iommu/arm-smmu: Defer probe of clients after smmu device bound

BugLink: https://bugs.launchpad.net/bugs/2095283

commit 229e6ee upstream.

Null pointer dereference occurs due to a race between smmu
driver probe and client driver probe, when of_dma_configure()
for client is called after the iommu_device_register() for smmu driver
probe has executed but before the driver_bound() for smmu driver
has been called.

Following is how the race occurs:

T1:Smmu device probe		T2: Client device probe

really_probe()
arm_smmu_device_probe()
iommu_device_register()
					really_probe()
					platform_dma_configure()
					of_dma_configure()
					of_dma_configure_id()
					of_iommu_configure()
					iommu_probe_device()
					iommu_init_device()
					arm_smmu_probe_device()
					arm_smmu_get_by_fwnode()
						driver_find_device_by_fwnode()
						driver_find_device()
						next_device()
						klist_next()
						    /* null ptr
						       assigned to smmu */
					/* null ptr dereference
					   while smmu->streamid_mask */
driver_bound()
	klist_add_tail()

When this null smmu pointer is dereferenced later in
arm_smmu_probe_device, the device crashes.

Fix this by deferring the probe of the client device
until the smmu device has bound to the arm smmu driver.

Fixes: 021bb84 ("iommu/arm-smmu: Wire up generic configuration support")
Cc: stable@vger.kernel.org
Co-developed-by: Prakash Gupta <quic_guptap@quicinc.com>
Signed-off-by: Prakash Gupta <quic_guptap@quicinc.com>
Signed-off-by: Pratyush Brahma <quic_pbrahma@quicinc.com>
Link: https://lore.kernel.org/r/20241004090428.2035-1-quic_pbrahma@quicinc.com
[will: Add comment]
Signed-off-by: Will Deacon <will@kernel.org>
[rm: backport for context conflict prior to 6.8]
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
CVE-2024-56568
Signed-off-by: Koichiro Den <koichiro.den@canonical.com>

Author: Pratyush Brahma

drivers/iommu/arm/arm-smmu/arm-smmu.c

@@ -1375,6 +1375,17 @@ static struct iommu_device *arm_smmu_probe_device(struct device *dev)
 			goto out_free;
 	} else if (fwspec && fwspec->ops == &arm_smmu_ops) {
 		smmu = arm_smmu_get_by_fwnode(fwspec->iommu_fwnode);
+
+		/*
+		 * Defer probe if the relevant SMMU instance hasn't finished
+		 * probing yet. This is a fragile hack and we'd ideally
+		 * avoid this race in the core code. Until that's ironed
+		 * out, however, this is the most pragmatic option on the
+		 * table.
+		 */
+		if (!smmu)
+			return ERR_PTR(dev_err_probe(dev, -EPROBE_DEFER,
+						"smmu dev has not bound yet\n"));
 	} else {
 		return ERR_PTR(-ENODEV);
 	}