Re: [RFC v9 15/29] vfio: Set up nested stage mappings

[Auger Eric <eric.auger@redhat.com>]

Hi Kunkun,

On 4/15/21 4:03 AM, Kunkun Jiang wrote:
> Hi Eric,
> 
> On 2021/4/14 16:05, Auger Eric wrote:
>> Hi Kunkun,
>>
>> On 4/14/21 3:45 AM, Kunkun Jiang wrote:
>>> On 2021/4/13 20:57, Auger Eric wrote:
>>>> Hi Kunkun,
>>>>
>>>> On 4/13/21 2:10 PM, Kunkun Jiang wrote:
>>>>> Hi Eric,
>>>>>
>>>>> On 2021/4/11 20:08, Eric Auger wrote:
>>>>>> In nested mode, legacy vfio_iommu_map_notify cannot be used as
>>>>>> there is no "caching" mode and we do not trap on map.
>>>>>>
>>>>>> On Intel, vfio_iommu_map_notify was used to DMA map the RAM
>>>>>> through the host single stage.
>>>>>>
>>>>>> With nested mode, we need to setup the stage 2 and the stage 1
>>>>>> separately. This patch introduces a prereg_listener to setup
>>>>>> the stage 2 mapping.
>>>>>>
>>>>>> The stage 1 mapping, owned by the guest, is passed to the host
>>>>>> when the guest invalidates the stage 1 configuration, through
>>>>>> a dedicated PCIPASIDOps callback. Guest IOTLB invalidations
>>>>>> are cascaded downto the host through another IOMMU MR UNMAP
>>>>>> notifier.
>>>>>>
>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>>
>>>>>> ---
>>>>>>
>>>>>> v7 -> v8:
>>>>>> - properly handle new IOMMUTLBEntry fields and especially
>>>>>>      propagate DOMAIN and PASID based invalidations
>>>>>>
>>>>>> v6 -> v7:
>>>>>> - remove PASID based invalidation
>>>>>>
>>>>>> v5 -> v6:
>>>>>> - add error_report_err()
>>>>>> - remove the abort in case of nested stage case
>>>>>>
>>>>>> v4 -> v5:
>>>>>> - use VFIO_IOMMU_SET_PASID_TABLE
>>>>>> - use PCIPASIDOps for config notification
>>>>>>
>>>>>> v3 -> v4:
>>>>>> - use iommu_inv_pasid_info for ASID invalidation
>>>>>>
>>>>>> v2 -> v3:
>>>>>> - use VFIO_IOMMU_ATTACH_PASID_TABLE
>>>>>> - new user API
>>>>>> - handle leaf
>>>>>>
>>>>>> v1 -> v2:
>>>>>> - adapt to uapi changes
>>>>>> - pass the asid
>>>>>> - pass IOMMU_NOTIFIER_S1_CFG when initializing the config notifier
>>>>>> ---
>>>>>>     hw/vfio/common.c     | 139
>>>>>> +++++++++++++++++++++++++++++++++++++++++--
>>>>>>     hw/vfio/pci.c        |  21 +++++++
>>>>>>     hw/vfio/trace-events |   2 +
>>>>>>     3 files changed, 157 insertions(+), 5 deletions(-)
>>>>>>
>>>>>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>>>>>> index 0cd7ef2139..e369d451e7 100644
>>>>>> --- a/hw/vfio/common.c
>>>>>> +++ b/hw/vfio/common.c
>>>>>> @@ -595,6 +595,73 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry
>>>>>> *iotlb, void **vaddr,
>>>>>>         return true;
>>>>>>     }
>>>>>>     +/* Propagate a guest IOTLB invalidation to the host (nested
>>>>>> mode) */
>>>>>> +static void vfio_iommu_unmap_notify(IOMMUNotifier *n, IOMMUTLBEntry
>>>>>> *iotlb)
>>>>>> +{
>>>>>> +    VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
>>>>>> +    struct vfio_iommu_type1_cache_invalidate ustruct = {};
>>>>>> +    VFIOContainer *container = giommu->container;
>>>>>> +    int ret;
>>>>>> +
>>>>>> +    assert(iotlb->perm == IOMMU_NONE);
>>>>>> +
>>>>>> +    ustruct.argsz = sizeof(ustruct);
>>>>>> +    ustruct.flags = 0;
>>>>>> +    ustruct.info.argsz = sizeof(struct iommu_cache_invalidate_info);
>>>>>> +    ustruct.info.version = IOMMU_CACHE_INVALIDATE_INFO_VERSION_1;
>>>>>> +    ustruct.info.cache = IOMMU_CACHE_INV_TYPE_IOTLB;
>>>>>> +
>>>>>> +    switch (iotlb->granularity) {
>>>>>> +    case IOMMU_INV_GRAN_DOMAIN:
>>>>>> +        ustruct.info.granularity = IOMMU_INV_GRANU_DOMAIN;
>>>>>> +        break;
>>>>>> +    case IOMMU_INV_GRAN_PASID:
>>>>>> +    {
>>>>>> +        struct iommu_inv_pasid_info *pasid_info;
>>>>>> +        int archid = -1;
>>>>>> +
>>>>>> +        pasid_info = &ustruct.info.granu.pasid_info;
>>>>>> +        ustruct.info.granularity = IOMMU_INV_GRANU_PASID;
>>>>>> +        if (iotlb->flags & IOMMU_INV_FLAGS_ARCHID) {
>>>>>> +            pasid_info->flags |= IOMMU_INV_ADDR_FLAGS_ARCHID;
>>>>>> +            archid = iotlb->arch_id;
>>>>>> +        }
>>>>>> +        pasid_info->archid = archid;
>>>>>> +        trace_vfio_iommu_asid_inv_iotlb(archid);
>>>>>> +        break;
>>>>>> +    }
>>>>>> +    case IOMMU_INV_GRAN_ADDR:
>>>>>> +    {
>>>>>> +        hwaddr start = iotlb->iova + giommu->iommu_offset;
>>>>>> +        struct iommu_inv_addr_info *addr_info;
>>>>>> +        size_t size = iotlb->addr_mask + 1;
>>>>>> +        int archid = -1;
>>>>>> +
>>>>>> +        addr_info = &ustruct.info.granu.addr_info;
>>>>>> +        ustruct.info.granularity = IOMMU_INV_GRANU_ADDR;
>>>>>> +        if (iotlb->leaf) {
>>>>>> +            addr_info->flags |= IOMMU_INV_ADDR_FLAGS_LEAF;
>>>>>> +        }
>>>>>> +        if (iotlb->flags & IOMMU_INV_FLAGS_ARCHID) {
>>>>>> +            addr_info->flags |= IOMMU_INV_ADDR_FLAGS_ARCHID;
>>>>>> +            archid = iotlb->arch_id;
>>>>>> +        }
>>>>>> +        addr_info->archid = archid;
>>>>>> +        addr_info->addr = start;
>>>>>> +        addr_info->granule_size = size;
>>>>>> +        addr_info->nb_granules = 1;
>>>>>> +        trace_vfio_iommu_addr_inv_iotlb(archid, start, size,
>>>>>> +                                        1, iotlb->leaf);
>>>>>> +        break;
>>>>>> +    }
>>>>> Should we pass a size to  host kernel here, even if vSMMU doesn't
>>>>> support
>>>>> RIL or guest kernel doesn't use RIL?
>>>>>
>>>>> It will cause TLBI issue in  this scenario: Guest kernel issues a
>>>>> TLBI cmd
>>>>> without "range" (tg = 0) to invalidate a 2M huge page. Then qemu
>>>>> passed
>>>>> the iova and size (4K) to host kernel. Finally, host kernel issues a
>>>>> TLBI cmd
>>>>> with "range" (4K) which can not invalidate the TLB entry of 2M huge
>>>>> page.
>>>>> (pSMMU supports RIL)
>>>> In that case the guest will loop over all 4K images belonging to the 2M
>>>> huge page and invalidate each of them. This should turn into qemu
>>>> notifications for each 4kB page, no? This is totally inefficient, hence
>>> The guest will not loop over all 4K images belonging to the 2M huge
>>> page.
>>> The iommu_iotlb_gather->pgsize will be 2M, if a page is 2M huge page.
>>> The
>>> gather->pgsize will be passed to __arm_smmu_tlb_inv_range as "granule":
>>>
>>> iommu_iotlb_gather_add_page
>>>      iommu_iotlb_sync
>>>          domain->ops->iotlb_sync
>>>              arm_smmu_iotlb_sync
>>>                  arm_smmu_tlb_inv_range_domain
>>>                      __arm_smmu_tlb_inv_range
>>>
>>> In the above mentioned scenario, guest kernel will issue a TLBI cmd only
>>> with
>>> "iova" (tg = 0).
>> OK I get it now. In that case I think I should do a TG=0 invalidation
>> on host even if the host does support RIL. Does that sound correct?
> Yeah, that's what I thought.
>> The trouble is the uapi struct does not convey such info explicitly.
>> Maybe I should use nb_granules = 0 to detect that case.
> It is troublesome to correct this issue. Using nb_granules = 0 may be
> a good choice.
>> I think for this use case RIL should be supported by the guest. Thoughts?
> Yes. If guest supports RIL, the scenario mentioned above does not exist.

After further study I really wonder if it is worth supporting the case
where the guest does not use range inval. Passing a nb_granules = 0 (~
size) would be OK to perform the TG=0 range invalidation on the host.
however host arm_smmu_tlb_inv_range_domain() then calls
arm_smmu_atc_inv_domain(smmu_domain, 0, iova, size) which needs a size.
We would need to trap guest CMD_ATC_INV and use different code paths on
host smmu driver to cascade the various cache invalidations. At the
moment, without maintainer giodance, I am a bit reluctant to add this
extra complexity.

So I would be inclined to fail in QEMU if we detect TG=0 is being used
by the guest. Recent guest kernels would be a prerequisite for this use
case. What do you think?

Thanks

Eric
> 
> Thanks,
> Kunkun Jiang
>>
>> Thanks
>>
>> Eric
>>> Thanks,
>>> Kunkun Jiang
>>>> the support of RIL on guest side and QEMU device.
>>>>
>>>> What do I miss?
>>>>
>>>> Thanks
>>>>
>>>> Eric
>>>>> Thanks,
>>>>> Kunkun Jiang
>>>>>> +    }
>>>>>> +
>>>>>> +    ret = ioctl(container->fd, VFIO_IOMMU_CACHE_INVALIDATE,
>>>>>> &ustruct);
>>>>>> +    if (ret) {
>>>>>> +        error_report("%p: failed to invalidate CACHE (%d)",
>>>>>> container, ret);
>>>>>> +    }
>>>>>> +}
>>>>>> +
>>>>>>     static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry
>>>>>> *iotlb)
>>>>>>     {
>>>>>>         VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
>>>>>> @@ -776,6 +843,35 @@ static void
>>>>>> vfio_dma_unmap_ram_section(VFIOContainer *container,
>>>>>>         }
>>>>>>     }
>>>>>>     +static void vfio_prereg_listener_region_add(MemoryListener
>>>>>> *listener,
>>>>>> +                                            MemoryRegionSection
>>>>>> *section)
>>>>>> +{
>>>>>> +    VFIOContainer *container =
>>>>>> +        container_of(listener, VFIOContainer, prereg_listener);
>>>>>> +    Error *err = NULL;
>>>>>> +
>>>>>> +    if (!memory_region_is_ram(section->mr)) {
>>>>>> +        return;
>>>>>> +    }
>>>>>> +
>>>>>> +    vfio_dma_map_ram_section(container, section, &err);
>>>>>> +    if (err) {
>>>>>> +        error_report_err(err);
>>>>>> +    }
>>>>>> +}
>>>>>> +static void vfio_prereg_listener_region_del(MemoryListener
>>>>>> *listener,
>>>>>> +                                     MemoryRegionSection *section)
>>>>>> +{
>>>>>> +    VFIOContainer *container =
>>>>>> +        container_of(listener, VFIOContainer, prereg_listener);
>>>>>> +
>>>>>> +    if (!memory_region_is_ram(section->mr)) {
>>>>>> +        return;
>>>>>> +    }
>>>>>> +
>>>>>> +    vfio_dma_unmap_ram_section(container, section);
>>>>>> +}
>>>>>> +
>>>>>>     static void vfio_listener_region_add(MemoryListener *listener,
>>>>>>                                          MemoryRegionSection
>>>>>> *section)
>>>>>>     {
>>>>>> @@ -879,9 +975,10 @@ static void
>>>>>> vfio_listener_region_add(MemoryListener *listener,
>>>>>>         memory_region_ref(section->mr);
>>>>>>           if (memory_region_is_iommu(section->mr)) {
>>>>>> +        IOMMUNotify notify;
>>>>>>             VFIOGuestIOMMU *giommu;
>>>>>>             IOMMUMemoryRegion *iommu_mr =
>>>>>> IOMMU_MEMORY_REGION(section->mr);
>>>>>> -        int iommu_idx;
>>>>>> +        int iommu_idx, flags;
>>>>>>               trace_vfio_listener_region_add_iommu(iova, end);
>>>>>>             /*
>>>>>> @@ -900,8 +997,18 @@ static void
>>>>>> vfio_listener_region_add(MemoryListener *listener,
>>>>>>             llend = int128_sub(llend, int128_one());
>>>>>>             iommu_idx = memory_region_iommu_attrs_to_index(iommu_mr,
>>>>>>                                                         
>>>>>> MEMTXATTRS_UNSPECIFIED);
>>>>>> -        iommu_notifier_init(&giommu->n, vfio_iommu_map_notify,
>>>>>> -                            IOMMU_NOTIFIER_IOTLB_EVENTS,
>>>>>> +
>>>>>> +        if (container->iommu_type == VFIO_TYPE1_NESTING_IOMMU) {
>>>>>> +            /* IOTLB unmap notifier to propagate guest IOTLB
>>>>>> invalidations */
>>>>>> +            flags = IOMMU_NOTIFIER_UNMAP;
>>>>>> +            notify = vfio_iommu_unmap_notify;
>>>>>> +        } else {
>>>>>> +            /* MAP/UNMAP IOTLB notifier */
>>>>>> +            flags = IOMMU_NOTIFIER_IOTLB_EVENTS;
>>>>>> +            notify = vfio_iommu_map_notify;
>>>>>> +        }
>>>>>> +
>>>>>> +        iommu_notifier_init(&giommu->n, notify, flags,
>>>>>>                                 section->offset_within_region,
>>>>>>                                 int128_get64(llend),
>>>>>>                                 iommu_idx);
>>>>>> @@ -921,7 +1028,9 @@ static void
>>>>>> vfio_listener_region_add(MemoryListener *listener,
>>>>>>                 goto fail;
>>>>>>             }
>>>>>>             QLIST_INSERT_HEAD(&container->giommu_list, giommu,
>>>>>> giommu_next);
>>>>>> -        memory_region_iommu_replay(giommu->iommu, &giommu->n);
>>>>>> +        if (flags & IOMMU_NOTIFIER_MAP) {
>>>>>> +            memory_region_iommu_replay(giommu->iommu, &giommu->n);
>>>>>> +        }
>>>>>>               return;
>>>>>>         }
>>>>>> @@ -1205,10 +1314,16 @@ static const MemoryListener
>>>>>> vfio_memory_listener = {
>>>>>>         .log_sync = vfio_listener_log_sync,
>>>>>>     };
>>>>>>     +static MemoryListener vfio_memory_prereg_listener = {
>>>>>> +    .region_add = vfio_prereg_listener_region_add,
>>>>>> +    .region_del = vfio_prereg_listener_region_del,
>>>>>> +};
>>>>>> +
>>>>>>     static void vfio_listener_release(VFIOContainer *container)
>>>>>>     {
>>>>>>         memory_listener_unregister(&container->listener);
>>>>>> -    if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
>>>>>> +    if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
>>>>>> +        container->iommu_type == VFIO_TYPE1_NESTING_IOMMU) {
>>>>>>             memory_listener_unregister(&container->prereg_listener);
>>>>>>         }
>>>>>>     }
>>>>>> @@ -1858,6 +1973,20 @@ static int vfio_connect_container(VFIOGroup
>>>>>> *group, AddressSpace *as,
>>>>>>                 vfio_get_iommu_info_migration(container, info);
>>>>>>             }
>>>>>>             g_free(info);
>>>>>> +
>>>>>> +        if (container->iommu_type == VFIO_TYPE1_NESTING_IOMMU) {
>>>>>> +            container->prereg_listener =
>>>>>> vfio_memory_prereg_listener;
>>>>>> +            memory_listener_register(&container->prereg_listener,
>>>>>> +                                     &address_space_memory);
>>>>>> +            if (container->error) {
>>>>>> +
>>>>>> memory_listener_unregister(&container->prereg_listener);
>>>>>> +                ret = -1;
>>>>>> +                error_propagate_prepend(errp, container->error,
>>>>>> +                                    "RAM memory listener
>>>>>> initialization failed "
>>>>>> +                                    "for container");
>>>>>> +                goto free_container_exit;
>>>>>> +            }
>>>>>> +        }
>>>>>>             break;
>>>>>>         }
>>>>>>         case VFIO_SPAPR_TCE_v2_IOMMU:
>>>>>> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
>>>>>> index 5c65aa0a98..cad7deec71 100644
>>>>>> --- a/hw/vfio/pci.c
>>>>>> +++ b/hw/vfio/pci.c
>>>>>> @@ -2773,6 +2773,25 @@ static void
>>>>>> vfio_unregister_req_notifier(VFIOPCIDevice *vdev)
>>>>>>         vdev->req_enabled = false;
>>>>>>     }
>>>>>>     +static int vfio_iommu_set_pasid_table(PCIBus *bus, int32_t
>>>>>> devfn,
>>>>>> +                                      IOMMUConfig *config)
>>>>>> +{
>>>>>> +    PCIDevice *pdev = bus->devices[devfn];
>>>>>> +    VFIOPCIDevice *vdev = DO_UPCAST(VFIOPCIDevice, pdev, pdev);
>>>>>> +    VFIOContainer *container = vdev->vbasedev.group->container;
>>>>>> +    struct vfio_iommu_type1_set_pasid_table info;
>>>>>> +
>>>>>> +    info.argsz = sizeof(info);
>>>>>> +    info.flags = VFIO_PASID_TABLE_FLAG_SET;
>>>>>> +    memcpy(&info.config, &config->pasid_cfg,
>>>>>> sizeof(config->pasid_cfg));
>>>>>> +
>>>>>> +    return ioctl(container->fd, VFIO_IOMMU_SET_PASID_TABLE, &info);
>>>>>> +}
>>>>>> +
>>>>>> +static PCIPASIDOps vfio_pci_pasid_ops = {
>>>>>> +    .set_pasid_table = vfio_iommu_set_pasid_table,
>>>>>> +};
>>>>>> +
>>>>>>     static void vfio_realize(PCIDevice *pdev, Error **errp)
>>>>>>     {
>>>>>>         VFIOPCIDevice *vdev = VFIO_PCI(pdev);
>>>>>> @@ -3084,6 +3103,8 @@ static void vfio_realize(PCIDevice *pdev, Error
>>>>>> **errp)
>>>>>>         vfio_register_req_notifier(vdev);
>>>>>>         vfio_setup_resetfn_quirk(vdev);
>>>>>>     +    pci_setup_pasid_ops(pdev, &vfio_pci_pasid_ops);
>>>>>> +
>>>>>>         return;
>>>>>>       out_deregister:
>>>>>> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
>>>>>> index 936d29d150..43696afc15 100644
>>>>>> --- a/hw/vfio/trace-events
>>>>>> +++ b/hw/vfio/trace-events
>>>>>> @@ -120,6 +120,8 @@ vfio_region_sparse_mmap_header(const char *name,
>>>>>> int index, int nr_areas) "Devic
>>>>>>     vfio_region_sparse_mmap_entry(int i, unsigned long start,
>>>>>> unsigned
>>>>>> long end) "sparse entry %d [0x%lx - 0x%lx]"
>>>>>>     vfio_get_dev_region(const char *name, int index, uint32_t type,
>>>>>> uint32_t subtype) "%s index %d, %08x/%0x8"
>>>>>>     vfio_dma_unmap_overflow_workaround(void) ""
>>>>>> +vfio_iommu_addr_inv_iotlb(int asid, uint64_t addr, uint64_t size,
>>>>>> uint64_t nb_granules, bool leaf) "nested IOTLB invalidate asid=%d,
>>>>>> addr=0x%"PRIx64" granule_size=0x%"PRIx64" nb_granules=0x%"PRIx64"
>>>>>> leaf=%d"
>>>>>> +vfio_iommu_asid_inv_iotlb(int asid) "nested IOTLB invalidate
>>>>>> asid=%d"
>>>>>>       # platform.c
>>>>>>     vfio_platform_base_device_init(char *name, int groupid) "%s
>>>>>> belongs
>>>>>> to group #%d"
>>>>>
>>>> .
>>>
>> .
> 
>