Re: [Xen-devel] [PATCH v2] VMX: Eliminate cr3 save/loading exiting when UG enabled

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 18.12.13 at 05:00, Yang Zhang <yang.z.zhang@xxxxxxxxx> wrote:
> From: Yang Zhang <yang.z.zhang@xxxxxxxxx>
> 
> With the feature of unrestricted guest, there should no vmexit
> be triggered when guest accesses the cr3 in non-paging mode.
> 
> Signed-off-by: Yang Zhang <yang.z.zhang@xxxxxxxxx>
> ---
> changes in v2:
> Fix the guest boot failure on non-UG platform.
> 
> The previous patch doesn't consider the non-UG platform and will cause guest 
> boot
> failure on non-ug platform.

Which "previous patch"? This one being v2, yet there not being a
v1 with the same title on the list back until the beginning of
October, what you say here is rather confusing. Please be a
little more considerate of other people's time and provide more
precise information: If a patch needing fixing went in already,
name its commit ID and title. If the patch needing fixing didn't go
in yet, integrate the fix with that patch (or series) and submit a
new version. And if you post a v2 of a patch under a different
title than the v1 was posted, please provide enough information
to be able to identify that patch.

Jan

Jan

> During developing this patch, i was thinking 
> whether
> the reading of GUEST_CR3 is necessary on each vmexit. I thought the answer 
> should
> be no. But current implemention relies the {hw,guest}_cr3 in many places and 
> it
> hard to do a cleanup.(If anyone interesting, welcome :))
> 
> BTW: this patch will improve performance, especially for context switching 
> intensive
> workload, we saw about 1% improvement. So please backport it to 4.3.
> ---
>  xen/arch/x86/hvm/vmx/vmx.c |    9 +++++----
>  1 files changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> index dfff628..f6409d6 100644
> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -1157,7 +1157,7 @@ static void vmx_update_guest_cr(struct vcpu *v, 
> unsigned int cr)
>              uint32_t cr3_ctls = (CPU_BASED_CR3_LOAD_EXITING |
>                                   CPU_BASED_CR3_STORE_EXITING);
>              v->arch.hvm_vmx.exec_control &= ~cr3_ctls;
> -            if ( !hvm_paging_enabled(v) )
> +            if ( !hvm_paging_enabled(v) && !vmx_unrestricted_guest(v) )
>                  v->arch.hvm_vmx.exec_control |= cr3_ctls;
>  
>              /* Trap CR3 updates if CR3 memory events are enabled. */
> @@ -1231,7 +1231,7 @@ static void vmx_update_guest_cr(struct vcpu *v, 
> unsigned int cr)
>      case 3:
>          if ( paging_mode_hap(v->domain) )
>          {
> -            if ( !hvm_paging_enabled(v) )
> +            if ( !hvm_paging_enabled(v) && !vmx_unrestricted_guest(v) )
>                  v->arch.hvm_vcpu.hw_cr[3] =
>                      v->domain->arch.hvm_domain.params[HVM_PARAM_IDENT_PT];
>              vmx_load_pdptrs(v);
> @@ -2487,10 +2487,11 @@ void vmx_vmexit_handler(struct cpu_user_regs *regs)
>  
>      hvm_invalidate_regs_fields(regs);
>  
> -    if ( paging_mode_hap(v->domain) && hvm_paging_enabled(v) )
> +    if ( paging_mode_hap(v->domain) )
>      {
>          __vmread(GUEST_CR3, &v->arch.hvm_vcpu.hw_cr[3]);
> -        v->arch.hvm_vcpu.guest_cr[3] = v->arch.hvm_vcpu.hw_cr[3];
> +        if ( vmx_unrestricted_guest(v) || hvm_paging_enabled(v) )
> +            v->arch.hvm_vcpu.guest_cr[3] = v->arch.hvm_vcpu.hw_cr[3];
>      }
>  
>      __vmread(VM_EXIT_REASON, &exit_reason);
> -- 
> 1.7.1




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel