[Devel] [PATCH RHEL7 COMMIT] ms/kvm: make halt_poll_ns per-vCPU

[Konstantin Khorenko]

The commit is pushed to "branch-rh7-3.10.0-327.10.1.vz7.12.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh7-3.10.0-327.10.1.vz7.12.11
------>
commit 434ec77f2d650b9a22157d922ee700d7e7375a0f
Author: Wanpeng Li <wanpeng.li at hotmail.com>
Date:   Tue Apr 19 14:22:48 2016 +0400

    ms/kvm: make halt_poll_ns per-vCPU
    
    Patchset description:
    KVM: introduce dynamic halt_poll_ns
    
    These series makes halt_poll_ns to be a per-cpu value, and add simple
    dynamics for auto-calculating it online.
    
    It is needed to fix performance problems:
    	https://jira.sw.ru/browse/PSBM-43100
    Actually, setting kvm parameter halt_poll_ns to appropriate value is
    enough, but for different nodes optimal halt_poll_ns will differ, so
    we need auto-calculating halt_poll_ns. Also, dynamic approach saves cpu
    cycles for cases when poll is not needed.
    
    All patches are ported from kernel upstream, originals:
    
    19020f8 KVM: make halt_poll_ns per-vCPU
    aca6ff2 KVM: dynamic halt-polling
    2cbd782 KVM: trace kvm_halt_poll_ns grow/shrink
    edb9272 KVM: fix polling for guest halt continued even if disable it
    62bea5b KVM: add halt_attempted_poll to VCPU stats
    920552b KVM: disable halt_poll_ns as default for s390x
    6b6de68 KVM: halt_polling: improve grow/shrink settings
    313f636 kvm: cap halt polling at exactly halt_poll_ns
    
    Conflicts were trivial. The most complicated is absence of some
    arch's in vz7 kernel, which are maintained in upstream.
    
    Christian Borntraeger (1):
      KVM: halt_polling: improve grow/shrink settings
    
    David Hildenbrand (1):
      KVM: disable halt_poll_ns as default for s390x
    
    David Matlack (1):
      kvm: cap halt polling at exactly halt_poll_ns
    
    Paolo Bonzini (1):
      KVM: add halt_attempted_poll to VCPU stats
    
    Wanpeng Li (4):
      KVM: make halt_poll_ns per-vCPU
      KVM: dynamic halt-polling
      KVM: trace kvm_halt_poll_ns grow/shrink
      KVM: fix polling for guest halt continued even if disable it
    
    ===========================
    This patch description:
    
    Change halt_poll_ns into per-VCPU variable, seeded from module parameter,
    to allow greater flexibility.
    
    Signed-off-by: Wanpeng Li <wanpeng.li at hotmail.com>
    Signed-off-by: Paolo Bonzini <pbonzini at redhat.com>
    (cherry picked from commit 19020f8ab83de9dc5a9c8af1f321a526f38bbc40)
    
    Conflicts:
    	virt/kvm/kvm_main.c
    
    Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov at virtuozzo.com>
---
 include/linux/kvm_host.h | 1 +
 virt/kvm/kvm_main.c      | 6 ++++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 46ff70c..cd6d1b9 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -247,6 +247,7 @@ struct kvm_vcpu {
 	int sigset_active;
 	sigset_t sigset;
 	struct kvm_vcpu_stat stat;
+	unsigned int halt_poll_ns;
 
 #ifdef CONFIG_HAS_IOMEM
 	int mmio_needed;
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index fa2124c..a320bf1 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -217,6 +217,7 @@ int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
 	vcpu->kvm = kvm;
 	vcpu->vcpu_id = id;
 	vcpu->pid = NULL;
+	vcpu->halt_poll_ns = 0;
 	init_waitqueue_head(&vcpu->wq);
 	kvm_async_pf_vcpu_init(vcpu);
 
@@ -1957,8 +1958,9 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 	bool waited = false;
 
 	start = cur = ktime_get();
-	if (halt_poll_ns) {
-		ktime_t stop = ktime_add_ns(ktime_get(), halt_poll_ns);
+	if (vcpu->halt_poll_ns) {
+		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
+
 		do {
 			/*
 			 * This sets KVM_REQ_UNHALT if an interrupt