What were you trying to do that didn't work?

A customer seems memory consumption of tuned growing over time, until it eats all the available memory. He encounters the issue on several of his systems.
For example, after 41 days of uptime, tuned consumes 2GB on one of his systems:

USER         PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
root        1092  0.6 55.5 2643692 2079060 ?     -    Mar26 362:48 /usr/libexec/platform-python -Es /usr/sbin/tuned -l -P

Since then the customer had to restart the service and I asked him to collect some coredumps regularly to see any trend, and clearly "strings" accumulate, e.g:

1. Coredump taken on May 7 (417MB coredump)

   ...
        997 .d/.
       998 cpu1
       998 cpu2
       998 cpu3
       998 fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon nopl xtopology tsc_reliable nonstop_tsc cpuid pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm invpcid_single pti ssbd ibrs ibpb stibp fsgsbase tsc_adjust bmi1 avx2 smep bmi2 invpcid xsaveopt arat md_clear flush_l1d arch_capabilities
   ...

2. Coredump taken on May 14 (422MB coredump)

   ...
      1165 .d/.
      1166 cpu0
      1166 cpu1
      1166 cpu2
      1166 cpu3
      1166 fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon nopl xtopology tsc_reliable nonstop_tsc cpuid pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm invpcid_single pti ssbd ibrs ibpb stibp fsgsbase tsc_adjust bmi1 avx2 smep bmi2 invpcid xsaveopt arat md_clear flush_l1d arch_capabilities
   ...

Customer system is running "virtual-guest" profile.

I'm trying to reproduce internally, but so far I don't have much insights what could be going on, I see some memory increase for my tuned instance, but it's not significative:

# grep RssAnon tuned.Wed\ May\ 15\ *
tuned.Wed May 15 07:06:33 CEST 2024:RssAnon:	   14948 kB
tuned.Wed May 15 08:00:02 CEST 2024:RssAnon:	   15188 kB
tuned.Wed May 15 09:00:01 CEST 2024:RssAnon:	   15272 kB
tuned.Wed May 15 10:00:01 CEST 2024:RssAnon:	   15356 kB
tuned.Wed May 15 11:00:01 CEST 2024:RssAnon:	   15256 kB

Additionally for me, I don't see any deduplication of "fpu vme de..." string (coming from /proc/cpuinfo I guess).

Please provide the package NVR for which bug is seen:

tuned-2.21.0-1.el8_9.noarch

How reproducible:

Always for the customer