kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Xiao Guangrong <guangrong.xiao@linux.intel.com>
To: Junaid Shahid <junaids@google.com>, kvm@vger.kernel.org
Cc: andreslc@google.com, pfeiner@google.com, pbonzini@redhat.com
Subject: Re: [PATCH v3 3/8] kvm: x86: mmu: Fast Page Fault path retries
Date: Thu, 15 Dec 2016 15:20:19 +0800	[thread overview]
Message-ID: <1948ea57-76c4-dfe1-bdd2-09dbf265747f@linux.intel.com> (raw)
In-Reply-To: <1481071577-40250-4-git-send-email-junaids@google.com>



On 12/07/2016 08:46 AM, Junaid Shahid wrote:
> This change adds retries into the Fast Page Fault path. Without the
> retries, the code still works, but if a retry does end up being needed,
> then it will result in a second page fault for the same memory access,
> which will cause much more overhead compared to just retrying within the
> original fault.
>
> This would be especially useful with the upcoming fast access tracking
> change, as that would make it more likely for retries to be needed
> (e.g. due to read and write faults happening on different CPUs at
> the same time).
>
> Signed-off-by: Junaid Shahid <junaids@google.com>
> ---
>  arch/x86/kvm/mmu.c | 124 +++++++++++++++++++++++++++++++----------------------
>  1 file changed, 73 insertions(+), 51 deletions(-)
>
> diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
> index 4d33275..bcf1b95 100644
> --- a/arch/x86/kvm/mmu.c
> +++ b/arch/x86/kvm/mmu.c
> @@ -2881,6 +2881,10 @@ static bool page_fault_can_be_fast(u32 error_code)
>  	return true;
>  }
>
> +/*
> + * Returns true if the SPTE was fixed successfully. Otherwise,
> + * someone else modified the SPTE from its original value.
> + */
>  static bool
>  fast_pf_fix_direct_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
>  			u64 *sptep, u64 spte)
> @@ -2907,8 +2911,10 @@ fast_pf_fix_direct_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
>  	 *
>  	 * Compare with set_spte where instead shadow_dirty_mask is set.
>  	 */
> -	if (cmpxchg64(sptep, spte, spte | PT_WRITABLE_MASK) == spte)
> -		kvm_vcpu_mark_page_dirty(vcpu, gfn);
> +	if (cmpxchg64(sptep, spte, spte | PT_WRITABLE_MASK) != spte)
> +		return false;
> +
> +	kvm_vcpu_mark_page_dirty(vcpu, gfn);
>
>  	return true;
>  }
> @@ -2923,8 +2929,9 @@ static bool fast_page_fault(struct kvm_vcpu *vcpu, gva_t gva, int level,
>  {
>  	struct kvm_shadow_walk_iterator iterator;
>  	struct kvm_mmu_page *sp;
> -	bool ret = false;
> +	bool fault_handled = false;
>  	u64 spte = 0ull;
> +	uint retry_count = 0;
>
>  	if (!VALID_PAGE(vcpu->arch.mmu.root_hpa))
>  		return false;
> @@ -2937,62 +2944,77 @@ static bool fast_page_fault(struct kvm_vcpu *vcpu, gva_t gva, int level,
>  		if (!is_shadow_present_pte(spte) || iterator.level < level)
>  			break;
>
> -	/*
> -	 * If the mapping has been changed, let the vcpu fault on the
> -	 * same address again.
> -	 */
> -	if (!is_shadow_present_pte(spte)) {
> -		ret = true;
> -		goto exit;
> -	}
> +	do {
> +		/*
> +		 * If the mapping has been changed, let the vcpu fault on the
> +		 * same address again.
> +		 */
> +		if (!is_shadow_present_pte(spte)) {
> +			fault_handled = true;
> +			break;
> +		}

Why not include lockless_walk into the loop, retry 4 times for a invalid sp is expensive.

I am curious that did you see this retry is really helpful?  :)


  reply	other threads:[~2016-12-15  7:28 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-27  2:19 [PATCH 0/4] Lockless Access Tracking for Intel CPUs without EPT A bits Junaid Shahid
2016-10-27  2:19 ` [PATCH 1/4] kvm: x86: mmu: Use symbolic constants for EPT Violation Exit Qualifications Junaid Shahid
2016-11-02 18:03   ` Paolo Bonzini
2016-11-02 21:40     ` Junaid Shahid
2016-10-27  2:19 ` [PATCH 2/4] kvm: x86: mmu: Rename spte_is_locklessly_modifiable() Junaid Shahid
2016-10-27  2:19 ` [PATCH 3/4] kvm: x86: mmu: Fast Page Fault path retries Junaid Shahid
2016-10-27  2:19 ` [PATCH 4/4] kvm: x86: mmu: Lockless access tracking for Intel CPUs without EPT A bits Junaid Shahid
2016-11-02 18:01   ` Paolo Bonzini
2016-11-02 21:42     ` Junaid Shahid
2016-11-08 23:00 ` [PATCH v2 0/5] Lockless Access Tracking " Junaid Shahid
2016-11-08 23:00   ` [PATCH v2 1/5] kvm: x86: mmu: Use symbolic constants for EPT Violation Exit Qualifications Junaid Shahid
2016-11-21 13:06     ` Paolo Bonzini
2016-11-08 23:00   ` [PATCH v2 2/5] kvm: x86: mmu: Rename spte_is_locklessly_modifiable() Junaid Shahid
2016-11-21 13:07     ` Paolo Bonzini
2016-11-08 23:00   ` [PATCH v2 3/5] kvm: x86: mmu: Fast Page Fault path retries Junaid Shahid
2016-11-21 13:13     ` Paolo Bonzini
2016-11-08 23:00   ` [PATCH v2 4/5] kvm: x86: mmu: Lockless access tracking for Intel CPUs without EPT A bits Junaid Shahid
2016-11-21 14:42     ` Paolo Bonzini
2016-11-24  3:50       ` Junaid Shahid
2016-11-25  9:45         ` Paolo Bonzini
2016-11-29  2:43           ` Junaid Shahid
2016-11-29  8:09             ` Paolo Bonzini
2016-11-30  0:59               ` Junaid Shahid
2016-11-30 11:09                 ` Paolo Bonzini
2016-12-01 22:54       ` Junaid Shahid
2016-12-02  8:33         ` Paolo Bonzini
2016-12-05 22:57           ` Junaid Shahid
2016-11-08 23:00   ` [PATCH v2 5/5] kvm: x86: mmu: Update documentation for fast page fault mechanism Junaid Shahid
2016-12-07  0:46 ` [PATCH v3 0/8] Lockless Access Tracking for Intel CPUs without EPT A bits Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 1/8] kvm: x86: mmu: Use symbolic constants for EPT Violation Exit Qualifications Junaid Shahid
2016-12-15  6:50     ` Xiao Guangrong
2016-12-15 23:06       ` Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 2/8] kvm: x86: mmu: Rename spte_is_locklessly_modifiable() Junaid Shahid
2016-12-15  6:51     ` Xiao Guangrong
2016-12-07  0:46   ` [PATCH v3 3/8] kvm: x86: mmu: Fast Page Fault path retries Junaid Shahid
2016-12-15  7:20     ` Xiao Guangrong [this message]
2016-12-15 23:36       ` Junaid Shahid
2016-12-16 13:13         ` Xiao Guangrong
2016-12-17  0:36           ` Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 4/8] kvm: x86: mmu: Refactor accessed/dirty checks in mmu_spte_update/clear Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 5/8] kvm: x86: mmu: Introduce a no-tracking version of mmu_spte_update Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 6/8] kvm: x86: mmu: Do not use bit 63 for tracking special SPTEs Junaid Shahid
2016-12-07  0:46   ` [PATCH v3 7/8] kvm: x86: mmu: Lockless access tracking for Intel CPUs without EPT A bits Junaid Shahid
2016-12-14 16:28     ` Paolo Bonzini
2016-12-14 22:36       ` Junaid Shahid
2016-12-14 23:35         ` Paolo Bonzini
2016-12-16 13:04     ` Xiao Guangrong
2016-12-16 15:23       ` Paolo Bonzini
2016-12-17  0:01         ` Junaid Shahid
2016-12-21  9:49         ` Xiao Guangrong
2016-12-21 18:00           ` Paolo Bonzini
2016-12-17  2:04       ` Junaid Shahid
2016-12-17 14:19         ` Paolo Bonzini
2016-12-20  3:36           ` Junaid Shahid
2016-12-20  9:01             ` Paolo Bonzini
2016-12-07  0:46   ` [PATCH v3 8/8] kvm: x86: mmu: Update documentation for fast page fault mechanism Junaid Shahid

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1948ea57-76c4-dfe1-bdd2-09dbf265747f@linux.intel.com \
    --to=guangrong.xiao@linux.intel.com \
    --cc=andreslc@google.com \
    --cc=junaids@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=pfeiner@google.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).