Description of problem: The qdrive3 boards are systems with multiple CPU modules. $ grep "CPU part" /proc/cpuinfo | sort -u CPU part : 0xd4b CPU part : 0xd4c The hex code is converted by lscpu to human readable name. Unfortunately, lscpu in RHEL9 is too old and it has no clue about 0xd4c aka "Cortex-X1C", so it prints "-" instead of the name. This issue will impact many scripts that uses lscpu from collecting accurate information about the test environment. Version-Release number of selected component (if applicable): util-linux 2.37.4 How reproducible: 100% Steps to Reproduce: 1. Run Linux on a qdrive3 board. 2. Install util-linux in the Linux. 3. Run and check the result of `lscpu`. Actual results: lscpu cannot convert 0xd4c to Cortex-X1C and include that information in the outputs. So the output looks confusing. Expected results: lscpu can display all CPU modules in its output. Additional info: 1. The initial discuss about this issue can be found at https://issues.redhat.com/browse/AUTOBUG-230
Thanks to Jiri Hladky for bringing this up, and to Karel and Radka for debugging!
I guess we can backport all human-readable ARM names from upstream. The number of new CPU identifiers (the hex) is huge: $ git diff HEAD^ upstream -- sys-utils/lscpu-arm.c | grep -c '+.*0x' 76