i have guest OS (linux kernel 18.104.22.168) running on host with linux kernel 22.214.171.124 using kvm. following is the qemu command line used to spawn the guest. Guest os has some application (multithreaded) which access TSC for delay/timing calculation. When guest is started with below command, it fails to come up due to some internal heartbeat failures… my guess is this is happening due to unsynchronized TSC clocks between cores. below are the message seen during bootup. Same guest OS comes up fine with no smp option. Anyone has any suggestions how to boot guest OS with SMP to avoid below issue? does TSC need to be synced between cpu cores? Any help is really appreciated. Thanks.
qemu-system-x86_64 -m 4096 -nographic -cdrom my.iso -hda vdisk1.img -smp 2 -no-acpi
Clocksource tsc unstable (delta = 4392634963547 ns)
tsc_sync: your timestamp counter is not reliable.
Slower fallback will be used for tracing. See the LTTng documentation to find an appropriate workaround for your architecture.
TSC unsynchronized : 649769653 cycles delta is over threshold 1000