On 9/16/20 12:17 PM, Suravee Suthikulpanit wrote:
> Commit e52d58d54a32 ("iommu/amd: Use cmpxchg_double() when updating
> 128-bit IRTE") removed an assumption that modify_irte_ga always set
> the valid bit, which requires the callers to set the appropriate value
> for the struct irte_ga.valid bit before calling the function.
>
> Similar to the commit 26e495f34107 ("iommu/amd: Restore IRTE.RemapEn
> bit after programming IRTE"), which is for the function
> amd_iommu_deactivate_guest_mode().
>
> The same change is also needed for the amd_iommu_activate_guest_mode().
> Otherwise, this could trigger IO_PAGE_FAULT for the VFIO based VMs with
> AVIC enabled.
>
> Reported-by: Maxim Levitsky <mlevi...@redhat.com>
> Tested-by: Maxim Levitsky <mlevi...@redhat.com>
> Cc: Joao Martins <joao.m.mart...@oracle.com>
> Fixes: e52d58d54a321 ("iommu/amd: Use cmpxchg_double() when updating 128-bit
> IRTE")
> Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpa...@amd.com>
> ---
> drivers/iommu/amd/iommu.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
> index e938677af8bc..db4fb840c59c 100644
> --- a/drivers/iommu/amd/iommu.c
> +++ b/drivers/iommu/amd/iommu.c
> @@ -3900,14 +3900,18 @@ int amd_iommu_activate_guest_mode(void *data)
> {
> struct amd_ir_data *ir_data = (struct amd_ir_data *)data;
> struct irte_ga *entry = (struct irte_ga *) ir_data->entry;
> + u64 valid;
>
> if (!AMD_IOMMU_GUEST_IR_VAPIC(amd_iommu_guest_ir) ||
> !entry || entry->lo.fields_vapic.guest_mode)
> return 0;
>
> + valid = entry->lo.fields_vapic.valid;
> +
> entry->lo.val = 0;
> entry->hi.val = 0;
>
> + entry->lo.fields_vapic.valid = valid;
> entry->lo.fields_vapic.guest_mode = 1;
> entry->lo.fields_vapic.ga_log_intr = 1;
> entry->hi.fields.ga_root_ptr = ir_data->ga_root_ptr;
>
FWIW,
Reviewed-by: Joao Martins <joao.m.mart...@oracle.com>
