Description of problem: msr: Write to unrecognized MSR 0x49 by perl Version-Release number of selected component (if applicable): Linux knut 5.9.13-200.fc33.x86_64 #1 SMP Tue Dec 8 15:42:52 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux How reproducible: every time Steps to Reproduce: 1. start system 2. se in dmesg 3. Actual results: Please report to x86 [24388.843824] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.846210] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.848557] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.850886] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.853237] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.855569] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.857872] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.860237] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 [24388.862585] msr: Write to unrecognized MSR 0x49 by perl Please report to x86 Expected results: It should not be there Additional info:
Yeah, I see this in my Fedora 33 (Linux 5.9 kernel) with Perl. I think the Linux 5.9 kernel is trying to filter now all the direct access to Machine Specific Registers (MSRs) from userland. Have not straced exactly what the Perl interpreter is doing in the background, in my case it's cruncing mostly a MariaDB database with DBI. You should be able to regulate dmesg noise by using the kernel msr.allow_writes=on/off/default boot time parameter. Does it work for you? GM
Close it is not a a bug it cause by spectre-meltdown-checker