Merge tag 'irq-msi-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull MSI irq updates from Thomas Gleixner:

 - Switch the MSI descriptor locking to guards

 - Replace the broken PCI/TPH implementation, which lacks any form of
   serialization against concurrent modifications with a properly
   serialized mechanism in the PCI/MSI core code

 - Replace the MSI descriptor abuse in the SCSI/UFS Qualcom driver with
   dedicated driver internal storage

* tag 'irq-msi-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  genirq/msi: Rename msi_[un]lock_descs()
  scsi: ufs: qcom: Remove the MSI descriptor abuse
  PCI/TPH: Replace the broken MSI-X control word update
  PCI/MSI: Provide a sane mechanism for TPH
  PCI: hv: Switch MSI descriptor locking to guard()
  PCI/MSI: Switch to MSI descriptor locking to guard()
  NTB/msi: Switch MSI descriptor locking to lock guard()
  soc: ti: ti_sci_inta_msi: Switch MSI descriptor locking to guard()
  genirq/msi: Use lock guards for MSI descriptor locking
  cleanup: Provide retain_ptr()
  genirq/msi: Make a few functions static

Author: torvalds

drivers/ntb/msi.c

@@ -106,10 +106,10 @@ int ntb_msi_setup_mws(struct ntb_dev *ntb)
 	if (!ntb->msi)
 		return -EINVAL;
 
-	msi_lock_descs(&ntb->pdev->dev);
-	desc = msi_first_desc(&ntb->pdev->dev, MSI_DESC_ASSOCIATED);
-	addr = desc->msg.address_lo + ((uint64_t)desc->msg.address_hi << 32);
-	msi_unlock_descs(&ntb->pdev->dev);
+	scoped_guard (msi_descs_lock, &ntb->pdev->dev) {
+		desc = msi_first_desc(&ntb->pdev->dev, MSI_DESC_ASSOCIATED);
+		addr = desc->msg.address_lo + ((uint64_t)desc->msg.address_hi << 32);
+	}
 
 	for (peer = 0; peer < ntb_peer_port_count(ntb); peer++) {
 		peer_widx = ntb_peer_highest_mw_idx(ntb, peer);
@@ -289,7 +289,7 @@ int ntbm_msi_request_threaded_irq(struct ntb_dev *ntb, irq_handler_t handler,
 	if (!ntb->msi)
 		return -EINVAL;
 
-	msi_lock_descs(dev);
+	guard(msi_descs_lock)(dev);
 	msi_for_each_desc(entry, dev, MSI_DESC_ASSOCIATED) {
 		if (irq_has_action(entry->irq))
 			continue;
@@ -307,17 +307,11 @@ int ntbm_msi_request_threaded_irq(struct ntb_dev *ntb, irq_handler_t handler,
 		ret = ntbm_msi_setup_callback(ntb, entry, msi_desc);
 		if (ret) {
 			devm_free_irq(&ntb->dev, entry->irq, dev_id);
-			goto unlock;
+			return ret;
 		}
-
-		ret = entry->irq;
-		goto unlock;
+		return entry->irq;
 	}
-	ret = -ENODEV;
-
-unlock:
-	msi_unlock_descs(dev);
-	return ret;
+	return -ENODEV;
 }
 EXPORT_SYMBOL(ntbm_msi_request_threaded_irq);
 

drivers/pci/controller/pci-hyperv.c

@@ -3975,24 +3975,18 @@ static int hv_pci_restore_msi_msg(struct pci_dev *pdev, void *arg)
 {
 	struct irq_data *irq_data;
 	struct msi_desc *entry;
-	int ret = 0;
 
 	if (!pdev->msi_enabled && !pdev->msix_enabled)
 		return 0;
 
-	msi_lock_descs(&pdev->dev);
+	guard(msi_descs_lock)(&pdev->dev);
 	msi_for_each_desc(entry, &pdev->dev, MSI_DESC_ASSOCIATED) {
 		irq_data = irq_get_irq_data(entry->irq);
-		if (WARN_ON_ONCE(!irq_data)) {
-			ret = -EINVAL;
-			break;
-		}
-
+		if (WARN_ON_ONCE(!irq_data))
+			return -EINVAL;
 		hv_compose_msi_msg(irq_data, &entry->msg);
 	}
-	msi_unlock_descs(&pdev->dev);
-
-	return ret;
+	return 0;
 }
 
 /*

drivers/pci/msi/api.c

@@ -53,10 +53,9 @@ void pci_disable_msi(struct pci_dev *dev)
 	if (!pci_msi_enabled() || !dev || !dev->msi_enabled)
 		return;
 
-	msi_lock_descs(&dev->dev);
+	guard(msi_descs_lock)(&dev->dev);
 	pci_msi_shutdown(dev);
 	pci_free_msi_irqs(dev);
-	msi_unlock_descs(&dev->dev);
 }
 EXPORT_SYMBOL(pci_disable_msi);
 
@@ -196,10 +195,9 @@ void pci_disable_msix(struct pci_dev *dev)
 	if (!pci_msi_enabled() || !dev || !dev->msix_enabled)
 		return;
 
-	msi_lock_descs(&dev->dev);
+	guard(msi_descs_lock)(&dev->dev);
 	pci_msix_shutdown(dev);
 	pci_free_msi_irqs(dev);
-	msi_unlock_descs(&dev->dev);
 }
 EXPORT_SYMBOL(pci_disable_msix);
 

drivers/pci/msi/msi.c

@@ -336,41 +336,11 @@ static int msi_verify_entries(struct pci_dev *dev)
 	return !entry ? 0 : -EIO;
 }
 
-/**
- * msi_capability_init - configure device's MSI capability structure
- * @dev: pointer to the pci_dev data structure of MSI device function
- * @nvec: number of interrupts to allocate
- * @affd: description of automatic IRQ affinity assignments (may be %NULL)
- *
- * Setup the MSI capability structure of the device with the requested
- * number of interrupts.  A return value of zero indicates the successful
- * setup of an entry with the new MSI IRQ.  A negative return value indicates
- * an error, and a positive return value indicates the number of interrupts
- * which could have been allocated.
- */
-static int msi_capability_init(struct pci_dev *dev, int nvec,
-			       struct irq_affinity *affd)
+static int __msi_capability_init(struct pci_dev *dev, int nvec, struct irq_affinity_desc *masks)
 {
-	struct irq_affinity_desc *masks = NULL;
+	int ret = msi_setup_msi_desc(dev, nvec, masks);
 	struct msi_desc *entry, desc;
-	int ret;
 
-	/* Reject multi-MSI early on irq domain enabled architectures */
-	if (nvec > 1 && !pci_msi_domain_supports(dev, MSI_FLAG_MULTI_PCI_MSI, ALLOW_LEGACY))
-		return 1;
-
-	/*
-	 * Disable MSI during setup in the hardware, but mark it enabled
-	 * so that setup code can evaluate it.
-	 */
-	pci_msi_set_enable(dev, 0);
-	dev->msi_enabled = 1;
-
-	if (affd)
-		masks = irq_create_affinity_masks(nvec, affd);
-
-	msi_lock_descs(&dev->dev);
-	ret = msi_setup_msi_desc(dev, nvec, masks);
 	if (ret)
 		goto fail;
 
@@ -399,19 +369,48 @@ static int msi_capability_init(struct pci_dev *dev, int nvec,
 
 	pcibios_free_irq(dev);
 	dev->irq = entry->irq;
-	goto unlock;
-
+	return 0;
 err:
 	pci_msi_unmask(&desc, msi_multi_mask(&desc));
 	pci_free_msi_irqs(dev);
 fail:
 	dev->msi_enabled = 0;
-unlock:
-	msi_unlock_descs(&dev->dev);
-	kfree(masks);
 	return ret;
 }
 
+/**
+ * msi_capability_init - configure device's MSI capability structure
+ * @dev: pointer to the pci_dev data structure of MSI device function
+ * @nvec: number of interrupts to allocate
+ * @affd: description of automatic IRQ affinity assignments (may be %NULL)
+ *
+ * Setup the MSI capability structure of the device with the requested
+ * number of interrupts.  A return value of zero indicates the successful
+ * setup of an entry with the new MSI IRQ.  A negative return value indicates
+ * an error, and a positive return value indicates the number of interrupts
+ * which could have been allocated.
+ */
+static int msi_capability_init(struct pci_dev *dev, int nvec,
+			       struct irq_affinity *affd)
+{
+	/* Reject multi-MSI early on irq domain enabled architectures */
+	if (nvec > 1 && !pci_msi_domain_supports(dev, MSI_FLAG_MULTI_PCI_MSI, ALLOW_LEGACY))
+		return 1;
+
+	/*
+	 * Disable MSI during setup in the hardware, but mark it enabled
+	 * so that setup code can evaluate it.
+	 */
+	pci_msi_set_enable(dev, 0);
+	dev->msi_enabled = 1;
+
+	struct irq_affinity_desc *masks __free(kfree) =
+		affd ? irq_create_affinity_masks(nvec, affd) : NULL;
+
+	guard(msi_descs_lock)(&dev->dev);
+	return __msi_capability_init(dev, nvec, masks);
+}
+
 int __pci_enable_msi_range(struct pci_dev *dev, int minvec, int maxvec,
 			   struct irq_affinity *affd)
 {
@@ -666,40 +665,41 @@ static void msix_mask_all(void __iomem *base, int tsize)
 		writel(ctrl, base + PCI_MSIX_ENTRY_VECTOR_CTRL);
 }
 
-static int msix_setup_interrupts(struct pci_dev *dev, struct msix_entry *entries,
-				 int nvec, struct irq_affinity *affd)
+static int __msix_setup_interrupts(struct pci_dev *dev, struct msix_entry *entries,
+				   int nvec, struct irq_affinity_desc *masks)
 {
-	struct irq_affinity_desc *masks = NULL;
-	int ret;
-
-	if (affd)
-		masks = irq_create_affinity_masks(nvec, affd);
+	int ret = msix_setup_msi_descs(dev, entries, nvec, masks);
 
-	msi_lock_descs(&dev->dev);
-	ret = msix_setup_msi_descs(dev, entries, nvec, masks);
 	if (ret)
-		goto out_free;
+		goto fail;
 
 	ret = pci_msi_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSIX);
 	if (ret)
-		goto out_free;
+		goto fail;
 
 	/* Check if all MSI entries honor device restrictions */
 	ret = msi_verify_entries(dev);
 	if (ret)
-		goto out_free;
+		goto fail;
 
 	msix_update_entries(dev, entries);
-	goto out_unlock;
+	return 0;
 
-out_free:
+fail:
 	pci_free_msi_irqs(dev);
-out_unlock:
-	msi_unlock_descs(&dev->dev);
-	kfree(masks);
 	return ret;
 }
 
+static int msix_setup_interrupts(struct pci_dev *dev, struct msix_entry *entries,
+				 int nvec, struct irq_affinity *affd)
+{
+	struct irq_affinity_desc *masks __free(kfree) =
+		affd ? irq_create_affinity_masks(nvec, affd) : NULL;
+
+	guard(msi_descs_lock)(&dev->dev);
+	return __msix_setup_interrupts(dev, entries, nvec, masks);
+}
+
 /**
  * msix_capability_init - configure device's MSI-X capability
  * @dev: pointer to the pci_dev data structure of MSI-X device function
@@ -871,13 +871,13 @@ void __pci_restore_msix_state(struct pci_dev *dev)
 
 	write_msg = arch_restore_msi_irqs(dev);
 
-	msi_lock_descs(&dev->dev);
-	msi_for_each_desc(entry, &dev->dev, MSI_DESC_ALL) {
-		if (write_msg)
-			__pci_write_msi_msg(entry, &entry->msg);
-		pci_msix_write_vector_ctrl(entry, entry->pci.msix_ctrl);
+	scoped_guard (msi_descs_lock, &dev->dev) {
+		msi_for_each_desc(entry, &dev->dev, MSI_DESC_ALL) {
+			if (write_msg)
+				__pci_write_msi_msg(entry, &entry->msg);
+			pci_msix_write_vector_ctrl(entry, entry->pci.msix_ctrl);
+		}
 	}
-	msi_unlock_descs(&dev->dev);
 
 	pci_msix_clear_and_set_ctrl(dev, PCI_MSIX_FLAGS_MASKALL, 0);
 }
@@ -916,6 +916,53 @@ void pci_free_msi_irqs(struct pci_dev *dev)
 	}
 }
 
+#ifdef CONFIG_PCIE_TPH
+/**
+ * pci_msix_write_tph_tag - Update the TPH tag for a given MSI-X vector
+ * @pdev:	The PCIe device to update
+ * @index:	The MSI-X index to update
+ * @tag:	The tag to write
+ *
+ * Returns: 0 on success, error code on failure
+ */
+int pci_msix_write_tph_tag(struct pci_dev *pdev, unsigned int index, u16 tag)
+{
+	struct msi_desc *msi_desc;
+	struct irq_desc *irq_desc;
+	unsigned int virq;
+
+	if (!pdev->msix_enabled)
+		return -ENXIO;
+
+	guard(msi_descs_lock)(&pdev->dev);
+	virq = msi_get_virq(&pdev->dev, index);
+	if (!virq)
+		return -ENXIO;
+	/*
+	 * This is a horrible hack, but short of implementing a PCI
+	 * specific interrupt chip callback and a huge pile of
+	 * infrastructure, this is the minor nuissance. It provides the
+	 * protection against concurrent operations on this entry and keeps
+	 * the control word cache in sync.
+	 */
+	irq_desc = irq_to_desc(virq);
+	if (!irq_desc)
+		return -ENXIO;
+
+	guard(raw_spinlock_irq)(&irq_desc->lock);
+	msi_desc = irq_data_get_msi_desc(&irq_desc->irq_data);
+	if (!msi_desc || msi_desc->pci.msi_attrib.is_virtual)
+		return -ENXIO;
+
+	msi_desc->pci.msix_ctrl &= ~PCI_MSIX_ENTRY_CTRL_ST;
+	msi_desc->pci.msix_ctrl |= FIELD_PREP(PCI_MSIX_ENTRY_CTRL_ST, tag);
+	pci_msix_write_vector_ctrl(msi_desc, msi_desc->pci.msix_ctrl);
+	/* Flush the write */
+	readl(pci_msix_desc_addr(msi_desc));
+	return 0;
+}
+#endif
+
 /* Misc. infrastructure */
 
 struct pci_dev *msi_desc_to_pci_dev(struct msi_desc *desc)

drivers/pci/pci.h

@@ -989,6 +989,15 @@ int pcim_request_region_exclusive(struct pci_dev *pdev, int bar,
 				  const char *name);
 void pcim_release_region(struct pci_dev *pdev, int bar);
 
+#ifdef CONFIG_PCI_MSI
+int pci_msix_write_tph_tag(struct pci_dev *pdev, unsigned int index, u16 tag);
+#else
+static inline int pci_msix_write_tph_tag(struct pci_dev *pdev, unsigned int index, u16 tag)
+{
+	return -ENODEV;
+}
+#endif
+
 /*
  * Config Address for PCI Configuration Mechanism #1
  *