BUG #18732: Segfault in pgbench on max_connections starvation - Mailing list pgsql-bugs

The following bug has been logged on the website:

Bug reference:      18732
Logged by:          Mikhail Kot
Email address:      mikhail@neon.tech
PostgreSQL version: 16.6
Operating system:   Debian 12
Description:

When --client connections in pgbench exceed max_connections in postgres,
pgbench 16 sometimes exits with segfault when a (presumably) ssl
certificate
validation error occurs.

OpenSSL version: 3.0.15-1~deb12u1
pgbench version: 16.6 (f5cfc6fa898544050e821ac688adafece1ac3cff)
pgbench params: pgbench postgresql://REDACTED/neondb?sslmode=require -c 2000
-T 60 -P 1 -j 20 --protocol=prepared

#0  0x00007f097342d3f0 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3
#1  0x00007f097342da0a in OPENSSL_LH_retrieve () from
/lib/x86_64-linux-gnu/libcrypto.so.3
#2  0x00007f097346a283 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3
#3  0x00007f097340bced in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3
#4  0x00007f097340c122 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3
#5  0x00007f09733f60ba in EVP_MD_fetch () from
/lib/x86_64-linux-gnu/libcrypto.so.3
#6  0x00007f09733f67f0 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3
#7  0x00007f097342899f in HMAC_Init_ex () from
/lib/x86_64-linux-gnu/libcrypto.so.3
#8  0x00007f09737ba60a in pg_hmac_init (ctx=0x7f09484a7360,
key=0x7f09484a7880 "R1M6EFcoABKs", len=12)
    at /home/myrrc/neon/vendor/postgres-v16/src/common/hmac_openssl.c:174
#9  0x00007f09737b54ef in scram_SaltedPassword (password=0x7f09484a7880
"R1M6EFcoABKs", 
    hash_type=PG_SHA256, key_length=32, salt=0x7f09484a78c0
"\260\376E\302@\0341Z\025%'\244H", 
    saltlen=16, iterations=4096, 
    result=0x7f09484a4118
"\235\245\371\260\203\243矮\357\224\305F\204F\341K\212\025$#\030CL\"\325ɑ\247\021os\340=IH\t\177",
errstr=0x7f09719fead8)
    at /home/myrrc/neon/vendor/postgres-v16/src/common/scram-common.c:87
#10 0x00007f097379452c in calculate_client_proof (state=0x7f09484a40f0, 
    client_final_message_without_proof=0x7f0948489920

"c=cD10bHMtc2VydmVyLWVuZC1wb2ludCwstoyKkoGIYqGK5C4vgGtRjvNeDwvmGQlaYHBXl8ZybAA=,r=RbPpYlql+b/rBgDtitBWxtAdW9BcFuPI9WsP7VCILEORedB6",

    result=0x7f09719feaf0 "0", errstr=0x7f09719fead8)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-auth-scram.c:788
#11 0x00007f0973793e55 in build_client_final_message
(state=0x7f09484a40f0)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-auth-scram.c:565
#12 0x00007f0973793403 in scram_exchange (opaq=0x7f09484a40f0, 
    input=0x7f09484a60a0
"r=RbPpYlql+b/rBgDtitBWxtAdW9BcFuPI9WsP7VCILEORedB6", inputlen=84, 
    output=0x7f09719febe0, outputlen=0x7f09719febdc, done=0x7f09719febdb,
success=0x7f09719febda)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-auth-scram.c:255
#13 0x00007f09737b002e in pg_SASL_continue (conn=0x7f0948486540,
payloadlen=84, final=false)
#14 0x00007f09737af729 in pg_fe_sendauth (areq=11, payloadlen=84,
conn=0x7f0948486540)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-auth.c:1139
#15 0x00007f0973798c5d in PQconnectPoll (conn=0x7f0948486540)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-connect.c:3802
#16 0x00007f0973794c9c in connectDBComplete (conn=0x7f0948486540)
    at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-connect.c:2511
#17 0x00007f09737949b4 in PQconnectdbParams (keywords=0x7f09719ff890,
values=0x7f09719ff850, 
    expand_dbname=1) at
/home/myrrc/neon/vendor/postgres-v16/src/interfaces/libpq/fe-connect.c:685
#18 0x0000558da1510b5e in doConnect ()
    at /home/myrrc/neon/vendor/postgres-v16/src/bin/pgbench/pgbench.c:1560
#19 0x0000558da15113d0 in threadRun (arg=0x558db50ebce0)
    at /home/myrrc/neon/vendor/postgres-v16/src/bin/pgbench/pgbench.c:7384
#20 0x00007f09730a81c4 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#21 0x00007f097312885c in ?? () from /lib/x86_64-linux-gnu/libc.so.6

Steps to reproduce:
1. Launch a postgres server with max_connections=900
2. Launch pgbench a couple of times with -c 2000

I was also able to reproduce this error by running multiple pgbench
instances
with same launch parameters. This error doesn't reproduce on pgbench 17.2 or
15.10
I can provide the coredump upon request.