Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Paolo Bonzini <pbonzini@redhat.com>
To:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc:     rkrcmar@redhat.com, Vitaly Kuznetsov <vkuznets@redhat.com>,
        Junaid Shahid <junaids@google.com>,
        Xiao Guangrong <xiaoguangrong@tencent.com>
Subject: [PATCH] KVM: try __get_user_pages_fast even if not in atomic context
Date:   Fri, 27 Jul 2018 17:46:47 +0200
Message-Id: <1532706407-11380-1-git-send-email-pbonzini@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

We are currently cutting hva_to_pfn_fast short if we do not want an
immediate exit, which is represented by !async && !atomic.  However,
this is unnecessary, and __get_user_pages_fast is *much* faster
because the regular get_user_pages takes pmd_lock/pte_lock.
In fact, when many CPUs take a nested vmexit at the same time
the contention on those locks is visible, and this patch removes
about 25% (compared to 4.18) from vmexit.flat on a 16 vCPU
nested guest.

Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 virt/kvm/kvm_main.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 861bb20e8451..0f26ff7ddedb 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1343,18 +1343,16 @@ static inline int check_user_page_hwpoison(unsigned long addr)
 }
 
 /*
- * The atomic path to get the writable pfn which will be stored in @pfn,
- * true indicates success, otherwise false is returned.
+ * The fast path to get the writable pfn which will be stored in @pfn,
+ * true indicates success, otherwise false is returned.  It's also the
+ * only part that runs if we can are in atomic context.
  */
-static bool hva_to_pfn_fast(unsigned long addr, bool atomic, bool *async,
-			    bool write_fault, bool *writable, kvm_pfn_t *pfn)
+static bool hva_to_pfn_fast(unsigned long addr, bool write_fault,
+			    bool *writable, kvm_pfn_t *pfn)
 {
 	struct page *page[1];
 	int npages;
 
-	if (!(async || atomic))
-		return false;
-
 	/*
 	 * Fast pin a writable pfn only if it is a write fault request
 	 * or the caller allows to map a writable pfn for a read fault
@@ -1498,7 +1496,7 @@ static kvm_pfn_t hva_to_pfn(unsigned long addr, bool atomic, bool *async,
 	/* we can do it either atomically or asynchronously, not both */
 	BUG_ON(atomic && async);
 
-	if (hva_to_pfn_fast(addr, atomic, async, write_fault, writable, &pfn))
+	if (hva_to_pfn_fast(addr, write_fault, writable, &pfn))
 		return pfn;
 
 	if (atomic)
-- 
1.8.3.1