Segfault in websocket test

[gdamore]

websocket test failure:

35/82 Test #35: nng.supplemental.websocket_test .............***Failed 0.79 sec
Test websocket stream wildcard:
websocket_test.c:36: Check nng_stream_listener_alloc(&l, "ws://127.0.0.1:0/test") succeeds... ok
websocket_test.c:38: Check nng_stream_listener_listen(l) succeeds... ok
websocket_test.c:43: Check nng_stream_listener_get(l, NNG_OPT_LOCADDR, &sa1, &sz) succeeds... ok
websocket_test.c:44: Check sz == sizeof(sa1)... ok
websocket_test.c:45: Check sa1.s_in.sa_port != 0... ok
websocket_test.c:46: Check sa1.s_family == NNG_AF_INET... ok
websocket_test.c:47: Check nuts_be32(sa1.s_in.sa_addr) == 0x7F000001u... ok
websocket_test.c:52: Check nng_stream_dialer_alloc(&d, uri) succeeds... ok
websocket_test.c:53: Check nng_aio_alloc(&daio, NULL, NULL) succeeds... ok
websocket_test.c:54: Check nng_aio_alloc(&laio, NULL, NULL) succeeds... ok
websocket_test.c:55: Check nng_aio_alloc(&aio1, NULL, NULL) succeeds... ok
websocket_test.c:56: Check nng_aio_alloc(&aio2, NULL, NULL) succeeds... ok
websocket_test.c:68: Check nng_aio_result(laio) succeeds... ok
websocket_test.c:69: Check nng_aio_result(daio) succeeds... ok
websocket_test.c:72: Check c1 != NULL... ok
websocket_test.c:73: Check c2 != NULL... ok
websocket_test.c:76: Check nng_stream_get_addr(c2, NNG_OPT_REMADDR, &sa2) succeeds... ok
websocket_test.c:77: Check sa1.s_family == sa2.s_family... ok
websocket_test.c:78: Check sa1.s_in.sa_addr == sa2.s_in.sa_addr... ok
websocket_test.c:79: Check sa1.s_in.sa_port == sa2.s_in.sa_port... ok
websocket_test.c:81: Check nng_stream_get_addr(c1, NNG_OPT_REMADDR, &sa1) succeeds... ok
websocket_test.c:82: Check nng_stream_get_addr(c2, NNG_OPT_LOCADDR, &sa2) succeeds... ok
websocket_test.c:83: Check sa1.s_family == sa2.s_family... ok
websocket_test.c:84: Check sa1.s_in.sa_addr == sa2.s_in.sa_addr... ok
websocket_test.c:85: Check sa1.s_in.sa_port == sa2.s_in.sa_port... ok
websocket_test.c:95: Check nng_aio_set_iov(aio1, 1, &iov) succeeds... ok
websocket_test.c:99: Check nng_aio_set_iov(aio2, 1, &iov) succeeds... ok
websocket_test.c:106: Check nng_aio_result(aio1) succeeds... ok
websocket_test.c:107: Check nng_aio_count(aio1) == 5... ok
websocket_test.c:109: Check nng_aio_result(aio2) succeeds... ok
websocket_test.c:110: Check nng_aio_count(aio2) == 5... ok
websocket_test.c:111: Check memcmp(buf1, buf2, 5) == 0... ok
SUCCESS: All conditions have passed.
Duration: 0.049140 secs

Test websocket conn properties:
ThreadSanitizer:DEADLYSIGNAL
==18807==ERROR: ThreadSanitizer: SEGV on unknown address 0x000000000020 (pc 0x0000004e0a9e bp 0x7ffd438aea20 sp 0x7ffd438ae9b0 T18807)
==18807==The signal is caused by a READ memory access.
==18807==Hint: address points to the zero page.
#0 nni_stream_get /home/runner/work/nng/nng/build/../src/core/stream.c:132:13 (websocket_test+0x4e0a9d)
#1 nng_stream_get_addr /home/runner/work/nng/nng/build/../src/core/stream.c:336:10 (websocket_test+0x4e1923)
#2 test_websocket_conn_props /home/runner/work/nng/nng/build/../src/supplemental/websocket/websocket_test.c:167:2 (websocket_test+0x4b9875)
#3 test_do_run_ /home/runner/work/nng/nng/build/../src/testing/acutest.h:1026:9 (websocket_test+0x4be396)
#4 test_run_ /home/runner/work/nng/nng/build/../src/testing/acutest.h:1122:23 (websocket_test+0x4b7bf6)
#5 main /home/runner/work/nng/nng/build/../src/testing/acutest.h:1724:13 (websocket_test+0x4b5284)
#6 __libc_start_main (libc.so.6+0x21bf6)
#7 _start (websocket_test+0x41d519)

ThreadSanitizer can not provide additional info.
SUMMARY: ThreadSanitizer: SEGV /home/runner/work/nng/nng/build/../src/core/stream.c:132:13 in nni_stream_get
==18807==ABORTING
ERROR: Unexpected exit code [66]

[gdamore]

This specific panic seems hard to understand, because of the test at websocket_test.c:164 -- the stream pointer should not be NULL, but it looks for all the world like it is NULL in this test failure.

[gdamore]

I suspect other corruption elsewhere. I could understand if the function pointer was NULL, but not the handle itself.

[gdamore]

I haven't seen this one again. I'm not going to hold 1.4 for it.

[gdamore]

Have not seen this one since this instance. I'm tempted to close it as not-reproducible.