Skip to contentSkip to conversation

iommu: Fix pci_dev_reset_iommu_prepare/done()

PATCH In review Nicolin Chen 4 participants 14 messages Apr 25 – May 11 IOMMU

lore
Share & tools
lore
#0 lore raw

v8 cover iommu: Fix pci_dev_reset_iommu_prepare/done()

Shuai and Kevin found a few bugs in the pci_dev_reset_iommu_prepare/done() helpers when used to handle some corner cases:
- Nested callbacks
- Multi-device groups
- WARN_ON/UAF due to concurrent detach

This needs some substantial rework by tracking device reset states on a per gdev basis. This series includes a few patches addressing them. Most of the patches are reviewed previously in a single patch v6. As we found more bugs during the reviews, I split that v6 to smaller patches so each of them will be cleaner.

This is on Github:
https://github.com/nicolinc/iommufd/commits/fix_iommu_reset-v8

Note that concurrent reset of two DMA alias siblings (sharing the same RID) might prematurely unblock when one device is done while the other is still resetting. And it's a bit convoluted to support this case. Given that it's unclear whether real ATS devices might share RID, for now, add a warning in the done(). A future work can fix it properly if someone hits it.

Changelog
v8:
* Add Reviewed-by tags
* Fix NULL group->domain in done()
* Tidy goto cleanup when using guard()
* Update patch subject and commit message
* Add warning on premature unblocking in DMA alias cases
* Drop unreachable skip in __iommu_group_set_domain_internal() error path
v7:
https://lore.kernel.org/all/[email protected]/
* Add Reviewed-by tags
* Split v6 into smaller patches
* Add one patch to fix UAF during detach()
* Add one patch to fix unnecessary ATS invalidation
v6:
https://lore.kernel.org/all/[email protected]/
* Update inline comments and commit message
* Add "max_pasids > 0" condition in both helpers
v5:
https://lore.kernel.org/all/[email protected]/
* Add 'blocked' to fix iommu_driver_get_domain_for_dev() return.
v4:
https://lore.kernel.org/all/[email protected]/
* Rename 'reset_cnt' to 'recovery_cnt'
v3:
https://lore.kernel.org/all/[email protected]/
* Turn prepare()/done() to be per-gdev
* Use reset_depth to track nested re-entries
* Replace group->resetting_domain with a reset_cnt
v2:
https://lore.kernel.org/all/[email protected]/
* Fix in the helpers by allowing re-entry
v1:
https://lore.kernel.org/all/[email protected]/

Nicolin Chen (8):
iommu: Fix NULL group->domain dereference in
pci_dev_reset_iommu_done()
iommu: Fix kdocs of pci_dev_reset_iommu_done()
iommu: Replace per-group resetting_domain with per-gdev blocked flag
iommu: Fix pasid attach in pci_dev_reset_iommu_prepare/done()
iommu: Fix nested pci_dev_reset_iommu_prepare/done()
iommu: Fix ATS invalidation timeouts during
__iommu_remove_group_pasid()
iommu: Fix WARN_ON in __iommu_group_set_domain_nofail() due to reset
iommu: Warn on premature unblock during DMA aliased sibling reset

 drivers/iommu/iommu.c | 223 ++++++++++++++++++++++++++++++++++--------
 1 file changed, 181 insertions(+), 42 deletions(-)
Signature · 1 line
2.43.0
#7 lore raw

PATCH v8 5/8 iommu: Fix nested pci_dev_reset_iommu_prepare/done()

Shuai found that cxl_reset_bus_function() calls pci_reset_bus_function() internally while both are calling pci_dev_reset_iommu_prepare/done().

As pci_dev_reset_iommu_prepare() doesn't support re-entry, the inner call will trigger a WARN_ON and return -EBUSY, resulting in failing the entire device reset.

On the other hand, removing the outer calls in the PCI callers is unsafe. As pointed out by Kevin, device-specific quirks like reset_hinic_vf_dev() execute custom firmware waits after their inner pcie_flr() completes. If the IOMMU protection relies solely on the inner reset, the IOMMU will be unblocked prematurely while the device is still resetting.

Instead, fix this by making pci_dev_reset_iommu_prepare/done() reentrant.

Introduce gdev->reset_depth to handle the re-entries on the same device.

---
 drivers/iommu/iommu.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)
drivers/iommu/iommu.c +13 -6
index 2907d76c39c68..7a5a5d3aabb65 100644
@@ -82,6 +82,7 @@ struct group_device {
 	 *  - Device is undergoing a reset
 	 */
 	bool blocked;
+	unsigned int reset_depth;
 };
 
 /* Iterate over each struct group_device in a struct iommu_group */
@@ -3997,20 +3998,23 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 	if (WARN_ON(!gdev))
 		return -ENODEV;
 
-	/* Re-entry is not allowed */
-	if (WARN_ON(gdev->blocked))
-		return -EBUSY;
+	if (gdev->reset_depth++)
+		return 0;
 
 	ret = __iommu_group_alloc_blocking_domain(group);
-	if (ret)
+	if (ret) {
+		gdev->reset_depth--;
 		return ret;
+	}
 
 	/* Stage RID domain at blocking_domain while retaining group->domain */
 	if (group->domain != group->blocking_domain) {
 		ret = __iommu_attach_device(group->blocking_domain, &pdev->dev,
 					    group->domain);
-		if (ret)
+		if (ret) {
+			gdev->reset_depth--;
 			return ret;
+		}
 	}
 
 	/*
@@ -4070,7 +4074,10 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 	if (WARN_ON(!gdev))
 		return;
 
-	if (!gdev->blocked)
+	/* Unbalanced done() calls would underflow the counter */
+	if (WARN_ON(gdev->reset_depth == 0))
+		return;
+	if (--gdev->reset_depth)
 		return;
 
 	if (WARN_ON(!group->blocking_domain))
Signature · 1 line
2.43.0
#1 lore raw

PATCH v8 1/8 iommu: Fix NULL group->domain dereference in pci_dev_reset_iommu_done()

In reply to Nicolin Chen

Local sashiko review pointed it out that group->domain could be NULL when a default domain fails to allocate during the first probe, which can crash at domain->ops->attach_dev dereference in __iommu_attach_device() invoked by pci_dev_reset_iommu_done().

pci_dev_reset_iommu_prepare() is fine as an old_domain pointer can be NULL.

Skip the re-attach in pci_dev_reset_iommu_done() to fix the bug.

Fixes c279e83953d9 ("iommu: Introduce pci_dev_reset_iommu_prepare/done()")
---
 drivers/iommu/iommu.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/iommu/iommu.c +7 -2
index 35db517809540..00b6a33515398 100644
@@ -4025,8 +4025,13 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 	if (WARN_ON(!group->blocking_domain))
 		return;
 
-	/* Re-attach RID domain back to group->domain */
-	if (group->domain != group->blocking_domain) {
+	/*
+	 * Re-attach RID domain back to group->domain
+	 *
+	 * Leave the device parked in the blocking_domain if group->domain isn't
+	 * initialized yet
+	 */
+	if (group->domain && group->domain != group->blocking_domain) {
 		WARN_ON(__iommu_attach_device(group->domain, &pdev->dev,
 					      group->blocking_domain));
 	}
Signature · 1 line
2.43.0
#3 lore raw

PATCH v8 2/8 iommu: Fix kdocs of pci_dev_reset_iommu_done()

In reply to Nicolin Chen

Remove the duplicated word. No functional change.

Fixes c279e83953d9 ("iommu: Introduce pci_dev_reset_iommu_prepare/done()")
---
 drivers/iommu/iommu.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/iommu/iommu.c +3 -3
index 00b6a33515398..82dd806e5e6a6 100644
@@ -3997,9 +3997,9 @@ EXPORT_SYMBOL_GPL(pci_dev_reset_iommu_prepare);
  * @pdev: PCI device that has finished a reset routine
  *
  * After a PCIe device finishes a reset routine, it wants to restore its IOMMU
- * IOMMU activity, including new translation as well as cache invalidation, by
- * re-attaching all RID/PASID of the device's back to the domains retained in
- * the core-level structure.
+ * activity, including new translation and cache invalidation, by re-attaching
+ * all RID/PASID of the device back to the domains retained in the core-level
+ * structure.
  *
  * Caller must pair it with a successful pci_dev_reset_iommu_prepare().
  *
Signature · 1 line
2.43.0
#5 lore raw

PATCH v8 4/8 iommu: Fix pasid attach in pci_dev_reset_iommu_prepare/done()

In reply to Nicolin Chen

Now the helpers handle per-gdev resets. Replace __iommu_set_group_pasid() with set_dev_pasid() accordingly, in the pci_dev_reset_iommu_done().

Also add max_pasids check as other callers.

---
 drivers/iommu/iommu.c | 25 ++++++++++++++++++-------
 1 file changed, 18 insertions(+), 7 deletions(-)
drivers/iommu/iommu.c +18 -7
index 5b784e43ca592..2907d76c39c68 100644
@@ -4026,9 +4026,14 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 	 * The pasid_array is mostly fenced by group->mutex, except one reader
 	 * in iommu_attach_handle_get(), so it's safe to read without xa_lock.
 	 */
-	xa_for_each_start(&group->pasid_array, pasid, entry, 1)
-		iommu_remove_dev_pasid(&pdev->dev, pasid,
-				       pasid_array_entry_to_domain(entry));
+	if (pdev->dev.iommu->max_pasids > 0) {
+		xa_for_each_start(&group->pasid_array, pasid, entry, 1) {
+			struct iommu_domain *pasid_dom =
+				pasid_array_entry_to_domain(entry);
+
+			iommu_remove_dev_pasid(&pdev->dev, pasid, pasid_dom);
+		}
+	}
 
 	group->recovery_cnt++;
 	return ret;
@@ -4095,10 +4100,16 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 	 * The pasid_array is mostly fenced by group->mutex, except one reader
 	 * in iommu_attach_handle_get(), so it's safe to read without xa_lock.
 	 */
-	xa_for_each_start(&group->pasid_array, pasid, entry, 1)
-		WARN_ON(__iommu_set_group_pasid(
-			pasid_array_entry_to_domain(entry), group, pasid,
-			group->blocking_domain));
+	if (pdev->dev.iommu->max_pasids > 0) {
+		xa_for_each_start(&group->pasid_array, pasid, entry, 1) {
+			struct iommu_domain *pasid_dom =
+				pasid_array_entry_to_domain(entry);
+
+			WARN_ON(pasid_dom->ops->set_dev_pasid(
+				pasid_dom, &pdev->dev, pasid,
+				group->blocking_domain));
+		}
+	}
 
 	if (!WARN_ON(group->recovery_cnt == 0))
 		group->recovery_cnt--;
Signature · 1 line
2.43.0
#4 lore raw

PATCH v8 3/8 iommu: Replace per-group resetting_domain with per-gdev blocked flag

In reply to Nicolin Chen

The core tracks device resetting states with a per-group resetting_domain, while a reset is actually per group-device. Such a mismatch might lead to confusion and even difficulty to untangle per-gdev handling requirement.

Shuai found that cxl_reset_bus_function() calls pci_reset_bus_function() internally while both are calling pci_dev_reset_iommu_prepare/done(). And the solution requires the core to track at the group_device level as well.

Introduce a 'blocked' flag to struct group_device, to allow a multi-device group to isolate concurrent device resets independently.

As the reset routine is per gdev, it cannot clear group->resetting_domain without iterating over the device list to ensure no other device is being reset. Simplify it by replacing the resetting_domain with a 'recovery_cnt' in the struct iommu_group.

No functional change. But this is essential to apply following bug fixes.

---
 drivers/iommu/iommu.c | 102 ++++++++++++++++++++++++++++++++----------
 1 file changed, 78 insertions(+), 24 deletions(-)
drivers/iommu/iommu.c +78 -24
index 82dd806e5e6a6..5b784e43ca592 100644
@@ -61,14 +61,14 @@ struct iommu_group {
 	int id;
 	struct iommu_domain *default_domain;
 	struct iommu_domain *blocking_domain;
-	/*
-	 * During a group device reset, @resetting_domain points to the physical
-	 * domain, while @domain points to the attached domain before the reset.
-	 */
-	struct iommu_domain *resetting_domain;
 	struct iommu_domain *domain;
 	struct list_head entry;
 	unsigned int owner_cnt;
+	/*
+	 * Number of devices in the group undergoing or awaiting recovery.
+	 * If non-zero, concurrent domain attachments are rejected.
+	 */
+	unsigned int recovery_cnt;
 	void *owner;
 };
 
@@ -76,12 +76,32 @@ struct group_device {
 	struct list_head list;
 	struct device *dev;
 	char *name;
+	/*
+	 * Device is blocked for a pending recovery while its group->domain is
+	 * retained. This can happen when:
+	 *  - Device is undergoing a reset
+	 */
+	bool blocked;
 };
 
 /* Iterate over each struct group_device in a struct iommu_group */
 #define for_each_group_device(group, pos) \
 	list_for_each_entry(pos, &(group)->devices, list)
 
+static struct group_device *__dev_to_gdev(struct device *dev)
+{
+	struct iommu_group *group = dev->iommu_group;
+	struct group_device *gdev;
+
+	lockdep_assert_held(&group->mutex);
+
+	for_each_group_device(group, gdev) {
+		if (gdev->dev == dev)
+			return gdev;
+	}
+	return NULL;
+}
+
 struct iommu_group_attribute {
 	struct attribute attr;
 	ssize_t (*show)(struct iommu_group *group, char *buf);
@@ -2191,6 +2211,8 @@ EXPORT_SYMBOL_GPL(iommu_attach_device);
 
 int iommu_deferred_attach(struct device *dev, struct iommu_domain *domain)
 {
+	struct group_device *gdev;
+
 	/*
 	 * This is called on the dma mapping fast path so avoid locking. This is
 	 * racy, but we have an expectation that the driver will setup its DMAs
@@ -2201,14 +2223,18 @@ int iommu_deferred_attach(struct device *dev, struct iommu_domain *domain)
 
 	guard(mutex)(&dev->iommu_group->mutex);
 
+	gdev = __dev_to_gdev(dev);
+	if (WARN_ON(!gdev))
+		return -ENODEV;
+
 	/*
-	 * This is a concurrent attach during a device reset. Reject it until
+	 * This is a concurrent attach during device recovery. Reject it until
 	 * pci_dev_reset_iommu_done() attaches the device to group->domain.
 	 *
 	 * Note that this might fail the iommu_dma_map(). But there's nothing
 	 * more we can do here.
 	 */
-	if (dev->iommu_group->resetting_domain)
+	if (gdev->blocked)
 		return -EBUSY;
 	return __iommu_attach_device(domain, dev, NULL);
 }
@@ -2265,19 +2291,24 @@ EXPORT_SYMBOL_GPL(iommu_get_domain_for_dev);
 struct iommu_domain *iommu_driver_get_domain_for_dev(struct device *dev)
 {
 	struct iommu_group *group = dev->iommu_group;
+	struct group_device *gdev;
 
 	lockdep_assert_held(&group->mutex);
 
+	gdev = __dev_to_gdev(dev);
+	if (WARN_ON(!gdev))
+		return NULL;
+
 	/*
 	 * Driver handles the low-level __iommu_attach_device(), including the
 	 * one invoked by pci_dev_reset_iommu_done() re-attaching the device to
 	 * the cached group->domain. In this case, the driver must get the old
-	 * domain from group->resetting_domain rather than group->domain. This
+	 * domain from group->blocking_domain rather than group->domain. This
 	 * prevents it from re-attaching the device from group->domain (old) to
 	 * group->domain (new).
 	 */
-	if (group->resetting_domain)
-		return group->resetting_domain;
+	if (gdev->blocked)
+		return group->blocking_domain;
 
 	return group->domain;
 }
@@ -2436,10 +2467,10 @@ static int __iommu_group_set_domain_internal(struct iommu_group *group,
 		return -EINVAL;
 
 	/*
-	 * This is a concurrent attach during a device reset. Reject it until
+	 * This is a concurrent attach during device recovery. Reject it until
 	 * pci_dev_reset_iommu_done() attaches the device to group->domain.
 	 */
-	if (group->resetting_domain)
+	if (group->recovery_cnt)
 		return -EBUSY;
 
 	/*
@@ -3567,10 +3598,10 @@ int iommu_attach_device_pasid(struct iommu_domain *domain,
 	mutex_lock(&group->mutex);
 
 	/*
-	 * This is a concurrent attach during a device reset. Reject it until
+	 * This is a concurrent attach during device recovery. Reject it until
 	 * pci_dev_reset_iommu_done() attaches the device to group->domain.
 	 */
-	if (group->resetting_domain) {
+	if (group->recovery_cnt) {
 		ret = -EBUSY;
 		goto out_unlock;
 	}
@@ -3660,10 +3691,10 @@ int iommu_replace_device_pasid(struct iommu_domain *domain,
 	mutex_lock(&group->mutex);
 
 	/*
-	 * This is a concurrent attach during a device reset. Reject it until
+	 * This is a concurrent attach during device recovery. Reject it until
 	 * pci_dev_reset_iommu_done() attaches the device to group->domain.
 	 */
-	if (group->resetting_domain) {
+	if (group->recovery_cnt) {
 		ret = -EBUSY;
 		goto out_unlock;
 	}
@@ -3934,12 +3965,12 @@ EXPORT_SYMBOL_NS_GPL(iommu_replace_group_handle, "IOMMUFD_INTERNAL");
  * routine wants to block any IOMMU activity: translation and ATS invalidation.
  *
  * This function attaches the device's RID/PASID(s) the group->blocking_domain,
- * setting the group->resetting_domain. This allows the IOMMU driver pausing any
+ * incrementing the group->recovery_cnt, to allow the IOMMU driver pausing any
  * IOMMU activity while leaving the group->domain pointer intact. Later when the
  * reset is finished, pci_dev_reset_iommu_done() can restore everything.
  *
  * Caller must use pci_dev_reset_iommu_prepare() with pci_dev_reset_iommu_done()
- * before/after the core-level reset routine, to unset the resetting_domain.
+ * before/after the core-level reset routine, to decrement the recovery_cnt.
  *
  * Return: 0 on success or negative error code if the preparation failed.
  *
@@ -3952,6 +3983,7 @@ EXPORT_SYMBOL_NS_GPL(iommu_replace_group_handle, "IOMMUFD_INTERNAL");
 int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 {
 	struct iommu_group *group = pdev->dev.iommu_group;
+	struct group_device *gdev;
 	unsigned long pasid;
 	void *entry;
 	int ret;
@@ -3961,8 +3993,12 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 
 	guard(mutex)(&group->mutex);
 
+	gdev = __dev_to_gdev(&pdev->dev);
+	if (WARN_ON(!gdev))
+		return -ENODEV;
+
 	/* Re-entry is not allowed */
-	if (WARN_ON(group->resetting_domain))
+	if (WARN_ON(gdev->blocked))
 		return -EBUSY;
 
 	ret = __iommu_group_alloc_blocking_domain(group);
@@ -3977,6 +4013,13 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 			return ret;
 	}
 
+	/*
+	 * Update gdev->blocked upon the domain change, as it is used to return
+	 * the correct domain in iommu_driver_get_domain_for_dev() that might be
+	 * called in a set_dev_pasid callback function.
+	 */
+	gdev->blocked = true;
+
 	/*
 	 * Stage PASID domains at blocking_domain while retaining pasid_array.
 	 *
@@ -3987,7 +4030,7 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 		iommu_remove_dev_pasid(&pdev->dev, pasid,
 				       pasid_array_entry_to_domain(entry));
 
-	group->resetting_domain = group->blocking_domain;
+	group->recovery_cnt++;
 	return ret;
 }
 EXPORT_SYMBOL_GPL(pci_dev_reset_iommu_prepare);
@@ -4009,6 +4052,7 @@ EXPORT_SYMBOL_GPL(pci_dev_reset_iommu_prepare);
 void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 {
 	struct iommu_group *group = pdev->dev.iommu_group;
+	struct group_device *gdev;
 	unsigned long pasid;
 	void *entry;
 
@@ -4017,11 +4061,13 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 
 	guard(mutex)(&group->mutex);
 
-	/* pci_dev_reset_iommu_prepare() was bypassed for the device */
-	if (!group->resetting_domain)
+	gdev = __dev_to_gdev(&pdev->dev);
+	if (WARN_ON(!gdev))
+		return;
+
+	if (!gdev->blocked)
 		return;
 
-	/* pci_dev_reset_iommu_prepare() was not successfully called */
 	if (WARN_ON(!group->blocking_domain))
 		return;
 
@@ -4036,6 +4082,13 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 					      group->blocking_domain));
 	}
 
+	/*
+	 * Update gdev->blocked upon the domain change, as it is used to return
+	 * the correct domain in iommu_driver_get_domain_for_dev() that might be
+	 * called in a set_dev_pasid callback function.
+	 */
+	gdev->blocked = false;
+
 	/*
 	 * Re-attach PASID domains back to the domains retained in pasid_array.
 	 *
@@ -4047,7 +4100,8 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 			pasid_array_entry_to_domain(entry), group, pasid,
 			group->blocking_domain));
 
-	group->resetting_domain = NULL;
+	if (!WARN_ON(group->recovery_cnt == 0))
+		group->recovery_cnt--;
 }
 EXPORT_SYMBOL_GPL(pci_dev_reset_iommu_done);
 
Signature · 1 line
2.43.0
#8 lore raw

PATCH v8 6/8 iommu: Fix ATS invalidation timeouts during __iommu_remove_group_pasid()

In reply to Nicolin Chen

If a device is blocked, its PASID domains are already detached. Repeating iommu_remove_dev_pasid() is unnecessary and might trigger ATS invalidation timeouts.

Skip the iommu_remove_dev_pasid() call upon gdev->blocked.

Fixes c279e83953d9 ("iommu: Introduce pci_dev_reset_iommu_prepare/done()")
---
 drivers/iommu/iommu.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/iommu/iommu.c +6 -1
index 7a5a5d3aabb65..d0f32bd954a72 100644
@@ -3554,7 +3554,12 @@ static void __iommu_remove_group_pasid(struct iommu_group *group,
 	struct group_device *device;
 
 	for_each_group_device(group, device) {
-		if (device->dev->iommu->max_pasids > 0)
+		/*
+		 * A group-level detach cannot fail, even if there is a blocked
+		 * device. In fact, blocked devices must be already detached for
+		 * a pending device recovery.
+		 */
+		if (!device->blocked && device->dev->iommu->max_pasids > 0)
 			iommu_remove_dev_pasid(device->dev, pasid, domain);
 	}
 }
Signature · 1 line
2.43.0
#10 lore raw

PATCH v8 7/8 iommu: Fix WARN_ON in __iommu_group_set_domain_nofail() due to reset

In reply to Nicolin Chen

In __iommu_group_set_domain_internal(), concurrent domain attachments are rejected when any device in the group is recovering. This is necessary to fence concurrent attachments to a multi-device group where devices might share the same RID due to PCI DMA alias quirks, but triggers the WARN_ON in __iommu_group_set_domain_nofail().

Other IOMMU_SET_DOMAIN_MUST_SUCCEED callers in detach/teardown paths, such as __iommu_group_set_core_domain and __iommu_release_dma_ownership, should not be rejected, as the domain would be freed anyway in these nofail paths while group->domain is still pointing to it. So pci_dev_reset_iommu_done() could trigger a UAF when re-attaching group->domain.

Honor the IOMMU_SET_DOMAIN_MUST_SUCCEED flag, allowing the callers through the group->recovery_cnt fence, so as to update the group->domain pointer. Instead add a gdev->blocked check in the device iteration loop, to prevent any concurrent per-device detachment.

---
 drivers/iommu/iommu.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
drivers/iommu/iommu.c +10 -2
index d0f32bd954a72..f21d352a67f70 100644
@@ -2469,9 +2469,10 @@ static int __iommu_group_set_domain_internal(struct iommu_group *group,
 
 	/*
 	 * This is a concurrent attach during device recovery. Reject it until
-	 * pci_dev_reset_iommu_done() attaches the device to group->domain.
+	 * pci_dev_reset_iommu_done() attaches the device to group->domain, if
+	 * IOMMU_SET_DOMAIN_MUST_SUCCEED is not set.
 	 */
-	if (group->recovery_cnt)
+	if (group->recovery_cnt && !(flags & IOMMU_SET_DOMAIN_MUST_SUCCEED))
 		return -EBUSY;
 
 	/*
@@ -2482,6 +2483,13 @@ static int __iommu_group_set_domain_internal(struct iommu_group *group,
 	 */
 	result = 0;
 	for_each_group_device(group, gdev) {
+		/*
+		 * Device under recovery is attached to group->blocking_domain.
+		 * Don't change that. pci_dev_reset_iommu_done() will re-attach
+		 * its domain to the updated group->domain, after the recovery.
+		 */
+		if (gdev->blocked)
+			continue;
 		ret = __iommu_device_set_domain(group, gdev->dev, new_domain,
 						group->domain, flags);
 		if (ret) {
Signature · 1 line
2.43.0
#11 lore raw

PATCH v8 8/8 iommu: Warn on premature unblock during DMA aliased sibling reset

In reply to Nicolin Chen

When two aliased siblings are in the same iommu_group, they might share the same RID. The reset functions don't support this case, though it is unclear whether there is a real case of having an ATS capable device on a PCI/PCI-X bus.

Theoretically, however, if two aliased devices are resetting concurrently, one might be unblocked prematurely in the middle of the reset by the other sibling who completes the reset first.

This isn't a regression from this series but it's better to spit a warning, so we can know if such use case is common enough for us to make subsequent patches for its coverage.

---
 drivers/iommu/iommu.c | 49 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
drivers/iommu/iommu.c +49 -0
index f21d352a67f70..dd53cce12087c 100644
@@ -4057,6 +4057,41 @@ int pci_dev_reset_iommu_prepare(struct pci_dev *pdev)
 }
 EXPORT_SYMBOL_GPL(pci_dev_reset_iommu_prepare);
 
+static int __group_device_cmp_dma_alias(struct pci_dev *dev, u16 alias,
+					void *data)
+{
+	return alias == *(u16 *)data;
+}
+
+static int group_device_cmp_dma_alias(struct pci_dev *dev, u16 alias,
+				      void *data)
+{
+	return pci_for_each_dma_alias(data, __group_device_cmp_dma_alias,
+				      &alias);
+}
+
+static bool group_device_dma_alias_is_blocked(struct iommu_group *group,
+					      struct group_device *gdev)
+{
+	struct group_device *sibling;
+
+	lockdep_assert_held(&group->mutex);
+
+	if (!dev_is_pci(gdev->dev))
+		return false;
+
+	for_each_group_device(group, sibling) {
+		if (sibling == gdev || !sibling->blocked ||
+		    !dev_is_pci(sibling->dev))
+			continue;
+		if (pci_for_each_dma_alias(to_pci_dev(gdev->dev),
+					   group_device_cmp_dma_alias,
+					   to_pci_dev(sibling->dev)))
+			return true;
+	}
+	return false;
+}
+
 /**
  * pci_dev_reset_iommu_done() - Restore IOMMU after a PCI device reset is done
  * @pdev: PCI device that has finished a reset routine
@@ -4096,6 +4131,20 @@ void pci_dev_reset_iommu_done(struct pci_dev *pdev)
 	if (WARN_ON(!group->blocking_domain))
 		return;
 
+	if (group_device_dma_alias_is_blocked(group, gdev)) {
+		/*
+		 * FIXME: DMA aliased devices share the same RID, which would be
+		 * convoluted to handle, as "gdev->blocked" is not sufficient:
+		 *  - "blocked" state is effectively shared across these devices
+		 *  - if the core skipped the blocking on the second device, the
+		 *    IOMMU driver's attachment state would diverge from the HW
+		 *    state
+		 * For now, just warn and see whether real ATS use cases hit it.
+		 */
+		pci_warn(pdev,
+			 "DMA-aliased sibling may be prematurely unblocked\n");
+	}
+
 	/*
 	 * Re-attach RID domain back to group->domain
 	 *
Signature · 1 line
2.43.0
Reviewed-by on v8 1/8
#2 lore raw

Re: [PATCH rc v8 1/8] iommu: Fix NULL group->domain dereference in pci_dev_reset_iommu_done()

In reply to Nicolin Chen
Quoted 7 lines · Nicolin Chen

Local sashiko review pointed it out that group->domain could be NULL when a default domain fails to allocate during the first probe, which can crash at domain->ops->attach_dev dereference in __iommu_attach_device() invoked by pci_dev_reset_iommu_done().

pci_dev_reset_iommu_prepare() is fine as an old_domain pointer can be NULL.

Quoted 8 lines

Skip the re-attach in pci_dev_reset_iommu_done() to fix the bug.

Fixes c279e83953d9 ("iommu: Introduce pci_dev_reset_iommu_prepare/done()")
---
  drivers/iommu/iommu.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)
Reviewed-by on v8 4/8
#6 lore raw

Re: [PATCH rc v8 4/8] iommu: Fix pasid attach in pci_dev_reset_iommu_prepare/done()

In reply to Nicolin Chen
Quoted 13 lines · Nicolin Chen

Now the helpers handle per-gdev resets. Replace __iommu_set_group_pasid() with set_dev_pasid() accordingly, in the pci_dev_reset_iommu_done().

Also add max_pasids check as other callers.

Quoted 3 lines
---
  drivers/iommu/iommu.c | 25 ++++++++++++++++++-------
  1 file changed, 18 insertions(+), 7 deletions(-)
Reviewed-by on v8 6/8
#9 lore raw

Re: [PATCH rc v8 6/8] iommu: Fix ATS invalidation timeouts during __iommu_remove_group_pasid()

In reply to Nicolin Chen
Quoted 6 lines · Nicolin Chen

If a device is blocked, its PASID domains are already detached. Repeating iommu_remove_dev_pasid() is unnecessary and might trigger ATS invalidation timeouts.

Skip the iommu_remove_dev_pasid() call upon gdev->blocked.

Quoted 8 lines
Fixes c279e83953d9 ("iommu: Introduce pci_dev_reset_iommu_prepare/done()")
---
  drivers/iommu/iommu.c | 7 ++++++-
  1 file changed, 6 insertions(+), 1 deletion(-)
Reviewed-by on v8 8/8
#12 lore raw

RE: [PATCH rc v8 8/8] iommu: Warn on premature unblock during DMA aliased sibling reset

In reply to Nicolin Chen
Quoted 13 lines

From: Nicolin Chen <[email protected]>
Sent: Saturday, April 25, 2026 9:15 AM

When two aliased siblings are in the same iommu_group, they might share the
same RID. The reset functions don't support this case, though it is unclear whether there is a real case of having an ATS capable device on a PCI/PCI-X bus.

Theoretically, however, if two aliased devices are resetting concurrently, one might be unblocked prematurely in the middle of the reset by the other sibling who completes the reset first.

Quoted 6 lines

This isn't a regression from this series but it's better to spit a warning, so we can know if such use case is common enough for us to make subsequent
patches for its coverage.

maintainer
#13 lore raw
In reply to Nicolin Chen
Quoted 12 lines · Nicolin Chen

Nicolin Chen (8):
iommu: Fix NULL group->domain dereference in
pci_dev_reset_iommu_done()
iommu: Fix kdocs of pci_dev_reset_iommu_done()
iommu: Replace per-group resetting_domain with per-gdev blocked flag
iommu: Fix pasid attach in pci_dev_reset_iommu_prepare/done()
iommu: Fix nested pci_dev_reset_iommu_prepare/done()
iommu: Fix ATS invalidation timeouts during
__iommu_remove_group_pasid()
iommu: Fix WARN_ON in __iommu_group_set_domain_nofail() due to reset
iommu: Warn on premature unblock during DMA aliased sibling reset

Quoted 2 lines
 drivers/iommu/iommu.c | 223 ++++++++++++++++++++++++++++++++++--------
 1 file changed, 181 insertions(+), 42 deletions(-)

Applied for -rc, thanks.