All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] mlx4: Add support for EEH error recovery
@ 2012-07-20 19:55 Kleber Sacilotto de Souza
  2012-07-22 10:29 ` Or Gerlitz
                   ` (2 more replies)
  0 siblings, 3 replies; 24+ messages in thread
From: Kleber Sacilotto de Souza @ 2012-07-20 19:55 UTC (permalink / raw)
  To: David S. Miller
  Cc: netdev, Jack Morgenstein, Yevgeny Petrilin, Or Gerlitz, cascardo,
	brking, Kleber Sacilotto de Souza

Currently the mlx4 drivers don't have the necessary callbacks to
implement EEH errors detection and recovery, so the PCI layer uses the
probe and remove callbacks to try to recover the device after an error on
the bus. However, these callbacks have race conditions with the internal
catastrophic error recovery functions, which will also detect the error
and this can cause the system to crash if both EEH and catas functions
try to reset the device.

This patch adds the necessary error recovery callbacks and makes sure
that the internal catastrophic error functions will not try to reset the
device in such scenarios. It also adds some calls to
pci_channel_offline() to suppress reads/writes on the bus when the slot
cannot accept I/O operations so we prevent unnecessary accesses to the
bus and speed up the device removal.

Signed-off-by: Kleber Sacilotto de Souza <klebers@linux.vnet.ibm.com>
---
 drivers/net/ethernet/mellanox/mlx4/catas.c |   25 ++++++++++----
 drivers/net/ethernet/mellanox/mlx4/cmd.c   |   49 ++++++++++++++++++++++++++-
 drivers/net/ethernet/mellanox/mlx4/main.c  |   30 ++++++++++++++++-
 3 files changed, 93 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx4/catas.c b/drivers/net/ethernet/mellanox/mlx4/catas.c
index 915e947..9c656fe 100644
--- a/drivers/net/ethernet/mellanox/mlx4/catas.c
+++ b/drivers/net/ethernet/mellanox/mlx4/catas.c
@@ -69,16 +69,21 @@ static void poll_catas(unsigned long dev_ptr)
 	struct mlx4_priv *priv = mlx4_priv(dev);
 
 	if (readl(priv->catas_err.map)) {
-		dump_err_buf(dev);
-
-		mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
+		/* If the device is off-line, we cannot try to recover it */
+		if (pci_channel_offline(dev->pdev))
+			mod_timer(&priv->catas_err.timer,
+				  round_jiffies(jiffies + MLX4_CATAS_POLL_INTERVAL));
+		else {
+			dump_err_buf(dev);
+			mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
 
-		if (internal_err_reset) {
-			spin_lock(&catas_lock);
-			list_add(&priv->catas_err.list, &catas_list);
-			spin_unlock(&catas_lock);
+			if (internal_err_reset) {
+				spin_lock(&catas_lock);
+				list_add(&priv->catas_err.list, &catas_list);
+				spin_unlock(&catas_lock);
 
-			queue_work(mlx4_wq, &catas_work);
+				queue_work(mlx4_wq, &catas_work);
+			}
 		}
 	} else
 		mod_timer(&priv->catas_err.timer,
@@ -100,6 +105,10 @@ static void catas_reset(struct work_struct *work)
 	list_for_each_entry_safe(priv, tmppriv, &tlist, catas_err.list) {
 		struct pci_dev *pdev = priv->dev.pdev;
 
+		/* If the device is off-line, we cannot reset it */
+		if (pci_channel_offline(pdev))
+			continue;
+
 		ret = mlx4_restart_one(priv->dev.pdev);
 		/* 'priv' now is not valid */
 		if (ret)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 7e94987..c8fef43 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -296,7 +296,12 @@ int mlx4_comm_cmd(struct mlx4_dev *dev, u8 cmd, u16 param,
 
 static int cmd_pending(struct mlx4_dev *dev)
 {
-	u32 status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET);
+	u32 status;
+
+	if (pci_channel_offline(dev->pdev))
+		return -EIO;
+
+	status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET);
 
 	return (status & swab32(1 << HCR_GO_BIT)) ||
 		(mlx4_priv(dev)->cmd.toggle ==
@@ -314,11 +319,29 @@ static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
 
 	mutex_lock(&cmd->hcr_mutex);
 
+	if (pci_channel_offline(dev->pdev)) {
+		/*
+		 * Device is going through error recovery
+		 * and cannot accept commands.
+		 */
+		ret = -EIO;
+		goto out;
+	}
+
 	end = jiffies;
 	if (event)
 		end += msecs_to_jiffies(GO_BIT_TIMEOUT_MSECS);
 
 	while (cmd_pending(dev)) {
+		if (pci_channel_offline(dev->pdev)) {
+			/*
+			 * Device is going through error recovery
+			 * and cannot accept commands.
+			 */
+			ret = -EIO;
+			goto out;
+		}
+
 		if (time_after_eq(jiffies, end)) {
 			mlx4_err(dev, "%s:cmd_pending failed\n", __func__);
 			goto out;
@@ -431,14 +454,33 @@ static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 
 	down(&priv->cmd.poll_sem);
 
+	if (pci_channel_offline(dev->pdev)) {
+		/*
+		 * Device is going through error recovery
+		 * and cannot accept commands.
+		 */
+		err = -EIO;
+		goto out;
+	}
+
 	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
 			    in_modifier, op_modifier, op, CMD_POLL_TOKEN, 0);
 	if (err)
 		goto out;
 
 	end = msecs_to_jiffies(timeout) + jiffies;
-	while (cmd_pending(dev) && time_before(jiffies, end))
+	while (cmd_pending(dev) && time_before(jiffies, end)) {
+		if (pci_channel_offline(dev->pdev)) {
+			/*
+			 * Device is going through error recovery
+			 * and cannot accept commands.
+			 */
+			err = -EIO;
+			goto out;
+		}
+
 		cond_resched();
+	}
 
 	if (cmd_pending(dev)) {
 		err = -ETIMEDOUT;
@@ -532,6 +574,9 @@ int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 	       int out_is_imm, u32 in_modifier, u8 op_modifier,
 	       u16 op, unsigned long timeout, int native)
 {
+	if (pci_channel_offline(dev->pdev))
+		return -EIO;
+
 	if (!mlx4_is_mfunc(dev) || (native && mlx4_is_master(dev))) {
 		if (mlx4_priv(dev)->cmd.use_events)
 			return mlx4_cmd_wait(dev, in_param, out_param,
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index 4264516..e717091 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -1775,6 +1775,9 @@ static int mlx4_get_ownership(struct mlx4_dev *dev)
 	void __iomem *owner;
 	u32 ret;
 
+	if (pci_channel_offline(dev->pdev))
+		return -EIO;
+
 	owner = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_OWNER_BASE,
 			MLX4_OWNER_SIZE);
 	if (!owner) {
@@ -1791,6 +1794,9 @@ static void mlx4_free_ownership(struct mlx4_dev *dev)
 {
 	void __iomem *owner;
 
+	if (pci_channel_offline(dev->pdev))
+		return;
+
 	owner = ioremap(pci_resource_start(dev->pdev, 0) + MLX4_OWNER_BASE,
 			MLX4_OWNER_SIZE);
 	if (!owner) {
@@ -2237,11 +2243,33 @@ static DEFINE_PCI_DEVICE_TABLE(mlx4_pci_table) = {
 
 MODULE_DEVICE_TABLE(pci, mlx4_pci_table);
 
+static pci_ers_result_t mlx4_pci_err_detected(struct pci_dev *pdev,
+					      pci_channel_state_t state)
+{
+	mlx4_remove_one(pdev);
+
+	return state == pci_channel_io_perm_failure ?
+		PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_NEED_RESET;
+}
+
+static pci_ers_result_t mlx4_pci_slot_reset(struct pci_dev *pdev)
+{
+	int ret = __mlx4_init_one(pdev, NULL);
+
+	return ret ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED;
+}
+
+static struct pci_error_handlers mlx4_err_handler = {
+	.error_detected = mlx4_pci_err_detected,
+	.slot_reset     = mlx4_pci_slot_reset,
+};
+
 static struct pci_driver mlx4_driver = {
 	.name		= DRV_NAME,
 	.id_table	= mlx4_pci_table,
 	.probe		= mlx4_init_one,
-	.remove		= __devexit_p(mlx4_remove_one)
+	.remove		= __devexit_p(mlx4_remove_one),
+	.err_handler    = &mlx4_err_handler,
 };
 
 static int __init mlx4_verify_params(void)
-- 
1.7.1

^ permalink raw reply related	[flat|nested] 24+ messages in thread
end of thread, other threads:[~2012-07-25 22:19 UTC | newest]

Thread overview: 24+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-07-20 19:55 [PATCH] mlx4: Add support for EEH error recovery Kleber Sacilotto de Souza
2012-07-22 10:29 ` Or Gerlitz
     [not found] ` <500BD558.2060803@mellanox.com>
2012-07-23  0:15   ` David Miller
2012-07-23 13:18     ` Kleber Sacilotto de Souza
2012-07-23 13:45       ` Or Gerlitz
2012-07-23 18:12         ` Kleber Sacilotto de Souza
2012-07-23 20:53           ` Kleber Sacilotto de Souza
2012-07-23 21:26             ` Or Gerlitz
2012-07-23 21:34               ` David Miller
2012-07-23 21:42                 ` Or Gerlitz
2012-07-23 21:44                   ` David Miller
2012-07-23 22:02                     ` Or Gerlitz
2012-07-23 22:21                       ` David Miller
2012-07-24 13:12               ` Kleber Sacilotto de Souza
2012-07-24 17:09                 ` Shlomo Pongartz
2012-07-24 17:35                   ` Kleber Sacilotto de Souza
2012-07-24 18:08                     ` Thadeu Lima de Souza Cascardo
2012-07-24 18:35                       ` Shlomo Pongratz
2012-07-24 18:39                       ` Shlomo Pongratz
2012-07-24 21:03 ` David Miller
2012-07-24 22:30   ` Or Gerlitz
2012-07-25 14:38     ` Shlomo Pongartz
     [not found]     ` <5010070B.5040405@mellanox.com>
2012-07-25 15:02       ` Kleber Sacilotto de Souza
2012-07-25 22:19       ` David Miller

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.