUT_DEVID_NODE(pls, i);
1862 devinfo = &args->devinfo[*dev_count];
1863 memcpy(&devinfo->dev_id, &dev->deviceid, NFS4_DEVICEID4_SIZE);
1864 devinfo->offset = pls->pls_range.offset;
1865 devinfo->length = pls->pls_range.length;
1866 /* well, we don't really know if IO is continuous or not! */
1867 devinfo->read_count = mirror->read_stat.io_stat.bytes_completed;
1868 devinfo->read_bytes = mirror->read_stat.io_stat.bytes_completed;
1869 devinfo->write_count = mirror->write_stat.io_stat.bytes_completed;
1870 devinfo->write_bytes = mirror->write_stat.io_stat.bytes_completed;
1871 devinfo->layout_type = LAYOUT_FLEX_FILES;
1872 devinfo->layoutstats_encode = ff_layout_encode_layoutstats;
1873 devinfo->layout_private = mirror;
1874 /* lseg refcount put in cleanup_layoutstats */
1875 pnfs_get_lseg(pls);
1876
1877 ++(*dev_count);
1878 }
1879
1880 return *dev_count < dev_limit;
1881 }
1882
1883 static int
1884 ff_layout_prepare_layoutstats(struct nfs42_layoutstat_args *args)
1885 {
1886 struct pnfs_layout_segment *pls;
1887 int dev_count = 0;
1888
1889 spin_lock(&args->inode->i_lock);
1890 list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
1891 dev_count += FF_LAYOUT_MIRROR_COUNT(pls);
1892 }
1893 spin_unlock(&args->inode->i_lock);
1894 /* For now, send at most PNFS_LAYOUTSTATS_MAXDEV statistics */
1895 if (dev_count > PNFS_LAYOUTSTATS_MAXDEV) {
1896 dprintk("%s: truncating devinfo to limit (%d:%d)\n",
1897 __func__, dev_count, PNFS_LAYOUTSTATS_MAXDEV);
1898 dev_count = PNFS_LAYOUTSTATS_MAXDEV;
1899 }
1900 args->devinfo = kmalloc(dev_count * sizeof(*args->devinfo), GFP_KERNEL);
1901 if (!args->devinfo)
1902 return -ENOMEM;
1903
1904 dev_count = 0;
1905 spin_lock(&args->inode->i_lock);
1906 list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
1907 if (!ff_layout_mirror_prepare_stats(args, pls, &dev_count,
1908 PNFS_LAYOUTSTATS_MAXDEV)) {
1909 break;
1910 }
1911 }
1912 spin_unlock(&args->inode->i_lock);
1913 args->num_dev = dev_count;
1914
1915 return 0;
1916 }
1917
1918 static void
1919 ff_layout_cleanup_layoutstats(struct nfs42_layoutstat_data *data)
1920 {
1921 struct nfs4_ff_layout_mirror *mirror;
1922 int i;
1923
1924 for (i = 0; i < data->args.num_dev; i++) {
1925 mirror = data->args.devinfo[i].layout_private;
1926 data->args.devinfo[i].layout_private = NULL;
1927 pnfs_put_lseg(mirror->lseg);
1928 }
1929 }
1930
1931 static struct pnfs_layoutdriver_type flexfilelayout_type = {
1932 .id = LAYOUT_FLEX_FILES,
1933 .name = "LAYOUT_FLEX_FILES",
1934 .owner = THIS_MODULE,
1935 .alloc_layout_hdr = ff_layout_alloc_layout_hdr,
1936 .free_layout_hdr = ff_layout_free_layout_hdr,
1937 .alloc_lseg = ff_layout_alloc_lseg,
1938 .free_lseg = ff_layout_free_lseg,
1939 .pg_read_ops = &ff_layout_pg_read_ops,
1940 .pg_write_ops = &ff_layout_pg_write_ops,
1941 .get_ds_info = ff_layout_get_ds_info,
1942 .free_deviceid_node = ff_layout_free_deviceid_node,
1943 .mark_request_commit = pnfs_layout_mark_request_commit,
1944 .clear_request_commit = pnfs_generic_clear_request_commit,
1945 .scan_commit_lists = pnfs_generic_scan_commit_lists,
1946 .recover_commit_reqs = pnfs_generic_recover_commit_reqs,
1947 .commit_pagelist = ff_layout_commit_pagelist,
1948 .read_pagelist = ff_layout_read_pagelist,
1949 .write_pagelist = ff_layout_write_pagelist,
1950 .alloc_deviceid_node = ff_layout_alloc_deviceid_node,
1951 .encode_layoutreturn = ff_layout_encode_layoutreturn,
1952 .sync = pnfs_nfs_generic_sync,
1953 .prepare_layoutstats = ff_layout_prepare_layoutstats,
1954 .cleanup_layoutstats = ff_layout_cleanup_layoutstats,
1955 };
1956
1957 static int __init nfs4flexfilelayout_init(void)
1958 {
1959 printk(KERN_INFO "%s: NFSv4 Flexfile Layout Driver Registering...\n",
1960 __func__);
1961 return pnfs_register_layoutdriver(&flexfilelayout_type);
1962 }
1963
1964 static void __exit nfs4flexfilelayout_exit(void)
1965 {
1966 printk(KERN_INFO "%s: NFSv4 Flexfile Layout Driver Unregistering...\n",
1967 __func__);
1968 pnfs_unregister_layoutdriver(&flexfilelayout_type);
1969 }
1970
1971 MODULE_ALIAS("nfs-layouttype4-4");
1972
1973 MODULE_LICENSE("GPL");
1974 MODULE_DESCRIPTION("The NFSv4 flexfile layout driver");
1975
1976 module_init(nfs4flexfilelayout_init);
1977 module_exit(nfs4flexfilelayout_exit);
Linux with added FPC-III support