Age Owner Branch data TLA Line data Source code
1 : : /*--------------------------------------------------------------------------
2 : : *
3 : : * setup.c
4 : : * Code to set up a dynamic shared memory segments and a specified
5 : : * number of background workers for shared memory message queue
6 : : * testing.
7 : : *
8 : : * Copyright (c) 2013-2025, PostgreSQL Global Development Group
9 : : *
10 : : * IDENTIFICATION
11 : : * src/test/modules/test_shm_mq/setup.c
12 : : *
13 : : * -------------------------------------------------------------------------
14 : : */
15 : :
16 : : #include "postgres.h"
17 : :
18 : : #include "miscadmin.h"
19 : : #include "pgstat.h"
20 : : #include "postmaster/bgworker.h"
21 : : #include "storage/shm_toc.h"
22 : : #include "test_shm_mq.h"
23 : : #include "utils/memutils.h"
24 : :
25 : : typedef struct
26 : : {
27 : : int nworkers;
28 : : BackgroundWorkerHandle *handle[FLEXIBLE_ARRAY_MEMBER];
29 : : } worker_state;
30 : :
31 : : static void setup_dynamic_shared_memory(int64 queue_size, int nworkers,
32 : : dsm_segment **segp,
33 : : test_shm_mq_header **hdrp,
34 : : shm_mq **outp, shm_mq **inp);
35 : : static worker_state *setup_background_workers(int nworkers,
36 : : dsm_segment *seg);
37 : : static void cleanup_background_workers(dsm_segment *seg, Datum arg);
38 : : static void wait_for_workers_to_become_ready(worker_state *wstate,
39 : : volatile test_shm_mq_header *hdr);
40 : : static bool check_worker_status(worker_state *wstate);
41 : :
42 : : /* value cached, fetched from shared memory */
43 : : static uint32 we_bgworker_startup = 0;
44 : :
45 : : /*
46 : : * Set up a dynamic shared memory segment and zero or more background workers
47 : : * for a test run.
48 : : */
49 : : void
4304 rhaas@postgresql.org 50 :CBC 5 : test_shm_mq_setup(int64 queue_size, int32 nworkers, dsm_segment **segp,
51 : : shm_mq_handle **output, shm_mq_handle **input)
52 : : {
53 : : dsm_segment *seg;
54 : : test_shm_mq_header *hdr;
4193 bruce@momjian.us 55 : 5 : shm_mq *outq = NULL; /* placate compiler */
56 : 5 : shm_mq *inq = NULL; /* placate compiler */
57 : : worker_state *wstate;
58 : :
59 : : /* Set up a dynamic shared memory segment. */
4305 rhaas@postgresql.org 60 : 5 : setup_dynamic_shared_memory(queue_size, nworkers, &seg, &hdr, &outq, &inq);
61 : 5 : *segp = seg;
62 : :
63 : : /* Register background workers. */
64 : 5 : wstate = setup_background_workers(nworkers, seg);
65 : :
66 : : /* Attach the queues. */
67 : 5 : *output = shm_mq_attach(outq, seg, wstate->handle[0]);
68 : 5 : *input = shm_mq_attach(inq, seg, wstate->handle[nworkers - 1]);
69 : :
70 : : /* Wait for workers to become ready. */
71 : 5 : wait_for_workers_to_become_ready(wstate, hdr);
72 : :
73 : : /*
74 : : * Once we reach this point, all workers are ready. We no longer need to
75 : : * kill them if we die; they'll die on their own as the message queues
76 : : * shut down.
77 : : */
78 : 5 : cancel_on_dsm_detach(seg, cleanup_background_workers,
79 : : PointerGetDatum(wstate));
80 : 5 : pfree(wstate);
81 : 5 : }
82 : :
83 : : /*
84 : : * Set up a dynamic shared memory segment.
85 : : *
86 : : * We set up a small control region that contains only a test_shm_mq_header,
87 : : * plus one region per message queue. There are as many message queues as
88 : : * the number of workers, plus one.
89 : : */
90 : : static void
4304 91 : 5 : setup_dynamic_shared_memory(int64 queue_size, int nworkers,
92 : : dsm_segment **segp, test_shm_mq_header **hdrp,
93 : : shm_mq **outp, shm_mq **inp)
94 : : {
95 : : shm_toc_estimator e;
96 : : int i;
97 : : Size segsize;
98 : : dsm_segment *seg;
99 : : shm_toc *toc;
100 : : test_shm_mq_header *hdr;
101 : :
102 : : /* Ensure a valid queue size. */
4305 103 [ + - - + ]: 5 : if (queue_size < 0 || ((uint64) queue_size) < shm_mq_minimum_size)
4305 rhaas@postgresql.org 104 [ # # ]:UBC 0 : ereport(ERROR,
105 : : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
106 : : errmsg("queue size must be at least %zu bytes",
107 : : shm_mq_minimum_size)));
108 : : if (queue_size != ((Size) queue_size))
109 : : ereport(ERROR,
110 : : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
111 : : errmsg("queue size overflows size_t")));
112 : :
113 : : /*
114 : : * Estimate how much shared memory we need.
115 : : *
116 : : * Because the TOC machinery may choose to insert padding of oddly-sized
117 : : * requests, we must estimate each chunk separately.
118 : : *
119 : : * We need one key to register the location of the header, and we need
120 : : * nworkers + 1 keys to track the locations of the message queues.
121 : : */
4305 rhaas@postgresql.org 122 :CBC 5 : shm_toc_initialize_estimator(&e);
123 : 5 : shm_toc_estimate_chunk(&e, sizeof(test_shm_mq_header));
124 [ + + ]: 17 : for (i = 0; i <= nworkers; ++i)
4242 tgl@sss.pgh.pa.us 125 : 12 : shm_toc_estimate_chunk(&e, (Size) queue_size);
4305 rhaas@postgresql.org 126 : 5 : shm_toc_estimate_keys(&e, 2 + nworkers);
127 : 5 : segsize = shm_toc_estimate(&e);
128 : :
129 : : /* Create the shared memory segment and establish a table of contents. */
3876 130 : 5 : seg = dsm_create(shm_toc_estimate(&e), 0);
4305 131 : 5 : toc = shm_toc_create(PG_TEST_SHM_MQ_MAGIC, dsm_segment_address(seg),
132 : : segsize);
133 : :
134 : : /* Set up the header region. */
135 : 5 : hdr = shm_toc_allocate(toc, sizeof(test_shm_mq_header));
136 : 5 : SpinLockInit(&hdr->mutex);
137 : 5 : hdr->workers_total = nworkers;
138 : 5 : hdr->workers_attached = 0;
139 : 5 : hdr->workers_ready = 0;
140 : 5 : shm_toc_insert(toc, 0, hdr);
141 : :
142 : : /* Set up one message queue per worker, plus one. */
143 [ + + ]: 17 : for (i = 0; i <= nworkers; ++i)
144 : : {
145 : : shm_mq *mq;
146 : :
4242 tgl@sss.pgh.pa.us 147 : 12 : mq = shm_mq_create(shm_toc_allocate(toc, (Size) queue_size),
148 : : (Size) queue_size);
4305 rhaas@postgresql.org 149 : 12 : shm_toc_insert(toc, i + 1, mq);
150 : :
151 [ + + ]: 12 : if (i == 0)
152 : : {
153 : : /* We send messages to the first queue. */
154 : 5 : shm_mq_set_sender(mq, MyProc);
155 : 5 : *outp = mq;
156 : : }
157 [ + + ]: 12 : if (i == nworkers)
158 : : {
159 : : /* We receive messages from the last queue. */
160 : 5 : shm_mq_set_receiver(mq, MyProc);
161 : 5 : *inp = mq;
162 : : }
163 : : }
164 : :
165 : : /* Return results to caller. */
166 : 5 : *segp = seg;
167 : 5 : *hdrp = hdr;
168 : 5 : }
169 : :
170 : : /*
171 : : * Register background workers.
172 : : */
173 : : static worker_state *
174 : 5 : setup_background_workers(int nworkers, dsm_segment *seg)
175 : : {
176 : : MemoryContext oldcontext;
177 : : BackgroundWorker worker;
178 : : worker_state *wstate;
179 : : int i;
180 : :
181 : : /*
182 : : * We need the worker_state object and the background worker handles to
183 : : * which it points to be allocated in CurTransactionContext rather than
184 : : * ExprContext; otherwise, they'll be destroyed before the on_dsm_detach
185 : : * hooks run.
186 : : */
187 : 5 : oldcontext = MemoryContextSwitchTo(CurTransactionContext);
188 : :
189 : : /* Create worker state object. */
190 : 5 : wstate = MemoryContextAlloc(TopTransactionContext,
191 : 5 : offsetof(worker_state, handle) +
192 : : sizeof(BackgroundWorkerHandle *) * nworkers);
193 : 5 : wstate->nworkers = 0;
194 : :
195 : : /*
196 : : * Arrange to kill all the workers if we abort before all workers are
197 : : * finished hooking themselves up to the dynamic shared memory segment.
198 : : *
199 : : * If we die after all the workers have finished hooking themselves up to
200 : : * the dynamic shared memory segment, we'll mark the two queues to which
201 : : * we're directly connected as detached, and the worker(s) connected to
202 : : * those queues will exit, marking any other queues to which they are
203 : : * connected as detached. This will cause any as-yet-unaware workers
204 : : * connected to those queues to exit in their turn, and so on, until
205 : : * everybody exits.
206 : : *
207 : : * But suppose the workers which are supposed to connect to the queues to
208 : : * which we're directly attached exit due to some error before they
209 : : * actually attach the queues. The remaining workers will have no way of
210 : : * knowing this. From their perspective, they're still waiting for those
211 : : * workers to start, when in fact they've already died.
212 : : */
213 : 5 : on_dsm_detach(seg, cleanup_background_workers,
214 : : PointerGetDatum(wstate));
215 : :
216 : : /* Configure a worker. */
3117 tgl@sss.pgh.pa.us 217 : 5 : memset(&worker, 0, sizeof(worker));
4305 rhaas@postgresql.org 218 : 5 : worker.bgw_flags = BGWORKER_SHMEM_ACCESS;
219 : 5 : worker.bgw_start_time = BgWorkerStart_ConsistentState;
220 : 5 : worker.bgw_restart_time = BGW_NEVER_RESTART;
221 : 5 : sprintf(worker.bgw_library_name, "test_shm_mq");
222 : 5 : sprintf(worker.bgw_function_name, "test_shm_mq_main");
2980 peter_e@gmx.net 223 : 5 : snprintf(worker.bgw_type, BGW_MAXLEN, "test_shm_mq");
4305 rhaas@postgresql.org 224 : 5 : worker.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
225 : : /* set bgw_notify_pid, so we can detect if the worker stops */
226 : 5 : worker.bgw_notify_pid = MyProcPid;
227 : :
228 : : /* Register the workers. */
229 [ + + ]: 12 : for (i = 0; i < nworkers; ++i)
230 : : {
231 [ - + ]: 7 : if (!RegisterDynamicBackgroundWorker(&worker, &wstate->handle[i]))
4305 rhaas@postgresql.org 232 [ # # ]:UBC 0 : ereport(ERROR,
233 : : (errcode(ERRCODE_INSUFFICIENT_RESOURCES),
234 : : errmsg("could not register background process"),
235 : : errhint("You may need to increase \"max_worker_processes\".")));
4305 rhaas@postgresql.org 236 :CBC 7 : ++wstate->nworkers;
237 : : }
238 : :
239 : : /* All done. */
240 : 5 : MemoryContextSwitchTo(oldcontext);
241 : 5 : return wstate;
242 : : }
243 : :
244 : : static void
4305 rhaas@postgresql.org 245 :UBC 0 : cleanup_background_workers(dsm_segment *seg, Datum arg)
246 : : {
247 : 0 : worker_state *wstate = (worker_state *) DatumGetPointer(arg);
248 : :
249 [ # # ]: 0 : while (wstate->nworkers > 0)
250 : : {
251 : 0 : --wstate->nworkers;
252 : 0 : TerminateBackgroundWorker(wstate->handle[wstate->nworkers]);
253 : : }
254 : 0 : }
255 : :
256 : : static void
4305 rhaas@postgresql.org 257 :CBC 5 : wait_for_workers_to_become_ready(worker_state *wstate,
258 : : volatile test_shm_mq_header *hdr)
259 : : {
4193 bruce@momjian.us 260 : 5 : bool result = false;
261 : :
262 : : for (;;)
4305 rhaas@postgresql.org 263 : 14 : {
264 : : int workers_ready;
265 : :
266 : : /* If all the workers are ready, we have succeeded. */
3672 267 [ - + ]: 19 : SpinLockAcquire(&hdr->mutex);
268 : 19 : workers_ready = hdr->workers_ready;
269 : 19 : SpinLockRelease(&hdr->mutex);
270 [ + + ]: 19 : if (workers_ready >= wstate->nworkers)
271 : : {
272 : 5 : result = true;
273 : 5 : break;
274 : : }
275 : :
276 : : /* If any workers (or the postmaster) have died, we have failed. */
277 [ - + ]: 14 : if (!check_worker_status(wstate))
278 : : {
3672 rhaas@postgresql.org 279 :UBC 0 : result = false;
280 : 0 : break;
281 : : }
282 : :
283 : : /* first time, allocate or get the custom wait event */
755 michael@paquier.xyz 284 [ + + ]:CBC 14 : if (we_bgworker_startup == 0)
285 : 1 : we_bgworker_startup = WaitEventExtensionNew("TestShmMqBgWorkerStartup");
286 : :
287 : : /* Wait to be signaled. */
2531 tmunro@postgresql.or 288 : 14 : (void) WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0,
289 : : we_bgworker_startup);
290 : :
291 : : /* Reset the latch so we don't spin. */
3672 rhaas@postgresql.org 292 : 14 : ResetLatch(MyLatch);
293 : :
294 : : /* An interrupt may have occurred while we were waiting. */
3375 tgl@sss.pgh.pa.us 295 [ - + ]: 14 : CHECK_FOR_INTERRUPTS();
296 : : }
297 : :
4305 rhaas@postgresql.org 298 [ - + ]: 5 : if (!result)
4305 rhaas@postgresql.org 299 [ # # ]:UBC 0 : ereport(ERROR,
300 : : (errcode(ERRCODE_INSUFFICIENT_RESOURCES),
301 : : errmsg("one or more background workers failed to start")));
4305 rhaas@postgresql.org 302 :CBC 5 : }
303 : :
304 : : static bool
305 : 14 : check_worker_status(worker_state *wstate)
306 : : {
307 : : int n;
308 : :
309 : : /* If any workers (or the postmaster) have died, we have failed. */
310 [ + + ]: 40 : for (n = 0; n < wstate->nworkers; ++n)
311 : : {
312 : : BgwHandleStatus status;
313 : : pid_t pid;
314 : :
315 : 26 : status = GetBackgroundWorkerPid(wstate->handle[n], &pid);
316 [ + - - + ]: 26 : if (status == BGWH_STOPPED || status == BGWH_POSTMASTER_DIED)
4305 rhaas@postgresql.org 317 :UBC 0 : return false;
318 : : }
319 : :
320 : : /* Otherwise, things still look OK. */
4305 rhaas@postgresql.org 321 :CBC 14 : return true;
322 : : }
|
