github/acrn-hypervisor
1
0
Fork 0
mirror of https://github.com/projectacrn/acrn-hypervisor.git synced 2025-08-04 01:41:08 +00:00
Code Issues Projects Releases Wiki Activity

hv: update guest RIP only if vcpu->arch.inst_len is non zero

Browse Source 
In very large number of VM extis, the VM-exit instruction length could be
zero, and it's no need to update VMX_GUEST_RIP.

Some examples:

- all external interrupt VM exits in non LAPIC passthru setup.
- for all the nested VM-exits that are reflecting to L1 hypervisor.

Tracked-On: #6289
Signed-off-by: Zide Chen <zide.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
This commit is contained in:
Zide Chen 2021-09-27 17:09:31 -07:00 committed by wenlingz
parent b7e9a68923
commit f801ba4ed7
2 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
hypervisor/arch/x86/guest/nested.c
View File

@ -1450,6 +1450,16 @@ int32_t nested_vmexit_handler(struct acrn_vcpu *vcpu)
vcpu->arch.nested.in_l2_guest = false;
}
/*
* For VM-exits that reflect to L1 hypervisor, ACRN can't advance to next guest RIP
* which is up to the L1 hypervisor to make the decision.
*
* The only case that doesn't need to be reflected is EPT violations that can be
* completely handled by ACRN, which requires L2 VM to re-execute the instruction
* after the shadow EPT is being properly setup.
* In either case, need to set vcpu->arch.inst_len to zero.
*/
vcpu_retain_rip(vcpu);
return 0;
}

10
hypervisor/arch/x86/guest/vcpu.c
View File

@ -657,8 +657,8 @@ static void write_cached_registers(struct acrn_vcpu *vcpu)
*/
int32_t run_vcpu(struct acrn_vcpu *vcpu)
{
uint32_t instlen, cs_attr;
uint64_t rip, ia32_efer, cr0;
uint32_t cs_attr;
uint64_t ia32_efer, cr0;
struct run_context *ctx =
&vcpu->arch.contexts[vcpu->arch.cur_context].run_ctx;
int32_t status = 0;
@ -718,9 +718,9 @@ int32_t run_vcpu(struct acrn_vcpu *vcpu)
/* This VCPU was already launched, check if the last guest
* instruction needs to be repeated and resume VCPU accordingly
*/
instlen = vcpu->arch.inst_len;
rip = vcpu_get_rip(vcpu);
exec_vmwrite(VMX_GUEST_RIP, rip + instlen);
if (vcpu->arch.inst_len != 0U) {
exec_vmwrite(VMX_GUEST_RIP, vcpu_get_rip(vcpu) + vcpu->arch.inst_len);
}
/* Resume the VM */
status = exec_vmentry(ctx, VM_RESUME, ibrs_type);