block: unify request timeout handling

Right now SCSI and others do their own command timeout handling.
Move those bits to the block layer.

Instead of having a timer per command, we try to be a bit more clever
and simply have one per-queue. This avoids the overhead of having to
tear down and setup a timer for each command, so it will result in a lot
less timer fiddling.

Signed-off-by: Mike Anderson <andmike@linux.vnet.ibm.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

block/Makefile

@@ -4,8 +4,8 @@
 
 obj-$(CONFIG_BLOCK) := elevator.o blk-core.o blk-tag.o blk-sysfs.o \
 			blk-barrier.o blk-settings.o blk-ioc.o blk-map.o \
-			blk-exec.o blk-merge.o blk-softirq.o ioctl.o genhd.o \
-			scsi_ioctl.o cmd-filter.o
+			blk-exec.o blk-merge.o blk-softirq.o blk-timeout.o \
+			ioctl.o genhd.o scsi_ioctl.o cmd-filter.o
 
 obj-$(CONFIG_BLK_DEV_BSG)	+= bsg.o
 obj-$(CONFIG_IOSCHED_NOOP)	+= noop-iosched.o

block/blk-core.c

@@ -110,6 +110,7 @@ void blk_rq_init(struct request_queue *q, struct request *rq)
 	memset(rq, 0, sizeof(*rq));
 
 	INIT_LIST_HEAD(&rq->queuelist);
+	INIT_LIST_HEAD(&rq->timeout_list);
 	rq->cpu = -1;
 	rq->q = q;
 	rq->sector = rq->hard_sector = (sector_t) -1;
@@ -490,6 +491,8 @@ struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
 	}
 
 	init_timer(&q->unplug_timer);
+	setup_timer(&q->timeout, blk_rq_timed_out_timer, (unsigned long) q);
+	INIT_LIST_HEAD(&q->timeout_list);
 
 	kobject_init(&q->kobj, &blk_queue_ktype);
 
@@ -897,6 +900,8 @@ EXPORT_SYMBOL(blk_start_queueing);
  */
 void blk_requeue_request(struct request_queue *q, struct request *rq)
 {
+	blk_delete_timer(rq);
+	blk_clear_rq_complete(rq);
 	blk_add_trace_rq(q, rq, BLK_TA_REQUEUE);
 
 	if (blk_rq_tagged(rq))
@@ -1650,6 +1655,8 @@ static void end_that_request_last(struct request *req, int error)
 {
 	struct gendisk *disk = req->rq_disk;
 
+	blk_delete_timer(req);
+
 	if (blk_rq_tagged(req))
 		blk_queue_end_tag(req->q, req);
 

block/blk-settings.c

@@ -77,6 +77,18 @@ void blk_queue_softirq_done(struct request_queue *q, softirq_done_fn *fn)
 }
 EXPORT_SYMBOL(blk_queue_softirq_done);
 
+void blk_queue_rq_timeout(struct request_queue *q, unsigned int timeout)
+{
+	q->rq_timeout = timeout;
+}
+EXPORT_SYMBOL_GPL(blk_queue_rq_timeout);
+
+void blk_queue_rq_timed_out(struct request_queue *q, rq_timed_out_fn *fn)
+{
+	q->rq_timed_out_fn = fn;
+}
+EXPORT_SYMBOL_GPL(blk_queue_rq_timed_out);
+
 /**
  * blk_queue_make_request - define an alternate make_request function for a device
  * @q:  the request queue for the device to be affected

block/blk-softirq.c

@@ -101,18 +101,7 @@ static struct notifier_block __cpuinitdata blk_cpu_notifier = {
 	.notifier_call	= blk_cpu_notify,
 };
 
-/**
- * blk_complete_request - end I/O on a request
- * @req:      the request being processed
- *
- * Description:
- *     Ends all I/O on a request. It does not handle partial completions,
- *     unless the driver actually implements this in its completion callback
- *     through requeueing. The actual completion happens out-of-order,
- *     through a softirq handler. The user must have registered a completion
- *     callback through blk_queue_softirq_done().
- **/
-void blk_complete_request(struct request *req)
+void __blk_complete_request(struct request *req)
 {
 	struct request_queue *q = req->q;
 	unsigned long flags;
@@ -151,6 +140,23 @@ do_local:
 
 	local_irq_restore(flags);
 }
+
+/**
+ * blk_complete_request - end I/O on a request
+ * @req:      the request being processed
+ *
+ * Description:
+ *     Ends all I/O on a request. It does not handle partial completions,
+ *     unless the driver actually implements this in its completion callback
+ *     through requeueing. The actual completion happens out-of-order,
+ *     through a softirq handler. The user must have registered a completion
+ *     callback through blk_queue_softirq_done().
+ **/
+void blk_complete_request(struct request *req)
+{
+	if (!blk_mark_rq_complete(req))
+		__blk_complete_request(req);
+}
 EXPORT_SYMBOL(blk_complete_request);
 
 __init int blk_softirq_init(void)

block/blk-timeout.c

@@ -0,0 +1,155 @@
+/*
+ * Functions related to generic timeout handling of requests.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/blkdev.h>
+
+#include "blk.h"
+
+/*
+ * blk_delete_timer - Delete/cancel timer for a given function.
+ * @req:	request that we are canceling timer for
+ *
+ */
+void blk_delete_timer(struct request *req)
+{
+	struct request_queue *q = req->q;
+
+	/*
+	 * Nothing to detach
+	 */
+	if (!q->rq_timed_out_fn || !req->deadline)
+		return;
+
+	list_del_init(&req->timeout_list);
+
+	if (list_empty(&q->timeout_list))
+		del_timer(&q->timeout);
+}
+
+static void blk_rq_timed_out(struct request *req)
+{
+	struct request_queue *q = req->q;
+	enum blk_eh_timer_return ret;
+
+	ret = q->rq_timed_out_fn(req);
+	switch (ret) {
+	case BLK_EH_HANDLED:
+		__blk_complete_request(req);
+		break;
+	case BLK_EH_RESET_TIMER:
+		blk_clear_rq_complete(req);
+		blk_add_timer(req);
+		break;
+	case BLK_EH_NOT_HANDLED:
+		/*
+		 * LLD handles this for now but in the future
+		 * we can send a request msg to abort the command
+		 * and we can move more of the generic scsi eh code to
+		 * the blk layer.
+		 */
+		break;
+	default:
+		printk(KERN_ERR "block: bad eh return: %d\n", ret);
+		break;
+	}
+}
+
+void blk_rq_timed_out_timer(unsigned long data)
+{
+	struct request_queue *q = (struct request_queue *) data;
+	unsigned long flags, uninitialized_var(next), next_set = 0;
+	struct request *rq, *tmp;
+
+	spin_lock_irqsave(q->queue_lock, flags);
+
+	list_for_each_entry_safe(rq, tmp, &q->timeout_list, timeout_list) {
+		if (time_after_eq(jiffies, rq->deadline)) {
+			list_del_init(&rq->timeout_list);
+
+			/*
+			 * Check if we raced with end io completion
+			 */
+			if (blk_mark_rq_complete(rq))
+				continue;
+			blk_rq_timed_out(rq);
+		}
+		if (!next_set) {
+			next = rq->deadline;
+			next_set = 1;
+		} else if (time_after(next, rq->deadline))
+			next = rq->deadline;
+	}
+
+	if (next_set && !list_empty(&q->timeout_list))
+		mod_timer(&q->timeout, round_jiffies(next));
+
+	spin_unlock_irqrestore(q->queue_lock, flags);
+}
+
+/**
+ * blk_abort_request -- Request request recovery for the specified command
+ * @req:	pointer to the request of interest
+ *
+ * This function requests that the block layer start recovery for the
+ * request by deleting the timer and calling the q's timeout function.
+ * LLDDs who implement their own error recovery MAY ignore the timeout
+ * event if they generated blk_abort_req. Must hold queue lock.
+ */
+void blk_abort_request(struct request *req)
+{
+	blk_delete_timer(req);
+	blk_rq_timed_out(req);
+}
+EXPORT_SYMBOL_GPL(blk_abort_request);
+
+/**
+ * blk_add_timer - Start timeout timer for a single request
+ * @req:	request that is about to start running.
+ *
+ * Notes:
+ *    Each request has its own timer, and as it is added to the queue, we
+ *    set up the timer. When the request completes, we cancel the timer.
+ */
+void blk_add_timer(struct request *req)
+{
+	struct request_queue *q = req->q;
+	unsigned long expiry;
+
+	if (!q->rq_timed_out_fn)
+		return;
+
+	BUG_ON(!list_empty(&req->timeout_list));
+	BUG_ON(test_bit(REQ_ATOM_COMPLETE, &req->atomic_flags));
+
+	if (req->timeout)
+		req->deadline = jiffies + req->timeout;
+	else {
+		req->deadline = jiffies + q->rq_timeout;
+		/*
+		 * Some LLDs, like scsi, peek at the timeout to prevent
+		 * a command from being retried forever.
+		 */
+		req->timeout = q->rq_timeout;
+	}
+	list_add_tail(&req->timeout_list, &q->timeout_list);
+
+	/*
+	 * If the timer isn't already pending or this timeout is earlier
+	 * than an existing one, modify the timer. Round to next nearest
+	 * second.
+	 */
+	expiry = round_jiffies(req->deadline);
+
+	/*
+	 * We use ->deadline == 0 to detect whether a timer was added or
+	 * not, so just increase to next jiffy for that specific case
+	 */
+	if (unlikely(!req->deadline))
+		req->deadline = 1;
+
+	if (!timer_pending(&q->timeout) ||
+	    time_before(expiry, q->timeout.expires))
+		mod_timer(&q->timeout, expiry);
+}

block/blk.h

@@ -17,6 +17,30 @@ void __blk_queue_free_tags(struct request_queue *q);
 
 void blk_unplug_work(struct work_struct *work);
 void blk_unplug_timeout(unsigned long data);
+void blk_rq_timed_out_timer(unsigned long data);
+void blk_delete_timer(struct request *);
+void blk_add_timer(struct request *);
+
+/*
+ * Internal atomic flags for request handling
+ */
+enum rq_atomic_flags {
+	REQ_ATOM_COMPLETE = 0,
+};
+
+/*
+ * EH timer and IO completion will both attempt to 'grab' the request, make
+ * sure that only one of them suceeds
+ */
+static inline int blk_mark_rq_complete(struct request *rq)
+{
+	return test_and_set_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags);
+}
+
+static inline void blk_clear_rq_complete(struct request *rq)
+{
+	clear_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags);
+}
 
 struct io_context *current_io_context(gfp_t gfp_flags, int node);
 

block/elevator.c

@@ -36,6 +36,8 @@
 #include <linux/hash.h>
 #include <linux/uaccess.h>
 
+#include "blk.h"
+
 static DEFINE_SPINLOCK(elv_list_lock);
 static LIST_HEAD(elv_list);
 
@@ -771,6 +773,12 @@ struct request *elv_next_request(struct request_queue *q)
 			 */
 			rq->cmd_flags |= REQ_STARTED;
 			blk_add_trace_rq(q, rq, BLK_TA_ISSUE);
+
+			/*
+			 * We are now handing the request to the hardware,
+			 * add the timeout handler
+			 */
+			blk_add_timer(rq);
 		}
 
 		if (!q->boundary_rq || q->boundary_rq == rq) {

drivers/ata/libata-eh.c

@@ -33,6 +33,7 @@
  */
 
 #include <linux/kernel.h>
+#include <linux/blkdev.h>
 #include <linux/pci.h>
 #include <scsi/scsi.h>
 #include <scsi/scsi_host.h>
@@ -457,29 +458,29 @@ static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
  *	RETURNS:
  *	EH_HANDLED or EH_NOT_HANDLED
  */
-enum scsi_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
+enum blk_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
 {
 	struct Scsi_Host *host = cmd->device->host;
 	struct ata_port *ap = ata_shost_to_port(host);
 	unsigned long flags;
 	struct ata_queued_cmd *qc;
-	enum scsi_eh_timer_return ret;
+	enum blk_eh_timer_return ret;
 
 	DPRINTK("ENTER\n");
 
 	if (ap->ops->error_handler) {
-		ret = EH_NOT_HANDLED;
+		ret = BLK_EH_NOT_HANDLED;
 		goto out;
 	}
 
-	ret = EH_HANDLED;
+	ret = BLK_EH_HANDLED;
 	spin_lock_irqsave(ap->lock, flags);
 	qc = ata_qc_from_tag(ap, ap->link.active_tag);
 	if (qc) {
 		WARN_ON(qc->scsicmd != cmd);
 		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
 		qc->err_mask |= AC_ERR_TIMEOUT;
-		ret = EH_NOT_HANDLED;
+		ret = BLK_EH_NOT_HANDLED;
 	}
 	spin_unlock_irqrestore(ap->lock, flags);
 
@@ -831,7 +832,7 @@ void ata_qc_schedule_eh(struct ata_queued_cmd *qc)
 	 * Note that ATA_QCFLAG_FAILED is unconditionally set after
 	 * this function completes.
 	 */
-	scsi_req_abort_cmd(qc->scsicmd);
+	blk_abort_request(qc->scsicmd->request);
 }
 
 /**

drivers/ata/libata.h

@@ -152,7 +152,7 @@ extern int ata_bus_probe(struct ata_port *ap);
 /* libata-eh.c */
 extern unsigned long ata_internal_cmd_timeout(struct ata_device *dev, u8 cmd);
 extern void ata_internal_cmd_timed_out(struct ata_device *dev, u8 cmd);
-extern enum scsi_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd);
+extern enum blk_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd);
 extern void ata_scsi_error(struct Scsi_Host *host);
 extern void ata_port_wait_eh(struct ata_port *ap);
 extern void ata_eh_fastdrain_timerfn(unsigned long arg);

drivers/scsi/aacraid/aachba.c

@@ -1139,7 +1139,7 @@ static struct aac_srb * aac_scsi_common(struct fib * fib, struct scsi_cmnd * cmd
 	srbcmd->id       = cpu_to_le32(scmd_id(cmd));
 	srbcmd->lun      = cpu_to_le32(cmd->device->lun);
 	srbcmd->flags    = cpu_to_le32(flag);
-	timeout = cmd->timeout_per_command/HZ;
+	timeout = cmd->request->timeout/HZ;
 	if (timeout == 0)
 		timeout = 1;
 	srbcmd->timeout  = cpu_to_le32(timeout);  // timeout in seconds

drivers/scsi/gdth.c

@@ -464,7 +464,6 @@ int __gdth_execute(struct scsi_device *sdev, gdth_cmd_str *gdtcmd, char *cmnd,
 
     /* use request field to save the ptr. to completion struct. */
     scp->request = (struct request *)&wait;
-    scp->timeout_per_command = timeout*HZ;
     scp->cmd_len = 12;
     scp->cmnd = cmnd;
     cmndinfo.priority = IOCTL_PRI;
@@ -1995,23 +1994,12 @@ static void gdth_putq(gdth_ha_str *ha, Scsi_Cmnd *scp, unchar priority)
     register Scsi_Cmnd *pscp;
     register Scsi_Cmnd *nscp;
     ulong flags;
-    unchar b, t;
 
     TRACE(("gdth_putq() priority %d\n",priority));
     spin_lock_irqsave(&ha->smp_lock, flags);
 
-    if (!cmndinfo->internal_command) {
+    if (!cmndinfo->internal_command)
         cmndinfo->priority = priority;
-        b = scp->device->channel;
-        t = scp->device->id;
-        if (priority >= DEFAULT_PRI) {
-            if ((b != ha->virt_bus && ha->raw[BUS_L2P(ha,b)].lock) ||
-                (b==ha->virt_bus && t<MAX_HDRIVES && ha->hdr[t].lock)) {
-                TRACE2(("gdth_putq(): locked IO ->update_timeout()\n"));
-                cmndinfo->timeout = gdth_update_timeout(scp, 0);
-            }
-        }
-    }
 
     if (ha->req_first==NULL) {
         ha->req_first = scp;                    /* queue was empty */
@@ -3899,6 +3887,39 @@ static const char *gdth_info(struct Scsi_Host *shp)
     return ((const char *)ha->binfo.type_string);
 }
 
+static enum blk_eh_timer_return gdth_timed_out(struct scsi_cmnd *scp)
+{
+	gdth_ha_str *ha = shost_priv(scp->device->host);
+	struct gdth_cmndinfo *cmndinfo = gdth_cmnd_priv(scp);
+	unchar b, t;
+	ulong flags;
+	enum blk_eh_timer_return retval = BLK_EH_NOT_HANDLED;
+
+	TRACE(("%s() cmd 0x%x\n", scp->cmnd[0], __func__));
+	b = scp->device->channel;
+	t = scp->device->id;
+
+	/*
+	 * We don't really honor the command timeout, but we try to
+	 * honor 6 times of the actual command timeout! So reset the
+	 * timer if this is less than 6th timeout on this command!
+	 */
+	if (++cmndinfo->timeout_count < 6)
+		retval = BLK_EH_RESET_TIMER;
+
+	/* Reset the timeout if it is locked IO */
+	spin_lock_irqsave(&ha->smp_lock, flags);
+	if ((b != ha->virt_bus && ha->raw[BUS_L2P(ha, b)].lock) ||
+	    (b == ha->virt_bus && t < MAX_HDRIVES && ha->hdr[t].lock)) {
+		TRACE2(("%s(): locked IO, reset timeout\n", __func__));
+		retval = BLK_EH_RESET_TIMER;
+	}
+	spin_unlock_irqrestore(&ha->smp_lock, flags);
+
+	return retval;
+}
+
+
 static int gdth_eh_bus_reset(Scsi_Cmnd *scp)
 {
     gdth_ha_str *ha = shost_priv(scp->device->host);
@@ -3992,7 +4013,7 @@ static int gdth_queuecommand(struct scsi_cmnd *scp,
     BUG_ON(!cmndinfo);
 
     scp->scsi_done = done;
-    gdth_update_timeout(scp, scp->timeout_per_command * 6);
+    cmndinfo->timeout_count = 0;
     cmndinfo->priority = DEFAULT_PRI;
 
     return __gdth_queuecommand(ha, scp, cmndinfo);
@@ -4096,12 +4117,10 @@ static int ioc_lockdrv(void __user *arg)
             ha->hdr[j].lock = 1;
             spin_unlock_irqrestore(&ha->smp_lock, flags);
             gdth_wait_completion(ha, ha->bus_cnt, j);
-            gdth_stop_timeout(ha, ha->bus_cnt, j);
         } else {
             spin_lock_irqsave(&ha->smp_lock, flags);
             ha->hdr[j].lock = 0;
             spin_unlock_irqrestore(&ha->smp_lock, flags);
-            gdth_start_timeout(ha, ha->bus_cnt, j);
             gdth_next(ha);
         }
     } 
@@ -4539,18 +4558,14 @@ static int gdth_ioctl(struct inode *inode, struct file *filep,
                 spin_lock_irqsave(&ha->smp_lock, flags);
                 ha->raw[i].lock = 1;
                 spin_unlock_irqrestore(&ha->smp_lock, flags);
-                for (j = 0; j < ha->tid_cnt; ++j) {
+		for (j = 0; j < ha->tid_cnt; ++j)
                     gdth_wait_completion(ha, i, j);
-                    gdth_stop_timeout(ha, i, j);
-                }
             } else {
                 spin_lock_irqsave(&ha->smp_lock, flags);
                 ha->raw[i].lock = 0;
                 spin_unlock_irqrestore(&ha->smp_lock, flags);
-                for (j = 0; j < ha->tid_cnt; ++j) {
-                    gdth_start_timeout(ha, i, j);
+		for (j = 0; j < ha->tid_cnt; ++j)
                     gdth_next(ha);
-                }
             }
         } 
         break;
@@ -4644,6 +4659,7 @@ static struct scsi_host_template gdth_template = {
         .slave_configure        = gdth_slave_configure,
         .bios_param             = gdth_bios_param,
         .proc_info              = gdth_proc_info,
+	.eh_timed_out		= gdth_timed_out,
         .proc_name              = "gdth",
         .can_queue              = GDTH_MAXCMDS,
         .this_id                = -1,

drivers/scsi/gdth.h

@@ -916,7 +916,7 @@ typedef struct {
         gdth_cmd_str *internal_cmd_str;         /* crier for internal messages*/
         dma_addr_t sense_paddr;                 /* sense dma-addr */
         unchar priority;
-        int timeout;
+	int timeout_count;			/* # of timeout calls */
         volatile int wait_for_completion;
         ushort status;
         ulong32 info;

drivers/scsi/gdth_proc.c

@@ -748,69 +748,3 @@ static void gdth_wait_completion(gdth_ha_str *ha, int busnum, int id)
     }
     spin_unlock_irqrestore(&ha->smp_lock, flags);
 }
-
-static void gdth_stop_timeout(gdth_ha_str *ha, int busnum, int id)
-{
-    ulong flags;
-    Scsi_Cmnd *scp;
-    unchar b, t;
-
-    spin_lock_irqsave(&ha->smp_lock, flags);
-
-    for (scp = ha->req_first; scp; scp = (Scsi_Cmnd *)scp->SCp.ptr) {
-        struct gdth_cmndinfo *cmndinfo = gdth_cmnd_priv(scp);
-        if (!cmndinfo->internal_command) {
-            b = scp->device->channel;
-            t = scp->device->id;
-            if (t == (unchar)id && b == (unchar)busnum) {
-                TRACE2(("gdth_stop_timeout(): update_timeout()\n"));
-                cmndinfo->timeout = gdth_update_timeout(scp, 0);
-            }
-        }
-    }
-    spin_unlock_irqrestore(&ha->smp_lock, flags);
-}
-
-static void gdth_start_timeout(gdth_ha_str *ha, int busnum, int id)
-{
-    ulong flags;
-    Scsi_Cmnd *scp;
-    unchar b, t;
-
-    spin_lock_irqsave(&ha->smp_lock, flags);
-
-    for (scp = ha->req_first; scp; scp = (Scsi_Cmnd *)scp->SCp.ptr) {
-        struct gdth_cmndinfo *cmndinfo = gdth_cmnd_priv(scp);
-        if (!cmndinfo->internal_command) {
-            b = scp->device->channel;
-            t = scp->device->id;
-            if (t == (unchar)id && b == (unchar)busnum) {
-                TRACE2(("gdth_start_timeout(): update_timeout()\n"));
-                gdth_update_timeout(scp, cmndinfo->timeout);
-            }
-        }
-    }
-    spin_unlock_irqrestore(&ha->smp_lock, flags);
-}
-
-static int gdth_update_timeout(Scsi_Cmnd *scp, int timeout)
-{
-    int oldto;
-
-    oldto = scp->timeout_per_command;
-    scp->timeout_per_command = timeout;
-
-    if (timeout == 0) {
-        del_timer(&scp->eh_timeout);
-        scp->eh_timeout.data = (unsigned long) NULL;
-        scp->eh_timeout.expires = 0;
-    } else {
-        if (scp->eh_timeout.data != (unsigned long) NULL) 
-            del_timer(&scp->eh_timeout);
-        scp->eh_timeout.data = (unsigned long) scp;
-        scp->eh_timeout.expires = jiffies + timeout;
-        add_timer(&scp->eh_timeout);
-    }
-
-    return oldto;
-}

drivers/scsi/gdth_proc.h

@@ -20,9 +20,6 @@ static char *gdth_ioctl_alloc(gdth_ha_str *ha, int size, int scratch,
                               ulong64 *paddr);
 static void gdth_ioctl_free(gdth_ha_str *ha, int size, char *buf, ulong64 paddr);
 static void gdth_wait_completion(gdth_ha_str *ha, int busnum, int id);
-static void gdth_stop_timeout(gdth_ha_str *ha, int busnum, int id);
-static void gdth_start_timeout(gdth_ha_str *ha, int busnum, int id);
-static int gdth_update_timeout(Scsi_Cmnd *scp, int timeout);
 
 #endif
 

drivers/scsi/ibmvscsi/ibmvscsi.c

@@ -756,7 +756,7 @@ static int ibmvscsi_queuecommand(struct scsi_cmnd *cmnd,
 	init_event_struct(evt_struct,
 			  handle_cmd_rsp,
 			  VIOSRP_SRP_FORMAT,
-			  cmnd->timeout_per_command/HZ);
+			  cmnd->request->timeout/HZ);
 
 	evt_struct->cmnd = cmnd;
 	evt_struct->cmnd_done = done;

drivers/scsi/ide-scsi.c

@@ -612,7 +612,7 @@ static int idescsi_queue (struct scsi_cmnd *cmd,
 	pc->req_xfer = pc->buf_size = scsi_bufflen(cmd);
 	pc->scsi_cmd = cmd;
 	pc->done = done;
-	pc->timeout = jiffies + cmd->timeout_per_command;
+	pc->timeout = jiffies + cmd->request->timeout;
 
 	if (test_bit(IDESCSI_LOG_CMD, &scsi->log)) {
 		printk ("ide-scsi: %s: que %lu, cmd = ", drive->name, cmd->serial_number);

drivers/scsi/ipr.c

@@ -3670,7 +3670,8 @@ static int ipr_slave_configure(struct scsi_device *sdev)
 			sdev->no_uld_attach = 1;
 		}
 		if (ipr_is_vset_device(res)) {
-			sdev->timeout = IPR_VSET_RW_TIMEOUT;
+			blk_queue_rq_timeout(sdev->request_queue,
+					     IPR_VSET_RW_TIMEOUT);
 			blk_queue_max_sectors(sdev->request_queue, IPR_VSET_MAX_SECTORS);
 		}
 		if (ipr_is_vset_device(res) || ipr_is_scsi_disk(res))

drivers/scsi/ips.c

@@ -3818,7 +3818,7 @@ ips_send_cmd(ips_ha_t * ha, ips_scb_t * scb)
 		scb->cmd.dcdb.segment_4G = 0;
 		scb->cmd.dcdb.enhanced_sg = 0;
 
-		TimeOut = scb->scsi_cmd->timeout_per_command;
+		TimeOut = scb->scsi_cmd->request->timeout;
 
 		if (ha->subsys->param[4] & 0x00100000) {	/* If NEW Tape DCDB is Supported */
 			if (!scb->sg_len) {

drivers/scsi/libiscsi.c

@@ -1476,12 +1476,12 @@ static void iscsi_start_tx(struct iscsi_conn *conn)
 		scsi_queue_work(conn->session->host, &conn->xmitwork);
 }
 
-static enum scsi_eh_timer_return iscsi_eh_cmd_timed_out(struct scsi_cmnd *scmd)
+static enum blk_eh_timer_return iscsi_eh_cmd_timed_out(struct scsi_cmnd *scmd)
 {
 	struct iscsi_cls_session *cls_session;
 	struct iscsi_session *session;
 	struct iscsi_conn *conn;
-	enum scsi_eh_timer_return rc = EH_NOT_HANDLED;
+	enum blk_eh_timer_return rc = BLK_EH_NOT_HANDLED;
 
 	cls_session = starget_to_session(scsi_target(scmd->device));
 	session = cls_session->dd_data;
@@ -1494,14 +1494,14 @@ static enum scsi_eh_timer_return iscsi_eh_cmd_timed_out(struct scsi_cmnd *scmd)
 		 * We are probably in the middle of iscsi recovery so let
 		 * that complete and handle the error.
 		 */
-		rc = EH_RESET_TIMER;
+		rc = BLK_EH_RESET_TIMER;
 		goto done;
 	}
 
 	conn = session->leadconn;
 	if (!conn) {
 		/* In the middle of shuting down */
-		rc = EH_RESET_TIMER;
+		rc = BLK_EH_RESET_TIMER;
 		goto done;
 	}
 
@@ -1513,20 +1513,21 @@ static enum scsi_eh_timer_return iscsi_eh_cmd_timed_out(struct scsi_cmnd *scmd)
 	 */
 	if (time_before_eq(conn->last_recv + (conn->recv_timeout * HZ) +
 			    (conn->ping_timeout * HZ), jiffies))
-		rc = EH_RESET_TIMER;
+		rc = BLK_EH_RESET_TIMER;
 	/*
 	 * if we are about to check the transport then give the command
 	 * more time
 	 */
 	if (time_before_eq(conn->last_recv + (conn->recv_timeout * HZ),
 			   jiffies))
-		rc = EH_RESET_TIMER;
+		rc = BLK_EH_RESET_TIMER;
 	/* if in the middle of checking the transport then give us more time */
 	if (conn->ping_task)
-		rc = EH_RESET_TIMER;
+		rc = BLK_EH_RESET_TIMER;
 done:
 	spin_unlock(&session->lock);
-	debug_scsi("return %s\n", rc == EH_RESET_TIMER ? "timer reset" : "nh");
+	debug_scsi("return %s\n", rc == BLK_EH_RESET_TIMER ?
+					"timer reset" : "nh");
 	return rc;
 }
 

drivers/scsi/libsas/sas_ata.c

@@ -398,7 +398,7 @@ void sas_ata_task_abort(struct sas_task *task)
 
 	/* Bounce SCSI-initiated commands to the SCSI EH */
 	if (qc->scsicmd) {
-		scsi_req_abort_cmd(qc->scsicmd);
+		blk_abort_request(qc->scsicmd->request);
 		scsi_schedule_eh(qc->scsicmd->device->host);
 		return;
 	}

drivers/scsi/libsas/sas_internal.h

@@ -55,7 +55,7 @@ void sas_unregister_phys(struct sas_ha_struct *sas_ha);
 int  sas_register_ports(struct sas_ha_struct *sas_ha);
 void sas_unregister_ports(struct sas_ha_struct *sas_ha);
 
-enum scsi_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *);
+enum blk_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *);
 
 int  sas_init_queue(struct sas_ha_struct *sas_ha);
 int  sas_init_events(struct sas_ha_struct *sas_ha);

drivers/scsi/libsas/sas_scsi_host.c

@@ -673,43 +673,43 @@ out:
 	return;
 }
 
-enum scsi_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *cmd)
+enum blk_eh_timer_return sas_scsi_timed_out(struct scsi_cmnd *cmd)
 {
 	struct sas_task *task = TO_SAS_TASK(cmd);
 	unsigned long flags;
 
 	if (!task) {
-		cmd->timeout_per_command /= 2;
+		cmd->request->timeout /= 2;
 		SAS_DPRINTK("command 0x%p, task 0x%p, gone: %s\n",
-			    cmd, task, (cmd->timeout_per_command ?
-			    "EH_RESET_TIMER" : "EH_NOT_HANDLED"));
-		if (!cmd->timeout_per_command)
-			return EH_NOT_HANDLED;
-		return EH_RESET_TIMER;
+			    cmd, task, (cmd->request->timeout ?
+			    "BLK_EH_RESET_TIMER" : "BLK_EH_NOT_HANDLED"));
+		if (!cmd->request->timeout)
+			return BLK_EH_NOT_HANDLED;
+		return BLK_EH_RESET_TIMER;
 	}
 
 	spin_lock_irqsave(&task->task_state_lock, flags);
 	BUG_ON(task->task_state_flags & SAS_TASK_STATE_ABORTED);
 	if (task->task_state_flags & SAS_TASK_STATE_DONE) {
 		spin_unlock_irqrestore(&task->task_state_lock, flags);
-		SAS_DPRINTK("command 0x%p, task 0x%p, timed out: EH_HANDLED\n",
-			    cmd, task);
-		return EH_HANDLED;
+		SAS_DPRINTK("command 0x%p, task 0x%p, timed out: "
+			    "BLK_EH_HANDLED\n", cmd, task);
+		return BLK_EH_HANDLED;
 	}
 	if (!(task->task_state_flags & SAS_TASK_AT_INITIATOR)) {
 		spin_unlock_irqrestore(&task->task_state_lock, flags);
 		SAS_DPRINTK("command 0x%p, task 0x%p, not at initiator: "
-			    "EH_RESET_TIMER\n",
+			    "BLK_EH_RESET_TIMER\n",
 			    cmd, task);
-		return EH_RESET_TIMER;
+		return BLK_EH_RESET_TIMER;
 	}
 	task->task_state_flags |= SAS_TASK_STATE_ABORTED;
 	spin_unlock_irqrestore(&task->task_state_lock, flags);
 
-	SAS_DPRINTK("command 0x%p, task 0x%p, timed out: EH_NOT_HANDLED\n",
+	SAS_DPRINTK("command 0x%p, task 0x%p, timed out: BLK_EH_NOT_HANDLED\n",
 		    cmd, task);
 
-	return EH_NOT_HANDLED;
+	return BLK_EH_NOT_HANDLED;
 }
 
 int sas_ioctl(struct scsi_device *sdev, int cmd, void __user *arg)
@@ -1039,7 +1039,7 @@ void sas_task_abort(struct sas_task *task)
 		return;
 	}
 
-	scsi_req_abort_cmd(sc);
+	blk_abort_request(sc->request);
 	scsi_schedule_eh(sc->device->host);
 }
 

drivers/scsi/megaraid/megaraid_sas.c

@@ -1167,7 +1167,7 @@ static int megasas_generic_reset(struct scsi_cmnd *scmd)
  * cmd has not been completed within the timeout period.
  */
 static enum
-scsi_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
+blk_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
 {
 	struct megasas_cmd *cmd = (struct megasas_cmd *)scmd->SCp.ptr;
 	struct megasas_instance *instance;
@@ -1175,7 +1175,7 @@ scsi_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
 
 	if (time_after(jiffies, scmd->jiffies_at_alloc +
 				(MEGASAS_DEFAULT_CMD_TIMEOUT * 2) * HZ)) {
-		return EH_NOT_HANDLED;
+		return BLK_EH_NOT_HANDLED;
 	}
 
 	instance = cmd->instance;
@@ -1189,7 +1189,7 @@ scsi_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
 
 		spin_unlock_irqrestore(instance->host->host_lock, flags);
 	}
-	return EH_RESET_TIMER;
+	return BLK_EH_RESET_TIMER;
 }
 
 /**

drivers/scsi/ncr53c8xx.c

@@ -4170,8 +4170,8 @@ static int ncr_queue_command (struct ncb *np, struct scsi_cmnd *cmd)
 	**
 	**----------------------------------------------------
 	*/
-	if (np->settle_time && cmd->timeout_per_command >= HZ) {
-		u_long tlimit = jiffies + cmd->timeout_per_command - HZ;
+	if (np->settle_time && cmd->request->timeout >= HZ) {
+		u_long tlimit = jiffies + cmd->request->timeout - HZ;
 		if (time_after(np->settle_time, tlimit))
 			np->settle_time = tlimit;
 	}

drivers/scsi/qla1280.c

@@ -2845,7 +2845,7 @@ qla1280_64bit_start_scsi(struct scsi_qla_host *ha, struct srb * sp)
 	memset(((char *)pkt + 8), 0, (REQUEST_ENTRY_SIZE - 8));
 
 	/* Set ISP command timeout. */
-	pkt->timeout = cpu_to_le16(cmd->timeout_per_command/HZ);
+	pkt->timeout = cpu_to_le16(cmd->request->timeout/HZ);
 
 	/* Set device target ID and LUN */
 	pkt->lun = SCSI_LUN_32(cmd);
@@ -3114,7 +3114,7 @@ qla1280_32bit_start_scsi(struct scsi_qla_host *ha, struct srb * sp)
 	memset(((char *)pkt + 8), 0, (REQUEST_ENTRY_SIZE - 8));
 
 	/* Set ISP command timeout. */
-	pkt->timeout = cpu_to_le16(cmd->timeout_per_command/HZ);
+	pkt->timeout = cpu_to_le16(cmd->request->timeout/HZ);
 
 	/* Set device target ID and LUN */
 	pkt->lun = SCSI_LUN_32(cmd);

drivers/scsi/qla4xxx/ql4_os.c

@@ -1542,7 +1542,7 @@ static int qla4xxx_eh_device_reset(struct scsi_cmnd *cmd)
 	DEBUG2(printk(KERN_INFO
 		      "scsi%ld: DEVICE_RESET cmd=%p jiffies = 0x%lx, to=%x,"
 		      "dpc_flags=%lx, status=%x allowed=%d\n", ha->host_no,
-		      cmd, jiffies, cmd->timeout_per_command / HZ,
+		      cmd, jiffies, cmd->request->timeout / HZ,
 		      ha->dpc_flags, cmd->result, cmd->allowed));
 
 	/* FIXME: wait for hba to go online */
@@ -1598,7 +1598,7 @@ static int qla4xxx_eh_target_reset(struct scsi_cmnd *cmd)
 	DEBUG2(printk(KERN_INFO
 		      "scsi%ld: TARGET_DEVICE_RESET cmd=%p jiffies = 0x%lx, "
 		      "to=%x,dpc_flags=%lx, status=%x allowed=%d\n",
-		      ha->host_no, cmd, jiffies, cmd->timeout_per_command / HZ,
+		      ha->host_no, cmd, jiffies, cmd->request->timeout / HZ,
 		      ha->dpc_flags, cmd->result, cmd->allowed));
 
 	stat = qla4xxx_reset_target(ha, ddb_entry);

drivers/scsi/scsi.c

@@ -291,7 +291,6 @@ struct scsi_cmnd *scsi_get_command(struct scsi_device *dev, gfp_t gfp_mask)
 		unsigned long flags;
 
 		cmd->device = dev;
-		init_timer(&cmd->eh_timeout);
 		INIT_LIST_HEAD(&cmd->list);
 		spin_lock_irqsave(&dev->list_lock, flags);
 		list_add_tail(&cmd->list, &dev->cmd_list);
@@ -652,14 +651,19 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 	unsigned long timeout;
 	int rtn = 0;
 
+	/*
+	 * We will use a queued command if possible, otherwise we will
+	 * emulate the queuing and calling of completion function ourselves.
+	 */
+	atomic_inc(&cmd->device->iorequest_cnt);
+
 	/* check if the device is still usable */
 	if (unlikely(cmd->device->sdev_state == SDEV_DEL)) {
 		/* in SDEV_DEL we error all commands. DID_NO_CONNECT
 		 * returns an immediate error upwards, and signals
 		 * that the device is no longer present */
 		cmd->result = DID_NO_CONNECT << 16;
-		atomic_inc(&cmd->device->iorequest_cnt);
-		__scsi_done(cmd);
+		scsi_done(cmd);
 		/* return 0 (because the command has been processed) */
 		goto out;
 	}
@@ -672,6 +676,7 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 		 * future requests should not occur until the device 
 		 * transitions out of the suspend state.
 		 */
+
 		scsi_queue_insert(cmd, SCSI_MLQUEUE_DEVICE_BUSY);
 
 		SCSI_LOG_MLQUEUE(3, printk("queuecommand : device blocked \n"));
@@ -714,20 +719,8 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 		host->resetting = 0;
 	}
 
-	/* 
-	 * AK: unlikely race here: for some reason the timer could
-	 * expire before the serial number is set up below.
-	 */
-	scsi_add_timer(cmd, cmd->timeout_per_command, scsi_times_out);
-
 	scsi_log_send(cmd);
 
-	/*
-	 * We will use a queued command if possible, otherwise we will
-	 * emulate the queuing and calling of completion function ourselves.
-	 */
-	atomic_inc(&cmd->device->iorequest_cnt);
-
 	/*
 	 * Before we queue this command, check if the command
 	 * length exceeds what the host adapter can handle.
@@ -744,6 +737,12 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 	}
 
 	spin_lock_irqsave(host->host_lock, flags);
+	/*
+	 * AK: unlikely race here: for some reason the timer could
+	 * expire before the serial number is set up below.
+	 *
+	 * TODO: kill serial or move to blk layer
+	 */
 	scsi_cmd_get_serial(host, cmd); 
 
 	if (unlikely(host->shost_state == SHOST_DEL)) {
@@ -754,12 +753,8 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 	}
 	spin_unlock_irqrestore(host->host_lock, flags);
 	if (rtn) {
-		if (scsi_delete_timer(cmd)) {
-			atomic_inc(&cmd->device->iodone_cnt);
-			scsi_queue_insert(cmd,
-					  (rtn == SCSI_MLQUEUE_DEVICE_BUSY) ?
-					  rtn : SCSI_MLQUEUE_HOST_BUSY);
-		}
+		scsi_queue_insert(cmd, (rtn == SCSI_MLQUEUE_DEVICE_BUSY) ?
+						rtn : SCSI_MLQUEUE_HOST_BUSY);
 		SCSI_LOG_MLQUEUE(3,
 		    printk("queuecommand : request rejected\n"));
 	}
@@ -769,24 +764,6 @@ int scsi_dispatch_cmd(struct scsi_cmnd *cmd)
 	return rtn;
 }
 
-/**
- * scsi_req_abort_cmd -- Request command recovery for the specified command
- * @cmd: pointer to the SCSI command of interest
- *
- * This function requests that SCSI Core start recovery for the
- * command by deleting the timer and adding the command to the eh
- * queue.  It can be called by either LLDDs or SCSI Core.  LLDDs who
- * implement their own error recovery MAY ignore the timeout event if
- * they generated scsi_req_abort_cmd.
- */
-void scsi_req_abort_cmd(struct scsi_cmnd *cmd)
-{
-	if (!scsi_delete_timer(cmd))
-		return;
-	scsi_times_out(cmd);
-}
-EXPORT_SYMBOL(scsi_req_abort_cmd);
-
 /**
  * scsi_done - Enqueue the finished SCSI command into the done queue.
  * @cmd: The SCSI Command for which a low-level device driver (LLDD) gives
@@ -802,42 +779,7 @@ EXPORT_SYMBOL(scsi_req_abort_cmd);
  */
 static void scsi_done(struct scsi_cmnd *cmd)
 {
-	/*
-	 * We don't have to worry about this one timing out anymore.
-	 * If we are unable to remove the timer, then the command
-	 * has already timed out.  In which case, we have no choice but to
-	 * let the timeout function run, as we have no idea where in fact
-	 * that function could really be.  It might be on another processor,
-	 * etc, etc.
-	 */
-	if (!scsi_delete_timer(cmd))
-		return;
-	__scsi_done(cmd);
-}
-
-/* Private entry to scsi_done() to complete a command when the timer
- * isn't running --- used by scsi_times_out */
-void __scsi_done(struct scsi_cmnd *cmd)
-{
-	struct request *rq = cmd->request;
-
-	/*
-	 * Set the serial numbers back to zero
-	 */
-	cmd->serial_number = 0;
-
-	atomic_inc(&cmd->device->iodone_cnt);
-	if (cmd->result)
-		atomic_inc(&cmd->device->ioerr_cnt);
-
-	BUG_ON(!rq);
-
-	/*
-	 * The uptodate/nbytes values don't matter, as we allow partial
-	 * completes and thus will check this in the softirq callback
-	 */
-	rq->completion_data = cmd;
-	blk_complete_request(rq);
+	blk_complete_request(cmd->request);
 }
 
 /* Move this to a header if it becomes more generally useful */

drivers/scsi/scsi_error.c

@@ -111,70 +111,9 @@ int scsi_eh_scmd_add(struct scsi_cmnd *scmd, int eh_flag)
 	return ret;
 }
 
-/**
- * scsi_add_timer - Start timeout timer for a single scsi command.
- * @scmd:	scsi command that is about to start running.
- * @timeout:	amount of time to allow this command to run.
- * @complete:	timeout function to call if timer isn't canceled.
- *
- * Notes:
- *    This should be turned into an inline function.  Each scsi command
- *    has its own timer, and as it is added to the queue, we set up the
- *    timer.  When the command completes, we cancel the timer.
- */
-void scsi_add_timer(struct scsi_cmnd *scmd, int timeout,
-		    void (*complete)(struct scsi_cmnd *))
-{
-
-	/*
-	 * If the clock was already running for this command, then
-	 * first delete the timer.  The timer handling code gets rather
-	 * confused if we don't do this.
-	 */
-	if (scmd->eh_timeout.function)
-		del_timer(&scmd->eh_timeout);
-
-	scmd->eh_timeout.data = (unsigned long)scmd;
-	scmd->eh_timeout.expires = jiffies + timeout;
-	scmd->eh_timeout.function = (void (*)(unsigned long)) complete;
-
-	SCSI_LOG_ERROR_RECOVERY(5, printk("%s: scmd: %p, time:"
-					  " %d, (%p)\n", __func__,
-					  scmd, timeout, complete));
-
-	add_timer(&scmd->eh_timeout);
-}
-
-/**
- * scsi_delete_timer - Delete/cancel timer for a given function.
- * @scmd:	Cmd that we are canceling timer for
- *
- * Notes:
- *     This should be turned into an inline function.
- *
- * Return value:
- *     1 if we were able to detach the timer.  0 if we blew it, and the
- *     timer function has already started to run.
- */
-int scsi_delete_timer(struct scsi_cmnd *scmd)
-{
-	int rtn;
-
-	rtn = del_timer(&scmd->eh_timeout);
-
-	SCSI_LOG_ERROR_RECOVERY(5, printk("%s: scmd: %p,"
-					 " rtn: %d\n", __func__,
-					 scmd, rtn));
-
-	scmd->eh_timeout.data = (unsigned long)NULL;
-	scmd->eh_timeout.function = NULL;
-
-	return rtn;
-}
-
 /**
  * scsi_times_out - Timeout function for normal scsi commands.
- * @scmd:	Cmd that is timing out.
+ * @req:	request that is timing out.
  *
  * Notes:
  *     We do not need to lock this.  There is the potential for a race
@@ -182,9 +121,11 @@ int scsi_delete_timer(struct scsi_cmnd *scmd)
  *     normal completion function determines that the timer has already
  *     fired, then it mustn't do anything.
  */
-void scsi_times_out(struct scsi_cmnd *scmd)
+enum blk_eh_timer_return scsi_times_out(struct request *req)
 {
-	enum scsi_eh_timer_return (* eh_timed_out)(struct scsi_cmnd *);
+	struct scsi_cmnd *scmd = req->special;
+	enum blk_eh_timer_return (*eh_timed_out)(struct scsi_cmnd *);
+	enum blk_eh_timer_return rtn = BLK_EH_NOT_HANDLED;
 
 	scsi_log_completion(scmd, TIMEOUT_ERROR);
 
@@ -196,22 +137,20 @@ void scsi_times_out(struct scsi_cmnd *scmd)
 		eh_timed_out = NULL;
 
 	if (eh_timed_out)
-		switch (eh_timed_out(scmd)) {
-		case EH_HANDLED:
-			__scsi_done(scmd);
-			return;
-		case EH_RESET_TIMER:
-			scsi_add_timer(scmd, scmd->timeout_per_command,
-				       scsi_times_out);
-			return;
-		case EH_NOT_HANDLED:
+		rtn = eh_timed_out(scmd);
+		switch (rtn) {
+		case BLK_EH_NOT_HANDLED:
 			break;
+		default:
+			return rtn;
 		}
 
 	if (unlikely(!scsi_eh_scmd_add(scmd, SCSI_EH_CANCEL_CMD))) {
 		scmd->result |= DID_TIME_OUT << 16;
-		__scsi_done(scmd);
+		return BLK_EH_HANDLED;
 	}
+
+	return BLK_EH_NOT_HANDLED;
 }
 
 /**
@@ -1793,7 +1732,6 @@ scsi_reset_provider(struct scsi_device *dev, int flag)
 
 	blk_rq_init(NULL, &req);
 	scmd->request = &req;
-	memset(&scmd->eh_timeout, 0, sizeof(scmd->eh_timeout));
 
 	scmd->cmnd = req.cmd;
 
@@ -1804,8 +1742,6 @@ scsi_reset_provider(struct scsi_device *dev, int flag)
 
 	scmd->sc_data_direction		= DMA_BIDIRECTIONAL;
 
-	init_timer(&scmd->eh_timeout);
-
 	spin_lock_irqsave(shost->host_lock, flags);
 	shost->tmf_in_progress = 1;
 	spin_unlock_irqrestore(shost->host_lock, flags);

drivers/scsi/scsi_lib.c

@@ -1181,7 +1181,6 @@ int scsi_setup_blk_pc_cmnd(struct scsi_device *sdev, struct request *req)
 	
 	cmd->transfersize = req->data_len;
 	cmd->allowed = req->retries;
-	cmd->timeout_per_command = req->timeout;
 	return BLKPREP_OK;
 }
 EXPORT_SYMBOL(scsi_setup_blk_pc_cmnd);
@@ -1416,17 +1415,26 @@ static void scsi_kill_request(struct request *req, struct request_queue *q)
 	spin_unlock(shost->host_lock);
 	spin_lock(sdev->request_queue->queue_lock);
 
-	__scsi_done(cmd);
+	blk_complete_request(req);
 }
 
 static void scsi_softirq_done(struct request *rq)
 {
-	struct scsi_cmnd *cmd = rq->completion_data;
-	unsigned long wait_for = (cmd->allowed + 1) * cmd->timeout_per_command;
+	struct scsi_cmnd *cmd = rq->special;
+	unsigned long wait_for = (cmd->allowed + 1) * rq->timeout;
 	int disposition;
 
 	INIT_LIST_HEAD(&cmd->eh_entry);
 
+	/*
+	 * Set the serial numbers back to zero
+	 */
+	cmd->serial_number = 0;
+
+	atomic_inc(&cmd->device->iodone_cnt);
+	if (cmd->result)
+		atomic_inc(&cmd->device->ioerr_cnt);
+
 	disposition = scsi_decide_disposition(cmd);
 	if (disposition != SUCCESS &&
 	    time_before(cmd->jiffies_at_alloc + wait_for, jiffies)) {
@@ -1675,6 +1683,7 @@ struct request_queue *scsi_alloc_queue(struct scsi_device *sdev)
 
 	blk_queue_prep_rq(q, scsi_prep_fn);
 	blk_queue_softirq_done(q, scsi_softirq_done);
+	blk_queue_rq_timed_out(q, scsi_times_out);
 	return q;
 }
 

drivers/scsi/scsi_priv.h

@@ -4,6 +4,7 @@
 #include <linux/device.h>
 
 struct request_queue;
+struct request;
 struct scsi_cmnd;
 struct scsi_device;
 struct scsi_host_template;
@@ -27,7 +28,6 @@ extern void scsi_exit_hosts(void);
 extern int scsi_dispatch_cmd(struct scsi_cmnd *cmd);
 extern int scsi_setup_command_freelist(struct Scsi_Host *shost);
 extern void scsi_destroy_command_freelist(struct Scsi_Host *shost);
-extern void __scsi_done(struct scsi_cmnd *cmd);
 #ifdef CONFIG_SCSI_LOGGING
 void scsi_log_send(struct scsi_cmnd *cmd);
 void scsi_log_completion(struct scsi_cmnd *cmd, int disposition);
@@ -49,10 +49,7 @@ extern int __init scsi_init_devinfo(void);
 extern void scsi_exit_devinfo(void);
 
 /* scsi_error.c */
-extern void scsi_add_timer(struct scsi_cmnd *, int,
-		void (*)(struct scsi_cmnd *));
-extern int scsi_delete_timer(struct scsi_cmnd *);
-extern void scsi_times_out(struct scsi_cmnd *cmd);
+extern enum blk_eh_timer_return scsi_times_out(struct request *req);
 extern int scsi_error_handler(void *host);
 extern int scsi_decide_disposition(struct scsi_cmnd *cmd);
 extern void scsi_eh_wakeup(struct Scsi_Host *shost);

drivers/scsi/scsi_sysfs.c

@@ -560,12 +560,15 @@ sdev_rd_attr (vendor, "%.8s\n");
 sdev_rd_attr (model, "%.16s\n");
 sdev_rd_attr (rev, "%.4s\n");
 
+/*
+ * TODO: can we make these symlinks to the block layer ones?
+ */
 static ssize_t
 sdev_show_timeout (struct device *dev, struct device_attribute *attr, char *buf)
 {
 	struct scsi_device *sdev;
 	sdev = to_scsi_device(dev);
-	return snprintf (buf, 20, "%d\n", sdev->timeout / HZ);
+	return snprintf(buf, 20, "%d\n", sdev->request_queue->rq_timeout / HZ);
 }
 
 static ssize_t
@@ -576,7 +579,7 @@ sdev_store_timeout (struct device *dev, struct device_attribute *attr,
 	int timeout;
 	sdev = to_scsi_device(dev);
 	sscanf (buf, "%d\n", &timeout);
-	sdev->timeout = timeout * HZ;
+	blk_queue_rq_timeout(sdev->request_queue, timeout * HZ);
 	return count;
 }
 static DEVICE_ATTR(timeout, S_IRUGO | S_IWUSR, sdev_show_timeout, sdev_store_timeout);

drivers/scsi/scsi_transport_fc.c

@@ -1950,15 +1950,15 @@ static int fc_vport_match(struct attribute_container *cont,
  * Notes:
  *	This routine assumes no locks are held on entry.
  */
-static enum scsi_eh_timer_return
+static enum blk_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 BLK_EH_RESET_TIMER;
 
-	return EH_NOT_HANDLED;
+	return BLK_EH_NOT_HANDLED;
 }
 
 /*

drivers/scsi/sd.c

@@ -383,7 +383,6 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq)
 	sector_t block = rq->sector;
 	sector_t threshold;
 	unsigned int this_count = rq->nr_sectors;
-	unsigned int timeout = sdp->timeout;
 	int ret;
 
 	if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
@@ -584,7 +583,6 @@ static int sd_prep_fn(struct request_queue *q, struct request *rq)
 	SCpnt->transfersize = sdp->sector_size;
 	SCpnt->underflow = this_count << 9;
 	SCpnt->allowed = SD_MAX_RETRIES;
-	SCpnt->timeout_per_command = timeout;
 
 	/*
 	 * This indicates that the command is ready from our end to be
@@ -1878,11 +1876,12 @@ static int sd_probe(struct device *dev)
 	sdkp->openers = 0;
 	sdkp->previous_state = 1;
 
-	if (!sdp->timeout) {
+	if (!sdp->request_queue->rq_timeout) {
 		if (sdp->type != TYPE_MOD)
-			sdp->timeout = SD_TIMEOUT;
+			blk_queue_rq_timeout(sdp->request_queue, SD_TIMEOUT);
 		else
-			sdp->timeout = SD_MOD_TIMEOUT;
+			blk_queue_rq_timeout(sdp->request_queue,
+					     SD_MOD_TIMEOUT);
 	}
 
 	device_initialize(&sdkp->dev);

drivers/scsi/sr.c

@@ -331,7 +331,7 @@ static int sr_done(struct scsi_cmnd *SCpnt)
 
 static int sr_prep_fn(struct request_queue *q, struct request *rq)
 {
-	int block=0, this_count, s_size, timeout = SR_TIMEOUT;
+	int block = 0, this_count, s_size;
 	struct scsi_cd *cd;
 	struct scsi_cmnd *SCpnt;
 	struct scsi_device *sdp = q->queuedata;
@@ -461,7 +461,6 @@ static int sr_prep_fn(struct request_queue *q, struct request *rq)
 	SCpnt->transfersize = cd->device->sector_size;
 	SCpnt->underflow = this_count << 9;
 	SCpnt->allowed = MAX_RETRIES;
-	SCpnt->timeout_per_command = timeout;
 
 	/*
 	 * This indicates that the command is ready from our end to be
@@ -620,6 +619,8 @@ static int sr_probe(struct device *dev)
 	disk->fops = &sr_bdops;
 	disk->flags = GENHD_FL_CD;
 
+	blk_queue_rq_timeout(sdev->request_queue, SR_TIMEOUT);
+
 	cd->device = sdev;
 	cd->disk = disk;
 	cd->driver = &sr_template;

drivers/scsi/sym53c8xx_2/sym_glue.c

@@ -519,8 +519,8 @@ static int sym53c8xx_queue_command(struct scsi_cmnd *cmd,
 	 *  Shorten our settle_time if needed for 
 	 *  this command not to time out.
 	 */
-	if (np->s.settle_time_valid && cmd->timeout_per_command) {
-		unsigned long tlimit = jiffies + cmd->timeout_per_command;
+	if (np->s.settle_time_valid && cmd->request->timeout) {
+		unsigned long tlimit = jiffies + cmd->request->timeout;
 		tlimit -= SYM_CONF_TIMER_INTERVAL*2;
 		if (time_after(np->s.settle_time, tlimit)) {
 			np->s.settle_time = tlimit;

include/linux/blkdev.h

@@ -147,6 +147,7 @@ struct request {
 
 	unsigned int cmd_flags;
 	enum rq_cmd_type_bits cmd_type;
+	unsigned long atomic_flags;
 
 	/* Maintain bio traversal state for part by part I/O submission.
 	 * hard_* are block layer internals, no driver should touch them!
@@ -214,6 +215,8 @@ struct request {
 	void *data;
 	void *sense;
 
+	unsigned long deadline;
+	struct list_head timeout_list;
 	unsigned int timeout;
 	int retries;
 
@@ -266,6 +269,14 @@ typedef void (prepare_flush_fn) (struct request_queue *, struct request *);
 typedef void (softirq_done_fn)(struct request *);
 typedef int (dma_drain_needed_fn)(struct request *);
 
+enum blk_eh_timer_return {
+	BLK_EH_NOT_HANDLED,
+	BLK_EH_HANDLED,
+	BLK_EH_RESET_TIMER,
+};
+
+typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
+
 enum blk_queue_state {
 	Queue_down,
 	Queue_up,
@@ -311,6 +322,7 @@ struct request_queue
 	merge_bvec_fn		*merge_bvec_fn;
 	prepare_flush_fn	*prepare_flush_fn;
 	softirq_done_fn		*softirq_done_fn;
+	rq_timed_out_fn		*rq_timed_out_fn;
 	dma_drain_needed_fn	*dma_drain_needed;
 
 	/*
@@ -386,6 +398,10 @@ struct request_queue
 	unsigned int		nr_sorted;
 	unsigned int		in_flight;
 
+	unsigned int		rq_timeout;
+	struct timer_list	timeout;
+	struct list_head	timeout_list;
+
 	/*
 	 * sg stuff
 	 */
@@ -770,6 +786,8 @@ extern int blk_end_request_callback(struct request *rq, int error,
 				unsigned int nr_bytes,
 				int (drv_callback)(struct request *));
 extern void blk_complete_request(struct request *);
+extern void __blk_complete_request(struct request *);
+extern void blk_abort_request(struct request *);
 
 /*
  * blk_end_request() takes bytes instead of sectors as a complete size.
@@ -811,6 +829,8 @@ extern void blk_queue_dma_alignment(struct request_queue *, int);
 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
 extern void blk_queue_set_discard(struct request_queue *, prepare_discard_fn *);
+extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
+extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
 extern int blk_queue_ordered(struct request_queue *, unsigned, prepare_flush_fn *);
 extern int blk_do_ordered(struct request_queue *, struct request **);

include/scsi/scsi_cmnd.h

@@ -75,7 +75,6 @@ struct scsi_cmnd {
 
 	int retries;
 	int allowed;
-	int timeout_per_command;
 
 	unsigned char prot_op;
 	unsigned char prot_type;
@@ -86,7 +85,6 @@ struct scsi_cmnd {
 	/* These elements define the operation we are about to perform */
 	unsigned char *cmnd;
 
-	struct timer_list eh_timeout;	/* Used to time out the command. */
 
 	/* These elements define the operation we ultimately want to perform */
 	struct scsi_data_buffer sdb;
@@ -139,7 +137,6 @@ extern void scsi_put_command(struct scsi_cmnd *);
 extern void __scsi_put_command(struct Scsi_Host *, struct scsi_cmnd *,
 			       struct device *);
 extern void scsi_finish_command(struct scsi_cmnd *cmd);
-extern void scsi_req_abort_cmd(struct scsi_cmnd *cmd);
 
 extern void *scsi_kmap_atomic_sg(struct scatterlist *sg, int sg_count,
 				 size_t *offset, size_t *len);

include/scsi/scsi_host.h

@@ -43,13 +43,6 @@ struct blk_queue_tags;
 #define DISABLE_CLUSTERING 0
 #define ENABLE_CLUSTERING 1
 
-enum scsi_eh_timer_return {
-	EH_NOT_HANDLED,
-	EH_HANDLED,
-	EH_RESET_TIMER,
-};
-
-
 struct scsi_host_template {
 	struct module *module;
 	const char *name;
@@ -347,7 +340,7 @@ struct scsi_host_template {
 	 *
 	 * Status: OPTIONAL
 	 */
-	enum scsi_eh_timer_return (* eh_timed_out)(struct scsi_cmnd *);
+	enum blk_eh_timer_return (*eh_timed_out)(struct scsi_cmnd *);
 
 	/*
 	 * Name of proc directory

include/scsi/scsi_transport.h

@@ -21,6 +21,7 @@
 #define SCSI_TRANSPORT_H
 
 #include <linux/transport_class.h>
+#include <linux/blkdev.h>
 #include <scsi/scsi_host.h>
 #include <scsi/scsi_device.h>
 
@@ -64,7 +65,7 @@ struct scsi_transport_template {
 	 *			begin counting again
 	 * EH_NOT_HANDLED	Begin normal error recovery
 	 */
-	enum scsi_eh_timer_return (* eh_timed_out)(struct scsi_cmnd *);
+	enum blk_eh_timer_return (*eh_timed_out)(struct scsi_cmnd *);
 
 	/*
 	 * Used as callback for the completion of i_t_nexus request