projects/linux·blob v2.6.21 · refs

198 lines · 5.1 KiB

1/******************************************************************************
2*******************************************************************************
3**
4** Copyright (C) 2005 Red Hat, Inc. All rights reserved.
5**
6** This copyrighted material is made available to anyone wishing to use,
7** modify, copy, or redistribute it subject to the terms and conditions
8** of the GNU General Public License v.2.
9**
10*******************************************************************************
11******************************************************************************/
12
13#include "dlm_internal.h"
14#include "member.h"
15#include "lock.h"
16#include "dir.h"
17#include "config.h"
18#include "requestqueue.h"
19
20struct rq_entry {
21 struct list_head list;
22 int nodeid;
23 char request[1];
24};
25
26/*
27 * Requests received while the lockspace is in recovery get added to the
28 * request queue and processed when recovery is complete. This happens when
29 * the lockspace is suspended on some nodes before it is on others, or the
30 * lockspace is enabled on some while still suspended on others.
31 */
32
33int dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd)
34{
35 struct rq_entry *e;
36 int length = hd->h_length;
37 int rv = 0;
38
39 e = kmalloc(sizeof(struct rq_entry) + length, GFP_KERNEL);
40 if (!e) {
41 log_print("dlm_add_requestqueue: out of memory\n");
42 return 0;
43 }
44
45 e->nodeid = nodeid;
46 memcpy(e->request, hd, length);
47
48 /* We need to check dlm_locking_stopped() after taking the mutex to
49 avoid a race where dlm_recoverd enables locking and runs
50 process_requestqueue between our earlier dlm_locking_stopped check
51 and this addition to the requestqueue. */
52
53 mutex_lock(&ls->ls_requestqueue_mutex);
54 if (dlm_locking_stopped(ls))
55 list_add_tail(&e->list, &ls->ls_requestqueue);
56 else {
57 log_debug(ls, "dlm_add_requestqueue skip from %d", nodeid);
58 kfree(e);
59 rv = -EAGAIN;
60 }
61 mutex_unlock(&ls->ls_requestqueue_mutex);
62 return rv;
63}
64
65int dlm_process_requestqueue(struct dlm_ls *ls)
66{
67 struct rq_entry *e;
68 struct dlm_header *hd;
69 int error = 0;
70
71 mutex_lock(&ls->ls_requestqueue_mutex);
72
73 for (;;) {
74 if (list_empty(&ls->ls_requestqueue)) {
75 mutex_unlock(&ls->ls_requestqueue_mutex);
76 error = 0;
77 break;
78 }
79 e = list_entry(ls->ls_requestqueue.next, struct rq_entry, list);
80 mutex_unlock(&ls->ls_requestqueue_mutex);
81
82 hd = (struct dlm_header *) e->request;
83 error = dlm_receive_message(hd, e->nodeid, 1);
84
85 if (error == -EINTR) {
86 /* entry is left on requestqueue */
87 log_debug(ls, "process_requestqueue abort eintr");
88 break;
89 }
90
91 mutex_lock(&ls->ls_requestqueue_mutex);
92 list_del(&e->list);
93 kfree(e);
94
95 if (dlm_locking_stopped(ls)) {
96 log_debug(ls, "process_requestqueue abort running");
97 mutex_unlock(&ls->ls_requestqueue_mutex);
98 error = -EINTR;
99 break;
100 }
101 schedule();
102 }
103
104 return error;
105}
106
107/*
108 * After recovery is done, locking is resumed and dlm_recoverd takes all the
109 * saved requests and processes them as they would have been by dlm_recvd. At
110 * the same time, dlm_recvd will start receiving new requests from remote
111 * nodes. We want to delay dlm_recvd processing new requests until
112 * dlm_recoverd has finished processing the old saved requests.
113 */
114
115void dlm_wait_requestqueue(struct dlm_ls *ls)
116{
117 for (;;) {
118 mutex_lock(&ls->ls_requestqueue_mutex);
119 if (list_empty(&ls->ls_requestqueue))
120 break;
121 if (dlm_locking_stopped(ls))
122 break;
123 mutex_unlock(&ls->ls_requestqueue_mutex);
124 schedule();
125 }
126 mutex_unlock(&ls->ls_requestqueue_mutex);
127}
128
129static int purge_request(struct dlm_ls *ls, struct dlm_message *ms, int nodeid)
130{
131 uint32_t type = ms->m_type;
132
133 /* the ls is being cleaned up and freed by release_lockspace */
134 if (!ls->ls_count)
135 return 1;
136
137 if (dlm_is_removed(ls, nodeid))
138 return 1;
139
140 /* directory operations are always purged because the directory is
141 always rebuilt during recovery and the lookups resent */
142
143 if (type == DLM_MSG_REMOVE ||
144 type == DLM_MSG_LOOKUP ||
145 type == DLM_MSG_LOOKUP_REPLY)
146 return 1;
147
148 if (!dlm_no_directory(ls))
149 return 0;
150
151 /* with no directory, the master is likely to change as a part of
152 recovery; requests to/from the defunct master need to be purged */
153
154 switch (type) {
155 case DLM_MSG_REQUEST:
156 case DLM_MSG_CONVERT:
157 case DLM_MSG_UNLOCK:
158 case DLM_MSG_CANCEL:
159 /* we're no longer the master of this resource, the sender
160 will resend to the new master (see waiter_needs_recovery) */
161
162 if (dlm_hash2nodeid(ls, ms->m_hash) != dlm_our_nodeid())
163 return 1;
164 break;
165
166 case DLM_MSG_REQUEST_REPLY:
167 case DLM_MSG_CONVERT_REPLY:
168 case DLM_MSG_UNLOCK_REPLY:
169 case DLM_MSG_CANCEL_REPLY:
170 case DLM_MSG_GRANT:
171 /* this reply is from the former master of the resource,
172 we'll resend to the new master if needed */
173
174 if (dlm_hash2nodeid(ls, ms->m_hash) != nodeid)
175 return 1;
176 break;
177 }
178
179 return 0;
180}
181
182void dlm_purge_requestqueue(struct dlm_ls *ls)
183{
184 struct dlm_message *ms;
185 struct rq_entry *e, *safe;
186
187 mutex_lock(&ls->ls_requestqueue_mutex);
188 list_for_each_entry_safe(e, safe, &ls->ls_requestqueue, list) {
189 ms = (struct dlm_message *) e->request;
190
191 if (purge_request(ls, ms, e->nodeid)) {
192 list_del(&e->list);
193 kfree(e);
194 }
195 }
196 mutex_unlock(&ls->ls_requestqueue_mutex);
197}
198