1 // SPDX-License-Identifier: GPL-2.0-only 1 // SPDX-License-Identifier: GPL-2.0-only
2 /********************************************* 2 /******************************************************************************
3 ********************************************** 3 *******************************************************************************
4 ** 4 **
5 ** Copyright (C) 2005-2011 Red Hat, Inc. All 5 ** Copyright (C) 2005-2011 Red Hat, Inc. All rights reserved.
6 ** 6 **
7 ** 7 **
8 ********************************************** 8 *******************************************************************************
9 ********************************************** 9 ******************************************************************************/
10 10
11 #include "dlm_internal.h" 11 #include "dlm_internal.h"
12 #include "lockspace.h" 12 #include "lockspace.h"
13 #include "member.h" 13 #include "member.h"
14 #include "recoverd.h" 14 #include "recoverd.h"
15 #include "recover.h" 15 #include "recover.h"
16 #include "rcom.h" 16 #include "rcom.h"
17 #include "config.h" 17 #include "config.h"
18 #include "midcomms.h" 18 #include "midcomms.h"
19 #include "lowcomms.h" 19 #include "lowcomms.h"
20 20
21 int dlm_slots_version(const struct dlm_header !! 21 int dlm_slots_version(struct dlm_header *h)
22 { 22 {
23 if ((le32_to_cpu(h->h_version) & 0x000 23 if ((le32_to_cpu(h->h_version) & 0x0000FFFF) < DLM_HEADER_SLOTS)
24 return 0; 24 return 0;
25 return 1; 25 return 1;
26 } 26 }
27 27
28 void dlm_slot_save(struct dlm_ls *ls, struct d 28 void dlm_slot_save(struct dlm_ls *ls, struct dlm_rcom *rc,
29 struct dlm_member *memb) 29 struct dlm_member *memb)
30 { 30 {
31 struct rcom_config *rf = (struct rcom_ 31 struct rcom_config *rf = (struct rcom_config *)rc->rc_buf;
32 32
33 if (!dlm_slots_version(&rc->rc_header) 33 if (!dlm_slots_version(&rc->rc_header))
34 return; 34 return;
35 35
36 memb->slot = le16_to_cpu(rf->rf_our_sl 36 memb->slot = le16_to_cpu(rf->rf_our_slot);
37 memb->generation = le32_to_cpu(rf->rf_ 37 memb->generation = le32_to_cpu(rf->rf_generation);
38 } 38 }
39 39
40 void dlm_slots_copy_out(struct dlm_ls *ls, str 40 void dlm_slots_copy_out(struct dlm_ls *ls, struct dlm_rcom *rc)
41 { 41 {
42 struct dlm_slot *slot; 42 struct dlm_slot *slot;
43 struct rcom_slot *ro; 43 struct rcom_slot *ro;
44 int i; 44 int i;
45 45
46 ro = (struct rcom_slot *)(rc->rc_buf + 46 ro = (struct rcom_slot *)(rc->rc_buf + sizeof(struct rcom_config));
47 47
48 /* ls_slots array is sparse, but not r 48 /* ls_slots array is sparse, but not rcom_slots */
49 49
50 for (i = 0; i < ls->ls_slots_size; i++ 50 for (i = 0; i < ls->ls_slots_size; i++) {
51 slot = &ls->ls_slots[i]; 51 slot = &ls->ls_slots[i];
52 if (!slot->nodeid) 52 if (!slot->nodeid)
53 continue; 53 continue;
54 ro->ro_nodeid = cpu_to_le32(sl 54 ro->ro_nodeid = cpu_to_le32(slot->nodeid);
55 ro->ro_slot = cpu_to_le16(slot 55 ro->ro_slot = cpu_to_le16(slot->slot);
56 ro++; 56 ro++;
57 } 57 }
58 } 58 }
59 59
60 #define SLOT_DEBUG_LINE 128 60 #define SLOT_DEBUG_LINE 128
61 61
62 static void log_slots(struct dlm_ls *ls, uint3 62 static void log_slots(struct dlm_ls *ls, uint32_t gen, int num_slots,
63 struct rcom_slot *ro0, s 63 struct rcom_slot *ro0, struct dlm_slot *array,
64 int array_size) 64 int array_size)
65 { 65 {
66 char line[SLOT_DEBUG_LINE]; 66 char line[SLOT_DEBUG_LINE];
67 int len = SLOT_DEBUG_LINE - 1; 67 int len = SLOT_DEBUG_LINE - 1;
68 int pos = 0; 68 int pos = 0;
69 int ret, i; 69 int ret, i;
70 70
71 memset(line, 0, sizeof(line)); 71 memset(line, 0, sizeof(line));
72 72
73 if (array) { 73 if (array) {
74 for (i = 0; i < array_size; i+ 74 for (i = 0; i < array_size; i++) {
75 if (!array[i].nodeid) 75 if (!array[i].nodeid)
76 continue; 76 continue;
77 77
78 ret = snprintf(line + 78 ret = snprintf(line + pos, len - pos, " %d:%d",
79 array[i 79 array[i].slot, array[i].nodeid);
80 if (ret >= len - pos) 80 if (ret >= len - pos)
81 break; 81 break;
82 pos += ret; 82 pos += ret;
83 } 83 }
84 } else if (ro0) { 84 } else if (ro0) {
85 for (i = 0; i < num_slots; i++ 85 for (i = 0; i < num_slots; i++) {
86 ret = snprintf(line + 86 ret = snprintf(line + pos, len - pos, " %d:%d",
87 ro0[i]. 87 ro0[i].ro_slot, ro0[i].ro_nodeid);
88 if (ret >= len - pos) 88 if (ret >= len - pos)
89 break; 89 break;
90 pos += ret; 90 pos += ret;
91 } 91 }
92 } 92 }
93 93
94 log_rinfo(ls, "generation %u slots %d% 94 log_rinfo(ls, "generation %u slots %d%s", gen, num_slots, line);
95 } 95 }
96 96
97 int dlm_slots_copy_in(struct dlm_ls *ls) 97 int dlm_slots_copy_in(struct dlm_ls *ls)
98 { 98 {
99 struct dlm_member *memb; 99 struct dlm_member *memb;
100 struct dlm_rcom *rc = ls->ls_recover_b 100 struct dlm_rcom *rc = ls->ls_recover_buf;
101 struct rcom_config *rf = (struct rcom_ 101 struct rcom_config *rf = (struct rcom_config *)rc->rc_buf;
102 struct rcom_slot *ro0, *ro; 102 struct rcom_slot *ro0, *ro;
103 int our_nodeid = dlm_our_nodeid(); 103 int our_nodeid = dlm_our_nodeid();
104 int i, num_slots; 104 int i, num_slots;
105 uint32_t gen; 105 uint32_t gen;
106 106
107 if (!dlm_slots_version(&rc->rc_header) 107 if (!dlm_slots_version(&rc->rc_header))
108 return -1; 108 return -1;
109 109
110 gen = le32_to_cpu(rf->rf_generation); 110 gen = le32_to_cpu(rf->rf_generation);
111 if (gen <= ls->ls_generation) { 111 if (gen <= ls->ls_generation) {
112 log_error(ls, "dlm_slots_copy_ 112 log_error(ls, "dlm_slots_copy_in gen %u old %u",
113 gen, ls->ls_generati 113 gen, ls->ls_generation);
114 } 114 }
115 ls->ls_generation = gen; 115 ls->ls_generation = gen;
116 116
117 num_slots = le16_to_cpu(rf->rf_num_slo 117 num_slots = le16_to_cpu(rf->rf_num_slots);
118 if (!num_slots) 118 if (!num_slots)
119 return -1; 119 return -1;
120 120
121 ro0 = (struct rcom_slot *)(rc->rc_buf 121 ro0 = (struct rcom_slot *)(rc->rc_buf + sizeof(struct rcom_config));
122 122
123 log_slots(ls, gen, num_slots, ro0, NUL 123 log_slots(ls, gen, num_slots, ro0, NULL, 0);
124 124
125 list_for_each_entry(memb, &ls->ls_node 125 list_for_each_entry(memb, &ls->ls_nodes, list) {
126 for (i = 0, ro = ro0; i < num_ 126 for (i = 0, ro = ro0; i < num_slots; i++, ro++) {
127 if (le32_to_cpu(ro->ro 127 if (le32_to_cpu(ro->ro_nodeid) != memb->nodeid)
128 continue; 128 continue;
129 memb->slot = le16_to_c 129 memb->slot = le16_to_cpu(ro->ro_slot);
130 memb->slot_prev = memb 130 memb->slot_prev = memb->slot;
131 break; 131 break;
132 } 132 }
133 133
134 if (memb->nodeid == our_nodeid 134 if (memb->nodeid == our_nodeid) {
135 if (ls->ls_slot && ls- 135 if (ls->ls_slot && ls->ls_slot != memb->slot) {
136 log_error(ls, 136 log_error(ls, "dlm_slots_copy_in our slot "
137 "cha 137 "changed %d %d", ls->ls_slot,
138 memb 138 memb->slot);
139 return -1; 139 return -1;
140 } 140 }
141 141
142 if (!ls->ls_slot) 142 if (!ls->ls_slot)
143 ls->ls_slot = 143 ls->ls_slot = memb->slot;
144 } 144 }
145 145
146 if (!memb->slot) { 146 if (!memb->slot) {
147 log_error(ls, "dlm_slo 147 log_error(ls, "dlm_slots_copy_in nodeid %d no slot",
148 memb->nodei 148 memb->nodeid);
149 return -1; 149 return -1;
150 } 150 }
151 } 151 }
152 152
153 return 0; 153 return 0;
154 } 154 }
155 155
156 /* for any nodes that do not support slots, we 156 /* for any nodes that do not support slots, we will not have set memb->slot
157 in wait_status_all(), so memb->slot will re 157 in wait_status_all(), so memb->slot will remain -1, and we will not
158 assign slots or set ls_num_slots here */ 158 assign slots or set ls_num_slots here */
159 159
160 int dlm_slots_assign(struct dlm_ls *ls, int *n 160 int dlm_slots_assign(struct dlm_ls *ls, int *num_slots, int *slots_size,
161 struct dlm_slot **slots_o 161 struct dlm_slot **slots_out, uint32_t *gen_out)
162 { 162 {
163 struct dlm_member *memb; 163 struct dlm_member *memb;
164 struct dlm_slot *array; 164 struct dlm_slot *array;
165 int our_nodeid = dlm_our_nodeid(); 165 int our_nodeid = dlm_our_nodeid();
166 int array_size, max_slots, i; 166 int array_size, max_slots, i;
167 int need = 0; 167 int need = 0;
168 int max = 0; 168 int max = 0;
169 int num = 0; 169 int num = 0;
170 uint32_t gen = 0; 170 uint32_t gen = 0;
171 171
172 /* our own memb struct will have slot 172 /* our own memb struct will have slot -1 gen 0 */
173 173
174 list_for_each_entry(memb, &ls->ls_node 174 list_for_each_entry(memb, &ls->ls_nodes, list) {
175 if (memb->nodeid == our_nodeid 175 if (memb->nodeid == our_nodeid) {
176 memb->slot = ls->ls_sl 176 memb->slot = ls->ls_slot;
177 memb->generation = ls- 177 memb->generation = ls->ls_generation;
178 break; 178 break;
179 } 179 }
180 } 180 }
181 181
182 list_for_each_entry(memb, &ls->ls_node 182 list_for_each_entry(memb, &ls->ls_nodes, list) {
183 if (memb->generation > gen) 183 if (memb->generation > gen)
184 gen = memb->generation 184 gen = memb->generation;
185 185
186 /* node doesn't support slots 186 /* node doesn't support slots */
187 187
188 if (memb->slot == -1) 188 if (memb->slot == -1)
189 return -1; 189 return -1;
190 190
191 /* node needs a slot assigned 191 /* node needs a slot assigned */
192 192
193 if (!memb->slot) 193 if (!memb->slot)
194 need++; 194 need++;
195 195
196 /* node has a slot assigned */ 196 /* node has a slot assigned */
197 197
198 num++; 198 num++;
199 199
200 if (!max || max < memb->slot) 200 if (!max || max < memb->slot)
201 max = memb->slot; 201 max = memb->slot;
202 202
203 /* sanity check, once slot is 203 /* sanity check, once slot is assigned it shouldn't change */
204 204
205 if (memb->slot_prev && memb->s 205 if (memb->slot_prev && memb->slot && memb->slot_prev != memb->slot) {
206 log_error(ls, "nodeid 206 log_error(ls, "nodeid %d slot changed %d %d",
207 memb->nodeid 207 memb->nodeid, memb->slot_prev, memb->slot);
208 return -1; 208 return -1;
209 } 209 }
210 memb->slot_prev = memb->slot; 210 memb->slot_prev = memb->slot;
211 } 211 }
212 212
213 array_size = max + need; 213 array_size = max + need;
214 array = kcalloc(array_size, sizeof(*ar 214 array = kcalloc(array_size, sizeof(*array), GFP_NOFS);
215 if (!array) 215 if (!array)
216 return -ENOMEM; 216 return -ENOMEM;
217 217
218 num = 0; 218 num = 0;
219 219
220 /* fill in slots (offsets) that are us 220 /* fill in slots (offsets) that are used */
221 221
222 list_for_each_entry(memb, &ls->ls_node 222 list_for_each_entry(memb, &ls->ls_nodes, list) {
223 if (!memb->slot) 223 if (!memb->slot)
224 continue; 224 continue;
225 225
226 if (memb->slot > array_size) { 226 if (memb->slot > array_size) {
227 log_error(ls, "invalid 227 log_error(ls, "invalid slot number %d", memb->slot);
228 kfree(array); 228 kfree(array);
229 return -1; 229 return -1;
230 } 230 }
231 231
232 array[memb->slot - 1].nodeid = 232 array[memb->slot - 1].nodeid = memb->nodeid;
233 array[memb->slot - 1].slot = m 233 array[memb->slot - 1].slot = memb->slot;
234 num++; 234 num++;
235 } 235 }
236 236
237 /* assign new slots from unused offset 237 /* assign new slots from unused offsets */
238 238
239 list_for_each_entry(memb, &ls->ls_node 239 list_for_each_entry(memb, &ls->ls_nodes, list) {
240 if (memb->slot) 240 if (memb->slot)
241 continue; 241 continue;
242 242
243 for (i = 0; i < array_size; i+ 243 for (i = 0; i < array_size; i++) {
244 if (array[i].nodeid) 244 if (array[i].nodeid)
245 continue; 245 continue;
246 246
247 memb->slot = i + 1; 247 memb->slot = i + 1;
248 memb->slot_prev = memb 248 memb->slot_prev = memb->slot;
249 array[i].nodeid = memb 249 array[i].nodeid = memb->nodeid;
250 array[i].slot = memb-> 250 array[i].slot = memb->slot;
251 num++; 251 num++;
252 252
253 if (!ls->ls_slot && me 253 if (!ls->ls_slot && memb->nodeid == our_nodeid)
254 ls->ls_slot = 254 ls->ls_slot = memb->slot;
255 break; 255 break;
256 } 256 }
257 257
258 if (!memb->slot) { 258 if (!memb->slot) {
259 log_error(ls, "no free 259 log_error(ls, "no free slot found");
260 kfree(array); 260 kfree(array);
261 return -1; 261 return -1;
262 } 262 }
263 } 263 }
264 264
265 gen++; 265 gen++;
266 266
267 log_slots(ls, gen, num, NULL, array, a 267 log_slots(ls, gen, num, NULL, array, array_size);
268 268
269 max_slots = (DLM_MAX_APP_BUFSIZE - siz 269 max_slots = (DLM_MAX_APP_BUFSIZE - sizeof(struct dlm_rcom) -
270 sizeof(struct rcom_config 270 sizeof(struct rcom_config)) / sizeof(struct rcom_slot);
271 271
272 if (num > max_slots) { 272 if (num > max_slots) {
273 log_error(ls, "num_slots %d ex 273 log_error(ls, "num_slots %d exceeds max_slots %d",
274 num, max_slots); 274 num, max_slots);
275 kfree(array); 275 kfree(array);
276 return -1; 276 return -1;
277 } 277 }
278 278
279 *gen_out = gen; 279 *gen_out = gen;
280 *slots_out = array; 280 *slots_out = array;
281 *slots_size = array_size; 281 *slots_size = array_size;
282 *num_slots = num; 282 *num_slots = num;
283 return 0; 283 return 0;
284 } 284 }
285 285
286 static void add_ordered_member(struct dlm_ls * 286 static void add_ordered_member(struct dlm_ls *ls, struct dlm_member *new)
287 { 287 {
288 struct dlm_member *memb = NULL; 288 struct dlm_member *memb = NULL;
289 struct list_head *tmp; 289 struct list_head *tmp;
290 struct list_head *newlist = &new->list 290 struct list_head *newlist = &new->list;
291 struct list_head *head = &ls->ls_nodes 291 struct list_head *head = &ls->ls_nodes;
292 292
293 list_for_each(tmp, head) { 293 list_for_each(tmp, head) {
294 memb = list_entry(tmp, struct 294 memb = list_entry(tmp, struct dlm_member, list);
295 if (new->nodeid < memb->nodeid 295 if (new->nodeid < memb->nodeid)
296 break; 296 break;
297 } 297 }
298 298
299 if (!memb) 299 if (!memb)
300 list_add_tail(newlist, head); 300 list_add_tail(newlist, head);
301 else { 301 else {
302 /* FIXME: can use list macro h 302 /* FIXME: can use list macro here */
303 newlist->prev = tmp->prev; 303 newlist->prev = tmp->prev;
304 newlist->next = tmp; 304 newlist->next = tmp;
305 tmp->prev->next = newlist; 305 tmp->prev->next = newlist;
306 tmp->prev = newlist; 306 tmp->prev = newlist;
307 } 307 }
308 } 308 }
309 309
310 static int add_remote_member(int nodeid) <<
311 { <<
312 int error; <<
313 <<
314 if (nodeid == dlm_our_nodeid()) <<
315 return 0; <<
316 <<
317 error = dlm_lowcomms_connect_node(node <<
318 if (error < 0) <<
319 return error; <<
320 <<
321 dlm_midcomms_add_member(nodeid); <<
322 return 0; <<
323 } <<
324 <<
325 static int dlm_add_member(struct dlm_ls *ls, s 310 static int dlm_add_member(struct dlm_ls *ls, struct dlm_config_node *node)
326 { 311 {
327 struct dlm_member *memb; 312 struct dlm_member *memb;
328 int error; 313 int error;
329 314
330 memb = kzalloc(sizeof(*memb), GFP_NOFS 315 memb = kzalloc(sizeof(*memb), GFP_NOFS);
331 if (!memb) 316 if (!memb)
332 return -ENOMEM; 317 return -ENOMEM;
333 318
334 memb->nodeid = node->nodeid; !! 319 error = dlm_lowcomms_connect_node(node->nodeid);
335 memb->weight = node->weight; <<
336 memb->comm_seq = node->comm_seq; <<
337 <<
338 error = add_remote_member(node->nodeid <<
339 if (error < 0) { 320 if (error < 0) {
340 kfree(memb); 321 kfree(memb);
341 return error; 322 return error;
342 } 323 }
343 324
>> 325 memb->nodeid = node->nodeid;
>> 326 memb->weight = node->weight;
>> 327 memb->comm_seq = node->comm_seq;
>> 328 dlm_midcomms_add_member(node->nodeid);
344 add_ordered_member(ls, memb); 329 add_ordered_member(ls, memb);
345 ls->ls_num_nodes++; 330 ls->ls_num_nodes++;
346 return 0; 331 return 0;
347 } 332 }
348 333
349 static struct dlm_member *find_memb(struct lis 334 static struct dlm_member *find_memb(struct list_head *head, int nodeid)
350 { 335 {
351 struct dlm_member *memb; 336 struct dlm_member *memb;
352 337
353 list_for_each_entry(memb, head, list) 338 list_for_each_entry(memb, head, list) {
354 if (memb->nodeid == nodeid) 339 if (memb->nodeid == nodeid)
355 return memb; 340 return memb;
356 } 341 }
357 return NULL; 342 return NULL;
358 } 343 }
359 344
360 int dlm_is_member(struct dlm_ls *ls, int nodei 345 int dlm_is_member(struct dlm_ls *ls, int nodeid)
361 { 346 {
362 if (find_memb(&ls->ls_nodes, nodeid)) 347 if (find_memb(&ls->ls_nodes, nodeid))
363 return 1; 348 return 1;
364 return 0; 349 return 0;
365 } 350 }
366 351
367 int dlm_is_removed(struct dlm_ls *ls, int node 352 int dlm_is_removed(struct dlm_ls *ls, int nodeid)
368 { 353 {
369 WARN_ON_ONCE(!nodeid || nodeid == -1); <<
370 <<
371 if (find_memb(&ls->ls_nodes_gone, node 354 if (find_memb(&ls->ls_nodes_gone, nodeid))
372 return 1; 355 return 1;
373 return 0; 356 return 0;
374 } 357 }
375 358
376 static void clear_memb_list(struct list_head * 359 static void clear_memb_list(struct list_head *head,
377 void (*after_del)( 360 void (*after_del)(int nodeid))
378 { 361 {
379 struct dlm_member *memb; 362 struct dlm_member *memb;
380 363
381 while (!list_empty(head)) { 364 while (!list_empty(head)) {
382 memb = list_entry(head->next, 365 memb = list_entry(head->next, struct dlm_member, list);
383 list_del(&memb->list); 366 list_del(&memb->list);
384 if (after_del) 367 if (after_del)
385 after_del(memb->nodeid 368 after_del(memb->nodeid);
386 kfree(memb); 369 kfree(memb);
387 } 370 }
388 } 371 }
389 372
390 static void remove_remote_member(int nodeid) !! 373 static void clear_members_cb(int nodeid)
391 { 374 {
392 if (nodeid == dlm_our_nodeid()) <<
393 return; <<
394 <<
395 dlm_midcomms_remove_member(nodeid); 375 dlm_midcomms_remove_member(nodeid);
396 } 376 }
397 377
398 void dlm_clear_members(struct dlm_ls *ls) 378 void dlm_clear_members(struct dlm_ls *ls)
399 { 379 {
400 clear_memb_list(&ls->ls_nodes, remove_ !! 380 clear_memb_list(&ls->ls_nodes, clear_members_cb);
401 ls->ls_num_nodes = 0; 381 ls->ls_num_nodes = 0;
402 } 382 }
403 383
404 void dlm_clear_members_gone(struct dlm_ls *ls) 384 void dlm_clear_members_gone(struct dlm_ls *ls)
405 { 385 {
406 clear_memb_list(&ls->ls_nodes_gone, NU 386 clear_memb_list(&ls->ls_nodes_gone, NULL);
407 } 387 }
408 388
409 static void make_member_array(struct dlm_ls *l 389 static void make_member_array(struct dlm_ls *ls)
410 { 390 {
411 struct dlm_member *memb; 391 struct dlm_member *memb;
412 int i, w, x = 0, total = 0, all_zero = 392 int i, w, x = 0, total = 0, all_zero = 0, *array;
413 393
414 kfree(ls->ls_node_array); 394 kfree(ls->ls_node_array);
415 ls->ls_node_array = NULL; 395 ls->ls_node_array = NULL;
416 396
417 list_for_each_entry(memb, &ls->ls_node 397 list_for_each_entry(memb, &ls->ls_nodes, list) {
418 if (memb->weight) 398 if (memb->weight)
419 total += memb->weight; 399 total += memb->weight;
420 } 400 }
421 401
422 /* all nodes revert to weight of 1 if 402 /* all nodes revert to weight of 1 if all have weight 0 */
423 403
424 if (!total) { 404 if (!total) {
425 total = ls->ls_num_nodes; 405 total = ls->ls_num_nodes;
426 all_zero = 1; 406 all_zero = 1;
427 } 407 }
428 408
429 ls->ls_total_weight = total; 409 ls->ls_total_weight = total;
430 array = kmalloc_array(total, sizeof(*a 410 array = kmalloc_array(total, sizeof(*array), GFP_NOFS);
431 if (!array) 411 if (!array)
432 return; 412 return;
433 413
434 list_for_each_entry(memb, &ls->ls_node 414 list_for_each_entry(memb, &ls->ls_nodes, list) {
435 if (!all_zero && !memb->weight 415 if (!all_zero && !memb->weight)
436 continue; 416 continue;
437 417
438 if (all_zero) 418 if (all_zero)
439 w = 1; 419 w = 1;
440 else 420 else
441 w = memb->weight; 421 w = memb->weight;
442 422
443 DLM_ASSERT(x < total, printk(" 423 DLM_ASSERT(x < total, printk("total %d x %d\n", total, x););
444 424
445 for (i = 0; i < w; i++) 425 for (i = 0; i < w; i++)
446 array[x++] = memb->nod 426 array[x++] = memb->nodeid;
447 } 427 }
448 428
449 ls->ls_node_array = array; 429 ls->ls_node_array = array;
450 } 430 }
451 431
452 /* send a status request to all members just t 432 /* send a status request to all members just to establish comms connections */
453 433
454 static int ping_members(struct dlm_ls *ls, uin !! 434 static int ping_members(struct dlm_ls *ls)
455 { 435 {
456 struct dlm_member *memb; 436 struct dlm_member *memb;
457 int error = 0; 437 int error = 0;
458 438
459 list_for_each_entry(memb, &ls->ls_node 439 list_for_each_entry(memb, &ls->ls_nodes, list) {
460 if (dlm_recovery_stopped(ls)) 440 if (dlm_recovery_stopped(ls)) {
461 error = -EINTR; 441 error = -EINTR;
462 break; 442 break;
463 } 443 }
464 error = dlm_rcom_status(ls, me !! 444 error = dlm_rcom_status(ls, memb->nodeid, 0);
465 if (error) 445 if (error)
466 break; 446 break;
467 } 447 }
468 if (error) 448 if (error)
469 log_rinfo(ls, "ping_members ab 449 log_rinfo(ls, "ping_members aborted %d last nodeid %d",
470 error, ls->ls_recove 450 error, ls->ls_recover_nodeid);
471 return error; 451 return error;
472 } 452 }
473 453
474 static void dlm_lsop_recover_prep(struct dlm_l 454 static void dlm_lsop_recover_prep(struct dlm_ls *ls)
475 { 455 {
476 if (!ls->ls_ops || !ls->ls_ops->recove 456 if (!ls->ls_ops || !ls->ls_ops->recover_prep)
477 return; 457 return;
478 ls->ls_ops->recover_prep(ls->ls_ops_ar 458 ls->ls_ops->recover_prep(ls->ls_ops_arg);
479 } 459 }
480 460
481 static void dlm_lsop_recover_slot(struct dlm_l 461 static void dlm_lsop_recover_slot(struct dlm_ls *ls, struct dlm_member *memb)
482 { 462 {
483 struct dlm_slot slot; 463 struct dlm_slot slot;
484 uint32_t seq; 464 uint32_t seq;
485 int error; 465 int error;
486 466
487 if (!ls->ls_ops || !ls->ls_ops->recove 467 if (!ls->ls_ops || !ls->ls_ops->recover_slot)
488 return; 468 return;
489 469
490 /* if there is no comms connection wit 470 /* if there is no comms connection with this node
491 or the present comms connection is 471 or the present comms connection is newer
492 than the one when this member was a 472 than the one when this member was added, then
493 we consider the node to have failed 473 we consider the node to have failed (versus
494 being removed due to dlm_release_lo 474 being removed due to dlm_release_lockspace) */
495 475
496 error = dlm_comm_seq(memb->nodeid, &se 476 error = dlm_comm_seq(memb->nodeid, &seq);
497 477
498 if (!error && seq == memb->comm_seq) 478 if (!error && seq == memb->comm_seq)
499 return; 479 return;
500 480
501 slot.nodeid = memb->nodeid; 481 slot.nodeid = memb->nodeid;
502 slot.slot = memb->slot; 482 slot.slot = memb->slot;
503 483
504 ls->ls_ops->recover_slot(ls->ls_ops_ar 484 ls->ls_ops->recover_slot(ls->ls_ops_arg, &slot);
505 } 485 }
506 486
507 void dlm_lsop_recover_done(struct dlm_ls *ls) 487 void dlm_lsop_recover_done(struct dlm_ls *ls)
508 { 488 {
509 struct dlm_member *memb; 489 struct dlm_member *memb;
510 struct dlm_slot *slots; 490 struct dlm_slot *slots;
511 int i, num; 491 int i, num;
512 492
513 if (!ls->ls_ops || !ls->ls_ops->recove 493 if (!ls->ls_ops || !ls->ls_ops->recover_done)
514 return; 494 return;
515 495
516 num = ls->ls_num_nodes; 496 num = ls->ls_num_nodes;
517 slots = kcalloc(num, sizeof(*slots), G 497 slots = kcalloc(num, sizeof(*slots), GFP_KERNEL);
518 if (!slots) 498 if (!slots)
519 return; 499 return;
520 500
521 i = 0; 501 i = 0;
522 list_for_each_entry(memb, &ls->ls_node 502 list_for_each_entry(memb, &ls->ls_nodes, list) {
523 if (i == num) { 503 if (i == num) {
524 log_error(ls, "dlm_lso 504 log_error(ls, "dlm_lsop_recover_done bad num %d", num);
525 goto out; 505 goto out;
526 } 506 }
527 slots[i].nodeid = memb->nodeid 507 slots[i].nodeid = memb->nodeid;
528 slots[i].slot = memb->slot; 508 slots[i].slot = memb->slot;
529 i++; 509 i++;
530 } 510 }
531 511
532 ls->ls_ops->recover_done(ls->ls_ops_ar 512 ls->ls_ops->recover_done(ls->ls_ops_arg, slots, num,
533 ls->ls_slot, 513 ls->ls_slot, ls->ls_generation);
534 out: 514 out:
535 kfree(slots); 515 kfree(slots);
536 } 516 }
537 517
538 static struct dlm_config_node *find_config_nod 518 static struct dlm_config_node *find_config_node(struct dlm_recover *rv,
539 519 int nodeid)
540 { 520 {
541 int i; 521 int i;
542 522
543 for (i = 0; i < rv->nodes_count; i++) 523 for (i = 0; i < rv->nodes_count; i++) {
544 if (rv->nodes[i].nodeid == nod 524 if (rv->nodes[i].nodeid == nodeid)
545 return &rv->nodes[i]; 525 return &rv->nodes[i];
546 } 526 }
547 return NULL; 527 return NULL;
548 } 528 }
549 529
550 int dlm_recover_members(struct dlm_ls *ls, str 530 int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
551 { 531 {
552 struct dlm_member *memb, *safe; 532 struct dlm_member *memb, *safe;
553 struct dlm_config_node *node; 533 struct dlm_config_node *node;
554 int i, error, neg = 0, low = -1; 534 int i, error, neg = 0, low = -1;
555 535
556 /* previously removed members that we' 536 /* previously removed members that we've not finished removing need to
557 * count as a negative change so the " 537 * count as a negative change so the "neg" recovery steps will happen
558 * 538 *
559 * This functionality must report all 539 * This functionality must report all member changes to lsops or
560 * midcomms layer and must never retur 540 * midcomms layer and must never return before.
561 */ 541 */
562 542
563 list_for_each_entry(memb, &ls->ls_node 543 list_for_each_entry(memb, &ls->ls_nodes_gone, list) {
564 log_rinfo(ls, "prev removed me 544 log_rinfo(ls, "prev removed member %d", memb->nodeid);
565 neg++; 545 neg++;
566 } 546 }
567 547
568 /* move departed members from ls_nodes 548 /* move departed members from ls_nodes to ls_nodes_gone */
569 549
570 list_for_each_entry_safe(memb, safe, & 550 list_for_each_entry_safe(memb, safe, &ls->ls_nodes, list) {
571 node = find_config_node(rv, me 551 node = find_config_node(rv, memb->nodeid);
572 if (node && !node->new) 552 if (node && !node->new)
573 continue; 553 continue;
574 554
575 if (!node) { 555 if (!node) {
576 log_rinfo(ls, "remove 556 log_rinfo(ls, "remove member %d", memb->nodeid);
577 } else { 557 } else {
578 /* removed and re-adde 558 /* removed and re-added */
579 log_rinfo(ls, "remove 559 log_rinfo(ls, "remove member %d comm_seq %u %u",
580 memb->nodeid 560 memb->nodeid, memb->comm_seq, node->comm_seq);
581 } 561 }
582 562
583 neg++; 563 neg++;
584 list_move(&memb->list, &ls->ls 564 list_move(&memb->list, &ls->ls_nodes_gone);
585 remove_remote_member(memb->nod !! 565 dlm_midcomms_remove_member(memb->nodeid);
586 ls->ls_num_nodes--; 566 ls->ls_num_nodes--;
587 dlm_lsop_recover_slot(ls, memb 567 dlm_lsop_recover_slot(ls, memb);
588 } 568 }
589 569
590 /* add new members to ls_nodes */ 570 /* add new members to ls_nodes */
591 571
592 for (i = 0; i < rv->nodes_count; i++) 572 for (i = 0; i < rv->nodes_count; i++) {
593 node = &rv->nodes[i]; 573 node = &rv->nodes[i];
594 if (dlm_is_member(ls, node->no 574 if (dlm_is_member(ls, node->nodeid))
595 continue; 575 continue;
596 error = dlm_add_member(ls, nod !! 576 dlm_add_member(ls, node);
597 if (error) <<
598 return error; <<
599 <<
600 log_rinfo(ls, "add member %d", 577 log_rinfo(ls, "add member %d", node->nodeid);
601 } 578 }
602 579
603 list_for_each_entry(memb, &ls->ls_node 580 list_for_each_entry(memb, &ls->ls_nodes, list) {
604 if (low == -1 || memb->nodeid 581 if (low == -1 || memb->nodeid < low)
605 low = memb->nodeid; 582 low = memb->nodeid;
606 } 583 }
607 ls->ls_low_nodeid = low; 584 ls->ls_low_nodeid = low;
608 585
609 make_member_array(ls); 586 make_member_array(ls);
610 *neg_out = neg; 587 *neg_out = neg;
611 588
612 error = ping_members(ls, rv->seq); !! 589 error = ping_members(ls);
613 log_rinfo(ls, "dlm_recover_members %d 590 log_rinfo(ls, "dlm_recover_members %d nodes", ls->ls_num_nodes);
614 return error; 591 return error;
615 } 592 }
616 593
617 /* Userspace guarantees that dlm_ls_stop() has 594 /* Userspace guarantees that dlm_ls_stop() has completed on all nodes before
618 dlm_ls_start() is called on any of them to 595 dlm_ls_start() is called on any of them to start the new recovery. */
619 596
620 int dlm_ls_stop(struct dlm_ls *ls) 597 int dlm_ls_stop(struct dlm_ls *ls)
621 { 598 {
622 int new; 599 int new;
623 600
624 /* 601 /*
625 * Prevent dlm_recv from being in the 602 * Prevent dlm_recv from being in the middle of something when we do
626 * the stop. This includes ensuring d 603 * the stop. This includes ensuring dlm_recv isn't processing a
627 * recovery message (rcom), while dlm_ 604 * recovery message (rcom), while dlm_recoverd is aborting and
628 * resetting things from an in-progres 605 * resetting things from an in-progress recovery. i.e. we want
629 * dlm_recoverd to abort its recovery 606 * dlm_recoverd to abort its recovery without worrying about dlm_recv
630 * processing an rcom at the same time 607 * processing an rcom at the same time. Stopping dlm_recv also makes
631 * it easy for dlm_receive_message() t 608 * it easy for dlm_receive_message() to check locking stopped and add a
632 * message to the requestqueue without 609 * message to the requestqueue without races.
633 */ 610 */
634 611
635 write_lock_bh(&ls->ls_recv_active); !! 612 down_write(&ls->ls_recv_active);
636 613
637 /* 614 /*
638 * Abort any recovery that's in progre 615 * Abort any recovery that's in progress (see RECOVER_STOP,
639 * dlm_recovery_stopped()) and tell an 616 * dlm_recovery_stopped()) and tell any other threads running in the
640 * dlm to quit any processing (see RUN 617 * dlm to quit any processing (see RUNNING, dlm_locking_stopped()).
641 */ 618 */
642 619
643 spin_lock_bh(&ls->ls_recover_lock); !! 620 spin_lock(&ls->ls_recover_lock);
644 set_bit(LSFL_RECOVER_STOP, &ls->ls_fla 621 set_bit(LSFL_RECOVER_STOP, &ls->ls_flags);
645 new = test_and_clear_bit(LSFL_RUNNING, 622 new = test_and_clear_bit(LSFL_RUNNING, &ls->ls_flags);
646 if (new) <<
647 timer_delete_sync(&ls->ls_scan <<
648 ls->ls_recover_seq++; 623 ls->ls_recover_seq++;
649 !! 624 spin_unlock(&ls->ls_recover_lock);
650 /* activate requestqueue and stop proc <<
651 write_lock_bh(&ls->ls_requestqueue_loc <<
652 set_bit(LSFL_RECV_MSG_BLOCKED, &ls->ls <<
653 write_unlock_bh(&ls->ls_requestqueue_l <<
654 spin_unlock_bh(&ls->ls_recover_lock); <<
655 625
656 /* 626 /*
657 * Let dlm_recv run again, now any nor 627 * Let dlm_recv run again, now any normal messages will be saved on the
658 * requestqueue for later. 628 * requestqueue for later.
659 */ 629 */
660 630
661 write_unlock_bh(&ls->ls_recv_active); !! 631 up_write(&ls->ls_recv_active);
662 632
663 /* 633 /*
664 * This in_recovery lock does two thin 634 * This in_recovery lock does two things:
665 * 1) Keeps this function from returni 635 * 1) Keeps this function from returning until all threads are out
666 * of locking routines and locking 636 * of locking routines and locking is truly stopped.
667 * 2) Keeps any new requests from bein 637 * 2) Keeps any new requests from being processed until it's unlocked
668 * when recovery is complete. 638 * when recovery is complete.
669 */ 639 */
670 640
671 if (new) { 641 if (new) {
672 set_bit(LSFL_RECOVER_DOWN, &ls 642 set_bit(LSFL_RECOVER_DOWN, &ls->ls_flags);
673 wake_up_process(ls->ls_recover 643 wake_up_process(ls->ls_recoverd_task);
674 wait_event(ls->ls_recover_lock 644 wait_event(ls->ls_recover_lock_wait,
675 test_bit(LSFL_RECOV 645 test_bit(LSFL_RECOVER_LOCK, &ls->ls_flags));
676 } 646 }
677 647
678 /* 648 /*
679 * The recoverd suspend/resume makes s 649 * The recoverd suspend/resume makes sure that dlm_recoverd (if
680 * running) has noticed RECOVER_STOP a 650 * running) has noticed RECOVER_STOP above and quit processing the
681 * previous recovery. 651 * previous recovery.
682 */ 652 */
683 653
684 dlm_recoverd_suspend(ls); 654 dlm_recoverd_suspend(ls);
685 655
686 spin_lock_bh(&ls->ls_recover_lock); !! 656 spin_lock(&ls->ls_recover_lock);
687 kfree(ls->ls_slots); 657 kfree(ls->ls_slots);
688 ls->ls_slots = NULL; 658 ls->ls_slots = NULL;
689 ls->ls_num_slots = 0; 659 ls->ls_num_slots = 0;
690 ls->ls_slots_size = 0; 660 ls->ls_slots_size = 0;
691 ls->ls_recover_status = 0; 661 ls->ls_recover_status = 0;
692 spin_unlock_bh(&ls->ls_recover_lock); !! 662 spin_unlock(&ls->ls_recover_lock);
693 663
694 dlm_recoverd_resume(ls); 664 dlm_recoverd_resume(ls);
695 665
696 if (!ls->ls_recover_begin) 666 if (!ls->ls_recover_begin)
697 ls->ls_recover_begin = jiffies 667 ls->ls_recover_begin = jiffies;
698 668
699 /* call recover_prep ops only once and 669 /* call recover_prep ops only once and not multiple times
700 * for each possible dlm_ls_stop() whe 670 * for each possible dlm_ls_stop() when recovery is already
701 * stopped. 671 * stopped.
702 * 672 *
703 * If we successful was able to clear 673 * If we successful was able to clear LSFL_RUNNING bit and
704 * it was set we know it is the first 674 * it was set we know it is the first dlm_ls_stop() call.
705 */ 675 */
706 if (new) 676 if (new)
707 dlm_lsop_recover_prep(ls); 677 dlm_lsop_recover_prep(ls);
708 678
709 return 0; 679 return 0;
710 } 680 }
711 681
712 int dlm_ls_start(struct dlm_ls *ls) 682 int dlm_ls_start(struct dlm_ls *ls)
713 { 683 {
714 struct dlm_recover *rv, *rv_old; 684 struct dlm_recover *rv, *rv_old;
715 struct dlm_config_node *nodes = NULL; 685 struct dlm_config_node *nodes = NULL;
716 int error, count; 686 int error, count;
717 687
718 rv = kzalloc(sizeof(*rv), GFP_NOFS); 688 rv = kzalloc(sizeof(*rv), GFP_NOFS);
719 if (!rv) 689 if (!rv)
720 return -ENOMEM; 690 return -ENOMEM;
721 691
722 error = dlm_config_nodes(ls->ls_name, 692 error = dlm_config_nodes(ls->ls_name, &nodes, &count);
723 if (error < 0) 693 if (error < 0)
724 goto fail_rv; 694 goto fail_rv;
725 695
726 spin_lock_bh(&ls->ls_recover_lock); !! 696 spin_lock(&ls->ls_recover_lock);
727 697
728 /* the lockspace needs to be stopped b 698 /* the lockspace needs to be stopped before it can be started */
729 699
730 if (!dlm_locking_stopped(ls)) { 700 if (!dlm_locking_stopped(ls)) {
731 spin_unlock_bh(&ls->ls_recover !! 701 spin_unlock(&ls->ls_recover_lock);
732 log_error(ls, "start ignored: 702 log_error(ls, "start ignored: lockspace running");
733 error = -EINVAL; 703 error = -EINVAL;
734 goto fail; 704 goto fail;
735 } 705 }
736 706
737 rv->nodes = nodes; 707 rv->nodes = nodes;
738 rv->nodes_count = count; 708 rv->nodes_count = count;
739 rv->seq = ++ls->ls_recover_seq; 709 rv->seq = ++ls->ls_recover_seq;
740 rv_old = ls->ls_recover_args; 710 rv_old = ls->ls_recover_args;
741 ls->ls_recover_args = rv; 711 ls->ls_recover_args = rv;
742 spin_unlock_bh(&ls->ls_recover_lock); !! 712 spin_unlock(&ls->ls_recover_lock);
743 713
744 if (rv_old) { 714 if (rv_old) {
745 log_error(ls, "unused recovery 715 log_error(ls, "unused recovery %llx %d",
746 (unsigned long long) 716 (unsigned long long)rv_old->seq, rv_old->nodes_count);
747 kfree(rv_old->nodes); 717 kfree(rv_old->nodes);
748 kfree(rv_old); 718 kfree(rv_old);
749 } 719 }
750 720
751 set_bit(LSFL_RECOVER_WORK, &ls->ls_fla 721 set_bit(LSFL_RECOVER_WORK, &ls->ls_flags);
752 wake_up_process(ls->ls_recoverd_task); 722 wake_up_process(ls->ls_recoverd_task);
753 return 0; 723 return 0;
754 724
755 fail: 725 fail:
756 kfree(nodes); 726 kfree(nodes);
757 fail_rv: 727 fail_rv:
758 kfree(rv); 728 kfree(rv);
759 return error; 729 return error;
760 } 730 }
761 731
762 732
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.