Hi everyone,

We are noticing that a function which uses a lot of gettime syscalls executes 
incorrectly after a cpuswitch out of kvm to the timing cpu model. It seems that 
the control path is altered at some return instruction inside libstdc++ code. 
The new pc, which is loaded by the return instruction from memory, is 
incorrect. This can be observed when running on linux4.4.0 and not when running 
on linux2.6.32.


We are running x86 in SE mode. Has anyone noticed this or encountered something 
similar?


Best regards,

Alex
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to