Merge: [RHEL-9.6] iommu/vt-d: a couple of late fixes

MR: https://gitlab.com/redhat/rhel/src/kernel/rhel-9/-/merge_requests/3471

JIRA: https://issues.redhat.com/browse/RHEL-70102
Upstream-Status: git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux.git
                 fixes branch

## Summary of Changes

This pulls in a couple late landing vt-d fixes. One to fix an issue that came in with the global static identity domain support from v6.12 which was no longer creating context entries for aliases, and another fixing up a suspicious RCU usage warning in the code to allocate dmar fault interrupts locally.

Signed-off-by: Jerry Snitselaar <jsnitsel@redhat.com>

Closes RHEL-70102

Approved-by: Donald Dutile <ddutile@redhat.com>
Approved-by: Eder Zulian <ezulian@redhat.com>
Approved-by: CKI KWF Bot <cki-ci-bot+kwf-gitlab-com@redhat.com>

Merged-by: CKI KWF Bot <cki-ci-bot+kwf-gitlab-com@redhat.com>

Author: CKI KWF Bot

drivers/iommu/intel/dmar.c

@@ -2056,6 +2056,7 @@ int enable_drhd_fault_handling(unsigned int cpu)
 	/*
 	 * Enable fault control interrupt.
 	 */
+	guard(rwsem_read)(&dmar_global_lock);
 	for_each_iommu(iommu, drhd) {
 		u32 fault_status;
 		int ret;

drivers/iommu/intel/iommu.c

@@ -3307,7 +3307,14 @@ int __init intel_iommu_init(void)
 		iommu_device_sysfs_add(&iommu->iommu, NULL,
 				       intel_iommu_groups,
 				       "%s", iommu->name);
+		/*
+		 * The iommu device probe is protected by the iommu_probe_device_lock.
+		 * Release the dmar_global_lock before entering the device probe path
+		 * to avoid unnecessary lock order splat.
+		 */
+		up_read(&dmar_global_lock);
 		iommu_device_register(&iommu->iommu, &intel_iommu_ops, NULL);
+		down_read(&dmar_global_lock);
 
 		iommu_pmu_register(iommu);
 	}
@@ -4547,9 +4554,6 @@ static int context_setup_pass_through_cb(struct pci_dev *pdev, u16 alias, void *
 {
 	struct device *dev = data;
 
-	if (dev != &pdev->dev)
-		return 0;
-
 	return context_setup_pass_through(dev, PCI_BUS_NUM(alias), alias & 0xff);
 }