Re: [PATCH 0/6] iommu/amd: Refactors for ATS updates

[Vasant Hegde <vasant.hegde@amd.com>]

Jason,


On 6/2/2026 5:38 PM, Jason Gunthorpe wrote:
> On Tue, Jun 02, 2026 at 02:19:54PM +0530, Vasant Hegde wrote:
>> -static int iommu_init_device(struct amd_iommu *iommu, struct device *dev)
>> +static struct iommu_dev_data *iommu_init_device(struct amd_iommu *iommu,
>> +						struct device *dev, u16 devid)
>>  {
>>  	struct iommu_dev_data *dev_data;
>> -	int devid, sbdf;
>> -
>> -	if (dev_iommu_priv_get(dev))
>> -		return 0;
>>
>> -	sbdf = get_device_sbdf_id(dev);
>> -	if (sbdf < 0)
>> -		return sbdf;
>> -
>> -	devid = PCI_SBDF_TO_DEVID(sbdf);
>>  	dev_data = find_dev_data(iommu, devid);
>>  	if (!dev_data)
>> -		return -ENOMEM;
>> +		return ERR_PTR(-ENOMEM);
> 
> Why does the AMD driver do this? The dev data is supposed to be
> allocated during probe and freed during remove, it doesn't make sense
> to search for it in probe - it should not exist if the kernel is
> working right.

Digging into git history and how its all evolved, I agree. We should fix up
find_dev_data() stuff. we should separate out search from allocation. I will add
separate fix once this series settles.

-Vasant