1 /******************************************************************************
2 *******************************************************************************
3 **
4 **  Copyright (C) 2005-2007 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 
20 struct rq_entry {
21 	struct list_head list;
22 	int nodeid;
23 	struct dlm_message request;
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 
dlm_add_requestqueue(struct dlm_ls * ls,int nodeid,struct dlm_message * ms)33 void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_message *ms)
34 {
35 	struct rq_entry *e;
36 	int length = ms->m_header.h_length - sizeof(struct dlm_message);
37 
38 	e = kmalloc(sizeof(struct rq_entry) + length, GFP_NOFS);
39 	if (!e) {
40 		log_print("dlm_add_requestqueue: out of memory len %d", length);
41 		return;
42 	}
43 
44 	e->nodeid = nodeid;
45 	memcpy(&e->request, ms, ms->m_header.h_length);
46 
47 	mutex_lock(&ls->ls_requestqueue_mutex);
48 	list_add_tail(&e->list, &ls->ls_requestqueue);
49 	mutex_unlock(&ls->ls_requestqueue_mutex);
50 }
51 
52 /*
53  * Called by dlm_recoverd to process normal messages saved while recovery was
54  * happening.  Normal locking has been enabled before this is called.  dlm_recv
55  * upon receiving a message, will wait for all saved messages to be drained
56  * here before processing the message it got.  If a new dlm_ls_stop() arrives
57  * while we're processing these saved messages, it may block trying to suspend
58  * dlm_recv if dlm_recv is waiting for us in dlm_wait_requestqueue.  In that
59  * case, we don't abort since locking_stopped is still 0.  If dlm_recv is not
60  * waiting for us, then this processing may be aborted due to locking_stopped.
61  */
62 
dlm_process_requestqueue(struct dlm_ls * ls)63 int dlm_process_requestqueue(struct dlm_ls *ls)
64 {
65 	struct rq_entry *e;
66 	int error = 0;
67 
68 	mutex_lock(&ls->ls_requestqueue_mutex);
69 
70 	for (;;) {
71 		if (list_empty(&ls->ls_requestqueue)) {
72 			mutex_unlock(&ls->ls_requestqueue_mutex);
73 			error = 0;
74 			break;
75 		}
76 		e = list_entry(ls->ls_requestqueue.next, struct rq_entry, list);
77 		mutex_unlock(&ls->ls_requestqueue_mutex);
78 
79 		dlm_receive_message_saved(ls, &e->request);
80 
81 		mutex_lock(&ls->ls_requestqueue_mutex);
82 		list_del(&e->list);
83 		kfree(e);
84 
85 		if (dlm_locking_stopped(ls)) {
86 			log_debug(ls, "process_requestqueue abort running");
87 			mutex_unlock(&ls->ls_requestqueue_mutex);
88 			error = -EINTR;
89 			break;
90 		}
91 		schedule();
92 	}
93 
94 	return error;
95 }
96 
97 /*
98  * After recovery is done, locking is resumed and dlm_recoverd takes all the
99  * saved requests and processes them as they would have been by dlm_recv.  At
100  * the same time, dlm_recv will start receiving new requests from remote nodes.
101  * We want to delay dlm_recv processing new requests until dlm_recoverd has
102  * finished processing the old saved requests.  We don't check for locking
103  * stopped here because dlm_ls_stop won't stop locking until it's suspended us
104  * (dlm_recv).
105  */
106 
dlm_wait_requestqueue(struct dlm_ls * ls)107 void dlm_wait_requestqueue(struct dlm_ls *ls)
108 {
109 	for (;;) {
110 		mutex_lock(&ls->ls_requestqueue_mutex);
111 		if (list_empty(&ls->ls_requestqueue))
112 			break;
113 		mutex_unlock(&ls->ls_requestqueue_mutex);
114 		schedule();
115 	}
116 	mutex_unlock(&ls->ls_requestqueue_mutex);
117 }
118 
purge_request(struct dlm_ls * ls,struct dlm_message * ms,int nodeid)119 static int purge_request(struct dlm_ls *ls, struct dlm_message *ms, int nodeid)
120 {
121 	uint32_t type = ms->m_type;
122 
123 	/* the ls is being cleaned up and freed by release_lockspace */
124 	if (!ls->ls_count)
125 		return 1;
126 
127 	if (dlm_is_removed(ls, nodeid))
128 		return 1;
129 
130 	/* directory operations are always purged because the directory is
131 	   always rebuilt during recovery and the lookups resent */
132 
133 	if (type == DLM_MSG_REMOVE ||
134 	    type == DLM_MSG_LOOKUP ||
135 	    type == DLM_MSG_LOOKUP_REPLY)
136 		return 1;
137 
138 	if (!dlm_no_directory(ls))
139 		return 0;
140 
141 	/* with no directory, the master is likely to change as a part of
142 	   recovery; requests to/from the defunct master need to be purged */
143 
144 	switch (type) {
145 	case DLM_MSG_REQUEST:
146 	case DLM_MSG_CONVERT:
147 	case DLM_MSG_UNLOCK:
148 	case DLM_MSG_CANCEL:
149 		/* we're no longer the master of this resource, the sender
150 		   will resend to the new master (see waiter_needs_recovery) */
151 
152 		if (dlm_hash2nodeid(ls, ms->m_hash) != dlm_our_nodeid())
153 			return 1;
154 		break;
155 
156 	case DLM_MSG_REQUEST_REPLY:
157 	case DLM_MSG_CONVERT_REPLY:
158 	case DLM_MSG_UNLOCK_REPLY:
159 	case DLM_MSG_CANCEL_REPLY:
160 	case DLM_MSG_GRANT:
161 		/* this reply is from the former master of the resource,
162 		   we'll resend to the new master if needed */
163 
164 		if (dlm_hash2nodeid(ls, ms->m_hash) != nodeid)
165 			return 1;
166 		break;
167 	}
168 
169 	return 0;
170 }
171 
dlm_purge_requestqueue(struct dlm_ls * ls)172 void dlm_purge_requestqueue(struct dlm_ls *ls)
173 {
174 	struct dlm_message *ms;
175 	struct rq_entry *e, *safe;
176 
177 	mutex_lock(&ls->ls_requestqueue_mutex);
178 	list_for_each_entry_safe(e, safe, &ls->ls_requestqueue, list) {
179 		ms =  &e->request;
180 
181 		if (purge_request(ls, ms, e->nodeid)) {
182 			list_del(&e->list);
183 			kfree(e);
184 		}
185 	}
186 	mutex_unlock(&ls->ls_requestqueue_mutex);
187 }
188 
189