Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Zhenyu Ye <yezhenyu2@huawei.com>
To:     <maz@kernel.org>, <james.morse@arm.com>,
        <julien.thierry.kdev@gmail.com>, <suzuki.poulose@arm.com>,
        <catalin.marinas@arm.com>, <will@kernel.org>,
        <steven.price@arm.com>
CC:     <yezhenyu2@huawei.com>, <linux-arm-kernel@lists.infradead.org>,
        <linux-kernel@vger.kernel.org>, <linux-arch@vger.kernel.org>,
        <linux-mm@kvack.org>, <arm@kernel.org>, <xiexiangyou@huawei.com>
Subject: [RFC PATCH v1] arm64: kvm: flush tlbs by range in unmap_stage2_range function
Date:   Thu, 9 Jul 2020 21:47:31 +0800
Message-ID: <20200709134731.2384-1-yezhenyu2@huawei.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 7BIT
Content-Type:   text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Now in unmap_stage2_range(), we unmap a page by the following
steps:
	p*d_clear();
	kvm_tlb_flush_vmid_ipa();  # take 2us;
	kvm_flush_dcache_p*d();    # take 0.5us;
	put_page();

When the range is very large, such as 1G, then unmap_stage2_range()
may take more than 500ms at one time. This may cause some performance
problems in the following case:

The VM that uses 1G hugepage memory, with high memory pressure (the
dirty page rate reaches 500MB/s), does migration with --live.  When
the bandwidth is less than dirty rate, the migration will failed and
VM will rollback to the source host.  unmap_stage2_range() will be
called to combine the scattered 4K pages -- then cause the vm's
downtime too long.

In my test, unmap_stage2_range() can take a maximum of 1.2s, and
the VM downtime reaches 7s. VM configuration is as follows:

  <memory unit='KiB'>201326592</memory>
  <vcpu placement='static'>48</vcpu>
  <memoryBacking>
    <hugepages>
      <page size='1' unit='GiB' nodeset='0'/>
    </hugepages>
  </memoryBacking>

The dirty rate is 500MB/s ~ 1000MB/s, and bandwidth is 500MB.

--
So, this patch move the kvm_tlb_flush_vmid_ipa() out of loop, and
flush tlbs by range after other operations are complete.  Because
we do not make new mapping for the pages, so this don't violate
the BBM rules.

After this change, the cost of unmap_stage2_range() can reduce to
16ms, and VM downtime can be less than 1s.

Signed-off-by: Zhenyu Ye <yezhenyu2@huawei.com>
---
 arch/arm64/include/asm/kvm_asm.h |  2 ++
 arch/arm64/kvm/hyp/tlb.c         | 36 ++++++++++++++++++++++++++++++++
 arch/arm64/kvm/mmu.c             | 11 +++++++---
 3 files changed, 46 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index 352aaebf4198..ef8203d3ca45 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -61,6 +61,8 @@ extern char __kvm_hyp_vector[];
 
 extern void __kvm_flush_vm_context(void);
 extern void __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa);
+extern void __kvm_tlb_flush_vmid_range(struct kvm *kvm, phys_addr_t start,
+				       phys_addr_t end);
 extern void __kvm_tlb_flush_vmid(struct kvm *kvm);
 extern void __kvm_tlb_flush_local_vmid(struct kvm_vcpu *vcpu);
 
diff --git a/arch/arm64/kvm/hyp/tlb.c b/arch/arm64/kvm/hyp/tlb.c
index d063a576d511..4f4737a7e588 100644
--- a/arch/arm64/kvm/hyp/tlb.c
+++ b/arch/arm64/kvm/hyp/tlb.c
@@ -189,6 +189,42 @@ void __hyp_text __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa)
 	__tlb_switch_to_host(kvm, &cxt);
 }
 
+void __hyp_text __kvm_tlb_flush_vmid_range(struct kvm *kvm, phys_addr_t start,
+					   phys_addr_t end)
+{
+	struct tlb_inv_context cxt;
+	unsigned long addr;
+
+	start = __TLBI_VADDR(start, 0);
+	end = __TLBI_VADDR(end, 0);
+
+	dsb(ishst);
+
+	/* Switch to requested VMID */
+	kvm = kern_hyp_va(kvm);
+	__tlb_switch_to_guest(kvm, &cxt);
+
+	if ((end - start) >= 512 << (PAGE_SHIFT - 12)) {
+		__tlbi(vmalls12e1is);
+		goto end;
+	}
+
+	for (addr = start; addr < end; addr += 1 << (PAGE_SHIFT - 12))
+		__tlbi(ipas2e1is, addr);
+
+	dsb(ish);
+	__tlbi(vmalle1is);
+
+end:
+	dsb(ish);
+	isb();
+
+	if (!has_vhe() && icache_is_vpipt())
+		__flush_icache_all();
+
+	__tlb_switch_to_host(kvm, &cxt);
+}
+
 void __hyp_text __kvm_tlb_flush_vmid(struct kvm *kvm)
 {
 	struct tlb_inv_context cxt;
diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index 8c0035cab6b6..bcc719c32921 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -63,6 +63,12 @@ static void kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa)
 	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, kvm, ipa);
 }
 
+static void kvm_tlb_flush_vmid_range(struct kvm *kvm, phys_addr_t start,
+				     phys_addr_t end)
+{
+	kvm_call_hyp(__kvm_tlb_flush_vmid_range, kvm, start, end);
+}
+
 /*
  * D-Cache management functions. They take the page table entries by
  * value, as they are flushing the cache using the kernel mapping (or
@@ -267,7 +273,6 @@ static void unmap_stage2_ptes(struct kvm *kvm, pmd_t *pmd,
 			pte_t old_pte = *pte;
 
 			kvm_set_pte(pte, __pte(0));
-			kvm_tlb_flush_vmid_ipa(kvm, addr);
 
 			/* No need to invalidate the cache for device mappings */
 			if (!kvm_is_device_pfn(pte_pfn(old_pte)))
@@ -295,7 +300,6 @@ static void unmap_stage2_pmds(struct kvm *kvm, pud_t *pud,
 				pmd_t old_pmd = *pmd;
 
 				pmd_clear(pmd);
-				kvm_tlb_flush_vmid_ipa(kvm, addr);
 
 				kvm_flush_dcache_pmd(old_pmd);
 
@@ -324,7 +328,6 @@ static void unmap_stage2_puds(struct kvm *kvm, p4d_t *p4d,
 				pud_t old_pud = *pud;
 
 				stage2_pud_clear(kvm, pud);
-				kvm_tlb_flush_vmid_ipa(kvm, addr);
 				kvm_flush_dcache_pud(old_pud);
 				put_page(virt_to_page(pud));
 			} else {
@@ -352,6 +355,8 @@ static void unmap_stage2_p4ds(struct kvm *kvm, pgd_t *pgd,
 
 	if (stage2_p4d_table_empty(kvm, start_p4d))
 		clear_stage2_pgd_entry(kvm, pgd, start_addr);
+
+	kvm_tlb_flush_vmid_range(kvm, start_addr, end);
 }
 
 /**
-- 
2.19.1