source: mainline/uspace/srv/bd/hr/raid0.c@ b883aa8

Last change on this file since b883aa8 was 80c760e, checked in by Miroslav Cimerman <mc@…>, 4 months ago

hr: remove truncated_blkno calculation from raid*.c

Calculate the initial truncated_blkno in hr_init_extents_from_cfg,
allowing the removal of hr_extent_t.blkno.

Also fixes double block_fini() on failed volume creation in
hr_util_try_assemble().
  • Property mode set to 100644
File size: 9.0 KB
Line 
1/*
2 * Copyright (c) 2025 Miroslav Cimerman
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 *
9 * - Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * - Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * - The name of the author may not be used to endorse or promote products
15 * derived from this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 */
28
29/** @addtogroup hr
30 * @{
31 */
32/**
33 * @file
34 */
35
36#include <abi/ipc/ipc.h>
37#include <bd_srv.h>
38#include <block.h>
39#include <errno.h>
40#include <hr.h>
41#include <io/log.h>
42#include <ipc/hr.h>
43#include <ipc/services.h>
44#include <loc.h>
45#include <task.h>
46#include <stdio.h>
47#include <stdlib.h>
48#include <str_error.h>
49
50#include "io.h"
51#include "superblock.h"
52#include "util.h"
53#include "var.h"
54
55static void hr_raid0_update_vol_status(hr_volume_t *);
56static void hr_raid0_state_callback(hr_volume_t *, size_t, errno_t);
57static errno_t hr_raid0_bd_op(hr_bd_op_type_t, bd_srv_t *, aoff64_t, size_t,
58 void *, const void *, size_t);
59
60/* bdops */
61static errno_t hr_raid0_bd_open(bd_srvs_t *, bd_srv_t *);
62static errno_t hr_raid0_bd_close(bd_srv_t *);
63static errno_t hr_raid0_bd_read_blocks(bd_srv_t *, aoff64_t, size_t, void *,
64 size_t);
65static errno_t hr_raid0_bd_sync_cache(bd_srv_t *, aoff64_t, size_t);
66static errno_t hr_raid0_bd_write_blocks(bd_srv_t *, aoff64_t, size_t,
67 const void *, size_t);
68static errno_t hr_raid0_bd_get_block_size(bd_srv_t *, size_t *);
69static errno_t hr_raid0_bd_get_num_blocks(bd_srv_t *, aoff64_t *);
70
71static bd_ops_t hr_raid0_bd_ops = {
72 .open = hr_raid0_bd_open,
73 .close = hr_raid0_bd_close,
74 .sync_cache = hr_raid0_bd_sync_cache,
75 .read_blocks = hr_raid0_bd_read_blocks,
76 .write_blocks = hr_raid0_bd_write_blocks,
77 .get_block_size = hr_raid0_bd_get_block_size,
78 .get_num_blocks = hr_raid0_bd_get_num_blocks
79};
80
81extern loc_srv_t *hr_srv;
82
83errno_t hr_raid0_create(hr_volume_t *new_volume)
84{
85 HR_DEBUG("%s()", __func__);
86
87 assert(new_volume->level == HR_LVL_0);
88
89 if (new_volume->extent_no < 2) {
90 HR_ERROR("RAID 0 array needs at least 2 devices\n");
91 return EINVAL;
92 }
93
94 hr_raid0_update_vol_status(new_volume);
95 if (new_volume->status != HR_VOL_ONLINE) {
96 HR_NOTE("\"%s\": unusable state, not creating\n",
97 new_volume->devname);
98 return EINVAL;
99 }
100
101 bd_srvs_init(&new_volume->hr_bds);
102 new_volume->hr_bds.ops = &hr_raid0_bd_ops;
103 new_volume->hr_bds.sarg = new_volume;
104
105 new_volume->state_callback = hr_raid0_state_callback;
106
107 return EOK;
108}
109
110/*
111 * Called only once in volume's lifetime.
112 */
113errno_t hr_raid0_init(hr_volume_t *vol)
114{
115 HR_DEBUG("%s()", __func__);
116
117 assert(vol->level == HR_LVL_0);
118
119 uint64_t total_blkno = vol->truncated_blkno * vol->extent_no;
120
121 vol->data_offset = vol->meta_ops->get_data_offset();
122
123 vol->data_blkno = total_blkno;
124 /* count md blocks */
125 vol->data_blkno -= vol->meta_ops->get_size() * vol->extent_no;
126
127 vol->strip_size = HR_STRIP_SIZE;
128
129 return EOK;
130}
131
132void hr_raid0_status_event(hr_volume_t *vol)
133{
134 HR_DEBUG("%s()", __func__);
135
136 hr_raid0_update_vol_status(vol);
137}
138
139static errno_t hr_raid0_bd_open(bd_srvs_t *bds, bd_srv_t *bd)
140{
141 HR_DEBUG("%s()", __func__);
142
143 hr_volume_t *vol = bd->srvs->sarg;
144
145 atomic_fetch_add_explicit(&vol->open_cnt, 1, memory_order_relaxed);
146
147 return EOK;
148}
149
150static errno_t hr_raid0_bd_close(bd_srv_t *bd)
151{
152 HR_DEBUG("%s()", __func__);
153
154 hr_volume_t *vol = bd->srvs->sarg;
155
156 atomic_fetch_sub_explicit(&vol->open_cnt, 1, memory_order_relaxed);
157
158 return EOK;
159}
160
161static errno_t hr_raid0_bd_sync_cache(bd_srv_t *bd, aoff64_t ba, size_t cnt)
162{
163 return hr_raid0_bd_op(HR_BD_SYNC, bd, ba, cnt, NULL, NULL, 0);
164}
165
166static errno_t hr_raid0_bd_read_blocks(bd_srv_t *bd, aoff64_t ba, size_t cnt,
167 void *buf, size_t size)
168{
169 return hr_raid0_bd_op(HR_BD_READ, bd, ba, cnt, buf, NULL, size);
170}
171
172static errno_t hr_raid0_bd_write_blocks(bd_srv_t *bd, aoff64_t ba, size_t cnt,
173 const void *data, size_t size)
174{
175 return hr_raid0_bd_op(HR_BD_WRITE, bd, ba, cnt, NULL, data, size);
176}
177
178static errno_t hr_raid0_bd_get_block_size(bd_srv_t *bd, size_t *rsize)
179{
180 hr_volume_t *vol = bd->srvs->sarg;
181
182 *rsize = vol->bsize;
183 return EOK;
184}
185
186static errno_t hr_raid0_bd_get_num_blocks(bd_srv_t *bd, aoff64_t *rnb)
187{
188 hr_volume_t *vol = bd->srvs->sarg;
189
190 *rnb = vol->data_blkno;
191 return EOK;
192}
193
194static void hr_raid0_update_vol_status(hr_volume_t *vol)
195{
196 fibril_mutex_lock(&vol->md_lock);
197
198 vol->meta_ops->inc_counter(vol->in_mem_md);
199 /* TODO: save right away */
200
201 fibril_mutex_unlock(&vol->md_lock);
202
203 fibril_rwlock_read_lock(&vol->states_lock);
204
205 hr_vol_status_t old_state = vol->status;
206
207 for (size_t i = 0; i < vol->extent_no; i++) {
208 if (vol->extents[i].status != HR_EXT_ONLINE) {
209 fibril_rwlock_read_unlock(&vol->states_lock);
210
211 if (old_state != HR_VOL_FAULTY) {
212 fibril_rwlock_write_lock(&vol->states_lock);
213 hr_update_vol_status(vol, HR_VOL_FAULTY);
214 fibril_rwlock_write_unlock(&vol->states_lock);
215 }
216 return;
217 }
218 }
219 fibril_rwlock_read_unlock(&vol->states_lock);
220
221 if (old_state != HR_VOL_ONLINE) {
222 fibril_rwlock_write_lock(&vol->states_lock);
223 hr_update_vol_status(vol, HR_VOL_ONLINE);
224 fibril_rwlock_write_unlock(&vol->states_lock);
225 }
226}
227
228static void hr_raid0_state_callback(hr_volume_t *vol, size_t extent, errno_t rc)
229{
230 if (rc == EOK)
231 return;
232
233 fibril_rwlock_write_lock(&vol->states_lock);
234
235 switch (rc) {
236 case ENOENT:
237 hr_update_ext_status(vol, extent, HR_EXT_MISSING);
238 break;
239 default:
240 hr_update_ext_status(vol, extent, HR_EXT_FAILED);
241 }
242
243 hr_update_vol_status(vol, HR_VOL_FAULTY);
244
245 fibril_rwlock_write_unlock(&vol->states_lock);
246}
247
248static errno_t hr_raid0_bd_op(hr_bd_op_type_t type, bd_srv_t *bd, aoff64_t ba,
249 size_t cnt, void *dst, const void *src, size_t size)
250{
251 hr_volume_t *vol = bd->srvs->sarg;
252 errno_t rc;
253 uint64_t phys_block, len;
254 size_t left;
255 const uint8_t *data_write = src;
256 uint8_t *data_read = dst;
257
258 fibril_rwlock_read_lock(&vol->states_lock);
259 if (vol->status != HR_VOL_ONLINE) {
260 fibril_rwlock_read_unlock(&vol->states_lock);
261 return EIO;
262 }
263 fibril_rwlock_read_unlock(&vol->states_lock);
264
265 /* propagate sync */
266 if (type == HR_BD_SYNC && ba == 0 && cnt == 0) {
267 hr_fgroup_t *group = hr_fgroup_create(vol->fge,
268 vol->extent_no);
269 if (group == NULL)
270 return ENOMEM;
271
272 for (size_t i = 0; i < vol->extent_no; i++) {
273 hr_io_t *io = hr_fgroup_alloc(group);
274 io->extent = i;
275 io->ba = ba;
276 io->cnt = cnt;
277 io->type = type;
278 io->vol = vol;
279
280 hr_fgroup_submit(group, hr_io_worker, io);
281 }
282
283 size_t bad;
284 rc = hr_fgroup_wait(group, NULL, &bad);
285 if (rc == ENOMEM)
286 return ENOMEM;
287
288 if (bad > 0)
289 return EIO;
290
291 return EOK;
292 }
293
294 if (type == HR_BD_READ || type == HR_BD_WRITE)
295 if (size < cnt * vol->bsize)
296 return EINVAL;
297
298 rc = hr_check_ba_range(vol, cnt, ba);
299 if (rc != EOK)
300 return rc;
301
302 uint64_t strip_size = vol->strip_size / vol->bsize; /* in blocks */
303 uint64_t strip_no = ba / strip_size;
304 uint64_t extent = strip_no % vol->extent_no;
305 uint64_t stripe = strip_no / vol->extent_no;
306 uint64_t strip_off = ba % strip_size;
307
308 left = cnt;
309
310 /* calculate how many strips does the IO span */
311 size_t end_strip_no = (ba + cnt - 1) / strip_size;
312 size_t span = end_strip_no - strip_no + 1;
313
314 hr_fgroup_t *group = hr_fgroup_create(vol->fge, span);
315 if (group == NULL)
316 return ENOMEM;
317
318 while (left != 0) {
319 phys_block = stripe * strip_size + strip_off;
320 cnt = min(left, strip_size - strip_off);
321 len = vol->bsize * cnt;
322 hr_add_ba_offset(vol, &phys_block);
323
324 hr_io_t *io = hr_fgroup_alloc(group);
325 io->extent = extent;
326 io->data_write = data_write;
327 io->data_read = data_read;
328 io->ba = phys_block;
329 io->cnt = cnt;
330 io->type = type;
331 io->vol = vol;
332
333 hr_fgroup_submit(group, hr_io_worker, io);
334
335 left -= cnt;
336 if (left == 0)
337 break;
338
339 if (type == HR_BD_READ)
340 data_read += len;
341 else if (type == HR_BD_WRITE)
342 data_write += len;
343
344 strip_off = 0;
345 extent++;
346 if (extent >= vol->extent_no) {
347 stripe++;
348 extent = 0;
349 }
350 }
351
352 size_t bad;
353 rc = hr_fgroup_wait(group, NULL, &bad);
354 if (rc == ENOMEM && type == HR_BD_READ)
355 return ENOMEM;
356
357 if (bad > 0)
358 return EIO;
359
360 return EOK;
361}
362
363/** @}
364 */
Note: See TracBrowser for help on using the repository browser.