iommu/vt-d: Use device rbtree in iopf reporting path

The existing I/O page fault handler currently locates the PCI device by
calling pci_get_domain_bus_and_slot(). This function searches the list
of all PCI devices until the desired device is found. To improve lookup
efficiency, replace it with device_rbtree_find() to search the device
within the probed device rbtree.

The I/O page fault is initiated by the device, which does not have any
synchronization mechanism with the software to ensure that the device
stays in the probed device tree. Theoretically, a device could be released
by the IOMMU subsystem after device_rbtree_find() and before
iopf_get_dev_fault_param(), which would cause a use-after-free problem.

Add a mutex to synchronize the I/O page fault reporting path and the IOMMU
release device path. This lock doesn't introduce any performance overhead,
as the conflict between I/O page fault reporting and device releasing is
very rare.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20240220065939.121116-3-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>

4 files changed, 15 insertions, 8 deletions

diff --git a/drivers/iommu/intel/dmar.c b/drivers/iommu/intel/dmar.c
index f9b63c2875f7..d14797aabb7a 100644
--- a/drivers/iommu/intel/dmar.c
+++ b/drivers/iommu/intel/dmar.c
@@ -1097,6 +1097,7 @@ static int alloc_iommu(struct dmar_drhd_unit *drhd)
 	iommu->segment = drhd->segment;
 	iommu->device_rbtree = RB_ROOT;
 	spin_lock_init(&iommu->device_rbtree_lock);
+	mutex_init(&iommu->iopf_lock);
 	iommu->node = NUMA_NO_NODE;
 
 	ver = readl(iommu->reg + DMAR_VER_REG);
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 025d7385cf58..60aa2dce32ef 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -4362,8 +4362,11 @@ free:
 static void intel_iommu_release_device(struct device *dev)
 {
 	struct device_domain_info *info = dev_iommu_priv_get(dev);
+	struct intel_iommu *iommu = info->iommu;
 
+	mutex_lock(&iommu->iopf_lock);
 	device_rbtree_remove(info);
+	mutex_unlock(&iommu->iopf_lock);
 	dmar_remove_one_dev_info(dev);
 	intel_pasid_free_table(dev);
 	intel_iommu_debugfs_remove_dev(info);
diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index 50d1e196db52..a0feab099f12 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -713,6 +713,8 @@ struct intel_iommu {
 #endif
 	struct iopf_queue *iopf_queue;
 	unsigned char iopfq_name[16];
+	/* Synchronization between fault report and iommu device release. */
+	struct mutex iopf_lock;
 	struct q_inval  *qi;            /* Queued invalidation info */
 	u32 iommu_state[MAX_SR_DMAR_REGS]; /* Store iommu states between suspend and resume.*/
 
diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c
index 1dd56d4eb88c..bdf3584ca0af 100644
--- a/drivers/iommu/intel/svm.c
+++ b/drivers/iommu/intel/svm.c
@@ -643,7 +643,7 @@ static irqreturn_t prq_event_thread(int irq, void *d)
 	struct intel_iommu *iommu = d;
 	struct page_req_dsc *req;
 	int head, tail, handled;
-	struct pci_dev *pdev;
+	struct device *dev;
 	u64 address;
 
 	/*
@@ -689,23 +689,24 @@ bad_req:
 		if (unlikely(req->lpig && !req->rd_req && !req->wr_req))
 			goto prq_advance;
 
-		pdev = pci_get_domain_bus_and_slot(iommu->segment,
-						   PCI_BUS_NUM(req->rid),
-						   req->rid & 0xff);
 		/*
 		 * If prq is to be handled outside iommu driver via receiver of
 		 * the fault notifiers, we skip the page response here.
 		 */
-		if (!pdev)
+		mutex_lock(&iommu->iopf_lock);
+		dev = device_rbtree_find(iommu, req->rid);
+		if (!dev) {
+			mutex_unlock(&iommu->iopf_lock);
 			goto bad_req;
+		}
 
-		if (intel_svm_prq_report(iommu, &pdev->dev, req))
+		if (intel_svm_prq_report(iommu, dev, req))
 			handle_bad_prq_event(iommu, req, QI_RESP_INVALID);
 		else
-			trace_prq_report(iommu, &pdev->dev, req->qw_0, req->qw_1,
+			trace_prq_report(iommu, dev, req->qw_0, req->qw_1,
 					 req->priv_data[0], req->priv_data[1],
 					 iommu->prq_seq_number++);
-		pci_dev_put(pdev);
+		mutex_unlock(&iommu->iopf_lock);
 prq_advance:
 		head = (head + sizeof(*req)) & PRQ_RING_MASK;
 	}