Re: [PATCH 2/2] iommu/vt-d: Fix NULL domain on device release

[Baolu Lu <baolu.lu@linux.intel.com>]

On 2/27/24 3:40 PM, Tian, Kevin wrote:
>> From: Lu Baolu<baolu.lu@linux.intel.com>
>> Sent: Friday, February 23, 2024 1:13 PM
>>
>> -static void dmar_remove_one_dev_info(struct device *dev)
>> -{
>> -	struct device_domain_info *info = dev_iommu_priv_get(dev);
>> -	struct dmar_domain *domain = info->domain;
>> -	struct intel_iommu *iommu = info->iommu;
>> -	unsigned long flags;
>> -
>> -	if (!dev_is_real_dma_subdevice(info->dev)) {
>> -		if (dev_is_pci(info->dev) && sm_supported(iommu))
>> -			intel_pasid_tear_down_entry(iommu, info->dev,
>> -					IOMMU_NO_PASID, false);
>> -
>> -		iommu_disable_pci_caps(info);
>> -		domain_context_clear(info);
>> -	}
>> -
>> -	spin_lock_irqsave(&domain->lock, flags);
>> -	list_del(&info->link);
>> -	spin_unlock_irqrestore(&domain->lock, flags);
>> -
>> -	domain_detach_iommu(domain, iommu);
>> -	info->domain = NULL;
>> -}
>> -
> what's required here is slightly different from device_block_translation()
> which leaves context entry uncleared in scalable mode (implying the
> pasid table must be valid). but in the release path the pasid table will
> be freed right after then leading to a use-after-free case.
> 
> let's add an explicit domain_context_clear() in intel_iommu_release_device().

Nice catch!

How about moving the scalable mode context entry management to probe and
release path? Currently, it's part of domain switch, that's really
irrelevant.

Best regards,
baolu