Re: [PATCH v7 13/23] intel_iommu: Bind/unbind guest page table to host

["Cédric Le Goater" <clg@redhat.com>]

On 10/24/25 10:43, Zhenzhong Duan wrote:
> This captures the guest PASID table entry modifications and propagates
> the changes to host to attach a hwpt with type determined per guest IOMMU
> PGTT configuration.
> 
> When PGTT=PT, attach PASID_0 to a second stage HWPT(GPA->HPA).
> When PGTT=FST, attach PASID_0 to nested HWPT with nesting parent HWPT
> coming from VFIO.
> 
> Co-Authored-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
>   include/hw/i386/intel_iommu.h |   1 +
>   hw/i386/intel_iommu.c         | 150 +++++++++++++++++++++++++++++++++-
>   hw/i386/trace-events          |   3 +
>   3 files changed, 151 insertions(+), 3 deletions(-)
> 
> diff --git a/include/hw/i386/intel_iommu.h b/include/hw/i386/intel_iommu.h
> index 3758ac239c..b5f8a9fc29 100644
> --- a/include/hw/i386/intel_iommu.h
> +++ b/include/hw/i386/intel_iommu.h
> @@ -104,6 +104,7 @@ struct VTDAddressSpace {
>       PCIBus *bus;
>       uint8_t devfn;
>       uint32_t pasid;
> +    uint32_t fs_hwpt;
>       AddressSpace as;
>       IOMMUMemoryRegion iommu;
>       MemoryRegion root;          /* The root container of the device */
> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
> index 871e6aad19..3789a36147 100644
> --- a/hw/i386/intel_iommu.c
> +++ b/hw/i386/intel_iommu.c
> @@ -20,6 +20,7 @@
>    */
>   
>   #include "qemu/osdep.h"
> +#include CONFIG_DEVICES /* CONFIG_IOMMUFD */
>   #include "qemu/error-report.h"
>   #include "qemu/main-loop.h"
>   #include "qapi/error.h"
> @@ -42,6 +43,9 @@
>   #include "migration/vmstate.h"
>   #include "trace.h"
>   #include "system/iommufd.h"
> +#ifdef CONFIG_IOMMUFD
> +#include <linux/iommufd.h>
> +#endif
>   
>   /* context entry operations */
>   #define PASID_0    0
> @@ -87,6 +91,7 @@ struct vtd_iotlb_key {
>   
>   static void vtd_address_space_refresh_all(IntelIOMMUState *s);
>   static void vtd_address_space_unmap(VTDAddressSpace *as, IOMMUNotifier *n);
> +static int vtd_bind_guest_pasid(VTDAddressSpace *vtd_as, Error **errp);
>   
>   static void vtd_pasid_cache_reset_locked(IntelIOMMUState *s)
>   {
> @@ -98,7 +103,11 @@ static void vtd_pasid_cache_reset_locked(IntelIOMMUState *s)
>       g_hash_table_iter_init(&as_it, s->vtd_address_spaces);
>       while (g_hash_table_iter_next(&as_it, NULL, (void **)&vtd_as)) {
>           VTDPASIDCacheEntry *pc_entry = &vtd_as->pasid_cache_entry;
> -        pc_entry->valid = false;
> +        if (pc_entry->valid) {
> +            pc_entry->valid = false;
> +            /* It's fatal to get failure during reset */
> +            vtd_bind_guest_pasid(vtd_as, &error_fatal);
> +        }
>       }
>   }
>   
> @@ -2380,6 +2389,128 @@ static void vtd_context_global_invalidate(IntelIOMMUState *s)
>       vtd_iommu_replay_all(s);
>   }
>   
> +#ifdef CONFIG_IOMMUFD
> +static int vtd_create_fs_hwpt(HostIOMMUDeviceIOMMUFD *idev,
> +                              VTDPASIDEntry *pe, uint32_t *fs_hwpt,
> +                              Error **errp)

Returning a bool is better. Same for the routines below.

> +{
> +    struct iommu_hwpt_vtd_s1 vtd = {};
> +
> +    vtd.flags = (VTD_SM_PASID_ENTRY_SRE_BIT(pe) ? IOMMU_VTD_S1_SRE : 0) |
> +                (VTD_SM_PASID_ENTRY_WPE_BIT(pe) ? IOMMU_VTD_S1_WPE : 0) |
> +                (VTD_SM_PASID_ENTRY_EAFE_BIT(pe) ? IOMMU_VTD_S1_EAFE : 0);
> +    vtd.addr_width = vtd_pe_get_fs_aw(pe);
> +    vtd.pgtbl_addr = (uint64_t)vtd_pe_get_fspt_base(pe);
> +
> +    return !iommufd_backend_alloc_hwpt(idev->iommufd, idev->devid,
> +                                       idev->hwpt_id, 0, IOMMU_HWPT_DATA_VTD_S1,
> +                                       sizeof(vtd), &vtd, fs_hwpt, errp);
> +}
> +
> +static void vtd_destroy_old_fs_hwpt(HostIOMMUDeviceIOMMUFD *idev,
> +                                    VTDAddressSpace *vtd_as)
> +{
> +    if (!vtd_as->fs_hwpt) {
> +        return;
> +    }
> +    iommufd_backend_free_id(idev->iommufd, vtd_as->fs_hwpt);
> +    vtd_as->fs_hwpt = 0;
> +}
> +
> +static int vtd_device_attach_iommufd(VTDHostIOMMUDevice *vtd_hiod,
> +                                     VTDAddressSpace *vtd_as, Error **errp)
> +{
> +    HostIOMMUDeviceIOMMUFD *idev = HOST_IOMMU_DEVICE_IOMMUFD(vtd_hiod->hiod);
> +    VTDPASIDEntry *pe = &vtd_as->pasid_cache_entry.pasid_entry;
> +    uint32_t hwpt_id;
> +    bool ret;
> +
> +    /*
> +     * We can get here only if flts=on, the supported PGTT is FST and PT.
> +     * Catch invalid PGTT when processing invalidation request to avoid
> +     * attaching to wrong hwpt.
> +     */
> +    if (!vtd_pe_pgtt_is_fst(pe) && !vtd_pe_pgtt_is_pt(pe)) {
> +        error_setg(errp, "Invalid PGTT type");
> +        return -EINVAL;
> +    }
> +
> +    if (vtd_pe_pgtt_is_pt(pe)) {
> +        hwpt_id = idev->hwpt_id;
> +    } else if (vtd_create_fs_hwpt(idev, pe, &hwpt_id, errp)) {
> +        return -EINVAL;
> +    }
> +
> +    ret = host_iommu_device_iommufd_attach_hwpt(idev, hwpt_id, errp);
> +    trace_vtd_device_attach_hwpt(idev->devid, vtd_as->pasid, hwpt_id, !ret);
> +    if (ret) {
> +        /* Destroy old fs_hwpt if it's a replacement */
> +        vtd_destroy_old_fs_hwpt(idev, vtd_as);
> +        if (vtd_pe_pgtt_is_fst(pe)) {
> +            vtd_as->fs_hwpt = hwpt_id;
> +        }
> +    } else if (vtd_pe_pgtt_is_fst(pe)) {
> +        iommufd_backend_free_id(idev->iommufd, hwpt_id);
> +    }
> +
> +    return !ret;
> +}
> +
> +static int vtd_device_detach_iommufd(VTDHostIOMMUDevice *vtd_hiod,
> +                                     VTDAddressSpace *vtd_as, Error **errp)
> +{
> +    HostIOMMUDeviceIOMMUFD *idev = HOST_IOMMU_DEVICE_IOMMUFD(vtd_hiod->hiod);
> +    IntelIOMMUState *s = vtd_as->iommu_state;
> +    uint32_t pasid = vtd_as->pasid;
> +    bool ret;
> +
> +    if (s->dmar_enabled && s->root_scalable) {
> +        ret = host_iommu_device_iommufd_detach_hwpt(idev, errp);
> +        trace_vtd_device_detach_hwpt(idev->devid, pasid, !ret);
> +    } else {
> +        /*
> +         * If DMAR remapping is disabled or guest switches to legacy mode,
> +         * we fallback to the default HWPT which contains shadow page table.
> +         * So guest DMA could still work.
> +         */
> +        ret = host_iommu_device_iommufd_attach_hwpt(idev, idev->hwpt_id, errp);
> +        trace_vtd_device_reattach_def_hwpt(idev->devid, pasid, idev->hwpt_id,
> +                                           !ret);
> +    }
> +
> +    if (ret) {
> +        vtd_destroy_old_fs_hwpt(idev, vtd_as);
> +    }
> +
> +    return !ret;
> +}
> +
> +static int vtd_bind_guest_pasid(VTDAddressSpace *vtd_as, Error **errp)
> +{
> +    VTDPASIDCacheEntry *pc_entry = &vtd_as->pasid_cache_entry;
> +    VTDHostIOMMUDevice *vtd_hiod = vtd_find_hiod_iommufd(vtd_as);
> +    int ret;
> +
> +    /* Ignore emulated device or legacy VFIO backed device */
> +    if (!vtd_hiod) {
> +        return 0;
> +    }
> +
> +    if (pc_entry->valid) {
> +        ret = vtd_device_attach_iommufd(vtd_hiod, vtd_as, errp);
> +    } else {
> +        ret = vtd_device_detach_iommufd(vtd_hiod, vtd_as, errp);
> +    }
> +
> +    return ret;
> +}
> +#else
> +static int vtd_bind_guest_pasid(VTDAddressSpace *vtd_as, Error **errp)
> +{
> +    return 0;
> +}
> +#endif
> +
>   /* Do a context-cache device-selective invalidation.
>    * @func_mask: FM field after shifting
>    */
> @@ -3134,6 +3265,8 @@ static void vtd_pasid_cache_sync_locked(gpointer key, gpointer value,
>       VTDPASIDEntry pe;
>       IOMMUNotifier *n;
>       uint16_t did;
> +    const char *err_prefix;

Setting this prefix looks a bit fragile. May be add a default value here.


Thanks,

C.


> +    Error *local_err = NULL;
>   
>       if (vtd_dev_get_pe_from_pasid(vtd_as, &pe)) {
>           if (!pc_entry->valid) {
> @@ -3154,7 +3287,9 @@ static void vtd_pasid_cache_sync_locked(gpointer key, gpointer value,
>               vtd_address_space_unmap(vtd_as, n);
>           }
>           vtd_switch_address_space(vtd_as);
> -        return;
> +
> +        err_prefix = "Detaching from HWPT failed: ";
> +        goto do_bind_unbind;
>       }
>   
>       /*
> @@ -3182,12 +3317,21 @@ static void vtd_pasid_cache_sync_locked(gpointer key, gpointer value,
>       if (!pc_entry->valid) {
>           pc_entry->pasid_entry = pe;
>           pc_entry->valid = true;
> -    } else if (!vtd_pasid_entry_compare(&pe, &pc_entry->pasid_entry)) {
> +        err_prefix = "Attaching to HWPT failed: ";
> +    } else if (vtd_pasid_entry_compare(&pe, &pc_entry->pasid_entry)) {
> +        err_prefix = "Replacing HWPT attachment failed: ";
> +    } else {
>           return;
>       }
>   
>       vtd_switch_address_space(vtd_as);
>       vtd_address_space_sync(vtd_as);
> +
> +do_bind_unbind:
> +    /* TODO: Fault event injection into guest, report error to QEMU for now */
> +    if (vtd_bind_guest_pasid(vtd_as, &local_err)) {
> +        error_reportf_err(local_err, "%s", err_prefix);
> +    }
>   }
>   
>   static void vtd_pasid_cache_sync(IntelIOMMUState *s, VTDPASIDCacheInfo *pc_info)
> diff --git a/hw/i386/trace-events b/hw/i386/trace-events
> index b704f4f90c..5a3ee1cf64 100644
> --- a/hw/i386/trace-events
> +++ b/hw/i386/trace-events
> @@ -73,6 +73,9 @@ vtd_warn_invalid_qi_tail(uint16_t tail) "tail 0x%"PRIx16
>   vtd_warn_ir_vector(uint16_t sid, int index, int vec, int target) "sid 0x%"PRIx16" index %d vec %d (should be: %d)"
>   vtd_warn_ir_trigger(uint16_t sid, int index, int trig, int target) "sid 0x%"PRIx16" index %d trigger %d (should be: %d)"
>   vtd_reset_exit(void) ""
> +vtd_device_attach_hwpt(uint32_t dev_id, uint32_t pasid, uint32_t hwpt_id, int ret) "dev_id %d pasid %d hwpt_id %d, ret: %d"
> +vtd_device_detach_hwpt(uint32_t dev_id, uint32_t pasid, int ret) "dev_id %d pasid %d ret: %d"
> +vtd_device_reattach_def_hwpt(uint32_t dev_id, uint32_t pasid, uint32_t hwpt_id, int ret) "dev_id %d pasid %d hwpt_id %d, ret: %d"
>   
>   # amd_iommu.c
>   amdvi_evntlog_fail(uint64_t addr, uint32_t head) "error: fail to write at addr 0x%"PRIx64" +  offset 0x%"PRIx32