1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
|
/****************************************************************
* *
* Copyright (c) 2005-2017 Fidelity National Information *
* Services, Inc. and/or its subsidiaries. All rights reserved. *
* *
* This source code contains the intellectual property *
* of its copyright holder(s), and is made available *
* under a license. If you do not know the terms of *
* the license, please stop and do not read further. *
* *
****************************************************************/
/* Process deferred stale timer interrupts.
Flushing of queues is deferred if the process is in crit. When a transaction
is complete, we will come here to process the deferred interrupt.
*/
#include "mdef.h"
#include <sys/mman.h>
#include "gdsroot.h"
#include "gdskill.h"
#include "gtm_facility.h"
#include "fileinfo.h"
#include "gdsbt.h"
#include "gdsblk.h"
#include "gdsfhead.h"
#include "filestruct.h"
#include "gdsbgtr.h"
#include "tp_change_reg.h"
#include "dpgbldir.h"
#include "process_deferred_stale.h"
#include "jnl.h"
#include "wcs_wt.h"
#include "repl_msg.h"
#include "gtmsource.h" /* for jnlpool_addrs_ptr_t */
GBLREF boolean_t unhandled_stale_timer_pop;
GBLREF gd_region *gv_cur_region;
GBLREF int process_exiting;
GBLREF jnl_gbls_t jgbl;
GBLREF jnlpool_addrs_ptr_t jnlpool;
void process_deferred_stale(void)
{
gd_region *r_cur, *r_top, *save_gv_cur_region;
gd_addr *addr_ptr;
sgmnt_addrs *csa;
jnlpool_addrs_ptr_t save_jnlpool;
int4 status;
assert(unhandled_stale_timer_pop);
/* If we are already exiting, do not worry about stale timer pop handling. Doing so could cause other issues
* e.g. EXIT from WAIT_FOR_REPL_INST_UNFREEZE_NOCSA macro because replication instance was frozen and we
* invoked "wcs_wtstart" on a journaled region which had a timer pop and that would terminate the process
* right away without doing any cleanup (e.g. csa->wbuf_dqd handling in secshr_db_clnup resulting in a
* shared memory state with a dirty cache-record that is not in the active queue). It is okay to skip this stale
* timer pop handling since we are anyways about to exit and will do any needed flushing as part of rundown.
*/
if (process_exiting)
return;
save_gv_cur_region = gv_cur_region;
save_jnlpool = jnlpool;
for (addr_ptr = get_next_gdr(NULL); NULL != addr_ptr; addr_ptr = get_next_gdr(addr_ptr))
{
for (r_cur = addr_ptr->regions, r_top = r_cur + addr_ptr->n_regions; r_cur < r_top; r_cur++)
{
if (r_cur->open)
{
csa = &FILE_INFO(r_cur)->s_addrs;
/* We don't expect to be in crit or commit when process_deferred_stale is invoked. The only
* exceptions are :
* (a) ONLINE ROLLBACK - holds the crit for the entire duration. In this case, do the flush anyways.
* While this might slowdown the online rollback run, with our current knowledge, we don't have
* any benchmarks to decide whether it is a good thing to let the buffer flush at the end (in
* gds_rundown) take care of all the updates done by rollback or do it periodically.
* (b) A concurrent process in t_end detected an ONLINE ROLLBACK in the final or penultimate retry
* and invoked gvcst_redo_root_search to redo the root search of the concerned gv_target. This
* root search ended up calling t_end and noticed an unhandled timer pop and thereby coming
* here. Since root search is invoked in the 2nd or 3rd retry, the process sets hold_onto_crit.
* So, assert that is indeed the case. Also, have a much stricter assert to ensure that
* gv_target->root = DIR_ROOT (indicating we are in gvcst_root_search). Also, since we want to
* avoid flushing buffers if already holding crit, continue to the next region.
* (c) In case of DSE commands (like DSE MAPS -RESTORE), we could come here with csa->hold_onto_crit
* being TRUE (see t_begin_crit which DSE invokes).
*/
assert(!T_IN_CRIT_OR_COMMIT_OR_WRITE(csa) || csa->hold_onto_crit
UNIX_ONLY(|| jgbl.onlnrlbk || (NULL == gv_target) || (DIR_ROOT == gv_target->root)));
if (UNIX_ONLY(!jgbl.onlnrlbk && )csa->now_crit)
continue;
if (csa->stale_defer && !FROZEN_CHILLED(csa))
{
gv_cur_region = r_cur;
tp_change_reg();
status = wcs_wtstart(r_cur, 0, NULL, NULL);
csa->stale_defer = FALSE;
BG_TRACE_ANY(csa, stale_defer_processed);
}
}
}
}
unhandled_stale_timer_pop = FALSE;
gv_cur_region = save_gv_cur_region;
tp_change_reg();
if (save_jnlpool != jnlpool)
jnlpool = save_jnlpool;
}
|
