Re: [PATCH 16/24] iommu/amd: Introduce IOMMUFD vDevice support for AMD

["Suthikulpanit, Suravee" <suravee.suthikulpanit@amd.com>]

On 8/19/2026 8:34 PM, guanghuifeng@linux.alibaba.com wrote:
> 
> 在 2026/7/27 21:29, Suravee Suthikulpanit 写道:
>> Initialize vDevice for AMD vIOMMU by setting up the Device ID Mapping
>> table using the guest device ID.
>>
>> Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
>> ---
>>   drivers/iommu/amd/amd_iommu_types.h | 10 ++++++++++
>>   drivers/iommu/amd/iommufd.c         | 31 +++++++++++++++++++++++++++++
>>   drivers/iommu/amd/nested.c          | 27 ++++++++++++++++++++++---
>>   3 files changed, 65 insertions(+), 3 deletions(-)
>>
>> ....
>>
>> diff --git a/drivers/iommu/amd/iommufd.c b/drivers/iommu/amd/iommufd.c
>> index bfc4b0ec22a9..1975b2932107 100644
>> --- a/drivers/iommu/amd/iommufd.c
>> +++ b/drivers/iommu/amd/iommufd.c
>> @@ -9,6 +9,7 @@
>>   #include "amd_iommu.h"
>>   #include "amd_viommu.h"
>>   #include "amd_iommu_types.h"
>> +#include "../iommufd/iommufd_private.h"
>>   static const struct iommufd_viommu_ops amd_viommu_ops;
>> @@ -126,6 +127,34 @@ static void amd_iommufd_viommu_destroy(struct 
>> iommufd_viommu *viommu)
>>       amd_iommu_gid_free(iommu, aviommu->gid);
>>   }
>> +/*
>> + * Called from drivers/iommu/iommufd/viommu.c: 
>> iommufd_vdevice_alloc_ioctl()
>> + */
>> +static int _amd_viommu_vdevice_init(struct iommufd_vdevice *vdev)
>> +{
>> +    struct iommu_dev_data *dev_data;
>> +    struct pci_dev *pdev = to_pci_dev(vdev->idev->dev);
>> +    struct iommufd_viommu *viommu = vdev->viommu;
>> +    struct amd_iommu_viommu *aviommu = container_of(viommu, struct 
>> amd_iommu_viommu, core);
>> +
>> +    if (!pdev) {
>> +        pr_err("%s: not a PCI device\n", __func__);
>> +        return -EINVAL;
>> +    }
>> +
> 
> to_pci_dev() is a container_of(), so it can never return NULL and the
> !pdev check is dead code. Worse, for a non-PCI device it yields a bogus
> pointer which is then dereferenced through dev_iommu_priv_get() below.

I will fix the code to check dev_is_pci() properly.

Thanks,
Suravee