Issue #21771 has been updated by byroot (Jean Boussier). So I'm only able to reproduce in release mode. I'm not yet clear about what is going on, but it feels like the compiler is optimizing the check out somehow. ---------------------------------------- Bug #21771: pg-1.6.2 testsuite segfaults in ruby4.0.0dev https://bugs.ruby-lang.org/issues/21771#change-115553 * Author: mtasaka (Mamoru TASAKA) * Status: Assigned * Assignee: byroot (Jean Boussier) * ruby -v: ruby 4.0.0dev (2025-12-09 master edca81a1bb) +PRISM [x86_64-linux] * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN ---------------------------------------- Looks like pg-1.6.2 https://github.com/ged/ruby-pg/tree/c75ed8da19c356b52d83da2627295760780d3d36 testsuite began to segfault, especially the following ``` #!/bin/bash export PGPORT=54321 export RUBY_PG_TEST_DIR=/builddir/build/BUILD/tmp export RUBYLIB=/builddir/build/BUILD/rubygem-pg-1.6.2-build/pg-1.6.2/usr/share/gems/gems/pg-1.6.2/lib:/builddir/build/BUILD/rubygem-pg-1.6.2-build/pg-1.6.2/usr/lib64/gems/ruby/pg-1.6.2 # The above path should be replaced properly rm -rf $RUBY_PG_TEST_DIR exec ruby -S --verbose \ -r ./spec/helpers \ rspec -f d \ spec/pg/connection_spec.rb -E '' \ --seed 1 ``` segfaults like below (will write later) It seems that segfaults began to happen between commit:706d80830b9f0a5a2eac66251d1417abb2ff143c (no problem) and commit:d7e55f84f2bd62d302b29513d4c4dc8ae9aef96f (segfaults) Tried with commit:edca81a1bb72a9dc54a37766d2c80790dec13884 it still segfaults. Note that ruby-pg *compiled* with commit:706d80830b9f0a5a2eac66251d1417abb2ff143 does *not* segfault with ruby commit:edca81a1bb72a9dc54a37766d2c80790dec13884 , so I suspect that there is something wrong with "inlined" function or so. -- https://bugs.ruby-lang.org/