What is a Hypercall?

Hypercall is an explicit call from Virtual Machine (VM) into the underlying VM Moniter (VMM). That is, a hypercall issued from the guest will cause a VM exit from the guest into the VMM. In this sense, it is kinda like syscall, which will result in a trap from userspace into  kernel space.  Thus, hypercall can serve as a communication approach between the guest and the underlying VMM. The guest can issue hypercalls via VMMCALL instruction on AMD SVM machine or VMCALL on Intel VMX machine.

Adding a Hypercall in Palacios

Palacios maintains a global hypercall map, which maintains the mapping between the unique hypercall number and the hypercall handler. (Each hypercall is associated with a unique hypercall number.). Therefore, to add a hypercal in Palacios, you need to:

• Implement a handler for the hypercall.

• Assign a unique hypercall number for the hypercall.

• Register the hypercall handler and the hypercall number with Palacios.

Sample Example:

Note: The hypercall management APIs in Palacios are contained in palacios/include/palacios/vmm_hypercall.h and palacios/src/palacios/vmm_hypercall.c.

Issuing Hypercalls from Palacios Guest

“Up to four arguments may be passed in rbx, rcx, rdx, and rsi respectively. The hypercall number should be placed in rax and the return value will be placed in rax.  No other registers will be clobbered unless explicitly stated by the particular hypercall.” (quoted from [1])

For exmaple, on a AMD SVM machine, a hypercall with 3 parameters can be issued via the following code:

References:

[1] Linux KVM Hypercall