mirror of
https://github.com/projectacrn/acrn-hypervisor.git
synced 2025-04-29 04:04:05 +00:00
The current implement will trigger shutdown vm request on the BSP VCPU on the VM, not the VCPU will trap out because triple fault. However, if the BSP VCPU on the VM is handling another IO emulation, it may overwrite the triple fault IO request on the vhm_request_buffer in function acrn_insert_request. The atomic operation of get_vhm_req_state can't guarantee the vhm_request_buffer will not access by another IO request if it is not running on the corresponding VCPU. So it should trigger triple fault shutdown VM IO request on the VCPU which trap out because of triple fault exception. Besides, rt_vm_pm1a_io_write will do the right thing which we shouldn't do it in triple_fault_shutdown_vm. Tracked-On: #1842 Signed-off-by: Li, Fei1 <fei1.li@intel.com> |
||
---|---|---|
.. | ||
acpi_ext.c | ||
dmar_parse.c |