Re: [Qemu-devel] [PATCH] [KVM] Needless to update msi route when only msi-x entry "control" section changed

[Michael S. Tsirkin]

On Tue, May 07, 2013 at 01:53:03AM +0000, Zhanghaoyu (A) wrote:
> >> >> With regard to old version linux guest(e.g., rhel-5.5), in ISR 
> >> >> processing, mask and unmask msi-x vector every time, which result in 
> >> >> VMEXIT, then QEMU will invoke kvm_irqchip_update_msi_route() to ask KVM 
> >> >> hypervisor to update the VM irq routing table. In KVM hypervisor, 
> >> >> synchronizing RCU needed after updating routing table, so much time 
> >> >> consumed for waiting in wait_rcu_gp(). So CPU usage in VM is so high, 
> >> >> while from the view of host, VM's total CPU usage is so low. 
> >> >> Masking/unmasking msi-x vector only set msi-x entry "control" section, 
> >> >> needless to update VM irq routing table.
> >> >> 
> >> >> Signed-off-by: Zhang Haoyu <address@hidden>
> >> >> Signed-off-by: Huang Weidong <address@hidden>
> >> >> Signed-off-by: Qin Chuanyu <address@hidden>
> >> >> ---
> >> >> hw/i386/kvm/pci-assign.c | 3 +++
> >> >> 1 files changed, 3 insertions(+)
> >> >> 
> >> >> --- a/hw/i386/kvm/pci-assign.c  2013-05-04 15:53:18.000000000 +0800
> >> >> +++ b/hw/i386/kvm/pci-assign.c  2013-05-04 15:50:46.000000000 +0800
> >> >> @@ -1576,6 +1576,8 @@ static void assigned_dev_msix_mmio_write
> >> >>                  MSIMessage msg;
> >> >>                  int ret;
> >> >> 
> >> >> +                /* Needless to update msi route when only msi-x entry 
> >> >> "control" section changed */
> >> >> +                if ((addr & (PCI_MSIX_ENTRY_SIZE - 1)) != 
> >> >> + PCI_MSIX_ENTRY_VECTOR_CTRL){
> >> >>                  msg.address = entry->addr_lo |
> >> >>                      ((uint64_t)entry->addr_hi << 32);
> >> >>                  msg.data = entry->data; @@ -1585,6 +1587,7 @@ 
> >> >> static void assigned_dev_msix_mmio_write
> >> >>                  if (ret) {
> >> >>                      error_report("Error updating irq routing entry 
> >> >> (%d)", ret);
> >> >>                  }
> >> >> +                }
> >> >>              }
> >> >>          }
> >> >>      }
> >> >> 
> >> >> Thanks,
> >> >> Zhang Haoyu
> >> >
> >> >
> >> >If guest wants to update the vector, it does it like this:
> >> >mask
> >> >update
> >> >unmask
> >> >and it looks like the only point where we update the vector is on unmask, 
> >> >so this patch will mean we don't update the vector ever.
> >> >
> >> >I'm not sure this combination (old guest + legacy device assignment
> >> >framework) is worth optimizing. Can you try VFIO instead?
> >> >
> >> >But if it is, the right way to do this is probably along the lines of the 
> >> >below patch. Want to try it out?
> >> >
> >> >diff --git a/kvm-all.c b/kvm-all.c
> >> >index 2d92721..afe2327 100644
> >> >--- a/kvm-all.c
> >> >+++ b/kvm-all.c
> >> >@@ -1006,6 +1006,11 @@ static int kvm_update_routing_entry(KVMState *s,
> >> >             continue;
> >> >         }
> >> > 
> >> >+        if (entry->type == new_entry->type &&
> >> >+            entry->flags == new_entry->flags &&
> >> >+            entry->u == new_entry->u) {
> >> >+            return 0;
> >> >+        }
> >> >         entry->type = new_entry->type;
> >> >         entry->flags = new_entry->flags;
> >> >         entry->u = new_entry->u;
> >> >
> >> 
> >> union type cannot be directly compared, I tried out below patch 
> >> instead,
> >> --- a/kvm-all.c 2013-05-06 09:56:38.000000000 +0800
> >> +++ b/kvm-all.c 2013-05-06 09:56:45.000000000 +0800
> >> @@ -1008,6 +1008,12 @@ static int kvm_update_routing_entry(KVMS
> >>              continue;
> >>          }
> >> 
> >> +        if (entry->type == new_entry->type &&
> >> +            entry->flags == new_entry->flags &&
> >> +            !memcmp(&entry->u, &new_entry->u, sizeof(entry->u))) {
> >> +            return 0;
> >> +        }
> >> +
> >>          entry->type = new_entry->type;
> >>          entry->flags = new_entry->flags;
> >>          entry->u = new_entry->u;
> >> 
> >> MST's patch is more universal than my first patch fixed in 
> >> assigned_dev_msix_mmio_write().
> >> On the case that the msix entry's other section but "control" section is 
> >> set to the identical value with old entry's, MST's patch also works.
> >> MST's patch also works on the non-passthrough scenario.
> >
> >Any numbers for either case?
> >
> I'm not sure what you said exactly means. 
> Do you want me to make a further statement for comparison between above two 
> patches?
> If yes, no other comments.


Sorry.
What I mean is that we'll need to know what is the
performance impact of my patch before we can
device whether to apply it.
Could you send this information please?

> >> And, after MST's patch applied, the below check in 
> >> virtio_pci_vq_vector_unmask() can be removed.
> >> --- a/hw/virtio/virtio-pci.c    2013-05-04 15:53:20.000000000 +0800
> >> +++ b/hw/virtio/virtio-pci.c    2013-05-06 10:25:58.000000000 +0800
> >> @@ -619,12 +619,10 @@ static int virtio_pci_vq_vector_unmask(V
> >> 
> >>      if (proxy->vector_irqfd) {
> >>          irqfd = &proxy->vector_irqfd[vector];
> >> -        if (irqfd->msg.data != msg.data || irqfd->msg.address != 
> >> msg.address) {
> >>              ret = kvm_irqchip_update_msi_route(kvm_state, irqfd->virq, 
> >> msg);
> >>              if (ret < 0) {
> >>                  return ret;
> >>              }
> >> -        }
> >>      }
> >> 
> >>      /* If guest supports masking, irqfd is already setup, unmask it.
> >> 
> >> Thanks,
> >> Zhang Haoyu