Hi Eddie, Do you mean the space between "include" and "<per_cpu.h>"? I checked coding_guidelines.rst and didn't find the statement about this situation. We use "space" instead of "TAB" in other existi

Hi Fei, Thanks for your comment. Actually, it is a TODO of https://lists.projectacrn.org/g/acrn-dev/topic/patch_v3_2_2_hv_reset/31028081?p=,,,20,0,0,0::recentpostdate%2Fsticky,,,20,2,160,31028081. IMH

This patch just removes unnecessary indent before #include in pm.c. Signed-off-by: Kaige Fu <kaige.fu@...> --- hypervisor/arch/x86/pm.c | 32 ++++++++++++++++---------------- 1 file changed, 16 i

This patch adds prefix 'p' before 'cpu' to physical cpu related functions. And there is no code logic change. Signed-off-by: Kaige Fu <kaige.fu@...> --- hypervisor/arch/x86/boot/cpu_primary.S |

It is better to use an unified MACRO. So, this patch moves AP_MASK to cpu.h and removes the defination of the MACRO spanning different source files. Signed-off-by: Kaige Fu <kaige.fu@...> --- hy

Hi Eddie, Sure. Will check and update it.

This patch removes dead loop in stop_cpus because timeout never occurs. If target cpu received a NMI and panic, it has called cpu_dead and stop_cpus success. If target cpu is running, an IPI will be d

The physical core of lapic_pt vm should be reset for security and correctness when shutdown the vm. Signed-off-by: Kaige Fu <kaige.fu@...> --- hypervisor/arch/x86/cpu.c | 14 +++++++++++++- hyper

This patch makes the following changes: - Add one parameter 'mask' to start_cpus for later use. - Set cpu state as DEAD instead of dead loop when fail to start cpu. - Panic when there are any failures

This patch remove the over design CPU_UP_TIMEOUT and use CPU_UP_TIMEOUT and CPU_DOWN_TIMEOUT instead. Signed-off-by: Kaige Fu <kaige.fu@...> Acked-by: Eddie Dong <eddie.dong@...> --- hyper

Currently, the previous configurations about guest_flags set by DM will not be cleared when shutdown the vm. Then it might bring issue for the next dm-launched vm. For example, if we create one vm wit

This patchset aims at fixing the following issues: - There will be some DM set flags remained after shutdown the vm. This patchset fix it by clear all DM set flags when shutdown the vm. - Since lapic

Got it. Will do in next version. > > >

Hi Eddie, In current stop_cpus, it do like this: 1) make_pcpu_offline (all APs) 2) Wait for all APs offlined 3) If timeout occurs, dead loop For reset_cpus in shutdown_vm, you commented that we should

Hi Eddie, The loop isn't necessary. Actually, this is ((1UL << phys_cpu_num) - 1UL) & ~(1UL << 0U)). I test it and it works for me. Yes. As we taked in previous version of the patchset. We will reuse

Hi Eddie, I got your comment. But here is what I think: I refine the stop_cpus to take one more parameter 'mask'. So, we can reuse the stop_cpus here directly instead of doing like the following: 1. m

Will use ms.

The physical core of lapic_pt vm should be reset for security and correctness when shutdown the vm. Signed-off-by: Kaige Fu <kaige.fu@...> --- hypervisor/arch/x86/guest/vm.c | 20 +++++++++++++++

This patch makes the following changes: - Add one parameter 'mask' for later use. - Set pcpu state as INVALID when fail to offline cpu. - Panic when there are any failures when stop_cpus in host_enter

This patch makes the following changes: - Add one parameter 'mask' to start_cpus for later use. - Set cpu state as INVALID instead of dead loop when fail to start cpu. - Panic when there are any failu