[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PULL 29/44] spapr/xive: introduce a VM state change ha
From: |
Cédric Le Goater |
Subject: |
Re: [Qemu-devel] [PULL 29/44] spapr/xive: introduce a VM state change handler |
Date: |
Tue, 4 Jun 2019 10:10:08 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.7.0 |
On 04/06/2019 09:49, Alexey Kardashevskiy wrote:
>
>
> On 29/05/2019 16:50, David Gibson wrote:
>> From: Cédric Le Goater <address@hidden>
>>
>> This handler is in charge of stabilizing the flow of event notifications
>> in the XIVE controller before migrating a guest. This is a requirement
>> before transferring the guest EQ pages to a destination.
>>
>> When the VM is stopped, the handler sets the source PQs to PENDING to
>> stop the flow of events and to possibly catch a triggered interrupt
>> occuring while the VM is stopped. Their previous state is saved. The
>> XIVE controller is then synced through KVM to flush any in-flight
>> event notification and to stabilize the EQs. At this stage, the EQ
>> pages are marked dirty to make sure the EQ pages are transferred if a
>> migration sequence is in progress.
>>
>> The previous configuration of the sources is restored when the VM
>> resumes, after a migration or a stop. If an interrupt was queued while
>> the VM was stopped, the handler simply generates the missing trigger.
>>
>> Signed-off-by: Cédric Le Goater <address@hidden>
>> Reviewed-by: David Gibson <address@hidden>
>> Message-Id: <address@hidden>
>> Signed-off-by: David Gibson <address@hidden>
>
> This one breaks my nvlink2 passthru setup. The host is v5.2-rc2.
> v5.2-rc3 fixes it though so it is backward compatibility issue which we
> care about to what degree here?
v5.2-rc2 had an ugly bug impacting passthru under some VM configuration,
XIVE + single CPU. See :
bcaa3110d584 ("KVM: PPC: Book3S HV: XIVE: Fix page offset when clearing
ESB pages")
passthru also had a serious issue impacting the XICS-over-XIVE and the
XIVE KVM devices :
ef9740204051 ("KVM: PPC: Book3S HV: XIVE: Do not clear IRQ data of
passthrough interrupts")
You need an v5.2-rc3 !
> I am forcing ic-mode=xive which is not the default so I am not so sure.
It should be OK.
C.
>
>
>
> address@hidden:~$ cat /proc/interrupts
> CPU0
> 16: 0 XIVE-IPI 0 Edge IPI
> 21: 0 XIVE-IRQ 4096 Edge RAS_EPOW
> 22: 0 XIVE-IRQ 4097 Edge RAS_HOTPLUG
> 257: 12372 XIVE-IRQ 4353 Edge ibmvscsi
> 258: 0 XIVE-IRQ 4864 Edge virtio0-config
> 259: 2157 XIVE-IRQ 4865 Edge virtio0-input.0
> 260: 1 XIVE-IRQ 4866 Edge virtio0-output.0
> 261: 0 XIVE-IRQ 4868 Edge xhci_hcd
> 262: 0 XIVE-IRQ 4869 Edge xhci_hcd
> 272: 1 XIVE-IRQ 4368 Edge hvc_console
> LOC: 10508 Local timer interrupts for timer event device
> BCT: 0 Broadcast timer interrupts for timer event device
> LOC: 0 Local timer interrupts for others
> SPU: 5 Spurious interrupts
> PMI: 0 Performance monitoring interrupts
> MCE: 0 Machine check exceptions
> NMI: 0 System Reset interrupts
> DBL: 0 Doorbell interrupts
>
>
> and 7bfc759c02b8 "spapr/xive: add state synchronization with KVM" works:
>
> CPU0
> 16: 0 XIVE-IPI 0 Edge IPI
> 19: 0 XIVE-IRQ 4610 Level NPU Device
> 20: 0 XIVE-IRQ 4611 Level NPU Device
> 21: 0 XIVE-IRQ 4096 Edge RAS_EPOW
> 22: 0 XIVE-IRQ 4097 Edge RAS_HOTPLUG
> 257: 11833 XIVE-IRQ 4353 Edge ibmvscsi
> 258: 0 XIVE-IRQ 4864 Edge virtio0-config
> 259: 1632 XIVE-IRQ 4865 Edge virtio0-input.0
> 260: 1 XIVE-IRQ 4866 Edge virtio0-output.0
> 261: 0 XIVE-IRQ 4868 Edge xhci_hcd
> 262: 0 XIVE-IRQ 4869 Edge xhci_hcd
> 263: 60 XIVE-IRQ 4867 Edge nvidia
> 272: 0 XIVE-IRQ 4368 Edge hvc_console
> LOC: 2236 Local timer interrupts for timer event device
> BCT: 0 Broadcast timer interrupts for timer event device
> LOC: 0 Local timer interrupts for others
> SPU: 2 Spurious interrupts
> PMI: 0 Performance monitoring interrupts
> MCE: 0 Machine check exceptions
> NMI: 0 System Reset interrupts
> DBL: 0 Doorbell interrupts
>
>
>
> Here is the command line:
>
> /home/aik/pbuild/qemu-aikrhel74alt-ppc64/ppc64-softmmu/qemu-system-ppc64 \
> -nodefaults \
> -chardev stdio,id=STDIO0,signal=off,mux=on \
> -device spapr-vty,id=svty0,reg=0x71000110,chardev=STDIO0 \
> -mon id=MON0,chardev=STDIO0,mode=readline -nographic -vga none \
> -enable-kvm \
> -device nec-usb-xhci,id=nec-usb-xhci0 -m 16G \
> -netdev "user,id=USER0,hostfwd=tcp::2223-:22" \
> -device "virtio-net-pci,id=vnet0,mac=C0:41:49:4b:00:00,netdev=USER0" \
> img/u1804-64G-cuda10.1-418.67-swiotlb.qcow2 \
> -machine pseries,cap-cfpc=broken,cap-htm=off,ic-mode=xive \
> -device "vfio-pci,id=vfio0006_00_00_1,host=0006:00:00.1" \
> -device "vfio-pci,id=vfio0006_00_00_0,host=0006:00:00.0" \
> -device "vfio-pci,id=vfio0004_04_00_0,host=0004:04:00.0" \
> -kernel ./vmldbg -append "root=/dev/sda2 console=hvc0 debug loglevel=8" \
> -snapshot \
> -smp 1,threads=1 -bios ./slof.bin \
> -L /home/aik/t/qemu-ppc64-bios/ \
> -trace events=qemu_trace_events -d guest_errors \
> -chardev socket,id=SOCKET0,server,nowait,path=qemu.mon.user2223.6_0_0_1 \
> -mon chardev=SOCKET0,mode=control
>
>
>
>
>> ---
>> hw/intc/spapr_xive_kvm.c | 96 ++++++++++++++++++++++++++++++++++++-
>> include/hw/ppc/spapr_xive.h | 1 +
>> 2 files changed, 96 insertions(+), 1 deletion(-)
>>
>> diff --git a/hw/intc/spapr_xive_kvm.c b/hw/intc/spapr_xive_kvm.c
>> index 8dd4f96e0b..735577a6f8 100644
>> --- a/hw/intc/spapr_xive_kvm.c
>> +++ b/hw/intc/spapr_xive_kvm.c
>> @@ -433,9 +433,100 @@ static void kvmppc_xive_get_queues(SpaprXive *xive,
>> Error **errp)
>> }
>> }
>>
>> +/*
>> + * The primary goal of the XIVE VM change handler is to mark the EQ
>> + * pages dirty when all XIVE event notifications have stopped.
>> + *
>> + * Whenever the VM is stopped, the VM change handler sets the source
>> + * PQs to PENDING to stop the flow of events and to possibly catch a
>> + * triggered interrupt occuring while the VM is stopped. The previous
>> + * state is saved in anticipation of a migration. The XIVE controller
>> + * is then synced through KVM to flush any in-flight event
>> + * notification and stabilize the EQs.
>> + *
>> + * At this stage, we can mark the EQ page dirty and let a migration
>> + * sequence transfer the EQ pages to the destination, which is done
>> + * just after the stop state.
>> + *
>> + * The previous configuration of the sources is restored when the VM
>> + * runs again. If an interrupt was queued while the VM was stopped,
>> + * simply generate a trigger.
>> + */
>> +static void kvmppc_xive_change_state_handler(void *opaque, int running,
>> + RunState state)
>> +{
>> + SpaprXive *xive = opaque;
>> + XiveSource *xsrc = &xive->source;
>> + Error *local_err = NULL;
>> + int i;
>> +
>> + /*
>> + * Restore the sources to their initial state. This is called when
>> + * the VM resumes after a stop or a migration.
>> + */
>> + if (running) {
>> + for (i = 0; i < xsrc->nr_irqs; i++) {
>> + uint8_t pq = xive_source_esb_get(xsrc, i);
>> + uint8_t old_pq;
>> +
>> + old_pq = xive_esb_read(xsrc, i, XIVE_ESB_SET_PQ_00 + (pq << 8));
>> +
>> + /*
>> + * An interrupt was queued while the VM was stopped,
>> + * generate a trigger.
>> + */
>> + if (pq == XIVE_ESB_RESET && old_pq == XIVE_ESB_QUEUED) {
>> + xive_esb_trigger(xsrc, i);
>> + }
>> + }
>> +
>> + return;
>> + }
>> +
>> + /*
>> + * Mask the sources, to stop the flow of event notifications, and
>> + * save the PQs locally in the XiveSource object. The XiveSource
>> + * state will be collected later on by its vmstate handler if a
>> + * migration is in progress.
>> + */
>> + for (i = 0; i < xsrc->nr_irqs; i++) {
>> + uint8_t pq = xive_esb_read(xsrc, i, XIVE_ESB_GET);
>> +
>> + /*
>> + * PQ is set to PENDING to possibly catch a triggered
>> + * interrupt occuring while the VM is stopped (hotplug event
>> + * for instance) .
>> + */
>> + if (pq != XIVE_ESB_OFF) {
>> + pq = xive_esb_read(xsrc, i, XIVE_ESB_SET_PQ_10);
>> + }
>> + xive_source_esb_set(xsrc, i, pq);
>> + }
>> +
>> + /*
>> + * Sync the XIVE controller in KVM, to flush in-flight event
>> + * notification that should be enqueued in the EQs and mark the
>> + * XIVE EQ pages dirty to collect all updates.
>> + */
>> + kvm_device_access(xive->fd, KVM_DEV_XIVE_GRP_CTRL,
>> + KVM_DEV_XIVE_EQ_SYNC, NULL, true, &local_err);
>> + if (local_err) {
>> + error_report_err(local_err);
>> + return;
>> + }
>> +}
>> +
>> void kvmppc_xive_synchronize_state(SpaprXive *xive, Error **errp)
>> {
>> - kvmppc_xive_source_get_state(&xive->source);
>> + /*
>> + * When the VM is stopped, the sources are masked and the previous
>> + * state is saved in anticipation of a migration. We should not
>> + * synchronize the source state in that case else we will override
>> + * the saved state.
>> + */
>> + if (runstate_is_running()) {
>> + kvmppc_xive_source_get_state(&xive->source);
>> + }
>>
>> /* EAT: there is no extra state to query from KVM */
>>
>> @@ -515,6 +606,9 @@ void kvmppc_xive_connect(SpaprXive *xive, Error **errp)
>> "xive.tima", tima_len, xive->tm_mmap);
>> sysbus_init_mmio(SYS_BUS_DEVICE(xive), &xive->tm_mmio);
>>
>> + xive->change = qemu_add_vm_change_state_handler(
>> + kvmppc_xive_change_state_handler, xive);
>> +
>> kvm_kernel_irqchip = true;
>> kvm_msi_via_irqfd_allowed = true;
>> kvm_gsi_direct_mapping = true;
>> diff --git a/include/hw/ppc/spapr_xive.h b/include/hw/ppc/spapr_xive.h
>> index 7e49badd8c..734662c12a 100644
>> --- a/include/hw/ppc/spapr_xive.h
>> +++ b/include/hw/ppc/spapr_xive.h
>> @@ -42,6 +42,7 @@ typedef struct SpaprXive {
>> /* KVM support */
>> int fd;
>> void *tm_mmap;
>> + VMChangeStateEntry *change;
>> } SpaprXive;
>>
>> /*
>>
>