[PATCH 0/6] ide: final bits of struct hwgroup_s -> struct ide_host change

[PATCH 0/6] ide: final bits of struct hwgroup_s -> struct ide_host change

[Bartlomiej Zolnierkiewicz]

In the process we fix years old bug, improve the locking for serialized ports
(making it simpler to grasp and more efficient at the same time) and simplify
the core IDE code significantly...

diffstat:
 drivers/ide/alim15x3.c     |    2 
 drivers/ide/au1xxx-ide.c   |    6 
 drivers/ide/icside.c       |    2 
 drivers/ide/ide-atapi.c    |    8 -
 drivers/ide/ide-cd.c       |   16 +-
 drivers/ide/ide-dma-sff.c  |    4 
 drivers/ide/ide-dma.c      |    2 
 drivers/ide/ide-floppy.c   |    2 
 drivers/ide/ide-io.c       |  297 +++++++++++++++++++++------------------------
 drivers/ide/ide-iops.c     |   67 ++++------
 drivers/ide/ide-lib.c      |    9 -
 drivers/ide/ide-park.c     |   16 +-
 drivers/ide/ide-pm.c       |    2 
 drivers/ide/ide-probe.c    |  198 ++++--------------------------
 drivers/ide/ide-tape.c     |    9 -
 drivers/ide/ide-taskfile.c |    4 
 drivers/ide/ide.c          |   24 ---
 drivers/ide/pdc202xx_old.c |    2 
 drivers/ide/pmac.c         |    2 
 drivers/ide/scc_pata.c     |    4 
 drivers/ide/sgiioc4.c      |    2 
 drivers/ide/tc86c001.c     |   10 -
 drivers/ide/trm290.c       |    2 
 drivers/ide/tx4939ide.c    |    2 
 drivers/ide/umc8672.c      |   13 -
 include/linux/ide.h        |  107 +++++-----------
 26 files changed, 299 insertions(+), 513 deletions(-)

[PATCH 1/6] ide: fix setting nIEN on idle devices

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: fix setting nIEN on idle devices

Fix do_ide_request() to operate on previous device / port instead of
the current one.  The original code was wrong since at least Feb 2002
(2.4.0 timeframe).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
all patches are for pata-2.6 tree

 drivers/ide/ide-io.c |   10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -724,7 +724,9 @@ void do_ide_request(struct request_queue
 	spin_lock_irq(&hwgroup->lock);
 
 	if (!ide_lock_hwgroup(hwgroup)) {
+		ide_hwif_t *prev_port;
 repeat:
+		prev_port = hwgroup->hwif;
 		hwgroup->rq = NULL;
 
 		if (drive->dev_flags & IDE_DFLAG_SLEEPING) {
@@ -734,13 +736,13 @@ repeat:
 			}
 		}
 
-		if (hwif != hwgroup->hwif) {
+		if (hwif != prev_port) {
 			/*
-			 * set nIEN for previous hwif, drives in the
+			 * set nIEN for previous port, drives in the
 			 * quirk_list may not like intr setups/cleanups
 			 */
-			if (drive->quirk_list == 0)
-				hwif->tp_ops->set_irq(hwif, 0);
+			if (hwgroup->drive->quirk_list == 0)
+				prev_port->tp_ops->set_irq(prev_port, 0);
 		}
 		hwgroup->hwif = hwif;
 		hwgroup->drive = drive;

[PATCH 2/6] ide: add ->cur_port to struct ide_host and use it for serialized hosts

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: add ->cur_port to struct ide_host and use it for serialized hosts

* Pass 'ide_hwif_t *' instead of 'ide_hwgroup_t *' to unexpected_intr().

* Cache pointer to the port currently being serviced in ->cur_port
  and use it instead of hwif->hwgroup on serialized hosts.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
 drivers/ide/ide-io.c |   25 +++++++++++++++----------
 include/linux/ide.h  |    1 +
 2 files changed, 16 insertions(+), 10 deletions(-)

Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -726,7 +726,7 @@ void do_ide_request(struct request_queue
 	if (!ide_lock_hwgroup(hwgroup)) {
 		ide_hwif_t *prev_port;
 repeat:
-		prev_port = hwgroup->hwif;
+		prev_port = hwif->host->cur_port;
 		hwgroup->rq = NULL;
 
 		if (drive->dev_flags & IDE_DFLAG_SLEEPING) {
@@ -736,15 +736,17 @@ repeat:
 			}
 		}
 
-		if (hwif != prev_port) {
+		if ((hwif->host->host_flags & IDE_HFLAG_SERIALIZE) &&
+		    hwif != prev_port) {
 			/*
 			 * set nIEN for previous port, drives in the
 			 * quirk_list may not like intr setups/cleanups
 			 */
-			if (hwgroup->drive->quirk_list == 0)
+			if (prev_port && hwgroup->drive->quirk_list == 0)
 				prev_port->tp_ops->set_irq(prev_port, 0);
+
+			hwif->host->cur_port = hwif;
 		}
-		hwgroup->hwif = hwif;
 		hwgroup->drive = drive;
 		drive->dev_flags &= ~(IDE_DFLAG_SLEEPING | IDE_DFLAG_PARKED);
 
@@ -976,7 +978,7 @@ void ide_timer_expiry (unsigned long dat
 /**
  *	unexpected_intr		-	handle an unexpected IDE interrupt
  *	@irq: interrupt line
- *	@hwgroup: hwgroup being processed
+ *	@hwif: port being processed
  *
  *	There's nothing really useful we can do with an unexpected interrupt,
  *	other than reading the status register (to clear it), and logging it.
@@ -1005,11 +1007,11 @@ void ide_timer_expiry (unsigned long dat
  *	is doing the current command, but we don't know which hwif burped
  *	mysteriously.
  */
- 
-static void unexpected_intr (int irq, ide_hwgroup_t *hwgroup)
+
+static void unexpected_intr(int irq, ide_hwif_t *hwif)
 {
+	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	u8 stat;
-	ide_hwif_t *hwif = hwgroup->hwif;
 
 	/*
 	 * handle the unexpected interrupt
@@ -1044,7 +1046,7 @@ static void unexpected_intr (int irq, id
  *	not need to override it. If you do be aware it is subtle in
  *	places
  *
- *	hwgroup->hwif is the interface in the group currently performing
+ *	hwif is the interface in the group currently performing
  *	a command. hwgroup->drive is the drive and hwgroup->handler is
  *	the IRQ handler to call. As we issue a command the handlers
  *	step through multiple states, reassigning the handler to the
@@ -1070,6 +1072,9 @@ irqreturn_t ide_intr (int irq, void *dev
 	irqreturn_t irq_ret = IRQ_NONE;
 	int plug_device = 0;
 
+	if (hwif->host->host_flags & IDE_HFLAG_SERIALIZE)
+		hwif = hwif->host->cur_port;
+
 	spin_lock_irqsave(&hwgroup->lock, flags);
 
 	if (!ide_ack_intr(hwif))
@@ -1099,7 +1104,7 @@ irqreturn_t ide_intr (int irq, void *dev
 			 * Probably not a shared PCI interrupt,
 			 * so we can safely try to do something about it:
 			 */
-			unexpected_intr(irq, hwgroup);
+			unexpected_intr(irq, hwif);
 #ifdef CONFIG_BLK_DEV_IDEPCI
 		} else {
 			/*
Index: b/include/linux/ide.h
===================================================================
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -852,6 +852,7 @@ struct ide_host {
 	unsigned int	(*init_chipset)(struct pci_dev *);
 	unsigned long	host_flags;
 	void		*host_priv;
+	ide_hwif_t	*cur_port;	/* for hosts requiring serialization */
 };
 
 /*

[PATCH 3/6] ide: use per-port IRQ handlers

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: use per-port IRQ handlers

Use hwif instead of hwgroup as {request,free}_irq()'s cookie,
teach ide_intr() to return early for non-active serialized ports,
modify unexpected_intr() accordingly and then use per-port IRQ
handlers instead of per-hwgroup ones.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
 drivers/ide/ide-io.c    |   56 ++++++++++++++++++------------------------------
 drivers/ide/ide-probe.c |   21 ++++++------------
 drivers/ide/ide.c       |   17 --------------
 include/linux/ide.h     |    4 +--
 4 files changed, 33 insertions(+), 65 deletions(-)

Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -723,7 +723,7 @@ void do_ide_request(struct request_queue
 	spin_unlock_irq(q->queue_lock);
 	spin_lock_irq(&hwgroup->lock);
 
-	if (!ide_lock_hwgroup(hwgroup)) {
+	if (!ide_lock_hwgroup(hwgroup, hwif)) {
 		ide_hwif_t *prev_port;
 repeat:
 		prev_port = hwif->host->cur_port;
@@ -1002,44 +1002,30 @@ void ide_timer_expiry (unsigned long dat
  *	before completing the issuance of any new drive command, so we will not
  *	be accidentally invoked as a result of any valid command completion
  *	interrupt.
- *
- *	Note that we must walk the entire hwgroup here. We know which hwif
- *	is doing the current command, but we don't know which hwif burped
- *	mysteriously.
  */
 
 static void unexpected_intr(int irq, ide_hwif_t *hwif)
 {
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
-	u8 stat;
+	u8 stat = hwif->tp_ops->read_status(hwif);
 
-	/*
-	 * handle the unexpected interrupt
-	 */
-	do {
-		if (hwif->irq == irq) {
-			stat = hwif->tp_ops->read_status(hwif);
-
-			if (!OK_STAT(stat, ATA_DRDY, BAD_STAT)) {
-				/* Try to not flood the console with msgs */
-				static unsigned long last_msgtime, count;
-				++count;
-				if (time_after(jiffies, last_msgtime + HZ)) {
-					last_msgtime = jiffies;
-					printk(KERN_ERR "%s%s: unexpected interrupt, "
-						"status=0x%02x, count=%ld\n",
-						hwif->name,
-						(hwif->next==hwgroup->hwif) ? "" : "(?)", stat, count);
-				}
-			}
+	if (!OK_STAT(stat, ATA_DRDY, BAD_STAT)) {
+		/* Try to not flood the console with msgs */
+		static unsigned long last_msgtime, count;
+		++count;
+
+		if (time_after(jiffies, last_msgtime + HZ)) {
+			last_msgtime = jiffies;
+			printk(KERN_ERR "%s: unexpected interrupt, "
+				"status=0x%02x, count=%ld\n",
+				hwif->name, stat, count);
 		}
-	} while ((hwif = hwif->next) != hwgroup->hwif);
+	}
 }
 
 /**
  *	ide_intr	-	default IDE interrupt handler
  *	@irq: interrupt number
- *	@dev_id: hwif group
+ *	@dev_id: hwif
  *	@regs: unused weirdness from the kernel irq layer
  *
  *	This is the default IRQ handler for the IDE layer. You should
@@ -1063,17 +1049,19 @@ static void unexpected_intr(int irq, ide
  
 irqreturn_t ide_intr (int irq, void *dev_id)
 {
-	unsigned long flags;
-	ide_hwgroup_t *hwgroup = (ide_hwgroup_t *)dev_id;
-	ide_hwif_t *hwif = hwgroup->hwif;
+	ide_hwif_t *hwif = (ide_hwif_t *)dev_id;
+	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	ide_drive_t *uninitialized_var(drive);
 	ide_handler_t *handler;
+	unsigned long flags;
 	ide_startstop_t startstop;
 	irqreturn_t irq_ret = IRQ_NONE;
 	int plug_device = 0;
 
-	if (hwif->host->host_flags & IDE_HFLAG_SERIALIZE)
-		hwif = hwif->host->cur_port;
+	if (hwif->host->host_flags & IDE_HFLAG_SERIALIZE) {
+		if (hwif != hwif->host->cur_port)
+			goto out_early;
+	}
 
 	spin_lock_irqsave(&hwgroup->lock, flags);
 
@@ -1172,7 +1160,7 @@ out_handled:
 	irq_ret = IRQ_HANDLED;
 out:
 	spin_unlock_irqrestore(&hwgroup->lock, flags);
-
+out_early:
 	if (plug_device)
 		ide_plug_device(drive);
 
Index: b/drivers/ide/ide-probe.c
===================================================================
--- a/drivers/ide/ide-probe.c
+++ b/drivers/ide/ide-probe.c
@@ -1022,6 +1022,7 @@ static int init_irq (ide_hwif_t *hwif)
 	unsigned int index;
 	ide_hwgroup_t *hwgroup;
 	ide_hwif_t *match = NULL;
+	int sa = 0;
 
 	mutex_lock(&ide_cfg_mtx);
 	hwif->hwgroup = NULL;
@@ -1076,24 +1077,18 @@ static int init_irq (ide_hwif_t *hwif)
 
 	ide_ports[hwif->index] = hwif;
 
-	/*
-	 * Allocate the irq, if not already obtained for another hwif
-	 */
-	if (!match || match->irq != hwif->irq) {
-		int sa = 0;
 #if defined(__mc68000__)
-		sa = IRQF_SHARED;
+	sa = IRQF_SHARED;
 #endif /* __mc68000__ */
 
-		if (hwif->chipset == ide_pci)
-			sa = IRQF_SHARED;
+	if (hwif->chipset == ide_pci)
+		sa = IRQF_SHARED;
 
-		if (io_ports->ctl_addr)
-			hwif->tp_ops->set_irq(hwif, 1);
+	if (io_ports->ctl_addr)
+		hwif->tp_ops->set_irq(hwif, 1);
 
-		if (request_irq(hwif->irq,&ide_intr,sa,hwif->name,hwgroup))
-	       		goto out_unlink;
-	}
+	if (request_irq(hwif->irq, &ide_intr, sa, hwif->name, hwif))
+		goto out_unlink;
 
 	if (!hwif->rqsize) {
 		if ((hwif->host_flags & IDE_HFLAG_NO_LBA48) ||
Index: b/drivers/ide/ide.c
===================================================================
--- a/drivers/ide/ide.c
+++ b/drivers/ide/ide.c
@@ -175,10 +175,6 @@ EXPORT_SYMBOL_GPL(ide_port_unregister_de
 
 void ide_unregister(ide_hwif_t *hwif)
 {
-	ide_hwif_t *g;
-	ide_hwgroup_t *hwgroup;
-	int irq_count = 0;
-
 	BUG_ON(in_interrupt());
 	BUG_ON(irqs_disabled());
 
@@ -191,18 +187,7 @@ void ide_unregister(ide_hwif_t *hwif)
 
 	ide_proc_unregister_port(hwif);
 
-	hwgroup = hwif->hwgroup;
-	/*
-	 * free the irq if we were the only hwif using it
-	 */
-	g = hwgroup->hwif;
-	do {
-		if (g->irq == hwif->irq)
-			++irq_count;
-		g = g->next;
-	} while (g != hwgroup->hwif);
-	if (irq_count == 1)
-		free_irq(hwif->irq, hwgroup);
+	free_irq(hwif->irq, hwif);
 
 	ide_remove_port_from_hwgroup(hwif);
 
Index: b/include/linux/ide.h
===================================================================
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -1274,14 +1274,14 @@ extern void ide_stall_queue(ide_drive_t 
 extern void ide_timer_expiry(unsigned long);
 extern irqreturn_t ide_intr(int irq, void *dev_id);
 
-static inline int ide_lock_hwgroup(ide_hwgroup_t *hwgroup)
+static inline int ide_lock_hwgroup(ide_hwgroup_t *hwgroup, ide_hwif_t *hwif)
 {
 	if (hwgroup->busy)
 		return 1;
 
 	hwgroup->busy = 1;
 	/* for atari only */
-	ide_get_lock(ide_intr, hwgroup);
+	ide_get_lock(ide_intr, hwif);
 
 	return 0;
 }

[PATCH 4/6] ide: remove hwgroup->hwif and {drive,hwif}->next

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: remove hwgroup->hwif and {drive,hwif}->next

* Add 'int port_count' field to ide_hwgroup_t to keep the track
  of the number of ports in the hwgroup.  Then update init_irq()
  and ide_remove_port_from_hwgroup() to use it.

* Remove no longer needed hwgroup->hwif, {drive,hwif}->next,
  ide_add_drive_to_hwgroup() and ide_remove_drive_from_hwgroup()
  (hwgroup->drive now only denotes the currently active device
   in the hwgroup).

* Update locking documentation in <linux/ide.h>.

While at it:

* Rename ->drive field in ide_hwgroup_t to ->cur_dev.

* Use __func__ in ide_timer_expiry().

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
 drivers/ide/ide-io.c    |   12 +++---
 drivers/ide/ide-probe.c |   90 +++---------------------------------------------
 include/linux/ide.h     |   13 +-----
 3 files changed, 15 insertions(+), 100 deletions(-)

Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -742,12 +742,12 @@ repeat:
 			 * set nIEN for previous port, drives in the
 			 * quirk_list may not like intr setups/cleanups
 			 */
-			if (prev_port && hwgroup->drive->quirk_list == 0)
+			if (prev_port && hwgroup->cur_dev->quirk_list == 0)
 				prev_port->tp_ops->set_irq(prev_port, 0);
 
 			hwif->host->cur_port = hwif;
 		}
-		hwgroup->drive = drive;
+		hwgroup->cur_dev = drive;
 		drive->dev_flags &= ~(IDE_DFLAG_SLEEPING | IDE_DFLAG_PARKED);
 
 		spin_unlock_irq(&hwgroup->lock);
@@ -913,9 +913,9 @@ void ide_timer_expiry (unsigned long dat
 		 * Either way, we don't really want to complain about anything.
 		 */
 	} else {
-		drive = hwgroup->drive;
+		drive = hwgroup->cur_dev;
 		if (!drive) {
-			printk(KERN_ERR "ide_timer_expiry: hwgroup->drive was NULL\n");
+			printk(KERN_ERR "%s: ->cur_dev was NULL\n", __func__);
 			hwgroup->handler = NULL;
 		} else {
 			ide_hwif_t *hwif;
@@ -1033,7 +1033,7 @@ static void unexpected_intr(int irq, ide
  *	places
  *
  *	hwif is the interface in the group currently performing
- *	a command. hwgroup->drive is the drive and hwgroup->handler is
+ *	a command. hwgroup->cur_dev is the drive and hwgroup->handler is
  *	the IRQ handler to call. As we issue a command the handlers
  *	step through multiple states, reassigning the handler to the
  *	next step in the process. Unlike a smart SCSI controller IDE
@@ -1105,7 +1105,7 @@ irqreturn_t ide_intr (int irq, void *dev
 		goto out;
 	}
 
-	drive = hwgroup->drive;
+	drive = hwgroup->cur_dev;
 	if (!drive) {
 		/*
 		 * This should NEVER happen, and there isn't much
Index: b/drivers/ide/ide-probe.c
===================================================================
--- a/drivers/ide/ide-probe.c
+++ b/drivers/ide/ide-probe.c
@@ -918,27 +918,9 @@ static int ide_init_queue(ide_drive_t *d
 	return 0;
 }
 
-static void ide_add_drive_to_hwgroup(ide_drive_t *drive)
-{
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
-
-	spin_lock_irq(&hwgroup->lock);
-	if (!hwgroup->drive) {
-		/* first drive for hwgroup. */
-		drive->next = drive;
-		hwgroup->drive = drive;
-		hwgroup->hwif = HWIF(hwgroup->drive);
-	} else {
-		drive->next = hwgroup->drive->next;
-		hwgroup->drive->next = drive;
-	}
-	spin_unlock_irq(&hwgroup->lock);
-}
-
 /*
  * For any present drive:
  * - allocate the block device queue
- * - link drive into the hwgroup
  */
 static int ide_port_setup_devices(ide_hwif_t *hwif)
 {
@@ -961,8 +943,6 @@ static int ide_port_setup_devices(ide_hw
 		}
 
 		j++;
-
-		ide_add_drive_to_hwgroup(drive);
 	}
 	mutex_unlock(&ide_cfg_mtx);
 
@@ -978,33 +958,9 @@ void ide_remove_port_from_hwgroup(ide_hw
 	ide_ports[hwif->index] = NULL;
 
 	spin_lock_irq(&hwgroup->lock);
-	/*
-	 * Remove us from the hwgroup, and free
-	 * the hwgroup if we were the only member
-	 */
-	if (hwif->next == hwif) {
-		BUG_ON(hwgroup->hwif != hwif);
+	/* Free the hwgroup if we were the only member. */
+	if (--hwgroup->port_count == 0)
 		kfree(hwgroup);
-	} else {
-		/* There is another interface in hwgroup.
-		 * Unlink us, and set hwgroup->drive and ->hwif to
-		 * something sane.
-		 */
-		ide_hwif_t *g = hwgroup->hwif;
-
-		while (g->next != hwif)
-			g = g->next;
-		g->next = hwif->next;
-		if (hwgroup->hwif == hwif) {
-			/* Chose a random hwif for hwgroup->hwif.
-			 * It's guaranteed that there are no drives
-			 * left in the hwgroup.
-			 */
-			BUG_ON(hwgroup->drive != NULL);
-			hwgroup->hwif = g;
-		}
-		BUG_ON(hwgroup->hwif == hwif);
-	}
 	spin_unlock_irq(&hwgroup->lock);
 }
 
@@ -1044,20 +1000,9 @@ static int init_irq (ide_hwif_t *hwif)
 	if (match) {
 		hwgroup = match->hwgroup;
 		hwif->hwgroup = hwgroup;
-		/*
-		 * Link us into the hwgroup.
-		 * This must be done early, do ensure that unexpected_intr
-		 * can find the hwif and prevent irq storms.
-		 * No drives are attached to the new hwif, choose_drive
-		 * can't do anything stupid (yet).
-		 * Add ourself as the 2nd entry to the hwgroup->hwif
-		 * linked list, the first entry is the hwif that owns
-		 * hwgroup->handler - do not change that.
-		 */
+
 		spin_lock_irq(&hwgroup->lock);
-		hwif->next = hwgroup->hwif->next;
-		hwgroup->hwif->next = hwif;
-		BUG_ON(hwif->next == hwif);
+		hwgroup->port_count++;
 		spin_unlock_irq(&hwgroup->lock);
 	} else {
 		hwgroup = kmalloc_node(sizeof(*hwgroup), GFP_KERNEL|__GFP_ZERO,
@@ -1068,7 +1013,8 @@ static int init_irq (ide_hwif_t *hwif)
 		spin_lock_init(&hwgroup->lock);
 
 		hwif->hwgroup = hwgroup;
-		hwgroup->hwif = hwif->next = hwif;
+
+		hwgroup->port_count = 1;
 
 		init_timer(&hwgroup->timer);
 		hwgroup->timer.function = &ide_timer_expiry;
@@ -1191,29 +1137,6 @@ void ide_init_disk(struct gendisk *disk,
 
 EXPORT_SYMBOL_GPL(ide_init_disk);
 
-static void ide_remove_drive_from_hwgroup(ide_drive_t *drive)
-{
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
-
-	if (drive == drive->next) {
-		/* special case: last drive from hwgroup. */
-		BUG_ON(hwgroup->drive != drive);
-		hwgroup->drive = NULL;
-	} else {
-		ide_drive_t *walk;
-
-		walk = hwgroup->drive;
-		while (walk->next != drive)
-			walk = walk->next;
-		walk->next = drive->next;
-		if (hwgroup->drive == drive) {
-			hwgroup->drive = drive->next;
-			hwgroup->hwif = hwgroup->drive->hwif;
-		}
-	}
-	BUG_ON(hwgroup->drive == drive);
-}
-
 static void drive_release_dev (struct device *dev)
 {
 	ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
@@ -1222,7 +1145,6 @@ static void drive_release_dev (struct de
 	ide_proc_unregister_device(drive);
 
 	spin_lock_irq(&hwgroup->lock);
-	ide_remove_drive_from_hwgroup(drive);
 	kfree(drive->id);
 	drive->id = NULL;
 	drive->dev_flags &= ~IDE_DFLAG_PRESENT;
Index: b/include/linux/ide.h
===================================================================
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -588,7 +588,6 @@ struct ide_drive_s {
 	struct request_queue	*queue;	/* request queue */
 
 	struct request		*rq;	/* current request */
-	struct ide_drive_s 	*next;	/* circular list of hwgroup drives */
 	void		*driver_data;	/* extra driver data */
 	u16			*id;	/* identification info */
 #ifdef CONFIG_IDE_PROC_FS
@@ -750,7 +749,6 @@ struct ide_dma_ops {
 struct ide_host;
 
 typedef struct hwif_s {
-	struct hwif_s *next;		/* for linked-list in ide_hwgroup_t */
 	struct hwif_s *mate;		/* other hwif from same PCI chip */
 	struct hwgroup_s *hwgroup;	/* actually (ide_hwgroup_t *) */
 	struct proc_dir_entry *proc;	/* /proc/ide/ directory entry */
@@ -874,9 +872,7 @@ typedef struct hwgroup_s {
 	unsigned int polling	: 1;
 
 		/* current drive */
-	ide_drive_t *drive;
-		/* ptr to current hwif in linked-list */
-	ide_hwif_t *hwif;
+	ide_drive_t *cur_dev;
 
 		/* current request */
 	struct request *rq;
@@ -892,6 +888,8 @@ typedef struct hwgroup_s {
 	int req_gen_timer;
 
 	spinlock_t lock;
+
+	int port_count;
 } ide_hwgroup_t;
 
 typedef struct ide_driver_s ide_driver_t;
@@ -1622,12 +1620,7 @@ extern struct mutex ide_cfg_mtx;
 /*
  * Structure locking:
  *
- * ide_cfg_mtx and hwgroup->lock together protect changes to
- * ide_hwif_t->next
- * ide_drive_t->next
- *
  * ide_hwgroup_t->busy: hwgroup->lock
- * ide_hwgroup_t->hwif: hwgroup->lock
  * ide_hwif_t->{hwgroup,mate}: constant, no locking
  * ide_drive_t->hwif: constant, no locking
  */

[PATCH 5/6] ide: use lock bitops for ports serialization

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: use lock bitops for ports serialization

* Add ->host_busy field to struct ide_host and use it's first bit
  together with lock bitops to provide new ports serialization method.

* Convert core IDE code to use new ide_[un]lock_host() helpers.

  This removes the need for taking hwgroup->lock if host is already
  busy on serialized hosts and makes it possible to merge ide_hwgroup_t
  into ide_hwif_t (done in the later patch).

* Remove no longer needed ide_hwgroup_t.busy and ide_[un]lock_hwgroup().

* Update do_ide_request() documentation.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
 drivers/ide/ide-io.c |   77 ++++++++++++++++++++++++---------------------------
 include/linux/ide.h  |   34 +++-------------------
 2 files changed, 42 insertions(+), 69 deletions(-)

Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -666,43 +666,36 @@ void ide_stall_queue (ide_drive_t *drive
 }
 EXPORT_SYMBOL(ide_stall_queue);
 
+static inline int ide_lock_host(struct ide_host *host, ide_hwif_t *hwif)
+{
+	int rc = 0;
+
+	if (host->host_flags & IDE_HFLAG_SERIALIZE) {
+		rc = test_and_set_bit_lock(IDE_HOST_BUSY, &host->host_busy);
+		if (rc == 0) {
+			/* for atari only */
+			ide_get_lock(ide_intr, hwif);
+		}
+	}
+	return rc;
+}
+
+static inline void ide_unlock_host(struct ide_host *host)
+{
+	/* for atari only */
+	ide_release_lock();
+	if (host->host_flags & IDE_HFLAG_SERIALIZE)
+		clear_bit_unlock(IDE_HOST_BUSY, &host->host_busy);
+}
+
 /*
  * Issue a new request to a drive from hwgroup
- *
- * A hwgroup is a serialized group of IDE interfaces.  Usually there is
- * exactly one hwif (interface) per hwgroup, but buggy controllers (eg. CMD640)
- * may have both interfaces in a single hwgroup to "serialize" access.
- * Or possibly multiple ISA interfaces can share a common IRQ by being grouped
- * together into one hwgroup for serialized access.
- *
- * Note also that several hwgroups can end up sharing a single IRQ,
- * possibly along with many other devices.  This is especially common in
- * PCI-based systems with off-board IDE controller cards.
- *
- * The IDE driver uses a per-hwgroup lock to protect the hwgroup->busy flag.
- *
- * The first thread into the driver for a particular hwgroup sets the
- * hwgroup->busy flag to indicate that this hwgroup is now active,
- * and then initiates processing of the top request from the request queue.
- *
- * Other threads attempting entry notice the busy setting, and will simply
- * queue their new requests and exit immediately.  Note that hwgroup->busy
- * remains set even when the driver is merely awaiting the next interrupt.
- * Thus, the meaning is "this hwgroup is busy processing a request".
- *
- * When processing of a request completes, the completing thread or IRQ-handler
- * will start the next request from the queue.  If no more work remains,
- * the driver will clear the hwgroup->busy flag and exit.
- *
- * The per-hwgroup spinlock is used to protect all access to the
- * hwgroup->busy flag, but is otherwise not needed for most processing in
- * the driver.  This makes the driver much more friendlier to shared IRQs
- * than previous designs, while remaining 100% (?) SMP safe and capable.
  */
 void do_ide_request(struct request_queue *q)
 {
 	ide_drive_t	*drive = q->queuedata;
 	ide_hwif_t	*hwif = drive->hwif;
+	struct ide_host *host = hwif->host;
 	ide_hwgroup_t	*hwgroup = hwif->hwgroup;
 	struct request	*rq;
 	ide_startstop_t	startstop;
@@ -721,9 +714,13 @@ void do_ide_request(struct request_queue
 		blk_remove_plug(q);
 
 	spin_unlock_irq(q->queue_lock);
+
+	if (ide_lock_host(host, hwif))
+		goto plug_device_2;
+
 	spin_lock_irq(&hwgroup->lock);
 
-	if (!ide_lock_hwgroup(hwgroup, hwif)) {
+	if (1) {
 		ide_hwif_t *prev_port;
 repeat:
 		prev_port = hwif->host->cur_port;
@@ -731,7 +728,7 @@ repeat:
 
 		if (drive->dev_flags & IDE_DFLAG_SLEEPING) {
 			if (time_before(drive->sleep, jiffies)) {
-				ide_unlock_hwgroup(hwgroup);
+				ide_unlock_host(host);
 				goto plug_device;
 			}
 		}
@@ -761,7 +758,7 @@ repeat:
 		spin_lock_irq(&hwgroup->lock);
 
 		if (!rq) {
-			ide_unlock_hwgroup(hwgroup);
+			ide_unlock_host(host);
 			goto out;
 		}
 
@@ -782,7 +779,7 @@ repeat:
 		    blk_pm_request(rq) == 0 &&
 		    (rq->cmd_flags & REQ_PREEMPT) == 0) {
 			/* there should be no pending command at this point */
-			ide_unlock_hwgroup(hwgroup);
+			ide_unlock_host(host);
 			goto plug_device;
 		}
 
@@ -794,8 +791,7 @@ repeat:
 
 		if (startstop == ide_stopped)
 			goto repeat;
-	} else
-		goto plug_device;
+	}
 out:
 	spin_unlock_irq(&hwgroup->lock);
 	spin_lock_irq(q->queue_lock);
@@ -803,6 +799,7 @@ out:
 
 plug_device:
 	spin_unlock_irq(&hwgroup->lock);
+plug_device_2:
 	spin_lock_irq(q->queue_lock);
 
 	if (!elv_queue_empty(q))
@@ -844,9 +841,9 @@ static ide_startstop_t ide_dma_timeout_r
 	ide_dma_off_quietly(drive);
 
 	/*
-	 * un-busy drive etc (hwgroup->busy is cleared on return) and
-	 * make sure request is sane
+	 * un-busy drive etc and make sure request is sane
 	 */
+
 	rq = HWGROUP(drive)->rq;
 
 	if (!rq)
@@ -964,7 +961,7 @@ void ide_timer_expiry (unsigned long dat
 			spin_lock_irq(&hwgroup->lock);
 			enable_irq(hwif->irq);
 			if (startstop == ide_stopped) {
-				ide_unlock_hwgroup(hwgroup);
+				ide_unlock_host(hwif->host);
 				plug_device = 1;
 			}
 		}
@@ -1150,7 +1147,7 @@ irqreturn_t ide_intr (int irq, void *dev
 	 */
 	if (startstop == ide_stopped) {
 		if (hwgroup->handler == NULL) {	/* paranoia */
-			ide_unlock_hwgroup(hwgroup);
+			ide_unlock_host(hwif->host);
 			plug_device = 1;
 		} else
 			printk(KERN_ERR "%s: %s: huh? expected NULL handler "
Index: b/include/linux/ide.h
===================================================================
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -851,8 +851,13 @@ struct ide_host {
 	unsigned long	host_flags;
 	void		*host_priv;
 	ide_hwif_t	*cur_port;	/* for hosts requiring serialization */
+
+	/* used for hosts requiring serialization */
+	volatile long	host_busy;
 };
 
+#define IDE_HOST_BUSY 0
+
 /*
  *  internal ide interrupt handler type
  */
@@ -866,8 +871,6 @@ typedef struct hwgroup_s {
 		/* irq handler, if active */
 	ide_startstop_t	(*handler)(ide_drive_t *);
 
-		/* BOOL: protects all fields below */
-	volatile int busy;
 		/* BOOL: polling active & poll_timeout field valid */
 	unsigned int polling	: 1;
 
@@ -1271,26 +1274,6 @@ extern void ide_stall_queue(ide_drive_t 
 
 extern void ide_timer_expiry(unsigned long);
 extern irqreturn_t ide_intr(int irq, void *dev_id);
-
-static inline int ide_lock_hwgroup(ide_hwgroup_t *hwgroup, ide_hwif_t *hwif)
-{
-	if (hwgroup->busy)
-		return 1;
-
-	hwgroup->busy = 1;
-	/* for atari only */
-	ide_get_lock(ide_intr, hwif);
-
-	return 0;
-}
-
-static inline void ide_unlock_hwgroup(ide_hwgroup_t *hwgroup)
-{
-	/* for atari only */
-	ide_release_lock();
-	hwgroup->busy = 0;
-}
-
 extern void do_ide_request(struct request_queue *);
 
 void ide_init_disk(struct gendisk *, ide_drive_t *);
@@ -1617,13 +1600,6 @@ static inline void ide_set_max_pio(ide_d
 
 extern spinlock_t ide_lock;
 extern struct mutex ide_cfg_mtx;
-/*
- * Structure locking:
- *
- * ide_hwgroup_t->busy: hwgroup->lock
- * ide_hwif_t->{hwgroup,mate}: constant, no locking
- * ide_drive_t->hwif: constant, no locking
- */
 
 #define local_irq_set(flags)	do { local_save_flags((flags)); local_irq_enable_in_hardirq(); } while (0)
 

[PATCH 6/6] ide: merge ide_hwgroup_t with ide_hwif_t

[Bartlomiej Zolnierkiewicz]

From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Subject: [PATCH] ide: merge ide_hwgroup_t with ide_hwif_t

* Merge ide_hwgroup_t with ide_hwif_t.

* Cleanup init_irq() accordingly, then remove no longer needed
  ide_remove_port_from_hwgroup() and ide_ports[].

* Remove now unused HWGROUP() macro.

While at it:

* ide_dump_ata_error() fixups

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
---
 drivers/ide/alim15x3.c     |    2 
 drivers/ide/au1xxx-ide.c   |    6 +-
 drivers/ide/icside.c       |    2 
 drivers/ide/ide-atapi.c    |    8 +--
 drivers/ide/ide-cd.c       |   16 ++----
 drivers/ide/ide-dma-sff.c  |    4 -
 drivers/ide/ide-dma.c      |    2 
 drivers/ide/ide-floppy.c   |    2 
 drivers/ide/ide-io.c       |  117 ++++++++++++++++++++++-----------------------
 drivers/ide/ide-iops.c     |   67 +++++++++++--------------
 drivers/ide/ide-lib.c      |    9 ++-
 drivers/ide/ide-park.c     |   16 +++---
 drivers/ide/ide-pm.c       |    2 
 drivers/ide/ide-probe.c    |   87 +++++----------------------------
 drivers/ide/ide-tape.c     |    9 +--
 drivers/ide/ide-taskfile.c |    4 -
 drivers/ide/ide.c          |    7 --
 drivers/ide/pdc202xx_old.c |    2 
 drivers/ide/pmac.c         |    2 
 drivers/ide/scc_pata.c     |    4 -
 drivers/ide/sgiioc4.c      |    2 
 drivers/ide/tc86c001.c     |   10 +--
 drivers/ide/trm290.c       |    2 
 drivers/ide/tx4939ide.c    |    2 
 drivers/ide/umc8672.c      |   13 ++---
 include/linux/ide.h        |   55 +++++++++------------
 26 files changed, 187 insertions(+), 265 deletions(-)

Index: b/drivers/ide/alim15x3.c
===================================================================
--- a/drivers/ide/alim15x3.c
+++ b/drivers/ide/alim15x3.c
@@ -198,7 +198,7 @@ static void ali_set_dma_mode(ide_drive_t
 static int ali15x3_dma_setup(ide_drive_t *drive)
 {
 	if (m5229_revision < 0xC2 && drive->media != ide_disk) {
-		if (rq_data_dir(drive->hwif->hwgroup->rq))
+		if (rq_data_dir(drive->hwif->rq))
 			return 1;	/* try PIO instead of DMA */
 	}
 	return ide_dma_setup(drive);
Index: b/drivers/ide/au1xxx-ide.c
===================================================================
--- a/drivers/ide/au1xxx-ide.c
+++ b/drivers/ide/au1xxx-ide.c
@@ -213,7 +213,7 @@ static int auide_build_dmatable(ide_driv
 {
 	int i, iswrite, count = 0;
 	ide_hwif_t *hwif = HWIF(drive);
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = hwif->rq;
 	_auide_hwif *ahwif = &auide_hwif;
 	struct scatterlist *sg;
 
@@ -309,8 +309,8 @@ static void auide_dma_exec_cmd(ide_drive
 }
 
 static int auide_dma_setup(ide_drive_t *drive)
-{       	
-	struct request *rq = HWGROUP(drive)->rq;
+{
+	struct request *rq = drive->hwif->rq;
 
 	if (!auide_build_dmatable(drive)) {
 		ide_map_sg(drive, rq);
Index: b/drivers/ide/icside.c
===================================================================
--- a/drivers/ide/icside.c
+++ b/drivers/ide/icside.c
@@ -312,7 +312,7 @@ static int icside_dma_setup(ide_drive_t 
 	ide_hwif_t *hwif = HWIF(drive);
 	struct expansion_card *ec = ECARD_DEV(hwif->dev);
 	struct icside_state *state = ecard_get_drvdata(ec);
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	unsigned int dma_mode;
 
 	if (rq_data_dir(rq))
Index: b/drivers/ide/ide-atapi.c
===================================================================
--- a/drivers/ide/ide-atapi.c
+++ b/drivers/ide/ide-atapi.c
@@ -241,7 +241,7 @@ EXPORT_SYMBOL_GPL(ide_retry_pc);
 
 int ide_cd_expiry(ide_drive_t *drive)
 {
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 	unsigned long wait = 0;
 
 	debug_log("%s: rq->cmd[0]: 0x%x\n", __func__, rq->cmd[0]);
@@ -292,7 +292,7 @@ static ide_startstop_t ide_pc_intr(ide_d
 {
 	struct ide_atapi_pc *pc = drive->pc;
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	const struct ide_tp_ops *tp_ops = hwif->tp_ops;
 	xfer_func_t *xferfunc;
 	unsigned int timeout, temp;
@@ -489,7 +489,7 @@ static ide_startstop_t ide_transfer_pc(i
 {
 	struct ide_atapi_pc *pc = drive->pc;
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	ide_expiry_t *expiry;
 	unsigned int timeout;
 	ide_startstop_t startstop;
@@ -560,7 +560,7 @@ ide_startstop_t ide_issue_pc(ide_drive_t
 
 	if (dev_is_idecd(drive)) {
 		tf_flags = IDE_TFLAG_OUT_NSECT | IDE_TFLAG_OUT_LBAL;
-		bcount = ide_cd_get_xferlen(hwif->hwgroup->rq);
+		bcount = ide_cd_get_xferlen(hwif->rq);
 		expiry = ide_cd_expiry;
 	} else {
 		tf_flags = IDE_TFLAG_OUT_DEVICE;
Index: b/drivers/ide/ide-cd.c
===================================================================
--- a/drivers/ide/ide-cd.c
+++ b/drivers/ide/ide-cd.c
@@ -239,7 +239,7 @@ static void cdrom_queue_request_sense(id
 
 static void cdrom_end_request(ide_drive_t *drive, int uptodate)
 {
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 	int nsectors = rq->hard_cur_sectors;
 
 	ide_debug_log(IDE_DBG_FUNC, "Call %s, cmd: 0x%x, uptodate: 0x%x, "
@@ -306,8 +306,7 @@ static void ide_dump_status_no_sense(ide
 static int cdrom_decode_status(ide_drive_t *drive, int good_stat, int *stat_ret)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
-	struct request *rq = hwgroup->rq;
+	struct request *rq = hwif->rq;
 	int stat, err, sense_key;
 
 	/* check for errors */
@@ -502,7 +501,7 @@ end_request:
 		blkdev_dequeue_request(rq);
 		spin_unlock_irqrestore(q->queue_lock, flags);
 
-		hwgroup->rq = NULL;
+		hwif->rq = NULL;
 
 		cdrom_queue_request_sense(drive, rq->sense, rq);
 	} else
@@ -728,7 +727,7 @@ static ide_startstop_t ide_cd_prepare_rw
  */
 static ide_startstop_t cdrom_start_rw_cont(ide_drive_t *drive)
 {
-	struct request *rq = drive->hwif->hwgroup->rq;
+	struct request *rq = drive->hwif->rq;
 
 	/* send the command to the drive and return */
 	return cdrom_transfer_packet_command(drive, rq, cdrom_newpc_intr);
@@ -865,8 +864,7 @@ static int cdrom_newpc_intr_dummy_cb(str
 static ide_startstop_t cdrom_newpc_intr(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
-	struct request *rq = hwgroup->rq;
+	struct request *rq = hwif->rq;
 	xfer_func_t *xferfunc;
 	ide_expiry_t *expiry = NULL;
 	int dma_error = 0, dma, stat, thislen, uptodate = 0;
@@ -1072,7 +1070,7 @@ end_request:
 		if (blk_end_request(rq, 0, dlen))
 			BUG();
 
-		hwgroup->rq = NULL;
+		hwif->rq = NULL;
 	} else {
 		if (!uptodate)
 			rq->cmd_flags |= REQ_FAILED;
@@ -1125,7 +1123,7 @@ static ide_startstop_t cdrom_start_rw(id
 
 static ide_startstop_t cdrom_do_newpc_cont(ide_drive_t *drive)
 {
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 
 	return cdrom_transfer_packet_command(drive, rq, cdrom_newpc_intr);
 }
Index: b/drivers/ide/ide-dma-sff.c
===================================================================
--- a/drivers/ide/ide-dma-sff.c
+++ b/drivers/ide/ide-dma-sff.c
@@ -175,7 +175,7 @@ EXPORT_SYMBOL_GPL(ide_build_dmatable);
 int ide_dma_setup(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	unsigned int reading = rq_data_dir(rq) ? 0 : ATA_DMA_WR;
 	u8 mmio = (hwif->host_flags & IDE_HFLAG_MMIO) ? 1 : 0;
 	u8 dma_stat;
@@ -240,7 +240,7 @@ static int dma_timer_expiry(ide_drive_t 
 	if ((dma_stat & 0x18) == 0x18)	/* BUSY Stupid Early Timer !! */
 		return WAIT_CMD;
 
-	hwif->hwgroup->expiry = NULL;	/* one free ride for now */
+	hwif->expiry = NULL;	/* one free ride for now */
 
 	if (dma_stat & ATA_DMA_ERR)	/* ERROR */
 		return -1;
Index: b/drivers/ide/ide-dma.c
===================================================================
--- a/drivers/ide/ide-dma.c
+++ b/drivers/ide/ide-dma.c
@@ -96,7 +96,7 @@ ide_startstop_t ide_dma_intr(ide_drive_t
 
 	if (OK_STAT(stat, DRIVE_READY, drive->bad_wstat | ATA_DRQ)) {
 		if (!dma_stat) {
-			struct request *rq = hwif->hwgroup->rq;
+			struct request *rq = hwif->rq;
 
 			task_end_request(drive, rq, stat);
 			return ide_stopped;
Index: b/drivers/ide/ide-floppy.c
===================================================================
--- a/drivers/ide/ide-floppy.c
+++ b/drivers/ide/ide-floppy.c
@@ -71,7 +71,7 @@
 static int ide_floppy_end_request(ide_drive_t *drive, int uptodate, int nsecs)
 {
 	struct ide_disk_obj *floppy = drive->driver_data;
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 	int error;
 
 	ide_debug_log(IDE_DBG_FUNC, "Call %s\n", __func__);
Index: b/drivers/ide/ide-io.c
===================================================================
--- a/drivers/ide/ide-io.c
+++ b/drivers/ide/ide-io.c
@@ -88,7 +88,7 @@ static int __ide_end_request(ide_drive_t
 		ret = 0;
 
 	if (ret == 0 && dequeue)
-		drive->hwif->hwgroup->rq = NULL;
+		drive->hwif->rq = NULL;
 
 	return ret;
 }
@@ -107,7 +107,7 @@ static int __ide_end_request(ide_drive_t
 int ide_end_request (ide_drive_t *drive, int uptodate, int nr_sectors)
 {
 	unsigned int nr_bytes = nr_sectors << 9;
-	struct request *rq = drive->hwif->hwgroup->rq;
+	struct request *rq = drive->hwif->rq;
 
 	if (!nr_bytes) {
 		if (blk_pc_request(rq))
@@ -160,8 +160,8 @@ EXPORT_SYMBOL_GPL(ide_end_dequeued_reque
  
 void ide_end_drive_cmd (ide_drive_t *drive, u8 stat, u8 err)
 {
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
-	struct request *rq = hwgroup->rq;
+	ide_hwif_t *hwif = drive->hwif;
+	struct request *rq = hwif->rq;
 
 	if (rq->cmd_type == REQ_TYPE_ATA_TASKFILE) {
 		ide_task_t *task = (ide_task_t *)rq->special;
@@ -186,7 +186,7 @@ void ide_end_drive_cmd (ide_drive_t *dri
 		return;
 	}
 
-	hwgroup->rq = NULL;
+	hwif->rq = NULL;
 
 	rq->errors = err;
 
@@ -321,7 +321,8 @@ ide_startstop_t ide_error (ide_drive_t *
 
 	err = ide_dump_status(drive, msg, stat);
 
-	if ((rq = HWGROUP(drive)->rq) == NULL)
+	rq = drive->hwif->rq;
+	if (rq == NULL)
 		return ide_stopped;
 
 	/* retry only "normal" I/O: */
@@ -654,7 +655,7 @@ kill_rq:
  *	@timeout: time to stall for (jiffies)
  *
  *	ide_stall_queue() can be used by a drive to give excess bandwidth back
- *	to the hwgroup by sleeping for timeout jiffies.
+ *	to the port by sleeping for timeout jiffies.
  */
  
 void ide_stall_queue (ide_drive_t *drive, unsigned long timeout)
@@ -689,14 +690,13 @@ static inline void ide_unlock_host(struc
 }
 
 /*
- * Issue a new request to a drive from hwgroup
+ * Issue a new request to a device.
  */
 void do_ide_request(struct request_queue *q)
 {
 	ide_drive_t	*drive = q->queuedata;
 	ide_hwif_t	*hwif = drive->hwif;
 	struct ide_host *host = hwif->host;
-	ide_hwgroup_t	*hwgroup = hwif->hwgroup;
 	struct request	*rq;
 	ide_startstop_t	startstop;
 
@@ -718,13 +718,13 @@ void do_ide_request(struct request_queue
 	if (ide_lock_host(host, hwif))
 		goto plug_device_2;
 
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 
 	if (1) {
 		ide_hwif_t *prev_port;
 repeat:
 		prev_port = hwif->host->cur_port;
-		hwgroup->rq = NULL;
+		hwif->rq = NULL;
 
 		if (drive->dev_flags & IDE_DFLAG_SLEEPING) {
 			if (time_before(drive->sleep, jiffies)) {
@@ -739,15 +739,15 @@ repeat:
 			 * set nIEN for previous port, drives in the
 			 * quirk_list may not like intr setups/cleanups
 			 */
-			if (prev_port && hwgroup->cur_dev->quirk_list == 0)
+			if (prev_port && hwif->cur_dev->quirk_list == 0)
 				prev_port->tp_ops->set_irq(prev_port, 0);
 
 			hwif->host->cur_port = hwif;
 		}
-		hwgroup->cur_dev = drive;
+		hwif->cur_dev = drive;
 		drive->dev_flags &= ~(IDE_DFLAG_SLEEPING | IDE_DFLAG_PARKED);
 
-		spin_unlock_irq(&hwgroup->lock);
+		spin_unlock_irq(&hwif->lock);
 		spin_lock_irq(q->queue_lock);
 		/*
 		 * we know that the queue isn't empty, but this can happen
@@ -755,7 +755,7 @@ repeat:
 		 */
 		rq = elv_next_request(drive->queue);
 		spin_unlock_irq(q->queue_lock);
-		spin_lock_irq(&hwgroup->lock);
+		spin_lock_irq(&hwif->lock);
 
 		if (!rq) {
 			ide_unlock_host(host);
@@ -783,22 +783,22 @@ repeat:
 			goto plug_device;
 		}
 
-		hwgroup->rq = rq;
+		hwif->rq = rq;
 
-		spin_unlock_irq(&hwgroup->lock);
+		spin_unlock_irq(&hwif->lock);
 		startstop = start_request(drive, rq);
-		spin_lock_irq(&hwgroup->lock);
+		spin_lock_irq(&hwif->lock);
 
 		if (startstop == ide_stopped)
 			goto repeat;
 	}
 out:
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 	spin_lock_irq(q->queue_lock);
 	return;
 
 plug_device:
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 plug_device_2:
 	spin_lock_irq(q->queue_lock);
 
@@ -807,7 +807,7 @@ plug_device_2:
 }
 
 /*
- * un-busy the hwgroup etc, and clear any pending DMA status. we want to
+ * un-busy the port etc, and clear any pending DMA status. we want to
  * retry the current request in pio mode instead of risking tossing it
  * all away
  */
@@ -844,12 +844,11 @@ static ide_startstop_t ide_dma_timeout_r
 	 * un-busy drive etc and make sure request is sane
 	 */
 
-	rq = HWGROUP(drive)->rq;
-
+	rq = hwif->rq;
 	if (!rq)
 		goto out;
 
-	HWGROUP(drive)->rq = NULL;
+	hwif->rq = NULL;
 
 	rq->errors = 0;
 
@@ -877,7 +876,7 @@ static void ide_plug_device(ide_drive_t 
 
 /**
  *	ide_timer_expiry	-	handle lack of an IDE interrupt
- *	@data: timer callback magic (hwgroup)
+ *	@data: timer callback magic (hwif)
  *
  *	An IDE command has timed out before the expected drive return
  *	occurred. At this point we attempt to clean up the current
@@ -891,18 +890,18 @@ static void ide_plug_device(ide_drive_t 
  
 void ide_timer_expiry (unsigned long data)
 {
-	ide_hwgroup_t	*hwgroup = (ide_hwgroup_t *) data;
+	ide_hwif_t	*hwif = (ide_hwif_t *)data;
 	ide_drive_t	*uninitialized_var(drive);
 	ide_handler_t	*handler;
-	ide_expiry_t	*expiry;
 	unsigned long	flags;
 	unsigned long	wait = -1;
 	int		plug_device = 0;
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
+
+	handler = hwif->handler;
 
-	if (((handler = hwgroup->handler) == NULL) ||
-	    (hwgroup->req_gen != hwgroup->req_gen_timer)) {
+	if (handler == NULL || hwif->req_gen != hwif->req_gen_timer) {
 		/*
 		 * Either a marginal timeout occurred
 		 * (got the interrupt just as timer expired),
@@ -910,39 +909,39 @@ void ide_timer_expiry (unsigned long dat
 		 * Either way, we don't really want to complain about anything.
 		 */
 	} else {
-		drive = hwgroup->cur_dev;
+		drive = hwif->cur_dev;
 		if (!drive) {
 			printk(KERN_ERR "%s: ->cur_dev was NULL\n", __func__);
-			hwgroup->handler = NULL;
+			hwif->handler = NULL;
 		} else {
-			ide_hwif_t *hwif;
+			ide_expiry_t *expiry = hwif->expiry;
 			ide_startstop_t startstop = ide_stopped;
 
-			if ((expiry = hwgroup->expiry) != NULL) {
+			if (expiry) {
 				/* continue */
 				if ((wait = expiry(drive)) > 0) {
 					/* reset timer */
-					hwgroup->timer.expires  = jiffies + wait;
-					hwgroup->req_gen_timer = hwgroup->req_gen;
-					add_timer(&hwgroup->timer);
-					spin_unlock_irqrestore(&hwgroup->lock, flags);
+					hwif->timer.expires  = jiffies + wait;
+					hwif->req_gen_timer = hwif->req_gen;
+					add_timer(&hwif->timer);
+					spin_unlock_irqrestore(&hwif->lock, flags);
 					return;
 				}
 			}
-			hwgroup->handler = NULL;
+			hwif->handler = NULL;
 			/*
 			 * We need to simulate a real interrupt when invoking
 			 * the handler() function, which means we need to
 			 * globally mask the specific IRQ:
 			 */
-			spin_unlock(&hwgroup->lock);
+			spin_unlock(&hwif->lock);
 			hwif  = HWIF(drive);
 			/* disable_irq_nosync ?? */
 			disable_irq(hwif->irq);
 			/* local CPU only,
 			 * as if we were handling an interrupt */
 			local_irq_disable();
-			if (hwgroup->polling) {
+			if (hwif->polling) {
 				startstop = handler(drive);
 			} else if (drive_is_ready(drive)) {
 				if (drive->waiting_for_dma)
@@ -958,7 +957,7 @@ void ide_timer_expiry (unsigned long dat
 					ide_error(drive, "irq timeout",
 						  hwif->tp_ops->read_status(hwif));
 			}
-			spin_lock_irq(&hwgroup->lock);
+			spin_lock_irq(&hwif->lock);
 			enable_irq(hwif->irq);
 			if (startstop == ide_stopped) {
 				ide_unlock_host(hwif->host);
@@ -966,7 +965,7 @@ void ide_timer_expiry (unsigned long dat
 			}
 		}
 	}
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 
 	if (plug_device)
 		ide_plug_device(drive);
@@ -1030,7 +1029,7 @@ static void unexpected_intr(int irq, ide
  *	places
  *
  *	hwif is the interface in the group currently performing
- *	a command. hwgroup->cur_dev is the drive and hwgroup->handler is
+ *	a command. hwif->cur_dev is the drive and hwif->handler is
  *	the IRQ handler to call. As we issue a command the handlers
  *	step through multiple states, reassigning the handler to the
  *	next step in the process. Unlike a smart SCSI controller IDE
@@ -1041,13 +1040,12 @@ static void unexpected_intr(int irq, ide
  *
  *	The handler eventually returns ide_stopped to indicate the
  *	request completed. At this point we issue the next request
- *	on the hwgroup and the process begins again.
+ *	on the port and the process begins again.
  */
- 
+
 irqreturn_t ide_intr (int irq, void *dev_id)
 {
 	ide_hwif_t *hwif = (ide_hwif_t *)dev_id;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	ide_drive_t *uninitialized_var(drive);
 	ide_handler_t *handler;
 	unsigned long flags;
@@ -1060,12 +1058,14 @@ irqreturn_t ide_intr (int irq, void *dev
 			goto out_early;
 	}
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
 
 	if (!ide_ack_intr(hwif))
 		goto out;
 
-	if ((handler = hwgroup->handler) == NULL || hwgroup->polling) {
+	handler = hwif->handler;
+
+	if (handler == NULL || hwif->polling) {
 		/*
 		 * Not expecting an interrupt from this drive.
 		 * That means this could be:
@@ -1102,7 +1102,7 @@ irqreturn_t ide_intr (int irq, void *dev
 		goto out;
 	}
 
-	drive = hwgroup->cur_dev;
+	drive = hwif->cur_dev;
 	if (!drive) {
 		/*
 		 * This should NEVER happen, and there isn't much
@@ -1123,10 +1123,10 @@ irqreturn_t ide_intr (int irq, void *dev
 		 */
 		goto out;
 
-	hwgroup->handler = NULL;
-	hwgroup->req_gen++;
-	del_timer(&hwgroup->timer);
-	spin_unlock(&hwgroup->lock);
+	hwif->handler = NULL;
+	hwif->req_gen++;
+	del_timer(&hwif->timer);
+	spin_unlock(&hwif->lock);
 
 	if (hwif->port_ops && hwif->port_ops->clear_irq)
 		hwif->port_ops->clear_irq(drive);
@@ -1137,7 +1137,7 @@ irqreturn_t ide_intr (int irq, void *dev
 	/* service this interrupt, may set handler for next interrupt */
 	startstop = handler(drive);
 
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 	/*
 	 * Note that handler() may have set things up for another
 	 * interrupt to occur soon, but it cannot happen until
@@ -1146,7 +1146,7 @@ irqreturn_t ide_intr (int irq, void *dev
 	 * won't allow another of the same (on any CPU) until we return.
 	 */
 	if (startstop == ide_stopped) {
-		if (hwgroup->handler == NULL) {	/* paranoia */
+		if (hwif->handler == NULL) {	/* paranoia */
 			ide_unlock_host(hwif->host);
 			plug_device = 1;
 		} else
@@ -1156,7 +1156,7 @@ irqreturn_t ide_intr (int irq, void *dev
 out_handled:
 	irq_ret = IRQ_HANDLED;
 out:
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 out_early:
 	if (plug_device)
 		ide_plug_device(drive);
@@ -1181,11 +1181,10 @@ out_early:
 
 void ide_do_drive_cmd(ide_drive_t *drive, struct request *rq)
 {
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
 	struct request_queue *q = drive->queue;
 	unsigned long flags;
 
-	hwgroup->rq = NULL;
+	drive->hwif->rq = NULL;
 
 	spin_lock_irqsave(q->queue_lock, flags);
 	__elv_add_request(q, rq, ELEVATOR_INSERT_FRONT, 0);
Index: b/drivers/ide/ide-iops.c
===================================================================
--- a/drivers/ide/ide-iops.c
+++ b/drivers/ide/ide-iops.c
@@ -822,25 +822,25 @@ int ide_config_drive_speed(ide_drive_t *
 static void __ide_set_handler (ide_drive_t *drive, ide_handler_t *handler,
 		      unsigned int timeout, ide_expiry_t *expiry)
 {
-	ide_hwgroup_t *hwgroup = HWGROUP(drive);
+	ide_hwif_t *hwif = drive->hwif;
 
-	BUG_ON(hwgroup->handler);
-	hwgroup->handler	= handler;
-	hwgroup->expiry		= expiry;
-	hwgroup->timer.expires	= jiffies + timeout;
-	hwgroup->req_gen_timer	= hwgroup->req_gen;
-	add_timer(&hwgroup->timer);
+	BUG_ON(hwif->handler);
+	hwif->handler		= handler;
+	hwif->expiry		= expiry;
+	hwif->timer.expires	= jiffies + timeout;
+	hwif->req_gen_timer	= hwif->req_gen;
+	add_timer(&hwif->timer);
 }
 
 void ide_set_handler (ide_drive_t *drive, ide_handler_t *handler,
 		      unsigned int timeout, ide_expiry_t *expiry)
 {
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
+	ide_hwif_t *hwif = drive->hwif;
 	unsigned long flags;
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
 	__ide_set_handler(drive, handler, timeout, expiry);
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 }
 
 EXPORT_SYMBOL(ide_set_handler);
@@ -863,10 +863,9 @@ void ide_execute_command(ide_drive_t *dr
 			 unsigned timeout, ide_expiry_t *expiry)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	unsigned long flags;
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
 	__ide_set_handler(drive, handler, timeout, expiry);
 	hwif->tp_ops->exec_command(hwif, cmd);
 	/*
@@ -876,26 +875,25 @@ void ide_execute_command(ide_drive_t *dr
 	 * FIXME: we could skip this delay with care on non shared devices
 	 */
 	ndelay(400);
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 }
 EXPORT_SYMBOL(ide_execute_command);
 
 void ide_execute_pkt_cmd(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	unsigned long flags;
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
 	hwif->tp_ops->exec_command(hwif, ATA_CMD_PACKET);
 	ndelay(400);
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 }
 EXPORT_SYMBOL_GPL(ide_execute_pkt_cmd);
 
 static inline void ide_complete_drive_reset(ide_drive_t *drive, int err)
 {
-	struct request *rq = drive->hwif->hwgroup->rq;
+	struct request *rq = drive->hwif->rq;
 
 	if (rq && blk_special_request(rq) && rq->cmd[0] == REQ_DRIVE_RESET)
 		ide_end_request(drive, err ? err : 1, 0);
@@ -913,7 +911,6 @@ static ide_startstop_t do_reset1 (ide_dr
 static ide_startstop_t atapi_reset_pollfunc (ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	u8 stat;
 
 	SELECT_DRIVE(drive);
@@ -923,20 +920,20 @@ static ide_startstop_t atapi_reset_pollf
 	if (OK_STAT(stat, 0, ATA_BUSY))
 		printk("%s: ATAPI reset complete\n", drive->name);
 	else {
-		if (time_before(jiffies, hwgroup->poll_timeout)) {
+		if (time_before(jiffies, hwif->poll_timeout)) {
 			ide_set_handler(drive, &atapi_reset_pollfunc, HZ/20, NULL);
 			/* continue polling */
 			return ide_started;
 		}
 		/* end of polling */
-		hwgroup->polling = 0;
+		hwif->polling = 0;
 		printk("%s: ATAPI reset timed-out, status=0x%02x\n",
 				drive->name, stat);
 		/* do it the old fashioned way */
 		return do_reset1(drive, 1);
 	}
 	/* done polling */
-	hwgroup->polling = 0;
+	hwif->polling = 0;
 	ide_complete_drive_reset(drive, 0);
 	return ide_stopped;
 }
@@ -968,7 +965,6 @@ static void ide_reset_report_error(ide_h
  */
 static ide_startstop_t reset_pollfunc (ide_drive_t *drive)
 {
-	ide_hwgroup_t *hwgroup	= HWGROUP(drive);
 	ide_hwif_t *hwif	= HWIF(drive);
 	const struct ide_port_ops *port_ops = hwif->port_ops;
 	u8 tmp;
@@ -986,7 +982,7 @@ static ide_startstop_t reset_pollfunc (i
 	tmp = hwif->tp_ops->read_status(hwif);
 
 	if (!OK_STAT(tmp, 0, ATA_BUSY)) {
-		if (time_before(jiffies, hwgroup->poll_timeout)) {
+		if (time_before(jiffies, hwif->poll_timeout)) {
 			ide_set_handler(drive, &reset_pollfunc, HZ/20, NULL);
 			/* continue polling */
 			return ide_started;
@@ -1007,7 +1003,7 @@ static ide_startstop_t reset_pollfunc (i
 		}
 	}
 out:
-	hwgroup->polling = 0;	/* done polling */
+	hwif->polling = 0;	/* done polling */
 	ide_complete_drive_reset(drive, err);
 	return ide_stopped;
 }
@@ -1081,7 +1077,6 @@ static void pre_reset(ide_drive_t *drive
 static ide_startstop_t do_reset1 (ide_drive_t *drive, int do_not_try_atapi)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	struct ide_io_ports *io_ports = &hwif->io_ports;
 	const struct ide_tp_ops *tp_ops = hwif->tp_ops;
 	const struct ide_port_ops *port_ops;
@@ -1089,10 +1084,10 @@ static ide_startstop_t do_reset1 (ide_dr
 	unsigned int unit;
 	DEFINE_WAIT(wait);
 
-	spin_lock_irqsave(&hwgroup->lock, flags);
+	spin_lock_irqsave(&hwif->lock, flags);
 
 	/* We must not reset with running handlers */
-	BUG_ON(hwgroup->handler != NULL);
+	BUG_ON(hwif->handler != NULL);
 
 	/* For an ATAPI device, first try an ATAPI SRST. */
 	if (drive->media != ide_disk && !do_not_try_atapi) {
@@ -1101,10 +1096,10 @@ static ide_startstop_t do_reset1 (ide_dr
 		udelay (20);
 		tp_ops->exec_command(hwif, ATA_CMD_DEV_RESET);
 		ndelay(400);
-		hwgroup->poll_timeout = jiffies + WAIT_WORSTCASE;
-		hwgroup->polling = 1;
+		hwif->poll_timeout = jiffies + WAIT_WORSTCASE;
+		hwif->polling = 1;
 		__ide_set_handler(drive, &atapi_reset_pollfunc, HZ/20, NULL);
-		spin_unlock_irqrestore(&hwgroup->lock, flags);
+		spin_unlock_irqrestore(&hwif->lock, flags);
 		return ide_started;
 	}
 
@@ -1127,9 +1122,9 @@ static ide_startstop_t do_reset1 (ide_dr
 		if (time_before_eq(timeout, now))
 			break;
 
-		spin_unlock_irqrestore(&hwgroup->lock, flags);
+		spin_unlock_irqrestore(&hwif->lock, flags);
 		timeout = schedule_timeout_uninterruptible(timeout - now);
-		spin_lock_irqsave(&hwgroup->lock, flags);
+		spin_lock_irqsave(&hwif->lock, flags);
 	} while (timeout);
 	finish_wait(&ide_park_wq, &wait);
 
@@ -1141,7 +1136,7 @@ static ide_startstop_t do_reset1 (ide_dr
 		pre_reset(&hwif->drives[unit]);
 
 	if (io_ports->ctl_addr == 0) {
-		spin_unlock_irqrestore(&hwgroup->lock, flags);
+		spin_unlock_irqrestore(&hwif->lock, flags);
 		ide_complete_drive_reset(drive, -ENXIO);
 		return ide_stopped;
 	}
@@ -1164,8 +1159,8 @@ static ide_startstop_t do_reset1 (ide_dr
 	tp_ops->set_irq(hwif, drive->quirk_list == 2);
 	/* more than enough time */
 	udelay(10);
-	hwgroup->poll_timeout = jiffies + WAIT_WORSTCASE;
-	hwgroup->polling = 1;
+	hwif->poll_timeout = jiffies + WAIT_WORSTCASE;
+	hwif->polling = 1;
 	__ide_set_handler(drive, &reset_pollfunc, HZ/20, NULL);
 
 	/*
@@ -1177,7 +1172,7 @@ static ide_startstop_t do_reset1 (ide_dr
 	if (port_ops && port_ops->resetproc)
 		port_ops->resetproc(drive);
 
-	spin_unlock_irqrestore(&hwgroup->lock, flags);
+	spin_unlock_irqrestore(&hwif->lock, flags);
 	return ide_started;
 }
 
Index: b/drivers/ide/ide-lib.c
===================================================================
--- a/drivers/ide/ide-lib.c
+++ b/drivers/ide/ide-lib.c
@@ -273,7 +273,7 @@ int ide_set_xfer_rate(ide_drive_t *drive
 
 static void ide_dump_opcode(ide_drive_t *drive)
 {
-	struct request *rq = drive->hwif->hwgroup->rq;
+	struct request *rq = drive->hwif->rq;
 	ide_task_t *task = NULL;
 
 	if (!rq)
@@ -346,10 +346,13 @@ static void ide_dump_ata_error(ide_drive
 	printk(KERN_CONT "}");
 	if ((err & (ATA_BBK | ATA_ABORTED)) == ATA_BBK ||
 	    (err & (ATA_UNC | ATA_IDNF | ATA_AMNF))) {
+		struct request *rq = drive->hwif->rq;
+
 		ide_dump_sector(drive);
-		if (HWGROUP(drive) && HWGROUP(drive)->rq)
+
+		if (rq)
 			printk(KERN_CONT ", sector=%llu",
-			       (unsigned long long)HWGROUP(drive)->rq->sector);
+			       (unsigned long long)rq->sector);
 	}
 	printk(KERN_CONT "\n");
 }
Index: b/drivers/ide/ide-park.c
===================================================================
--- a/drivers/ide/ide-park.c
+++ b/drivers/ide/ide-park.c
@@ -7,22 +7,22 @@ DECLARE_WAIT_QUEUE_HEAD(ide_park_wq);
 
 static void issue_park_cmd(ide_drive_t *drive, unsigned long timeout)
 {
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
+	ide_hwif_t *hwif = drive->hwif;
 	struct request_queue *q = drive->queue;
 	struct request *rq;
 	int rc;
 
 	timeout += jiffies;
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 	if (drive->dev_flags & IDE_DFLAG_PARKED) {
 		int reset_timer = time_before(timeout, drive->sleep);
 		int start_queue = 0;
 
 		drive->sleep = timeout;
 		wake_up_all(&ide_park_wq);
-		if (reset_timer && del_timer(&hwgroup->timer))
+		if (reset_timer && del_timer(&hwif->timer))
 			start_queue = 1;
-		spin_unlock_irq(&hwgroup->lock);
+		spin_unlock_irq(&hwif->lock);
 
 		if (start_queue) {
 			spin_lock_irq(q->queue_lock);
@@ -31,7 +31,7 @@ static void issue_park_cmd(ide_drive_t *
 		}
 		return;
 	}
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 
 	rq = blk_get_request(q, READ, __GFP_WAIT);
 	rq->cmd[0] = REQ_PARK_HEADS;
@@ -64,21 +64,21 @@ ssize_t ide_park_show(struct device *dev
 		      char *buf)
 {
 	ide_drive_t *drive = to_ide_device(dev);
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
+	ide_hwif_t *hwif = drive->hwif;
 	unsigned long now;
 	unsigned int msecs;
 
 	if (drive->dev_flags & IDE_DFLAG_NO_UNLOAD)
 		return -EOPNOTSUPP;
 
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 	now = jiffies;
 	if (drive->dev_flags & IDE_DFLAG_PARKED &&
 	    time_after(drive->sleep, now))
 		msecs = jiffies_to_msecs(drive->sleep - now);
 	else
 		msecs = 0;
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 
 	return snprintf(buf, 20, "%u\n", msecs);
 }
Index: b/drivers/ide/ide-pm.c
===================================================================
--- a/drivers/ide/ide-pm.c
+++ b/drivers/ide/ide-pm.c
@@ -194,7 +194,7 @@ void ide_complete_pm_request(ide_drive_t
 	}
 	spin_unlock_irqrestore(q->queue_lock, flags);
 
-	drive->hwif->hwgroup->rq = NULL;
+	drive->hwif->rq = NULL;
 
 	if (blk_end_request(rq, 0, 0))
 		BUG();
Index: b/drivers/ide/ide-probe.c
===================================================================
--- a/drivers/ide/ide-probe.c
+++ b/drivers/ide/ide-probe.c
@@ -949,79 +949,20 @@ static int ide_port_setup_devices(ide_hw
 	return j;
 }
 
-static ide_hwif_t *ide_ports[MAX_HWIFS];
-
-void ide_remove_port_from_hwgroup(ide_hwif_t *hwif)
-{
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
-
-	ide_ports[hwif->index] = NULL;
-
-	spin_lock_irq(&hwgroup->lock);
-	/* Free the hwgroup if we were the only member. */
-	if (--hwgroup->port_count == 0)
-		kfree(hwgroup);
-	spin_unlock_irq(&hwgroup->lock);
-}
-
 /*
- * This routine sets up the irq for an ide interface, and creates a new
- * hwgroup for the irq/hwif if none was previously assigned.
- *
- * Much of the code is for correctly detecting/handling irq sharing
- * and irq serialization situations.  This is somewhat complex because
- * it handles static as well as dynamic (PCMCIA) IDE interfaces.
+ * This routine sets up the IRQ for an IDE interface.
  */
 static int init_irq (ide_hwif_t *hwif)
 {
 	struct ide_io_ports *io_ports = &hwif->io_ports;
-	unsigned int index;
-	ide_hwgroup_t *hwgroup;
-	ide_hwif_t *match = NULL;
 	int sa = 0;
 
 	mutex_lock(&ide_cfg_mtx);
-	hwif->hwgroup = NULL;
-
-	for (index = 0; index < MAX_HWIFS; index++) {
-		ide_hwif_t *h = ide_ports[index];
-
-		if (h && h->hwgroup) {  /* scan only initialized ports */
-			if (hwif->host->host_flags & IDE_HFLAG_SERIALIZE) {
-				if (hwif->host == h->host)
-					match = h;
-			}
-		}
-	}
-
-	/*
-	 * If we are still without a hwgroup, then form a new one
-	 */
-	if (match) {
-		hwgroup = match->hwgroup;
-		hwif->hwgroup = hwgroup;
-
-		spin_lock_irq(&hwgroup->lock);
-		hwgroup->port_count++;
-		spin_unlock_irq(&hwgroup->lock);
-	} else {
-		hwgroup = kmalloc_node(sizeof(*hwgroup), GFP_KERNEL|__GFP_ZERO,
-				       hwif_to_node(hwif));
-		if (hwgroup == NULL)
-			goto out_up;
-
-		spin_lock_init(&hwgroup->lock);
+	spin_lock_init(&hwif->lock);
 
-		hwif->hwgroup = hwgroup;
-
-		hwgroup->port_count = 1;
-
-		init_timer(&hwgroup->timer);
-		hwgroup->timer.function = &ide_timer_expiry;
-		hwgroup->timer.data = (unsigned long) hwgroup;
-	}
-
-	ide_ports[hwif->index] = hwif;
+	init_timer(&hwif->timer);
+	hwif->timer.function = &ide_timer_expiry;
+	hwif->timer.data = (unsigned long)hwif;
 
 #if defined(__mc68000__)
 	sa = IRQF_SHARED;
@@ -1034,7 +975,7 @@ static int init_irq (ide_hwif_t *hwif)
 		hwif->tp_ops->set_irq(hwif, 1);
 
 	if (request_irq(hwif->irq, &ide_intr, sa, hwif->name, hwif))
-		goto out_unlink;
+		goto out_up;
 
 	if (!hwif->rqsize) {
 		if ((hwif->host_flags & IDE_HFLAG_NO_LBA48) ||
@@ -1052,14 +993,12 @@ static int init_irq (ide_hwif_t *hwif)
 	printk(KERN_INFO "%s at 0x%08lx on irq %d", hwif->name,
 		io_ports->data_addr, hwif->irq);
 #endif /* __mc68000__ */
-	if (match)
-		printk(KERN_CONT " (serialized with %s)", match->name);
+	if (hwif->host->host_flags & IDE_HFLAG_SERIALIZE)
+		printk(KERN_CONT " (serialized)");
 	printk(KERN_CONT "\n");
 
 	mutex_unlock(&ide_cfg_mtx);
 	return 0;
-out_unlink:
-	ide_remove_port_from_hwgroup(hwif);
 out_up:
 	mutex_unlock(&ide_cfg_mtx);
 	return 1;
@@ -1140,20 +1079,20 @@ EXPORT_SYMBOL_GPL(ide_init_disk);
 static void drive_release_dev (struct device *dev)
 {
 	ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
-	ide_hwgroup_t *hwgroup = drive->hwif->hwgroup;
+	ide_hwif_t *hwif = drive->hwif;
 
 	ide_proc_unregister_device(drive);
 
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 	kfree(drive->id);
 	drive->id = NULL;
 	drive->dev_flags &= ~IDE_DFLAG_PRESENT;
 	/* Messed up locking ... */
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 	blk_cleanup_queue(drive->queue);
-	spin_lock_irq(&hwgroup->lock);
+	spin_lock_irq(&hwif->lock);
 	drive->queue = NULL;
-	spin_unlock_irq(&hwgroup->lock);
+	spin_unlock_irq(&hwif->lock);
 
 	complete(&drive->gendev_rel_comp);
 }
Index: b/drivers/ide/ide-tape.c
===================================================================
--- a/drivers/ide/ide-tape.c
+++ b/drivers/ide/ide-tape.c
@@ -479,7 +479,7 @@ static void ide_tape_kfree_buffer(idetap
 
 static int idetape_end_request(ide_drive_t *drive, int uptodate, int nr_sects)
 {
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 	idetape_tape_t *tape = drive->driver_data;
 	unsigned long flags;
 	int error;
@@ -531,7 +531,7 @@ static void ide_tape_callback(ide_drive_
 			printk(KERN_ERR "ide-tape: Error in REQUEST SENSE "
 					"itself - Aborting request!\n");
 	} else if (pc->c[0] == READ_6 || pc->c[0] == WRITE_6) {
-		struct request *rq = drive->hwif->hwgroup->rq;
+		struct request *rq = drive->hwif->rq;
 		int blocks = pc->xferred / tape->blk_size;
 
 		tape->avg_size += blocks * tape->blk_size;
@@ -576,7 +576,7 @@ static void ide_tape_callback(ide_drive_
 
 /*
  * Postpone the current request so that ide.c will be able to service requests
- * from another device on the same hwgroup while we are polling for DSC.
+ * from another device on the same port while we are polling for DSC.
  */
 static void idetape_postpone_request(ide_drive_t *drive)
 {
@@ -584,7 +584,8 @@ static void idetape_postpone_request(ide
 
 	debug_log(DBG_PROCS, "Enter %s\n", __func__);
 
-	tape->postponed_rq = HWGROUP(drive)->rq;
+	tape->postponed_rq = drive->hwif->rq;
+
 	ide_stall_queue(drive, tape->dsc_poll_freq);
 }
 
Index: b/drivers/ide/ide-taskfile.c
===================================================================
--- a/drivers/ide/ide-taskfile.c
+++ b/drivers/ide/ide-taskfile.c
@@ -361,7 +361,7 @@ static ide_startstop_t task_in_unexpecte
 static ide_startstop_t task_in_intr(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	u8 stat = hwif->tp_ops->read_status(hwif);
 
 	/* Error? */
@@ -395,7 +395,7 @@ static ide_startstop_t task_in_intr(ide_
 static ide_startstop_t task_out_intr (ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = hwif->rq;
 	u8 stat = hwif->tp_ops->read_status(hwif);
 
 	if (!OK_STAT(stat, DRIVE_READY, drive->bad_wstat))
Index: b/drivers/ide/ide.c
===================================================================
--- a/drivers/ide/ide.c
+++ b/drivers/ide/ide.c
@@ -189,8 +189,6 @@ void ide_unregister(ide_hwif_t *hwif)
 
 	free_irq(hwif->irq, hwif);
 
-	ide_remove_port_from_hwgroup(hwif);
-
 	device_unregister(hwif->portdev);
 	device_unregister(&hwif->gendev);
 	wait_for_completion(&hwif->gendev_rel_comp);
@@ -315,7 +313,6 @@ static int set_pio_mode_abuse(ide_hwif_t
 static int set_pio_mode(ide_drive_t *drive, int arg)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *hwgroup = hwif->hwgroup;
 	const struct ide_port_ops *port_ops = hwif->port_ops;
 
 	if (arg < 0 || arg > 255)
@@ -330,9 +327,9 @@ static int set_pio_mode(ide_drive_t *dri
 			unsigned long flags;
 
 			/* take lock for IDE_DFLAG_[NO_]UNMASK/[NO_]IO_32BIT */
-			spin_lock_irqsave(&hwgroup->lock, flags);
+			spin_lock_irqsave(&hwif->lock, flags);
 			port_ops->set_pio_mode(drive, arg);
-			spin_unlock_irqrestore(&hwgroup->lock, flags);
+			spin_unlock_irqrestore(&hwif->lock, flags);
 		} else
 			port_ops->set_pio_mode(drive, arg);
 	} else {
Index: b/drivers/ide/pdc202xx_old.c
===================================================================
--- a/drivers/ide/pdc202xx_old.c
+++ b/drivers/ide/pdc202xx_old.c
@@ -169,8 +169,8 @@ static void pdc202xx_dma_start(ide_drive
 	if (drive->current_speed > XFER_UDMA_2)
 		pdc_old_enable_66MHz_clock(drive->hwif);
 	if (drive->media != ide_disk || (drive->dev_flags & IDE_DFLAG_LBA48)) {
-		struct request *rq	= HWGROUP(drive)->rq;
 		ide_hwif_t *hwif	= HWIF(drive);
+		struct request *rq	= hwif->rq;
 		unsigned long high_16	= hwif->extra_base - 16;
 		unsigned long atapi_reg	= high_16 + (hwif->channel ? 0x24 : 0x20);
 		u32 word_count	= 0;
Index: b/drivers/ide/pmac.c
===================================================================
--- a/drivers/ide/pmac.c
+++ b/drivers/ide/pmac.c
@@ -1516,7 +1516,7 @@ pmac_ide_dma_setup(ide_drive_t *drive)
 	ide_hwif_t *hwif = HWIF(drive);
 	pmac_ide_hwif_t *pmif =
 		(pmac_ide_hwif_t *)dev_get_drvdata(hwif->gendev.parent);
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = hwif->rq;
 	u8 unit = drive->dn & 1, ata4 = (pmif->kind == controller_kl_ata4);
 
 	if (!pmac_ide_build_dmatable(drive, rq)) {
Index: b/drivers/ide/scc_pata.c
===================================================================
--- a/drivers/ide/scc_pata.c
+++ b/drivers/ide/scc_pata.c
@@ -316,7 +316,7 @@ static void scc_dma_host_set(ide_drive_t
 static int scc_dma_setup(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = hwif->rq;
 	unsigned int reading;
 	u8 dma_stat;
 
@@ -405,7 +405,7 @@ static int scc_dma_end(ide_drive_t *driv
 			       drive->name);
 			data_loss = 1;
 			if (retry++) {
-				struct request *rq = HWGROUP(drive)->rq;
+				struct request *rq = hwif->rq;
 				int unit;
 				/* ERROR_RESET and drive->crc_count are needed
 				 * to reduce DMA transfer mode in retry process.
Index: b/drivers/ide/sgiioc4.c
===================================================================
--- a/drivers/ide/sgiioc4.c
+++ b/drivers/ide/sgiioc4.c
@@ -492,7 +492,7 @@ use_pio_instead:
 
 static int sgiioc4_dma_setup(ide_drive_t *drive)
 {
-	struct request *rq = HWGROUP(drive)->rq;
+	struct request *rq = drive->hwif->rq;
 	unsigned int count = 0;
 	int ddir;
 
Index: b/drivers/ide/tc86c001.c
===================================================================
--- a/drivers/ide/tc86c001.c
+++ b/drivers/ide/tc86c001.c
@@ -64,11 +64,10 @@ static int tc86c001_timer_expiry(ide_dri
 {
 	ide_hwif_t *hwif	= HWIF(drive);
 	ide_expiry_t *expiry	= ide_get_hwifdata(hwif);
-	ide_hwgroup_t *hwgroup	= HWGROUP(drive);
 	u8 dma_stat		= inb(hwif->dma_base + ATA_DMA_STATUS);
 
 	/* Restore a higher level driver's expiry handler first. */
-	hwgroup->expiry	= expiry;
+	hwif->expiry = expiry;
 
 	if ((dma_stat & 5) == 1) {	/* DMA active and no interrupt */
 		unsigned long sc_base	= hwif->config_data;
@@ -111,10 +110,9 @@ static int tc86c001_timer_expiry(ide_dri
 static void tc86c001_dma_start(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif	= HWIF(drive);
-	ide_hwgroup_t *hwgroup	= HWGROUP(drive);
 	unsigned long sc_base	= hwif->config_data;
 	unsigned long twcr_port	= sc_base + (drive->dn ? 0x06 : 0x04);
-	unsigned long nsectors	= hwgroup->rq->nr_sectors;
+	unsigned long nsectors	= hwif->rq->nr_sectors;
 
 	/*
 	 * We have to manually load the sector count and size into
@@ -125,8 +123,8 @@ static void tc86c001_dma_start(ide_drive
 	outw(SECTOR_SIZE / 2, twcr_port); /* Transfer Word Count 1/2 */
 
 	/* Install our timeout expiry hook, saving the current handler... */
-	ide_set_hwifdata(hwif, hwgroup->expiry);
-	hwgroup->expiry = &tc86c001_timer_expiry;
+	ide_set_hwifdata(hwif, hwif->expiry);
+	hwif->expiry = &tc86c001_timer_expiry;
 
 	ide_dma_start(drive);
 }
Index: b/drivers/ide/trm290.c
===================================================================
--- a/drivers/ide/trm290.c
+++ b/drivers/ide/trm290.c
@@ -184,7 +184,7 @@ static void trm290_dma_exec_cmd(ide_driv
 static int trm290_dma_setup(ide_drive_t *drive)
 {
 	ide_hwif_t *hwif = drive->hwif;
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	unsigned int count, rw;
 
 	if (rq_data_dir(rq)) {
Index: b/drivers/ide/tx4939ide.c
===================================================================
--- a/drivers/ide/tx4939ide.c
+++ b/drivers/ide/tx4939ide.c
@@ -287,7 +287,7 @@ static int tx4939ide_dma_setup(ide_drive
 {
 	ide_hwif_t *hwif = drive->hwif;
 	void __iomem *base = TX4939IDE_BASE(hwif);
-	struct request *rq = hwif->hwgroup->rq;
+	struct request *rq = hwif->rq;
 	u8 reading;
 	int nent;
 
Index: b/drivers/ide/umc8672.c
===================================================================
--- a/drivers/ide/umc8672.c
+++ b/drivers/ide/umc8672.c
@@ -106,22 +106,21 @@ static void umc_set_speeds(u8 speeds[])
 
 static void umc_set_pio_mode(ide_drive_t *drive, const u8 pio)
 {
-	ide_hwif_t *hwif = drive->hwif;
-	ide_hwgroup_t *mate_hwgroup = hwif->mate ? hwif->mate->hwgroup : NULL;
+	ide_hwif_t *hwif = drive->hwif, *mate = hwif->mate;
 	unsigned long uninitialized_var(flags);
 
 	printk("%s: setting umc8672 to PIO mode%d (speed %d)\n",
 		drive->name, pio, pio_to_umc[pio]);
-	if (mate_hwgroup)
-		spin_lock_irqsave(&mate_hwgroup->lock, flags);
-	if (mate_hwgroup && mate_hwgroup->handler) {
+	if (mate)
+		spin_lock_irqsave(&mate->lock, flags);
+	if (mate && mate->handler) {
 		printk(KERN_ERR "umc8672: other interface is busy: exiting tune_umc()\n");
 	} else {
 		current_speeds[drive->name[2] - 'a'] = pio_to_umc[pio];
 		umc_set_speeds(current_speeds);
 	}
-	if (mate_hwgroup)
-		spin_unlock_irqrestore(&mate_hwgroup->lock, flags);
+	if (mate)
+		spin_unlock_irqrestore(&mate->lock, flags);
 }
 
 static const struct ide_port_ops umc8672_port_ops = {
Index: b/include/linux/ide.h
===================================================================
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -42,7 +42,6 @@ typedef unsigned char	byte;	/* used ever
 #define ERROR_RECAL	1	/* Recalibrate every 2nd retry */
 
 #define HWIF(drive)		((ide_hwif_t *)((drive)->hwif))
-#define HWGROUP(drive)		((ide_hwgroup_t *)(HWIF(drive)->hwgroup))
 
 /*
  * Definitions for accessing IDE controller registers
@@ -750,7 +749,6 @@ struct ide_host;
 
 typedef struct hwif_s {
 	struct hwif_s *mate;		/* other hwif from same PCI chip */
-	struct hwgroup_s *hwgroup;	/* actually (ide_hwgroup_t *) */
 	struct proc_dir_entry *proc;	/* /proc/ide/ directory entry */
 
 	struct ide_host *host;
@@ -839,6 +837,30 @@ typedef struct hwif_s {
 #ifdef CONFIG_BLK_DEV_IDEACPI
 	struct ide_acpi_hwif_link *acpidata;
 #endif
+
+	/* IRQ handler, if active */
+	ide_startstop_t	(*handler)(ide_drive_t *);
+
+	/* BOOL: polling active & poll_timeout field valid */
+	unsigned int polling : 1;
+
+	/* current drive */
+	ide_drive_t *cur_dev;
+
+	/* current request */
+	struct request *rq;
+
+	/* failsafe timer */
+	struct timer_list timer;
+	/* timeout value during long polls */
+	unsigned long poll_timeout;
+	/* queried upon timeouts */
+	int (*expiry)(ide_drive_t *);
+
+	int req_gen;
+	int req_gen_timer;
+
+	spinlock_t lock;
 } ____cacheline_internodealigned_in_smp ide_hwif_t;
 
 #define MAX_HOST_PORTS 4
@@ -867,34 +889,6 @@ typedef int (ide_expiry_t)(ide_drive_t *
 /* used by ide-cd, ide-floppy, etc. */
 typedef void (xfer_func_t)(ide_drive_t *, struct request *rq, void *, unsigned);
 
-typedef struct hwgroup_s {
-		/* irq handler, if active */
-	ide_startstop_t	(*handler)(ide_drive_t *);
-
-		/* BOOL: polling active & poll_timeout field valid */
-	unsigned int polling	: 1;
-
-		/* current drive */
-	ide_drive_t *cur_dev;
-
-		/* current request */
-	struct request *rq;
-
-		/* failsafe timer */
-	struct timer_list timer;
-		/* timeout value during long polls */
-	unsigned long poll_timeout;
-		/* queried upon timeouts */
-	int (*expiry)(ide_drive_t *);
-
-	int req_gen;
-	int req_gen_timer;
-
-	spinlock_t lock;
-
-	int port_count;
-} ide_hwgroup_t;
-
 typedef struct ide_driver_s ide_driver_t;
 
 extern struct mutex ide_setting_mtx;
@@ -1511,7 +1505,6 @@ static inline void ide_acpi_port_init_de
 static inline void ide_acpi_set_state(ide_hwif_t *hwif, int on) {}
 #endif
 
-void ide_remove_port_from_hwgroup(ide_hwif_t *);
 void ide_unregister(ide_hwif_t *);
 
 void ide_register_region(struct gendisk *);

Re: [PATCH 1/6] ide: fix setting nIEN on idle devices

[Willy Tarreau]

Hi Bart,

On Sun, Dec 14, 2008 at 12:39:32AM +0100, Bartlomiej Zolnierkiewicz wrote:
> From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
> Subject: [PATCH] ide: fix setting nIEN on idle devices
> 
> Fix do_ide_request() to operate on previous device / port instead of
> the current one.  The original code was wrong since at least Feb 2002
> (2.4.0 timeframe).

What could be the impact of this bug ? Should I backport your fix to 2.4 ?

Thanks,
Willy

Re: [PATCH 1/6] ide: fix setting nIEN on idle devices

[Bartlomiej Zolnierkiewicz]

On Sunday 14 December 2008, Willy Tarreau wrote:
> Hi Bart,
> 
> On Sun, Dec 14, 2008 at 12:39:32AM +0100, Bartlomiej Zolnierkiewicz wrote:
> > From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
> > Subject: [PATCH] ide: fix setting nIEN on idle devices
> > 
> > Fix do_ide_request() to operate on previous device / port instead of
> > the current one.  The original code was wrong since at least Feb 2002
> > (2.4.0 timeframe).
> 
> What could be the impact of this bug ? Should I backport your fix to 2.4 ?

The fix results in better handling of unexpected IRQs on serialized IDE
ports.  However since unexpected IRQs coming from a device is not something
we should normally see and I don't recall seeing any bugreports that would
be potentially closed by this fix I don't think that there is an urgent need
to backport it (thanks for caring about 2.4.x users anyway!).

Thanks,
Bart

Re: [PATCH 1/6] ide: fix setting nIEN on idle devices

[Willy Tarreau]

On Tue, Dec 16, 2008 at 07:57:52PM +0100, Bartlomiej Zolnierkiewicz wrote:
> On Sunday 14 December 2008, Willy Tarreau wrote:
> > Hi Bart,
> > 
> > On Sun, Dec 14, 2008 at 12:39:32AM +0100, Bartlomiej Zolnierkiewicz wrote:
> > > From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
> > > Subject: [PATCH] ide: fix setting nIEN on idle devices
> > > 
> > > Fix do_ide_request() to operate on previous device / port instead of
> > > the current one.  The original code was wrong since at least Feb 2002
> > > (2.4.0 timeframe).
> > 
> > What could be the impact of this bug ? Should I backport your fix to 2.4 ?
> 
> The fix results in better handling of unexpected IRQs on serialized IDE
> ports.  However since unexpected IRQs coming from a device is not something
> we should normally see and I don't recall seeing any bugreports that would
> be potentially closed by this fix I don't think that there is an urgent need
> to backport it (thanks for caring about 2.4.x users anyway!).

OK, thanks Bart !
Willy