Description of problem: Enabling the legacy provider in openssl.cnf makes "squid -k parse" die with a SEGV. Version-Release number of selected component (if applicable): openssl-libs-3.0.1-43.el9_0.x86_64 squid-5.5-3.el9_1.x86_64 How reproducible: always (I think) Reproduced in a fresh minimal install of RHEL 9.1/x64 in Virtualbox 7 but it was first noticed in a VM hosted by VMware vSphere 7. Steps to Reproduce: 1. Enable (uncomment) the legacy provider in /etc/pki/tls/openssl.cnf [provider_sect] default = default_sect legacy = legacy_sect ## [default_sect] activate = 1 [legacy_sect] activate = 1 2. Try to parse squid config: squid -k parse -f /etc/squid/squid.conf.default Actual results: ... 2022/12/22 18:04:29| Initializing https:// proxy context 2022/12/22 18:04:29| Requiring client certificates. Segmentation fault (core dumped) Expected results: No SEGV by squid -k parse Additional info: The actual squid config does not matter. Also throws a SEGV with an empty squid.conf.
Dear Walter, Does squid introduce any atexit handlers?
No idea. We're just squid users and noticed the bug because we try to test the config before applying it via puppet. The openssl legacy provider is required for a different software unrelated to squid.
Changing the component then.
Same happens with nmap when fed a script that uses openssl (eg, ssl-enum-ciphers.nse)