Skip site navigation (1)Skip section navigation (2) 
Date:      Fri, 29 Jul 2011 20:30:28 +0000 (UTC)
From:      Alexander Motin <mav@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   svn commit: r224496 - head/sys/cam
Message-ID:  <201107292030.p6TKUSaf064895@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help 
Author: mav
Date: Fri Jul 29 20:30:28 2011
New Revision: 224496
URL: http://svn.freebsd.org/changeset/base/224496

Log:
  In some cases failed SATA disks may report their presence, but don't
  respond to any commands. I've found that because of multiple command
  retries, each of which cause 30s timeout, bus reset and another retry or
  requeue for many commands, it may take ages to eventually drop the
  failed device. The odd thing is that those retries continue even after
  XPT considered device as dead and invalidated it.
  
  This patch makes cam_periph_error() to block any command retries after
  periph was marked as invalid. With that patch all activity completes in
  1-2 minutes, just after several timeouts, required to consider device
  death. This should make ZFS, gmirror, graid, etc. operation more robust.
  
  Reviewed by:	mjacob@ on scsi@
  
  Approved by:	re (kib)

Modified:
  head/sys/cam/cam_periph.c

Modified: head/sys/cam/cam_periph.c
==============================================================================
--- head/sys/cam/cam_periph.c	Fri Jul 29 20:24:04 2011	(r224495)
+++ head/sys/cam/cam_periph.c	Fri Jul 29 20:30:28 2011	(r224496)
@@ -1550,7 +1550,8 @@ camperiphscsisenseerror(union ccb *ccb, 
 		 * make sure we actually have retries available.
 		 */
 		if ((err_action & SSQ_DECREMENT_COUNT) != 0) {
-		 	if (ccb->ccb_h.retry_count > 0)
+		 	if (ccb->ccb_h.retry_count > 0 &&
+			    (periph->flags & CAM_PERIPH_INVALID) == 0)
 		 		ccb->ccb_h.retry_count--;
 			else {
 				*action_string = "Retries exhausted";
@@ -1718,6 +1719,7 @@ int
 cam_periph_error(union ccb *ccb, cam_flags camflags,
 		 u_int32_t sense_flags, union ccb *save_ccb)
 {
+	struct cam_periph *periph;
 	const char *action_string;
 	cam_status  status;
 	int	    frozen;
@@ -1725,7 +1727,8 @@ cam_periph_error(union ccb *ccb, cam_fla
 	int         openings;
 	u_int32_t   relsim_flags;
 	u_int32_t   timeout = 0;
-	
+
+	periph = xpt_path_periph(ccb->ccb_h.path);
 	action_string = NULL;
 	status = ccb->ccb_h.status;
 	frozen = (status & CAM_DEV_QFRZN) != 0;
@@ -1787,9 +1790,9 @@ cam_periph_error(union ccb *ccb, cam_fla
 			xpt_print(ccb->ccb_h.path, "Data overrun\n");
 			printed++;
 		}
-		error = EIO;	/* we have to kill the command */
 		/* decrement the number of retries */
-		if (ccb->ccb_h.retry_count > 0) {
+		if (ccb->ccb_h.retry_count > 0 &&
+		    (periph->flags & CAM_PERIPH_INVALID) == 0) {
 			ccb->ccb_h.retry_count--;
 			error = ERESTART;
 		} else {
@@ -1808,7 +1811,8 @@ cam_periph_error(union ccb *ccb, cam_fla
 		struct cam_path *newpath;
 
 		if ((camflags & CAM_RETRY_SELTO) != 0) {
-			if (ccb->ccb_h.retry_count > 0) {
+			if (ccb->ccb_h.retry_count > 0 &&
+			    (periph->flags & CAM_PERIPH_INVALID) == 0) {
 
 				ccb->ccb_h.retry_count--;
 				error = ERESTART;
@@ -1826,10 +1830,11 @@ cam_periph_error(union ccb *ccb, cam_fla
 				timeout = periph_selto_delay;
 				break;
 			}
+			action_string = "Retries exhausted";
 		}
 		error = ENXIO;
 		/* Should we do more if we can't create the path?? */
-		if (xpt_create_path(&newpath, xpt_path_periph(ccb->ccb_h.path),
+		if (xpt_create_path(&newpath, periph,
 				    xpt_path_path_id(ccb->ccb_h.path),
 				    xpt_path_target_id(ccb->ccb_h.path),
 				    CAM_LUN_WILDCARD) != CAM_REQ_CMP) 
@@ -1874,11 +1879,16 @@ cam_periph_error(union ccb *ccb, cam_fla
 		/* FALLTHROUGH */
 	case CAM_REQUEUE_REQ:
 		/* Unconditional requeue */
-		error = ERESTART;
 		if (bootverbose && printed == 0) {
 			xpt_print(ccb->ccb_h.path, "Request requeued\n");
 			printed++;
 		}
+		if ((periph->flags & CAM_PERIPH_INVALID) == 0)
+			error = ERESTART;
+		else {
+			action_string = "Retries exhausted";
+			error = EIO;
+		}
 		break;
 	case CAM_RESRC_UNAVAIL:
 		/* Wait a bit for the resource shortage to abate. */
@@ -1893,7 +1903,8 @@ cam_periph_error(union ccb *ccb, cam_fla
 		/* FALLTHROUGH */
 	default:
 		/* decrement the number of retries */
-		if (ccb->ccb_h.retry_count > 0) {
+		if (ccb->ccb_h.retry_count > 0 &&
+		    (periph->flags & CAM_PERIPH_INVALID) == 0) {
 			ccb->ccb_h.retry_count--;
 			error = ERESTART;
 			if (bootverbose && printed == 0) {

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201107292030.p6TKUSaf064895>