linux-kernel - Re: [PATCH v4 3/3] KVM: X86: Don't use PV TLB flush with dedicated vCPUs and steal time disabled

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 13 Feb 2018 17:12:43 +0100
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Wanpeng Li <kernellwp@...il.com>, linux-kernel@...r.kernel.org,
        kvm@...r.kernel.org
Cc:     Radim Krčmář <rkrcmar@...hat.com>,
        Eduardo Habkost <ehabkost@...hat.com>
Subject: Re: [PATCH v4 3/3] KVM: X86: Don't use PV TLB flush with dedicated
 vCPUs and steal time disabled

On 13/02/2018 02:05, Wanpeng Li wrote:
> From: Wanpeng Li <wanpengli@...cent.com>
> 
> vCPUs are very unlikely to get preempted when they are the only task
> running on a CPU.  PV TLB flush is slower that the native flush in that
> case. In addition, avoid traversing all the cpus for pv tlb flush when
> steal time is disabled since pv tlb flush depends on the field in steal
> time for shared data.
> 
> Cc: Paolo Bonzini <pbonzini@...hat.com>
> Cc: Radim Krčmář <rkrcmar@...hat.com>
> Cc: Eduardo Habkost <ehabkost@...hat.com>
> Signed-off-by: Wanpeng Li <wanpengli@...cent.com>
> ---
>  arch/x86/kernel/kvm.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index c5566d9..285822f 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -545,7 +545,9 @@ static void __init kvm_guest_init(void)
>  		pv_time_ops.steal_clock = kvm_steal_clock;
>  	}
>  
> -	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH))
> +	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH) &&
> +		!kvm_para_has_feature(KVM_HINTS_DEDICATED) &&
> +		!kvm_para_has_feature(KVM_FEATURE_STEAL_TIME))
>  		pv_mmu_ops.flush_tlb_others = kvm_flush_tlb_others;
>  
>  	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
> @@ -638,7 +640,9 @@ static __init int kvm_setup_pv_tlb_flush(void)
>  {
>  	int cpu;
>  
> -	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH)) {
> +	if (kvm_para_has_feature(KVM_FEATURE_PV_TLB_FLUSH) &&
> +		!kvm_para_has_feature(KVM_HINTS_DEDICATED) &&

This should have checked the hints word.

In general, I'm going to change in the whole series
kvm_hint_has_feature with kvm_para_has_hint, and kvm_arch_hint_features
with kvm_arch_para_hints.  But apart from this small naming issue, the
series looks good, and I'm applying it to kvm/queue.

Paolo

> +		!kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
>  		for_each_possible_cpu(cpu) {
>  			zalloc_cpumask_var_node(per_cpu_ptr(&__pv_tlb_mask, cpu),
>  				GFP_KERNEL, cpu_to_node(cpu));
>