3rd party tool turbostat crashes with "Floating exception (core dumped)"

[thowe]

For testing the current CPU clock, I have been running turbostat for many years. This has worked over many versions of FreeBSD. Since version 13 the tool stops after startup with "Floating exception (core dumped)".

Here's what I do to install and run turbostat (worked so well before FreeBSD):

Code Select Expand


pkg add https://pkg.freebsd.org/FreeBSD:13:amd64/latest/All/turbostat-4.17_2.pkg
rehash
kldload cpuctl
turbostat --interval 3



Since FreeBSD 13 I now have the following output:

Code Select Expand


# turbostat --interval 3
turbostat version 17.06.23 - Len Brown <lenb@kernel.org>
CPUID(0): AuthenticAMD 13 CPUID levels; family:model:stepping 0xf:30:1 (15:48:1)
CPUID(1): SSE3 MONITOR - - - TSC MSR - -
CPUID(6): APERF, No-TURBO, No-DTS, No-PTM, No-HWP, No-HWPnotify, No-HWPwindow, No-HWPepp, No-HWPpkg, No-EPB
CPUID(7): No-SGX
NSFOD /sys/devices/system/cpu/cpu0/cpufreq/scaling_driver
Floating exception (core dumped)


My questions:

• Is there any reason that turbostat should no longer run with FreeBSD 13? (When installing I get a message that the package is no longer maintained.)
• Are there alternative similar tools that show the current CPU clock?

Thanks for any hints.

[franco]

Might be broken these days, kernel internals change all the time and maybe the software cannot cope with it. This is especially bad when internals change but the build still works, so nobody starts looking why that is.

Not aware of another tool, but I don't know much about this to be honest.


Cheers,
Franco

[Patrick M. Hausen]

FreeBSD 13.0 reached EOL just a couple of days ago. While it was still supported, all packages were built on 13.0. The general policy is to always build packages on the lowest supported release of a particular major version.

OPNsense on the other hand is based on FreeBSD 13.1.

Packages that rely on specific kernel interfaces sometimes break on the newer release. This is specifically troublesome for kernel modules because systems might crash at boot if they are configured to load the module automatically.

That being said with 13.0 officially retired the package clusters are currently busy rebuilding everything for 13.1. So most probably the package will work again in a couple of days when the binaries have been replaced.

HTH,
Patrick

[thowe]

Thanks a lot!  :)

So I'll reinstall the tool in a few days and see if it is working again. Or maybe I'll find another way to determine the current CPU clock.

[Patrick M. Hausen]

You can check if it was built for 13.1 like this:

Code Select Expand

$ file /usr/local/sbin/turbostat
/usr/local/sbin/turbostat: ELF 64-bit LSB executable, x86-64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 13.0 (1300139), FreeBSD-style, stripped


[thowe]

Perfect, thanks for the hint!

[franco]

Or try to build cleanly from ports tree:

# opnsense-code ports src
# cd /usr/ports/sysutils/turbostat
# make reinstall


Cheers,
Franco