Merge branch 'pci/bwctrl'

- Simplify link bandwidth controller by replacing the count of Link
  Bandwidth Management Status (LBMS) events with a PCI_LINK_LBMS_SEEN flag
  (Ilpo Järvinen)

- Update the Link Speed after retraining, since the Link Speed may have
  changed (Ilpo Järvinen)

* pci/bwctrl:
  PCI: Update Link Speed after retraining
  PCI/bwctrl: Replace lbms_count with PCI_LINK_LBMS_SEEN flag

5 files changed, 44 insertions, 83 deletions

diff --git a/drivers/pci/hotplug/pciehp_ctrl.c b/drivers/pci/hotplug/pciehp_ctrl.c
index d603a7aa7483..bcc938d4420f 100644
--- a/drivers/pci/hotplug/pciehp_ctrl.c
+++ b/drivers/pci/hotplug/pciehp_ctrl.c
@@ -131,7 +131,7 @@ static void remove_board(struct controller *ctrl, bool safe_removal)
 			      INDICATOR_NOOP);
 
 	/* Don't carry LBMS indications across */
-	pcie_reset_lbms_count(ctrl->pcie->port);
+	pcie_reset_lbms(ctrl->pcie->port);
 }
 
 static int pciehp_enable_slot(struct controller *ctrl);
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 4d7c9f64ea24..eb0c55078d5e 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -4718,6 +4718,11 @@ static int pcie_wait_for_link_status(struct pci_dev *pdev,
  * @pdev: Device whose link to retrain.
  * @use_lt: Use the LT bit if TRUE, or the DLLLA bit if FALSE, for status.
  *
+ * Trigger retraining of the PCIe Link and wait for the completion of the
+ * retraining. As link retraining is known to asserts LBMS and may change
+ * the Link Speed, LBMS is cleared after the retraining and the Link Speed
+ * of the subordinate bus is updated.
+ *
  * Retrain completion status is retrieved from the Link Status Register
  * according to @use_lt.  It is not verified whether the use of the DLLLA
  * bit is valid.
@@ -4757,7 +4762,19 @@ int pcie_retrain_link(struct pci_dev *pdev, bool use_lt)
 	 * to track link speed or width changes made by hardware itself
 	 * in attempt to correct unreliable link operation.
 	 */
-	pcie_reset_lbms_count(pdev);
+	pcie_reset_lbms(pdev);
+
+	/*
+	 * Ensure the Link Speed updates after retraining in case the Link
+	 * Speed was changed because of the retraining. While the bwctrl's
+	 * IRQ handler normally picks up the new Link Speed, clearing LBMS
+	 * races with the IRQ handler reading the Link Status register and
+	 * can result in the handler returning early without updating the
+	 * Link Speed.
+	 */
+	if (pdev->subordinate)
+		pcie_update_link_speed(pdev->subordinate);
+
 	return rc;
 }
 
diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
index eca2812cfd25..3ec7d7b4e812 100644
--- a/drivers/pci/pci.h
+++ b/drivers/pci/pci.h
@@ -557,6 +557,7 @@ static inline int pci_dev_set_disconnected(struct pci_dev *dev, void *unused)
 #define PCI_DPC_RECOVERED 1
 #define PCI_DPC_RECOVERING 2
 #define PCI_DEV_REMOVED 3
+#define PCI_LINK_LBMS_SEEN	6
 
 static inline void pci_dev_assign_added(struct pci_dev *dev)
 {
@@ -828,14 +829,9 @@ static inline void pcie_ecrc_get_policy(char *str) { }
 #endif
 
 #ifdef CONFIG_PCIEPORTBUS
-void pcie_reset_lbms_count(struct pci_dev *port);
-int pcie_lbms_count(struct pci_dev *port, unsigned long *val);
+void pcie_reset_lbms(struct pci_dev *port);
 #else
-static inline void pcie_reset_lbms_count(struct pci_dev *port) {}
-static inline int pcie_lbms_count(struct pci_dev *port, unsigned long *val)
-{
-	return -EOPNOTSUPP;
-}
+static inline void pcie_reset_lbms(struct pci_dev *port) {}
 #endif
 
 struct pci_dev_reset_methods {
diff --git a/drivers/pci/pcie/bwctrl.c b/drivers/pci/pcie/bwctrl.c
index d8d2aa85a229..36f939f23d34 100644
--- a/drivers/pci/pcie/bwctrl.c
+++ b/drivers/pci/pcie/bwctrl.c
@@ -38,24 +38,14 @@
 /**
  * struct pcie_bwctrl_data - PCIe bandwidth controller
  * @set_speed_mutex:	Serializes link speed changes
- * @lbms_count:		Count for LBMS (since last reset)
  * @cdev:		Thermal cooling device associated with the port
  */
 struct pcie_bwctrl_data {
 	struct mutex set_speed_mutex;
-	atomic_t lbms_count;
 	struct thermal_cooling_device *cdev;
 };
 
-/*
- * Prevent port removal during LBMS count accessors and Link Speed changes.
- *
- * These have to be differentiated because pcie_bwctrl_change_speed() calls
- * pcie_retrain_link() which uses LBMS count reset accessor on success
- * (using just one rwsem triggers "possible recursive locking detected"
- * warning).
- */
-static DECLARE_RWSEM(pcie_bwctrl_lbms_rwsem);
+/* Prevent port removal during Link Speed changes. */
 static DECLARE_RWSEM(pcie_bwctrl_setspeed_rwsem);
 
 static bool pcie_valid_speed(enum pci_bus_speed speed)
@@ -127,18 +117,7 @@ static int pcie_bwctrl_change_speed(struct pci_dev *port, u16 target_speed, bool
 	if (ret != PCIBIOS_SUCCESSFUL)
 		return pcibios_err_to_errno(ret);
 
-	ret = pcie_retrain_link(port, use_lt);
-	if (ret < 0)
-		return ret;
-
-	/*
-	 * Ensure link speed updates also with platforms that have problems
-	 * with notifications.
-	 */
-	if (port->subordinate)
-		pcie_update_link_speed(port->subordinate);
-
-	return 0;
+	return pcie_retrain_link(port, use_lt);
 }
 
 /**
@@ -202,15 +181,14 @@ int pcie_set_target_speed(struct pci_dev *port, enum pci_bus_speed speed_req,
 
 static void pcie_bwnotif_enable(struct pcie_device *srv)
 {
-	struct pcie_bwctrl_data *data = srv->port->link_bwctrl;
 	struct pci_dev *port = srv->port;
 	u16 link_status;
 	int ret;
 
-	/* Count LBMS seen so far as one */
+	/* Note if LBMS has been seen so far */
 	ret = pcie_capability_read_word(port, PCI_EXP_LNKSTA, &link_status);
 	if (ret == PCIBIOS_SUCCESSFUL && link_status & PCI_EXP_LNKSTA_LBMS)
-		atomic_inc(&data->lbms_count);
+		set_bit(PCI_LINK_LBMS_SEEN, &port->priv_flags);
 
 	pcie_capability_set_word(port, PCI_EXP_LNKCTL,
 				 PCI_EXP_LNKCTL_LBMIE | PCI_EXP_LNKCTL_LABIE);
@@ -233,7 +211,6 @@ static void pcie_bwnotif_disable(struct pci_dev *port)
 static irqreturn_t pcie_bwnotif_irq(int irq, void *context)
 {
 	struct pcie_device *srv = context;
-	struct pcie_bwctrl_data *data = srv->port->link_bwctrl;
 	struct pci_dev *port = srv->port;
 	u16 link_status, events;
 	int ret;
@@ -247,7 +224,7 @@ static irqreturn_t pcie_bwnotif_irq(int irq, void *context)
 		return IRQ_NONE;
 
 	if (events & PCI_EXP_LNKSTA_LBMS)
-		atomic_inc(&data->lbms_count);
+		set_bit(PCI_LINK_LBMS_SEEN, &port->priv_flags);
 
 	pcie_capability_write_word(port, PCI_EXP_LNKSTA, events);
 
@@ -262,31 +239,10 @@ static irqreturn_t pcie_bwnotif_irq(int irq, void *context)
 	return IRQ_HANDLED;
 }
 
-void pcie_reset_lbms_count(struct pci_dev *port)
+void pcie_reset_lbms(struct pci_dev *port)
 {
-	struct pcie_bwctrl_data *data;
-
-	guard(rwsem_read)(&pcie_bwctrl_lbms_rwsem);
-	data = port->link_bwctrl;
-	if (data)
-		atomic_set(&data->lbms_count, 0);
-	else
-		pcie_capability_write_word(port, PCI_EXP_LNKSTA,
-					   PCI_EXP_LNKSTA_LBMS);
-}
-
-int pcie_lbms_count(struct pci_dev *port, unsigned long *val)
-{
-	struct pcie_bwctrl_data *data;
-
-	guard(rwsem_read)(&pcie_bwctrl_lbms_rwsem);
-	data = port->link_bwctrl;
-	if (!data)
-		return -ENOTTY;
-
-	*val = atomic_read(&data->lbms_count);
-
-	return 0;
+	clear_bit(PCI_LINK_LBMS_SEEN, &port->priv_flags);
+	pcie_capability_write_word(port, PCI_EXP_LNKSTA, PCI_EXP_LNKSTA_LBMS);
 }
 
 static int pcie_bwnotif_probe(struct pcie_device *srv)
@@ -308,18 +264,16 @@ static int pcie_bwnotif_probe(struct pcie_device *srv)
 		return ret;
 
 	scoped_guard(rwsem_write, &pcie_bwctrl_setspeed_rwsem) {
-		scoped_guard(rwsem_write, &pcie_bwctrl_lbms_rwsem) {
-			port->link_bwctrl = data;
-
-			ret = request_irq(srv->irq, pcie_bwnotif_irq,
-					  IRQF_SHARED, "PCIe bwctrl", srv);
-			if (ret) {
-				port->link_bwctrl = NULL;
-				return ret;
-			}
+		port->link_bwctrl = data;
 
-			pcie_bwnotif_enable(srv);
+		ret = request_irq(srv->irq, pcie_bwnotif_irq,
+				  IRQF_SHARED, "PCIe bwctrl", srv);
+		if (ret) {
+			port->link_bwctrl = NULL;
+			return ret;
 		}
+
+		pcie_bwnotif_enable(srv);
 	}
 
 	pci_dbg(port, "enabled with IRQ %d\n", srv->irq);
@@ -339,13 +293,11 @@ static void pcie_bwnotif_remove(struct pcie_device *srv)
 	pcie_cooling_device_unregister(data->cdev);
 
 	scoped_guard(rwsem_write, &pcie_bwctrl_setspeed_rwsem) {
-		scoped_guard(rwsem_write, &pcie_bwctrl_lbms_rwsem) {
-			pcie_bwnotif_disable(srv->port);
+		pcie_bwnotif_disable(srv->port);
 
-			free_irq(srv->irq, srv);
+		free_irq(srv->irq, srv);
 
-			srv->port->link_bwctrl = NULL;
-		}
+		srv->port->link_bwctrl = NULL;
 	}
 }
 
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 8d610c17e0f2..64ac1ee944d3 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -38,14 +38,10 @@
 
 static bool pcie_lbms_seen(struct pci_dev *dev, u16 lnksta)
 {
-	unsigned long count;
-	int ret;
-
-	ret = pcie_lbms_count(dev, &count);
-	if (ret < 0)
-		return lnksta & PCI_EXP_LNKSTA_LBMS;
+	if (test_bit(PCI_LINK_LBMS_SEEN, &dev->priv_flags))
+		return true;
 
-	return count > 0;
+	return lnksta & PCI_EXP_LNKSTA_LBMS;
 }
 
 /*