Xen slower than Jailhouse??

[Sara Alonso <alonsosara44@xxxxxxxxx>]

Hello!

I'm running a bare-metal application with Xen and Jailhouse, measuring the time it takes the processor to complete it, to make a comparison between both hypervisors. From what I understand, jailhouse should be faster. However, there is a piece of code that with Xen runs in 4000 clock cycles and with jailhouse takes 14000 clock cycles, and this is what I don't understand.

This piece of code is the same in both cases (copied below). There is a periodic interrupt and this piece of code runs inside the interrupt handler. In Xen I am booting the application in dom1, which runs on 1 vCPU. And in Jailhouse, I run the application in a cell running in 1 core.

Do you have any idea what I might be doing wrong or why jailhouse might be slower than xen in that piece of code?

Thanks!

for (int q = 92; q > 0; q--) {

        state->a[q] = state->a[q-1];

}

state->a[0] = t33;

for (int q = 83; q > 0; q--) {

    state->b[q] = state->b[q-1];

}

state->b[0] = t11;

for (int q = 110; q > 0; q--) {

    state->c[q] = state->c[q-1];

}

state->c[0] = t22;