diff mbox series

[1/4] sym53c8xx_2: move PCI EEH handling to host reset

Message ID 20210819090716.94049-2-hare@suse.de
State New
Headers show
Series sym53c8xx_2: Fixes for SCSI EH rework | expand

Commit Message

Hannes Reinecke Aug. 19, 2021, 9:07 a.m. UTC 
When PCI EEH is active the entire card will be reset; which means
that we _have_ to do the equivalent of a eh_host_reset().
So trying to do it from other EH callbacks is pointless, and we
should delegate it to eh_host_reset().

Signed-off-by: Hannes Reinecke <hare@suse.de>
---
 drivers/scsi/sym53c8xx_2/sym_glue.c | 97 ++++++++++++++++++-----------
 1 file changed, 60 insertions(+), 37 deletions(-)

Comments

Christoph Hellwig Aug. 20, 2021, 5:13 a.m. UTC | #1 
On Thu, Aug 19, 2021 at 11:07:13AM +0200, Hannes Reinecke wrote:
> When PCI EEH is active the entire card will be reset; which means

> that we _have_ to do the equivalent of a eh_host_reset().

> So trying to do it from other EH callbacks is pointless, and we

> should delegate it to eh_host_reset().

> 

> Signed-off-by: Hannes Reinecke <hare@suse.de>


Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>
diff mbox series

Patch

diff --git a/drivers/scsi/sym53c8xx_2/sym_glue.c b/drivers/scsi/sym53c8xx_2/sym_glue.c
index 16b65fc4405c..dafdab5869fa 100644
--- a/drivers/scsi/sym53c8xx_2/sym_glue.c
+++ b/drivers/scsi/sym53c8xx_2/sym_glue.c
@@ -569,12 +569,11 @@  static void sym53c8xx_timer(struct timer_list *t)
  *  Generic method for our eh processing.
  *  The 'op' argument tells what we have to do.
  */
-static int sym_eh_handler(int op, char *opname, struct scsi_cmnd *cmd)
+static int sym_eh_handler(struct Scsi_Host *shost, int op,
+			  char *opname, struct scsi_cmnd *cmd)
 {
 	struct sym_ucmd *ucmd = SYM_UCMD_PTR(cmd);
-	struct Scsi_Host *shost = cmd->device->host;
 	struct sym_data *sym_data = shost_priv(shost);
-	struct pci_dev *pdev = sym_data->pdev;
 	struct sym_hcb *np = sym_data->ncb;
 	SYM_QUEHEAD *qp;
 	int cmd_queued = 0;
@@ -583,36 +582,6 @@  static int sym_eh_handler(int op, char *opname, struct scsi_cmnd *cmd)
 
 	scmd_printk(KERN_WARNING, cmd, "%s operation started\n", opname);
 
-	/* We may be in an error condition because the PCI bus
-	 * went down. In this case, we need to wait until the
-	 * PCI bus is reset, the card is reset, and only then
-	 * proceed with the scsi error recovery.  There's no
-	 * point in hurrying; take a leisurely wait.
-	 */
-#define WAIT_FOR_PCI_RECOVERY	35
-	if (pci_channel_offline(pdev)) {
-		int finished_reset = 0;
-		init_completion(&eh_done);
-		spin_lock_irq(shost->host_lock);
-		/* Make sure we didn't race */
-		if (pci_channel_offline(pdev)) {
-			BUG_ON(sym_data->io_reset);
-			sym_data->io_reset = &eh_done;
-		} else {
-			finished_reset = 1;
-		}
-		spin_unlock_irq(shost->host_lock);
-		if (!finished_reset)
-			finished_reset = wait_for_completion_timeout
-						(sym_data->io_reset,
-						WAIT_FOR_PCI_RECOVERY*HZ);
-		spin_lock_irq(shost->host_lock);
-		sym_data->io_reset = NULL;
-		spin_unlock_irq(shost->host_lock);
-		if (!finished_reset)
-			return SCSI_FAILED;
-	}
-
 	spin_lock_irq(shost->host_lock);
 	/* This one is queued in some place -> to wait for completion */
 	FOR_EACH_QUEUED_ELEMENT(&np->busy_ccbq, qp) {
@@ -672,22 +641,76 @@  static int sym_eh_handler(int op, char *opname, struct scsi_cmnd *cmd)
  */
 static int sym53c8xx_eh_abort_handler(struct scsi_cmnd *cmd)
 {
-	return sym_eh_handler(SYM_EH_ABORT, "ABORT", cmd);
+	struct Scsi_Host *shost = cmd->device->host;
+	struct sym_data *sym_data = shost_priv(shost);
+	struct pci_dev *pdev = sym_data->pdev;
+
+	if (pci_channel_offline(pdev))
+		return FAILED;
+	return sym_eh_handler(shost, SYM_EH_ABORT, "ABORT", cmd);
 }
 
 static int sym53c8xx_eh_device_reset_handler(struct scsi_cmnd *cmd)
 {
-	return sym_eh_handler(SYM_EH_DEVICE_RESET, "DEVICE RESET", cmd);
+	struct Scsi_Host *shost = cmd->device->host;
+	struct sym_data *sym_data = shost_priv(shost);
+	struct pci_dev *pdev = sym_data->pdev;
+
+	if (pci_channel_offline(pdev))
+		return FAILED;
+	return sym_eh_handler(shost, SYM_EH_DEVICE_RESET, "DEVICE RESET", cmd);
 }
 
 static int sym53c8xx_eh_bus_reset_handler(struct scsi_cmnd *cmd)
 {
-	return sym_eh_handler(SYM_EH_BUS_RESET, "BUS RESET", cmd);
+	struct Scsi_Host *shost = cmd->device->host;
+	struct sym_data *sym_data = shost_priv(shost);
+	struct pci_dev *pdev = sym_data->pdev;
+
+	if (pci_channel_offline(pdev))
+		return FAILED;
+	return sym_eh_handler(shost, SYM_EH_BUS_RESET, "BUS RESET", cmd);
 }
 
 static int sym53c8xx_eh_host_reset_handler(struct scsi_cmnd *cmd)
 {
-	return sym_eh_handler(SYM_EH_HOST_RESET, "HOST RESET", cmd);
+	struct Scsi_Host *shost = cmd->device->host;
+	struct sym_data *sym_data = shost_priv(shost);
+	struct pci_dev *pdev = sym_data->pdev;
+
+	/* We may be in an error condition because the PCI bus
+	 * went down. In this case, we need to wait until the
+	 * PCI bus is reset, the card is reset, and only then
+	 * proceed with the scsi error recovery.  There's no
+	 * point in hurrying; take a leisurely wait.
+	 */
+#define WAIT_FOR_PCI_RECOVERY	35
+	if (pci_channel_offline(pdev)) {
+		struct completion eh_done;
+		int finished_reset = 0;
+
+		init_completion(&eh_done);
+		spin_lock_irq(shost->host_lock);
+		/* Make sure we didn't race */
+		if (pci_channel_offline(pdev)) {
+			BUG_ON(sym_data->io_reset);
+			sym_data->io_reset = &eh_done;
+		} else {
+			finished_reset = 1;
+		}
+		spin_unlock_irq(shost->host_lock);
+		if (!finished_reset)
+			finished_reset = wait_for_completion_timeout
+						(sym_data->io_reset,
+						WAIT_FOR_PCI_RECOVERY*HZ);
+		spin_lock_irq(shost->host_lock);
+		sym_data->io_reset = NULL;
+		spin_unlock_irq(shost->host_lock);
+		if (!finished_reset)
+			return SCSI_FAILED;
+	}
+
+	return sym_eh_handler(shost, SYM_EH_HOST_RESET, "HOST RESET", cmd);
 }
 
 /*