iommu/vt-d: real DMA sub-device info allocation

Domain context mapping can encounter issues with sub-devices of a real DMA device. A sub-device cannot have a valid context entry due to it potentially aliasing another device's 16-bit ID. It's expected that sub-devices of the real DMA device uses the real DMA device's requester when context mapping.

This is an issue when a sub-device is removed where the context entry is cleared for all aliases. Other sub-devices are still valid, resulting in those sub-devices being stranded without valid context entries.

The correct approach is to use the real DMA device when programming the context entries. The insertion path is correct because device_to_iommu() will return the bus and devfn of the real DMA device. The removal path needs to only operate on the real DMA device, otherwise the entire context entry would be cleared for all sub-devices of the real DMA device.

This patch also adds a helper to determine if a struct device is a sub-device of a real DMA device.

Signed-off-by Jon Derrick <[email protected]>

Sub-devices of a real DMA device might exist on a separate segment than the real DMA device and its IOMMU. These devices should still have a valid device_domain_info, but the current dma alias model won't allocate info for the subdevice.

This patch adds a segment member to struct device_domain_info and uses the sub-device's BDF so that these sub-devices won't alias to other devices.

Signed-off-by Jon Derrick <[email protected]>

---
  drivers/iommu/intel-iommu.c | 19 +++++++++++++++----
  include/linux/intel-iommu.h |  1 +
  2 files changed, 16 insertions(+), 4 deletions(-)

drivers/iommu/intel-iommu.c +15 -4

index 1ff45b2..6d39b9b 100644
@@ -2463,7 +2463,7 @@ static void do_deferred_attach(struct device *dev)
  	struct device_domain_info *info;
 
  	list_for_each_entry(info, &device_domain_list, global)
-		if (info->iommu->segment == segment && info->bus == bus &&
+		if (info->segment == segment && info->bus == bus &&
  		    info->devfn == devfn)
  			return info;
 

@@ -2520,8 +2520,18 @@ static struct dmar_domain *dmar_insert_one_dev_info(struct intel_iommu *iommu,
  	if (!info)
  		return NULL;
 
-	info->bus = bus;
-	info->devfn = devfn;
+	if (!dev_is_real_dma_subdevice(dev)) {
+		info->bus = bus;
+		info->devfn = devfn;
+		info->segment = iommu->segment;
+	} else {
+		struct pci_dev *pdev = to_pci_dev(dev);
+
+		info->bus = pdev->bus->number;
+		info->devfn = pdev->devfn;
+		info->segment = pci_domain_nr(pdev->bus);
+	}
+
  	info->ats_supported = info->pasid_supported = info->pri_supported = 0;
  	info->ats_enabled = info->pasid_enabled = info->pri_enabled = 0;
  	info->ats_qdep = 0;

@@ -2561,7 +2571,8 @@ static struct dmar_domain *dmar_insert_one_dev_info(struct intel_iommu *iommu,
 
  	if (!found) {
  		struct device_domain_info *info2;
-		info2 = dmar_search_domain_by_dev_info(iommu->segment, bus, devfn);
+		info2 = dmar_search_domain_by_dev_info(info->segment, info->bus,
+						       info->devfn);
  		if (info2) {
  			found      = info2->domain;
  			info2->dev = dev;

include/linux/intel-iommu.h +1 -0

index 21633ce..4100bd2 100644
@@ -609,6 +609,7 @@ struct device_domain_info {
  	struct list_head auxiliary_domains; /* auxiliary domains
  					     * attached to this device
  					     */
+	u32 segment;		/* PCI segment number */
  	u8 bus;			/* PCI bus number */
  	u8 devfn;		/* PCI devfn number */
  	u16 pfsid;		/* SRIOV physical function source ID */

Quoted 5 lines · Jon Derrick

Sub-devices of a real DMA device might exist on a separate segment than the real DMA device and its IOMMU. These devices should still have a valid device_domain_info, but the current dma alias model won't allocate info for the subdevice.

Quoted 5 lines

This patch adds a segment member to struct device_domain_info and uses the sub-device's BDF so that these sub-devices won't alias to other devices.

Signed-off-by Jon Derrick <[email protected]>

Acked-by Lu Baolu <[email protected]>

By removing the real DMA indirection in find_domain(), we can allow sub-devices of a real DMA device to have their own valid device_domain_info. The dmar lookup and context entry removal paths have been fixed to account for sub-devices.

Fixes 2b0140c69637 ("iommu/vt-d: Use pci_real_dma_dev() for mapping")

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=207575

Signed-off-by Jon Derrick <[email protected]>

---
  drivers/iommu/intel-iommu.c | 3 ---
  1 file changed, 3 deletions(-)

drivers/iommu/intel-iommu.c +0 -3

index 6d39b9b..5767882 100644
@@ -2436,9 +2436,6 @@ struct dmar_domain *find_domain(struct device *dev)
  	if (unlikely(attach_deferred(dev) || iommu_dummy(dev)))
  		return NULL;
 
-	if (dev_is_pci(dev))
-		dev = &pci_real_dma_dev(to_pci_dev(dev))->dev;
-
  	/* No lock here, assumes no domain exit in normal case */
  	info = get_domain_info(dev);
  	if (likely(info))

Jon Derrick (3):
iommu/vt-d: Only clear real DMA device's context entries iommu/vt-d: Allocate domain info for real DMA sub-devices iommu/vt-d: Remove real DMA lookup in find_domain

 drivers/iommu/intel-iommu.c | 31 +++++++++++++++++++++++--------
 include/linux/intel-iommu.h |  1 +
 2 files changed, 24 insertions(+), 8 deletions(-)