[SCSI] FC transport : Avoid device offline cases by stalling aborts until device unblocked

This moves the eh_timed_out functionality from the scsi_host_template
to the transport_template. Given that this is now a transport function,
the EH_RESET_TIMER case no longer caps the timer reschedulings. The
transport guarantees that this is not an infinite condition.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

4 files changed, 51 insertions, 20 deletions

diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index 5cc97b721661..9cf020622134 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -29,6 +29,7 @@
 #include <scsi/scsi_dbg.h>
 #include <scsi/scsi_device.h>
 #include <scsi/scsi_eh.h>
+#include <scsi/scsi_transport.h>
 #include <scsi/scsi_host.h>
 #include <scsi/scsi_ioctl.h>
 #include <scsi/scsi_request.h>
@@ -163,16 +164,12 @@ void scsi_times_out(struct scsi_cmnd *scmd)
 {
 	scsi_log_completion(scmd, TIMEOUT_ERROR);
 
-	if (scmd->device->host->hostt->eh_timed_out)
-		switch (scmd->device->host->hostt->eh_timed_out(scmd)) {
+	if (scmd->device->host->transportt->eh_timed_out)
+		switch (scmd->device->host->transportt->eh_timed_out(scmd)) {
 		case EH_HANDLED:
 			__scsi_done(scmd);
 			return;
 		case EH_RESET_TIMER:
-			/* This allows a single retry even of a command
-			 * with allowed == 0 */
-			if (scmd->retries++ > scmd->allowed)
-				break;
 			scsi_add_timer(scmd, scmd->timeout_per_command,
 				       scsi_times_out);
 			return;
diff --git a/drivers/scsi/scsi_transport_fc.c b/drivers/scsi/scsi_transport_fc.c
index 56012b2694d2..3c3baa9f5f28 100644
--- a/drivers/scsi/scsi_transport_fc.c
+++ b/drivers/scsi/scsi_transport_fc.c
@@ -31,6 +31,7 @@
 #include <scsi/scsi_host.h>
 #include <scsi/scsi_transport.h>
 #include <scsi/scsi_transport_fc.h>
+#include <scsi/scsi_cmnd.h>
 #include "scsi_priv.h"
 
 /*
@@ -1090,6 +1091,40 @@ static int fc_rport_match(struct attribute_container *cont,
 }
 
 
+/**
+ * fc_timed_out - FC Transport I/O timeout intercept handler
+ *
+ * @scmd:	The SCSI command which timed out
+ *
+ * This routine protects against error handlers getting invoked while a
+ * rport is in a blocked state, typically due to a temporarily loss of
+ * connectivity. If the error handlers are allowed to proceed, requests
+ * to abort i/o, reset the target, etc will likely fail as there is no way
+ * to communicate with the device to perform the requested function. These
+ * failures may result in the midlayer taking the device offline, requiring
+ * manual intervention to restore operation.
+ *
+ * This routine, called whenever an i/o times out, validates the state of
+ * the underlying rport. If the rport is blocked, it returns
+ * EH_RESET_TIMER, which will continue to reschedule the timeout.
+ * Eventually, either the device will return, or devloss_tmo will fire,
+ * and when the timeout then fires, it will be handled normally.
+ * If the rport is not blocked, normal error handling continues.
+ *
+ * Notes:
+ *	This routine assumes no locks are held on entry.
+ **/
+static enum scsi_eh_timer_return
+fc_timed_out(struct scsi_cmnd *scmd)
+{
+	struct fc_rport *rport = starget_to_rport(scsi_target(scmd->device));
+
+	if (rport->port_state == FC_PORTSTATE_BLOCKED)
+		return EH_RESET_TIMER;
+
+	return EH_NOT_HANDLED;
+}
+
 /*
  * Must be called with shost->host_lock held
  */
@@ -1146,6 +1181,8 @@ fc_attach_transport(struct fc_function_template *ft)
 	/* Transport uses the shost workq for scsi scanning */
 	i->t.create_work_queue = 1;
 
+	i->t.eh_timed_out = fc_timed_out;
+
 	i->t.user_scan = fc_user_scan;
 	
 	/*
diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h
index 827992949c4b..a6cf3e535c0b 100644
--- a/include/scsi/scsi_host.h
+++ b/include/scsi/scsi_host.h
@@ -147,20 +147,6 @@ struct scsi_host_template {
 	int (* eh_host_reset_handler)(struct scsi_cmnd *);
 
 	/*
-	 * This is an optional routine to notify the host that the scsi
-	 * timer just fired.  The returns tell the timer routine what to
-	 * do about this:
-	 *
-	 * EH_HANDLED:		I fixed the error, please complete the command
-	 * EH_RESET_TIMER:	I need more time, reset the timer and
-	 *			begin counting again
-	 * EH_NOT_HANDLED	Begin normal error recovery
-	 *
-	 * Status: OPTIONAL
-	 */
-	enum scsi_eh_timer_return (* eh_timed_out)(struct scsi_cmnd *);
-
-	/*
 	 * Before the mid layer attempts to scan for a new device where none
 	 * currently exists, it will call this entry in your driver.  Should
 	 * your driver need to allocate any structs or perform any other init
diff --git a/include/scsi/scsi_transport.h b/include/scsi/scsi_transport.h
index e7b1054adf86..b3657f111937 100644
--- a/include/scsi/scsi_transport.h
+++ b/include/scsi/scsi_transport.h
@@ -48,6 +48,17 @@ struct scsi_transport_template {
 	 * True if the transport wants to use a host-based work-queue
 	 */
 	unsigned int create_work_queue : 1;
+
+	/*
+	 * This is an optional routine that allows the transport to become
+	 * involved when a scsi io timer fires. The return value tells the
+	 * timer routine how to finish the io timeout handling:
+	 * EH_HANDLED:		I fixed the error, please complete the command
+	 * EH_RESET_TIMER:	I need more time, reset the timer and
+	 *			begin counting again
+	 * EH_NOT_HANDLED	Begin normal error recovery
+	 */
+	enum scsi_eh_timer_return (* eh_timed_out)(struct scsi_cmnd *);
 };
 
 #define transport_class_to_shost(tc) \