[PATCHSET] new Power Management for libata, take 3

[PATCHSET] new Power Management for libata, take 3

[Tejun Heo]

Hello, all.

This is the third take of new-power-management patchset.  Changes from
the last take[1] are...

* Patches #1-7 are merged to #upstream and thus dropped from this
  patchset.

* libata-reimplement-controller-wide-PM patch has been reimplemented
  to include support for ap->port_suspend() and ap->port_resume().

This patchset contains 8 patches.

#01-04 : new core PM implementation
#05-08 : convert LLDs to use new PM

As before, ata_piix, sata_sil and sata_sil24 are converted.  Both mem
and disk sleeps are tested and verifed to work on the following
configurations.

* On P5LD2 + latest BIOS
  - onboard ICH7R in ata_piix mode
  - sil3112 PCI card
  - sil3114 PCI card
  - sil3124 PCI-x card on a PCI slot
  - sil3132 PCI-e card

* On Fujitsu lifebook P7120 (+ PATA enabled + patch for PCI ID)
  - onboard ICH6M in ata_piix mode
  - sil3112 PCMCIA card

As in the last take, most of PM operations are performed by EH and
fully synchronized with all other operations.  Putting to sleep and
waking up should be safe under any circumstances as far as libata is
concerned.

Thanks.

--
tejun

[1] http://article.gmane.org/gmane.linux.ide/11570

[PATCH 1/8] libata: implement PM EH actions

[Tejun Heo]

Implement two PM per-dev EH actions - ATA_EH_SUSPEND and
ATA_EH_RESUME.  These two actions put the target device into suspended
mode and resumes from it respectively.

Once a device is put to suspended mode, no EH operations other than
RESUME is allowed on the device.  The device will stay suspended till
it gets resumed and thus reset and revalidated.  To implement this, a
new device state helper - ata_dev_ready() - is implemented and used
over various EH action implementations to operate only on attached &
running devices.

Also, if all possible devices on a port are suspended, reset is
skipped too.  This prevents spurious events including hotplug events
from disrupting suspended devices.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/libata-core.c |    5 +
 drivers/scsi/libata-eh.c   |  192 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/libata.h     |   11 ++-
 3 files changed, 203 insertions(+), 5 deletions(-)

88a32f793e5b0eabf23a4a5944fcae2c6d106535
diff --git a/drivers/scsi/libata-core.c b/drivers/scsi/libata-core.c
index 47fff7b..5d34df6 100644
--- a/drivers/scsi/libata-core.c
+++ b/drivers/scsi/libata-core.c
@@ -2138,7 +2138,7 @@ int ata_set_mode(struct ata_port *ap, st
 		 * return error code and failing device on failure.
 		 */
 		for (i = 0; i < ATA_MAX_DEVICES; i++) {
-			if (ata_dev_enabled(&ap->device[i])) {
+			if (ata_dev_ready(&ap->device[i])) {
 				ap->ops->set_mode(ap);
 				break;
 			}
@@ -2204,7 +2204,8 @@ int ata_set_mode(struct ata_port *ap, st
 	for (i = 0; i < ATA_MAX_DEVICES; i++) {
 		dev = &ap->device[i];
 
-		if (!ata_dev_enabled(dev))
+		/* don't udpate suspended devices' xfer mode */
+		if (!ata_dev_ready(dev))
 			continue;
 
 		rc = ata_dev_set_mode(dev);
diff --git a/drivers/scsi/libata-eh.c b/drivers/scsi/libata-eh.c
index bf5a72a..f05334a 100644
--- a/drivers/scsi/libata-eh.c
+++ b/drivers/scsi/libata-eh.c
@@ -760,8 +760,13 @@ static void ata_eh_about_to_do(struct at
 	unsigned long flags;
 
 	spin_lock_irqsave(ap->lock, flags);
+
 	ata_eh_clear_action(dev, &ap->eh_info, action);
-	ap->flags |= ATA_FLAG_RECOVERED;
+
+	/* set RECOVERED iff we're gonna perform real recovery */
+	if (!(action & (ATA_EH_SUSPEND | ATA_EH_RESUME)))
+		ap->flags |= ATA_FLAG_RECOVERED;
+
 	spin_unlock_irqrestore(ap->lock, flags);
 }
 
@@ -1605,7 +1610,7 @@ static int ata_eh_revalidate_and_attach(
 		dev = &ap->device[i];
 		action = ata_eh_dev_action(dev);
 
-		if (action & ATA_EH_REVALIDATE && ata_dev_enabled(dev)) {
+		if (action & ATA_EH_REVALIDATE && ata_dev_ready(dev)) {
 			if (ata_port_offline(ap)) {
 				rc = -EIO;
 				break;
@@ -1648,6 +1653,164 @@ static int ata_eh_revalidate_and_attach(
 	return rc;
 }
 
+/**
+ *	ata_eh_suspend - handle suspend EH action
+ *	@ap: target host port
+ *	@r_failed_dev: result parameter to indicate failing device
+ *
+ *	Handle suspend EH action.  Disk devices are spinned down and
+ *	other types of devices are just marked suspended.  Once
+ *	suspended, no EH action to the device is allowed until it is
+ *	resumed.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ *
+ *	RETURNS:
+ *	0 on success, -errno otherwise
+ */
+static int ata_eh_suspend(struct ata_port *ap, struct ata_device **r_failed_dev)
+{
+	struct ata_device *dev;
+	int i, rc = 0;
+
+	DPRINTK("ENTER\n");
+
+	for (i = 0; i < ATA_MAX_DEVICES; i++) {
+		unsigned long flags;
+		unsigned int action, err_mask;
+
+		dev = &ap->device[i];
+		action = ata_eh_dev_action(dev);
+
+		if (!ata_dev_enabled(dev) || !(action & ATA_EH_SUSPEND))
+			continue;
+
+		WARN_ON(dev->flags & ATA_DFLAG_SUSPENDED);
+
+		ata_eh_about_to_do(ap, dev, ATA_EH_SUSPEND);
+
+		if (dev->class == ATA_DEV_ATA && !(action & ATA_EH_PM_FREEZE)) {
+			/* flush cache */
+			rc = ata_flush_cache(dev);
+			if (rc)
+				break;
+
+			/* spin down */
+			err_mask = ata_do_simple_cmd(dev, ATA_CMD_STANDBYNOW1);
+			if (err_mask) {
+				ata_dev_printk(dev, KERN_ERR, "failed to "
+					       "spin down (err_mask=0x%x)\n",
+					       err_mask);
+				rc = -EIO;
+				break;
+			}
+		}
+
+		spin_lock_irqsave(ap->lock, flags);
+		dev->flags |= ATA_DFLAG_SUSPENDED;
+		spin_unlock_irqrestore(ap->lock, flags);
+
+		ata_eh_done(ap, dev, ATA_EH_SUSPEND);
+	}
+
+	if (rc)
+		*r_failed_dev = dev;
+
+	DPRINTK("EXIT\n");
+	return 0;
+}
+
+/**
+ *	ata_eh_prep_resume - prep for resume EH action
+ *	@ap: target host port
+ *
+ *	Clear SUSPENDED in preparation for scheduled resume actions.
+ *	This allows other parts of EH to access the devices being
+ *	resumed.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ */
+static void ata_eh_prep_resume(struct ata_port *ap)
+{
+	struct ata_device *dev;
+	unsigned long flags;
+	int i;
+
+	DPRINTK("ENTER\n");
+
+	for (i = 0; i < ATA_MAX_DEVICES; i++) {
+		unsigned int action;
+
+		dev = &ap->device[i];
+		action = ata_eh_dev_action(dev);
+
+		if (!ata_dev_enabled(dev) || !(action & ATA_EH_RESUME))
+			continue;
+
+		spin_lock_irqsave(&ap->host_set->lock, flags);
+		dev->flags &= ~ATA_DFLAG_SUSPENDED;
+		spin_unlock_irqrestore(&ap->host_set->lock, flags);
+	}
+
+	DPRINTK("EXIT\n");
+}
+
+/**
+ *	ata_eh_resume - handle resume EH action
+ *	@ap: target host port
+ *	@r_failed_dev: result parameter to indicate failing device
+ *
+ *	Handle resume EH action.  Target devices are already reset and
+ *	revalidated.  Spinning up is the only operation left.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ *
+ *	RETURNS:
+ *	0 on success, -errno otherwise
+ */
+static int ata_eh_resume(struct ata_port *ap, struct ata_device **r_failed_dev)
+{
+	struct ata_device *dev;
+	int i, rc = 0;
+
+	DPRINTK("ENTER\n");
+
+	for (i = 0; i < ATA_MAX_DEVICES; i++) {
+		unsigned int action, err_mask;
+
+		dev = &ap->device[i];
+		action = ata_eh_dev_action(dev);
+
+		if (!ata_dev_enabled(dev) || !(action & ATA_EH_RESUME))
+			continue;
+
+		ata_eh_about_to_do(ap, dev, ATA_EH_RESUME);
+
+		if (dev->class == ATA_DEV_ATA && !(action & ATA_EH_PM_FREEZE)) {
+			err_mask = ata_do_simple_cmd(dev,
+						     ATA_CMD_IDLEIMMEDIATE);
+			if (err_mask) {
+				ata_dev_printk(dev, KERN_ERR, "failed to "
+					       "spin up (err_mask=0x%x)\n",
+					       err_mask);
+				rc = -EIO;
+				break;
+			}
+		}
+
+		ata_eh_done(ap, dev, ATA_EH_RESUME);
+	}
+
+	if (rc)
+		*r_failed_dev = dev;
+
+	DPRINTK("EXIT\n");
+	return 0;
+}
+
 static int ata_port_nr_enabled(struct ata_port *ap)
 {
 	int i, cnt = 0;
@@ -1673,6 +1836,18 @@ static int ata_eh_skip_recovery(struct a
 	struct ata_eh_context *ehc = &ap->eh_context;
 	int i;
 
+	/* skip if all possible devices are suspended */
+	for (i = 0; i < ata_port_max_devices(ap); i++) {
+		struct ata_device *dev = &ap->device[i];
+
+		if (ata_dev_absent(dev) || ata_dev_ready(dev))
+			break;
+	}
+
+	if (i == ata_port_max_devices(ap))
+		return 1;
+
+	/* always thaw frozen port and recover failed devices */
 	if (ap->flags & ATA_FLAG_FROZEN || ata_port_nr_enabled(ap))
 		return 0;
 
@@ -1747,6 +1922,9 @@ static int ata_eh_recover(struct ata_por
 	if (ap->flags & ATA_FLAG_UNLOADING)
 		goto out;
 
+	/* prep for resume */
+	ata_eh_prep_resume(ap);
+
 	/* skip EH if possible. */
 	if (ata_eh_skip_recovery(ap))
 		ehc->i.action = 0;
@@ -1774,6 +1952,11 @@ static int ata_eh_recover(struct ata_por
 	if (rc)
 		goto dev_fail;
 
+	/* resume devices */
+	rc = ata_eh_resume(ap, &dev);
+	if (rc)
+		goto dev_fail;
+
 	/* configure transfer mode if the port has been reset */
 	if (ehc->i.flags & ATA_EHI_DID_RESET) {
 		rc = ata_set_mode(ap, &dev);
@@ -1783,6 +1966,11 @@ static int ata_eh_recover(struct ata_por
 		}
 	}
 
+	/* suspend devices */
+	rc = ata_eh_suspend(ap, &dev);
+	if (rc)
+		goto dev_fail;
+
 	goto out;
 
  dev_fail:
diff --git a/include/linux/libata.h b/include/linux/libata.h
index f4284bf..78c1be0 100644
--- a/include/linux/libata.h
+++ b/include/linux/libata.h
@@ -131,6 +131,7 @@ enum {
 	ATA_DFLAG_CFG_MASK	= (1 << 8) - 1,
 
 	ATA_DFLAG_PIO		= (1 << 8), /* device currently in PIO mode */
+	ATA_DFLAG_SUSPENDED	= (1 << 9), /* device suspended */
 	ATA_DFLAG_INIT_MASK	= (1 << 16) - 1,
 
 	ATA_DFLAG_DETACH	= (1 << 16),
@@ -248,9 +249,12 @@ enum {
 	ATA_EH_REVALIDATE	= (1 << 0),
 	ATA_EH_SOFTRESET	= (1 << 1),
 	ATA_EH_HARDRESET	= (1 << 2),
+	ATA_EH_SUSPEND		= (1 << 3),
+	ATA_EH_RESUME		= (1 << 4),
 
 	ATA_EH_RESET_MASK	= ATA_EH_SOFTRESET | ATA_EH_HARDRESET,
-	ATA_EH_PERDEV_MASK	= ATA_EH_REVALIDATE,
+	ATA_EH_PERDEV_MASK	= ATA_EH_REVALIDATE |
+				  ATA_EH_SUSPEND | ATA_EH_RESUME,
 
 	/* ata_eh_info->flags */
 	ATA_EHI_HOTPLUGGED	= (1 << 0),  /* could have been hotplugged */
@@ -921,6 +925,11 @@ static inline unsigned int ata_dev_absen
 	return ata_class_absent(dev->class);
 }
 
+static inline unsigned int ata_dev_ready(const struct ata_device *dev)
+{
+	return ata_dev_enabled(dev) && !(dev->flags & ATA_DFLAG_SUSPENDED);
+}
+
 /*
  * port helpers
  */
-- 
1.3.2

[PATCH 4/8] libata: reimplement controller-wide PM

[Tejun Heo]

Reimplement controller-wide PM.  ata_host_set_suspend/resume() are
defined to suspend and resume a host_set.  While suspended, EHs for
all ports in the host_set are pegged using ATA_FLAG_SUSPENDED and
frozen.

Because SCSI device hotplug is done asynchronously against the rest of
libata EH and the same mutex is used when adding new device, suspend
cannot wait for hotplug to complete.  So, if SCSI device hotplug is in
progress, suspend fails with -EBUSY.

In most cases, host_set resume is followed by device resume.  As each
resume operation requires a reset, a single host_set-wide resume
operation may result in multiple resets.  To avoid this, resume waits
upto 1 second giving PM to request resume for devices.

There is no room left for additional port flag, but several bits are
schedueled to be removed.  This patch adds ATA_FLAG_RESUME as bit 31
which is reserved for LLD flags.  No in-kernel LLD uses bit 31 now, so
this works for the time being and we can change it once some of the
obsolete bits are removed.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/libata-core.c |  160 +++++++++++++++++++++++++++++++++++++++++++-
 drivers/scsi/libata-eh.c   |  128 +++++++++++++++++++++++++++++++++++
 include/linux/libata.h     |   16 ++++
 3 files changed, 297 insertions(+), 7 deletions(-)

18ffcde45ad19c30b9cc7d7c3b33db62c698888b
diff --git a/drivers/scsi/libata-core.c b/drivers/scsi/libata-core.c
index e42a7f6..1a4afe8 100644
--- a/drivers/scsi/libata-core.c
+++ b/drivers/scsi/libata-core.c
@@ -4981,6 +4981,117 @@ int ata_port_offline(struct ata_port *ap
 	return 0;
 }
 
+static int ata_host_set_request_pm(struct ata_host_set *host_set,
+				   pm_message_t mesg, int wait)
+{
+	unsigned long flags;
+	int i, rc;
+
+	for (i = 0; i < host_set->n_ports; i++) {
+		struct ata_port *ap = host_set->ports[i];
+
+		/* Previous resume operation might still be in
+		 * progress.  Wait for PM_PENDING to clear.
+		 */
+		if (ap->flags & ATA_FLAG_PM_PENDING) {
+			ata_port_wait_eh(ap);
+			WARN_ON(ap->flags & ATA_FLAG_PM_PENDING);
+		}
+
+		/* request PM ops to EH */
+		spin_lock_irqsave(ap->lock, flags);
+
+		ap->pm_mesg = mesg;
+		if (wait) {
+			rc = 0;
+			ap->pm_result = &rc;
+		}
+
+		ap->flags |= ATA_FLAG_PM_PENDING;
+		ata_port_schedule_eh(ap);
+
+		spin_unlock_irqrestore(ap->lock, flags);
+
+		/* wait and check result */
+		if (wait) {
+			ata_port_wait_eh(ap);
+			WARN_ON(ap->flags & ATA_FLAG_PM_PENDING);
+			if (rc)
+				return rc;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ *	ata_host_set_suspend - suspend host_set
+ *	@host_set: host_set to suspend
+ *	@mesg: PM message
+ *
+ *	Suspend @host_set.  Actual operation is performed by EH.  This
+ *	function requests EH to perform PM operations and waits for EH
+ *	to finish.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ *
+ *	RETURNS:
+ *	0 on success, -errno on failure.
+ */
+int ata_host_set_suspend(struct ata_host_set *host_set, pm_message_t mesg)
+{
+	int i, j, rc;
+
+	rc = ata_host_set_request_pm(host_set, mesg, 1);
+	if (rc)
+		goto fail;
+
+	/* EH is quiescent now.  Fail if we have any ready device.
+	 * This happens if hotplug occurs between completion of device
+	 * suspension and here.
+	 */
+	for (i = 0; i < host_set->n_ports; i++) {
+		struct ata_port *ap = host_set->ports[i];
+
+		for (j = 0; j < ATA_MAX_DEVICES; j++) {
+			struct ata_device *dev = &ap->device[j];
+
+			if (ata_dev_ready(dev)) {
+				ata_port_printk(ap, KERN_WARNING,
+						"suspend failed, device %d "
+						"still active\n", dev->devno);
+				rc = -EBUSY;
+				goto fail;
+			}
+		}
+	}
+
+	host_set->dev->power.power_state = mesg;
+	return 0;
+
+ fail:
+	ata_host_set_resume(host_set);
+	return rc;
+}
+
+/**
+ *	ata_host_set_resume - resume host_set
+ *	@host_set: host_set to resume
+ *
+ *	Resume @host_set.  Actual operation is performed by EH.  This
+ *	function requests EH to perform PM operations and returns.
+ *	Note that all resume operations are performed parallely.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ */
+void ata_host_set_resume(struct ata_host_set *host_set)
+{
+	ata_host_set_request_pm(host_set, PMSG_ON, 0);
+	host_set->dev->power.power_state = PMSG_ON;
+}
+
 /**
  *	ata_port_start - Set port up for dma.
  *	@ap: Port to initialize
@@ -5621,20 +5732,55 @@ int pci_test_config_bits(struct pci_dev 
 	return (tmp == bits->val) ? 1 : 0;
 }
 
-int ata_pci_device_suspend(struct pci_dev *pdev, pm_message_t state)
+void ata_pci_device_do_suspend(struct pci_dev *pdev, pm_message_t state)
 {
 	pci_save_state(pdev);
-	pci_disable_device(pdev);
-	pci_set_power_state(pdev, PCI_D3hot);
-	return 0;
+
+	if (state.event == PM_EVENT_SUSPEND) {
+		pci_disable_device(pdev);
+		pci_set_power_state(pdev, PCI_D3hot);
+	}
 }
 
-int ata_pci_device_resume(struct pci_dev *pdev)
+void ata_pci_device_do_resume(struct pci_dev *pdev)
 {
 	pci_set_power_state(pdev, PCI_D0);
 	pci_restore_state(pdev);
 	pci_enable_device(pdev);
 	pci_set_master(pdev);
+}
+
+int ata_pci_device_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	struct ata_host_set *host_set = dev_get_drvdata(&pdev->dev);
+	int rc = 0;
+
+	rc = ata_host_set_suspend(host_set, state);
+	if (rc)
+		return rc;
+
+	if (host_set->next) {
+		rc = ata_host_set_suspend(host_set->next, state);
+		if (rc) {
+			ata_host_set_resume(host_set);
+			return rc;
+		}
+	}
+
+	ata_pci_device_do_suspend(pdev, state);
+
+	return 0;
+}
+
+int ata_pci_device_resume(struct pci_dev *pdev)
+{
+	struct ata_host_set *host_set = dev_get_drvdata(&pdev->dev);
+
+	ata_pci_device_do_resume(pdev);
+	ata_host_set_resume(host_set);
+	if (host_set->next)
+		ata_host_set_resume(host_set->next);
+
 	return 0;
 }
 #endif /* CONFIG_PCI */
@@ -5814,6 +5960,8 @@ EXPORT_SYMBOL_GPL(sata_scr_write);
 EXPORT_SYMBOL_GPL(sata_scr_write_flush);
 EXPORT_SYMBOL_GPL(ata_port_online);
 EXPORT_SYMBOL_GPL(ata_port_offline);
+EXPORT_SYMBOL_GPL(ata_host_set_suspend);
+EXPORT_SYMBOL_GPL(ata_host_set_resume);
 EXPORT_SYMBOL_GPL(ata_id_string);
 EXPORT_SYMBOL_GPL(ata_id_c_string);
 EXPORT_SYMBOL_GPL(ata_scsi_simulate);
@@ -5828,6 +5976,8 @@ EXPORT_SYMBOL_GPL(ata_pci_host_stop);
 EXPORT_SYMBOL_GPL(ata_pci_init_native_mode);
 EXPORT_SYMBOL_GPL(ata_pci_init_one);
 EXPORT_SYMBOL_GPL(ata_pci_remove_one);
+EXPORT_SYMBOL_GPL(ata_pci_device_do_suspend);
+EXPORT_SYMBOL_GPL(ata_pci_device_do_resume);
 EXPORT_SYMBOL_GPL(ata_pci_device_suspend);
 EXPORT_SYMBOL_GPL(ata_pci_device_resume);
 EXPORT_SYMBOL_GPL(ata_pci_default_filter);
diff --git a/drivers/scsi/libata-eh.c b/drivers/scsi/libata-eh.c
index 1c2b48c..06294f0 100644
--- a/drivers/scsi/libata-eh.c
+++ b/drivers/scsi/libata-eh.c
@@ -47,6 +47,8 @@ #include "libata.h"
 
 static void __ata_port_freeze(struct ata_port *ap);
 static void ata_eh_finish(struct ata_port *ap);
+static void ata_eh_handle_port_suspend(struct ata_port *ap);
+static void ata_eh_handle_port_resume(struct ata_port *ap);
 
 static void ata_ering_record(struct ata_ering *ering, int is_io,
 			     unsigned int err_mask)
@@ -263,6 +265,9 @@ void ata_scsi_error(struct Scsi_Host *ho
  repeat:
 	/* invoke error handler */
 	if (ap->ops->error_handler) {
+		/* process port resume request */
+		ata_eh_handle_port_resume(ap);
+
 		/* fetch & clear EH info */
 		spin_lock_irqsave(ap_lock, flags);
 
@@ -275,12 +280,15 @@ void ata_scsi_error(struct Scsi_Host *ho
 
 		spin_unlock_irqrestore(ap_lock, flags);
 
-		/* invoke EH.  if unloading, just finish failed qcs */
-		if (!(ap->flags & ATA_FLAG_UNLOADING))
+		/* invoke EH, skip if unloading or suspended */
+		if (!(ap->flags & (ATA_FLAG_UNLOADING | ATA_FLAG_SUSPENDED)))
 			ap->ops->error_handler(ap);
 		else
 			ata_eh_finish(ap);
 
+		/* process port suspend request */
+		ata_eh_handle_port_suspend(ap);
+
 		/* Exception might have happend after ->error_handler
 		 * recovered the port but before this point.  Repeat
 		 * EH in such case.
@@ -2126,3 +2134,119 @@ void ata_do_eh(struct ata_port *ap, ata_
 	ata_eh_recover(ap, prereset, softreset, hardreset, postreset);
 	ata_eh_finish(ap);
 }
+
+/**
+ *	ata_eh_handle_port_suspend - perform port suspend operation
+ *	@ap: port to suspend
+ *
+ *	Suspend @ap.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ */
+static void ata_eh_handle_port_suspend(struct ata_port *ap)
+{
+	unsigned long flags;
+	int rc = 0;
+
+	/* are we suspending? */
+	spin_lock_irqsave(ap->lock, flags);
+	if (!(ap->flags & ATA_FLAG_PM_PENDING) ||
+	    ap->pm_mesg.event == PM_EVENT_ON) {
+		spin_unlock_irqrestore(ap->lock, flags);
+		return;
+	}
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	WARN_ON(ap->flags & ATA_FLAG_SUSPENDED);
+
+	/* suspend */
+	ata_eh_freeze_port(ap);
+
+	if (ap->ops->port_suspend)
+		rc = ap->ops->port_suspend(ap, ap->pm_mesg);
+
+	/* report result */
+	spin_lock_irqsave(ap->lock, flags);
+
+	ap->flags &= ~ATA_FLAG_PM_PENDING;
+	if (rc == 0)
+		ap->flags |= ATA_FLAG_SUSPENDED;
+	else
+		ata_port_schedule_eh(ap);
+
+	if (ap->pm_result) {
+		*ap->pm_result = rc;
+		ap->pm_result = NULL;
+	}
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	return;
+}
+
+/**
+ *	ata_eh_handle_port_resume - perform port resume operation
+ *	@ap: port to resume
+ *
+ *	Resume @ap.
+ *
+ *	This function also waits upto one second until all devices
+ *	hanging off this port requests resume EH action.  This is to
+ *	prevent invoking EH and thus reset multiple times on resume.
+ *
+ *	On DPM resume, where some of devices might not be resumed
+ *	together, this may delay port resume upto one second, but such
+ *	DPM resumes are rare and 1 sec delay isn't too bad.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ */
+static void ata_eh_handle_port_resume(struct ata_port *ap)
+{
+	unsigned long timeout;
+	unsigned long flags;
+	int i, rc = 0;
+
+	/* are we resuming? */
+	spin_lock_irqsave(ap->lock, flags);
+	if (!(ap->flags & ATA_FLAG_PM_PENDING) ||
+	    ap->pm_mesg.event != PM_EVENT_ON) {
+		spin_unlock_irqrestore(ap->lock, flags);
+		return;
+	}
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	/* spurious? */
+	if (!(ap->flags & ATA_FLAG_SUSPENDED))
+		goto done;
+
+	if (ap->ops->port_resume)
+		rc = ap->ops->port_resume(ap);
+
+	/* give devices time to request EH */
+	timeout = jiffies + HZ; /* 1s max */
+	while (1) {
+		for (i = 0; i < ATA_MAX_DEVICES; i++) {
+			struct ata_device *dev = &ap->device[i];
+			unsigned int action = ata_eh_dev_action(dev);
+
+			if ((dev->flags & ATA_DFLAG_SUSPENDED) &&
+			    !(action & ATA_EH_RESUME))
+				break;
+		}
+
+		if (i == ATA_MAX_DEVICES || time_after(jiffies, timeout))
+			break;
+		msleep(10);
+	}
+
+ done:
+	spin_lock_irqsave(ap->lock, flags);
+	ap->flags &= ~(ATA_FLAG_PM_PENDING | ATA_FLAG_SUSPENDED);
+	if (ap->pm_result) {
+		*ap->pm_result = rc;
+		ap->pm_result = NULL;
+	}
+	spin_unlock_irqrestore(ap->lock, flags);
+}
diff --git a/include/linux/libata.h b/include/linux/libata.h
index 33943e9..e840265 100644
--- a/include/linux/libata.h
+++ b/include/linux/libata.h
@@ -176,6 +176,11 @@ enum {
 	ATA_FLAG_DISABLED	= (1 << 22), /* port is disabled, ignore it */
 	ATA_FLAG_SUSPENDED	= (1 << 23), /* port is suspended (power) */
 
+	/* XXX - Bit 31 is reserved for LLDD but no LLDD is using it
+	 * ATM.  Steal it for the time being.  The following flag must
+	 * be updated once above obsolete flags are removed. */
+	ATA_FLAG_PM_PENDING	= (1 << 31), /* PM operation pending */
+
 	/* bits 24:31 of ap->flags are reserved for LLDD specific flags */
 
 	/* struct ata_queued_cmd flags */
@@ -540,6 +545,9 @@ struct ata_port {
 	struct list_head	eh_done_q;
 	wait_queue_head_t	eh_wait_q;
 
+	pm_message_t		pm_mesg;
+	int			*pm_result;
+
 	void			*private_data;
 
 	u8			sector_buf[ATA_SECT_SIZE]; /* owned by EH */
@@ -594,6 +602,9 @@ struct ata_port_operations {
 	void (*scr_write) (struct ata_port *ap, unsigned int sc_reg,
 			   u32 val);
 
+	int (*port_suspend) (struct ata_port *ap, pm_message_t mesg);
+	int (*port_resume) (struct ata_port *ap);
+
 	int (*port_start) (struct ata_port *ap);
 	void (*port_stop) (struct ata_port *ap);
 
@@ -649,6 +660,8 @@ #ifdef CONFIG_PCI
 extern int ata_pci_init_one (struct pci_dev *pdev, struct ata_port_info **port_info,
 			     unsigned int n_ports);
 extern void ata_pci_remove_one (struct pci_dev *pdev);
+extern void ata_pci_device_do_suspend(struct pci_dev *pdev, pm_message_t state);
+extern void ata_pci_device_do_resume(struct pci_dev *pdev);
 extern int ata_pci_device_suspend(struct pci_dev *pdev, pm_message_t state);
 extern int ata_pci_device_resume(struct pci_dev *pdev);
 extern int ata_pci_clear_simplex(struct pci_dev *pdev);
@@ -669,6 +682,9 @@ extern int ata_port_online(struct ata_po
 extern int ata_port_offline(struct ata_port *ap);
 extern int ata_scsi_device_resume(struct scsi_device *);
 extern int ata_scsi_device_suspend(struct scsi_device *, pm_message_t state);
+extern int ata_host_set_suspend(struct ata_host_set *host_set,
+				pm_message_t mesg);
+extern void ata_host_set_resume(struct ata_host_set *host_set);
 extern int ata_ratelimit(void);
 extern unsigned int ata_busy_sleep(struct ata_port *ap,
 				   unsigned long timeout_pat,
-- 
1.3.2

[PATCH 5/8] sata_sil: separate out sil_init_controller()

[Tejun Heo]

Separate out controller initialization from sil_init_one() into
sil_init_controller().  This will be used by resume.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/sata_sil.c |   86 ++++++++++++++++++++++++++---------------------
 1 files changed, 48 insertions(+), 38 deletions(-)

5b2f39eb5f2f70041cdcbc575ad3e71e477006aa
diff --git a/drivers/scsi/sata_sil.c b/drivers/scsi/sata_sil.c
index 51d86d7..04f2cfe 100644
--- a/drivers/scsi/sata_sil.c
+++ b/drivers/scsi/sata_sil.c
@@ -561,6 +561,52 @@ static void sil_dev_config(struct ata_po
 	}
 }
 
+static void sil_init_controller(struct pci_dev *pdev,
+				int n_ports, unsigned long host_flags,
+				void __iomem *mmio_base)
+{
+	u8 cls;
+	u32 tmp;
+	int i;
+
+	/* Initialize FIFO PCI bus arbitration */
+	cls = sil_get_device_cache_line(pdev);
+	if (cls) {
+		cls >>= 3;
+		cls++;  /* cls = (line_size/8)+1 */
+		for (i = 0; i < n_ports; i++)
+			writew(cls << 8 | cls,
+			       mmio_base + sil_port[i].fifo_cfg);
+	} else
+		dev_printk(KERN_WARNING, &pdev->dev,
+			   "cache line size not set.  Driver may not function\n");
+
+	/* Apply R_ERR on DMA activate FIS errata workaround */
+	if (host_flags & SIL_FLAG_RERR_ON_DMA_ACT) {
+		int cnt;
+
+		for (i = 0, cnt = 0; i < n_ports; i++) {
+			tmp = readl(mmio_base + sil_port[i].sfis_cfg);
+			if ((tmp & 0x3) != 0x01)
+				continue;
+			if (!cnt)
+				dev_printk(KERN_INFO, &pdev->dev,
+					   "Applying R_ERR on DMA activate "
+					   "FIS errata fix\n");
+			writel(tmp & ~0x3, mmio_base + sil_port[i].sfis_cfg);
+			cnt++;
+		}
+	}
+
+	if (n_ports == 4) {
+		/* flip the magic "make 4 ports work" bit */
+		tmp = readl(mmio_base + sil_port[2].bmdma);
+		if ((tmp & SIL_INTR_STEERING) == 0)
+			writel(tmp | SIL_INTR_STEERING,
+			       mmio_base + sil_port[2].bmdma);
+	}
+}
+
 static int sil_init_one (struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	static int printed_version;
@@ -570,8 +616,6 @@ static int sil_init_one (struct pci_dev 
 	int rc;
 	unsigned int i;
 	int pci_dev_busy = 0;
-	u32 tmp;
-	u8 cls;
 
 	if (!printed_version++)
 		dev_printk(KERN_DEBUG, &pdev->dev, "version " DRV_VERSION "\n");
@@ -630,42 +674,8 @@ static int sil_init_one (struct pci_dev 
 		ata_std_ports(&probe_ent->port[i]);
 	}
 
-	/* Initialize FIFO PCI bus arbitration */
-	cls = sil_get_device_cache_line(pdev);
-	if (cls) {
-		cls >>= 3;
-		cls++;  /* cls = (line_size/8)+1 */
-		for (i = 0; i < probe_ent->n_ports; i++)
-			writew(cls << 8 | cls,
-			       mmio_base + sil_port[i].fifo_cfg);
-	} else
-		dev_printk(KERN_WARNING, &pdev->dev,
-			   "cache line size not set.  Driver may not function\n");
-
-	/* Apply R_ERR on DMA activate FIS errata workaround */
-	if (probe_ent->host_flags & SIL_FLAG_RERR_ON_DMA_ACT) {
-		int cnt;
-
-		for (i = 0, cnt = 0; i < probe_ent->n_ports; i++) {
-			tmp = readl(mmio_base + sil_port[i].sfis_cfg);
-			if ((tmp & 0x3) != 0x01)
-				continue;
-			if (!cnt)
-				dev_printk(KERN_INFO, &pdev->dev,
-					   "Applying R_ERR on DMA activate "
-					   "FIS errata fix\n");
-			writel(tmp & ~0x3, mmio_base + sil_port[i].sfis_cfg);
-			cnt++;
-		}
-	}
-
-	if (ent->driver_data == sil_3114) {
-		/* flip the magic "make 4 ports work" bit */
-		tmp = readl(mmio_base + sil_port[2].bmdma);
-		if ((tmp & SIL_INTR_STEERING) == 0)
-			writel(tmp | SIL_INTR_STEERING,
-			       mmio_base + sil_port[2].bmdma);
-	}
+	sil_init_controller(pdev, probe_ent->n_ports, probe_ent->host_flags,
+			    mmio_base);
 
 	pci_set_master(pdev);
 
-- 
1.3.2

[PATCH 3/8] libata: move ata_flush_cache() from libata-core.c to libata-eh.c

[Tejun Heo]

With recent PM reimplementation, there's no user left in
libata-core.c.  Move the function to libata-eh.c and make it static.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/libata-core.c |   22 ----------------------
 drivers/scsi/libata-eh.c   |   22 ++++++++++++++++++++++
 drivers/scsi/libata.h      |    1 -
 3 files changed, 22 insertions(+), 23 deletions(-)

563390c48a0dfaee3eb26e36b1cd4bad877087e2
diff --git a/drivers/scsi/libata-core.c b/drivers/scsi/libata-core.c
index 7e584c9..e42a7f6 100644
--- a/drivers/scsi/libata-core.c
+++ b/drivers/scsi/libata-core.c
@@ -4981,28 +4981,6 @@ int ata_port_offline(struct ata_port *ap
 	return 0;
 }
 
-int ata_flush_cache(struct ata_device *dev)
-{
-	unsigned int err_mask;
-	u8 cmd;
-
-	if (!ata_try_flush_cache(dev))
-		return 0;
-
-	if (ata_id_has_flush_ext(dev->id))
-		cmd = ATA_CMD_FLUSH_EXT;
-	else
-		cmd = ATA_CMD_FLUSH;
-
-	err_mask = ata_do_simple_cmd(dev, cmd);
-	if (err_mask) {
-		ata_dev_printk(dev, KERN_ERR, "failed to flush cache\n");
-		return -EIO;
-	}
-
-	return 0;
-}
-
 /**
  *	ata_port_start - Set port up for dma.
  *	@ap: Port to initialize
diff --git a/drivers/scsi/libata-eh.c b/drivers/scsi/libata-eh.c
index f05334a..1c2b48c 100644
--- a/drivers/scsi/libata-eh.c
+++ b/drivers/scsi/libata-eh.c
@@ -1653,6 +1653,28 @@ static int ata_eh_revalidate_and_attach(
 	return rc;
 }
 
+static int ata_flush_cache(struct ata_device *dev)
+{
+	unsigned int err_mask;
+	u8 cmd;
+
+	if (!ata_try_flush_cache(dev))
+		return 0;
+
+	if (ata_id_has_flush_ext(dev->id))
+		cmd = ATA_CMD_FLUSH_EXT;
+	else
+		cmd = ATA_CMD_FLUSH;
+
+	err_mask = ata_do_simple_cmd(dev, cmd);
+	if (err_mask) {
+		ata_dev_printk(dev, KERN_ERR, "failed to flush cache\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
 /**
  *	ata_eh_suspend - handle suspend EH action
  *	@ap: target host port
diff --git a/drivers/scsi/libata.h b/drivers/scsi/libata.h
index c325679..990d0b4 100644
--- a/drivers/scsi/libata.h
+++ b/drivers/scsi/libata.h
@@ -65,7 +65,6 @@ extern int ata_check_atapi_dma(struct at
 extern void ata_dev_select(struct ata_port *ap, unsigned int device,
                            unsigned int wait, unsigned int can_sleep);
 extern void swap_buf_le16(u16 *buf, unsigned int buf_words);
-extern int ata_flush_cache(struct ata_device *dev);
 extern void ata_dev_init(struct ata_device *dev);
 extern int ata_task_ioctl(struct scsi_device *scsidev, void __user *arg);
 extern int ata_cmd_ioctl(struct scsi_device *scsidev, void __user *arg);
-- 
1.3.2

[PATCH 2/8] libata: reimplement per-dev PM

[Tejun Heo]

Reimplement per-dev PM.  The original implementation directly put the
device into suspended mode and didn't synchronize w/ EH operations
including hotplug.  This patch reimplements ata_scsi_device_suspend()
and ata_scsi_device_resume() such that they request EH to perform the
respective operations.  Both functions synchronize with hotplug such
that it doesn't affect already detached devices.

Suspend wait for completion but resume just issues request and
returns.  This allows parallel wake up of devices and thus speeds up
system resume.

Due to sdev detach synchronization, it's not feasible to separate out
EH requesting from sdev handling; thus, ata_device_suspend/resume()
are removed and everything is implemented in respective libata-scsi
functions.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/libata-core.c |   84 ---------------------------------
 drivers/scsi/libata-scsi.c |  112 ++++++++++++++++++++++++++++++++++++++++++--
 include/linux/libata.h     |   11 ++--
 3 files changed, 111 insertions(+), 96 deletions(-)

73557f86d2c055c6350fe254cf7af49798d44dac
diff --git a/drivers/scsi/libata-core.c b/drivers/scsi/libata-core.c
index 5d34df6..7e584c9 100644
--- a/drivers/scsi/libata-core.c
+++ b/drivers/scsi/libata-core.c
@@ -5003,88 +5003,6 @@ int ata_flush_cache(struct ata_device *d
 	return 0;
 }
 
-static int ata_standby_drive(struct ata_device *dev)
-{
-	unsigned int err_mask;
-
-	err_mask = ata_do_simple_cmd(dev, ATA_CMD_STANDBYNOW1);
-	if (err_mask) {
-		ata_dev_printk(dev, KERN_ERR, "failed to standby drive "
-			       "(err_mask=0x%x)\n", err_mask);
-		return -EIO;
-	}
-
-	return 0;
-}
-
-static int ata_start_drive(struct ata_device *dev)
-{
-	unsigned int err_mask;
-
-	err_mask = ata_do_simple_cmd(dev, ATA_CMD_IDLEIMMEDIATE);
-	if (err_mask) {
-		ata_dev_printk(dev, KERN_ERR, "failed to start drive "
-			       "(err_mask=0x%x)\n", err_mask);
-		return -EIO;
-	}
-
-	return 0;
-}
-
-/**
- *	ata_device_resume - wakeup a previously suspended devices
- *	@dev: the device to resume
- *
- *	Kick the drive back into action, by sending it an idle immediate
- *	command and making sure its transfer mode matches between drive
- *	and host.
- *
- */
-int ata_device_resume(struct ata_device *dev)
-{
-	struct ata_port *ap = dev->ap;
-
-	if (ap->flags & ATA_FLAG_SUSPENDED) {
-		struct ata_device *failed_dev;
-
-		ata_busy_sleep(ap, ATA_TMOUT_BOOT_QUICK, ATA_TMOUT_BOOT);
-		ata_busy_wait(ap, ATA_BUSY | ATA_DRQ, 200000);
-
-		ap->flags &= ~ATA_FLAG_SUSPENDED;
-		while (ata_set_mode(ap, &failed_dev))
-			ata_dev_disable(failed_dev);
-	}
-	if (!ata_dev_enabled(dev))
-		return 0;
-	if (dev->class == ATA_DEV_ATA)
-		ata_start_drive(dev);
-
-	return 0;
-}
-
-/**
- *	ata_device_suspend - prepare a device for suspend
- *	@dev: the device to suspend
- *	@state: target power management state
- *
- *	Flush the cache on the drive, if appropriate, then issue a
- *	standbynow command.
- */
-int ata_device_suspend(struct ata_device *dev, pm_message_t state)
-{
-	struct ata_port *ap = dev->ap;
-
-	if (!ata_dev_enabled(dev))
-		return 0;
-	if (dev->class == ATA_DEV_ATA)
-		ata_flush_cache(dev);
-
-	if (state.event != PM_EVENT_FREEZE)
-		ata_standby_drive(dev);
-	ap->flags |= ATA_FLAG_SUSPENDED;
-	return 0;
-}
-
 /**
  *	ata_port_start - Set port up for dma.
  *	@ap: Port to initialize
@@ -5938,8 +5856,6 @@ EXPORT_SYMBOL_GPL(ata_pci_default_filter
 EXPORT_SYMBOL_GPL(ata_pci_clear_simplex);
 #endif /* CONFIG_PCI */
 
-EXPORT_SYMBOL_GPL(ata_device_suspend);
-EXPORT_SYMBOL_GPL(ata_device_resume);
 EXPORT_SYMBOL_GPL(ata_scsi_device_suspend);
 EXPORT_SYMBOL_GPL(ata_scsi_device_resume);
 
diff --git a/drivers/scsi/libata-scsi.c b/drivers/scsi/libata-scsi.c
index 93d18a7..67a719c 100644
--- a/drivers/scsi/libata-scsi.c
+++ b/drivers/scsi/libata-scsi.c
@@ -399,20 +399,120 @@ void ata_dump_status(unsigned id, struct
 	}
 }
 
-int ata_scsi_device_resume(struct scsi_device *sdev)
+/**
+ *	ata_scsi_device_suspend - suspend ATA device associated with sdev
+ *	@sdev: the SCSI device to suspend
+ *	@state: target power management state
+ *
+ *	Request suspend EH action on the ATA device associated with
+ *	@sdev and wait for the operation to complete.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ *
+ *	RETURNS:
+ *	0 on success, -errno otherwise.
+ */
+int ata_scsi_device_suspend(struct scsi_device *sdev, pm_message_t state)
 {
 	struct ata_port *ap = ata_shost_to_port(sdev->host);
-	struct ata_device *dev = __ata_scsi_find_dev(ap, sdev);
+	struct ata_device *dev = ata_scsi_find_dev(ap, sdev);
+	unsigned long flags;
+	unsigned int action;
+	int rc = 0;
+
+	if (!dev)
+		goto out;
+
+	spin_lock_irqsave(ap->lock, flags);
+
+	/* wait for the previous resume to complete */
+	while (dev->flags & ATA_DFLAG_SUSPENDED) {
+		spin_unlock_irqrestore(ap->lock, flags);
+		ata_port_wait_eh(ap);
+		spin_lock_irqsave(ap->lock, flags);
+	}
+
+	/* if @sdev is already detached, nothing to do */
+	if (sdev->sdev_state == SDEV_OFFLINE ||
+	    sdev->sdev_state == SDEV_CANCEL || sdev->sdev_state == SDEV_DEL)
+		goto out_unlock;
+
+	/* request suspend */
+	action = ATA_EH_SUSPEND;
+	if (state.event != PM_EVENT_SUSPEND)
+		action |= ATA_EH_PM_FREEZE;
+	ap->eh_info.dev_action[dev->devno] |= action;
+	ata_port_schedule_eh(ap);
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	/* wait for EH to do the job */
+	ata_port_wait_eh(ap);
+
+	spin_lock_irqsave(ap->lock, flags);
 
-	return ata_device_resume(dev);
+	/* If @sdev is still attached but the associated ATA device
+	 * isn't suspended, the operation failed.
+	 */
+	if (sdev->sdev_state != SDEV_OFFLINE &&
+	    sdev->sdev_state != SDEV_CANCEL && sdev->sdev_state != SDEV_DEL &&
+	    !(dev->flags & ATA_DFLAG_SUSPENDED))
+		rc = -EIO;
+
+ out_unlock:
+	spin_unlock_irqrestore(ap->lock, flags);
+ out:
+	if (rc == 0)
+		sdev->sdev_gendev.power.power_state = state;
+	return rc;
 }
 
-int ata_scsi_device_suspend(struct scsi_device *sdev, pm_message_t state)
+/**
+ *	ata_scsi_device_resume - resume ATA device associated with sdev
+ *	@sdev: the SCSI device to resume
+ *
+ *	Request resume EH action on the ATA device associated with
+ *	@sdev and return immediately.  This enables parallel
+ *	wakeup/spinup of devices.
+ *
+ *	LOCKING:
+ *	Kernel thread context (may sleep).
+ *
+ *	RETURNS:
+ *	0.
+ */
+int ata_scsi_device_resume(struct scsi_device *sdev)
 {
 	struct ata_port *ap = ata_shost_to_port(sdev->host);
-	struct ata_device *dev = __ata_scsi_find_dev(ap, sdev);
+	struct ata_device *dev = ata_scsi_find_dev(ap, sdev);
+	unsigned long flags;
+	unsigned int action;
+
+	if (!dev)
+		goto out;
 
-	return ata_device_suspend(dev, state);
+	spin_lock_irqsave(ap->lock, flags);
+
+	/* if @sdev is already detached, nothing to do */
+	if (sdev->sdev_state == SDEV_OFFLINE ||
+	    sdev->sdev_state == SDEV_CANCEL || sdev->sdev_state == SDEV_DEL)
+		goto out_unlock;
+
+	/* request resume */
+	action = ATA_EH_RESUME;
+	if (sdev->sdev_gendev.power.power_state.event == PM_EVENT_SUSPEND)
+		ata_ehi_hotplugged(&ap->eh_info);
+	else
+		action |= ATA_EH_PM_FREEZE;
+	ap->eh_info.dev_action[dev->devno] |= action;
+	ata_port_schedule_eh(ap);
+
+ out_unlock:
+	spin_unlock_irqrestore(ap->lock, flags);
+ out:
+	sdev->sdev_gendev.power.power_state = PMSG_ON;
+	return 0;
 }
 
 /**
diff --git a/include/linux/libata.h b/include/linux/libata.h
index 78c1be0..33943e9 100644
--- a/include/linux/libata.h
+++ b/include/linux/libata.h
@@ -249,12 +249,13 @@ enum {
 	ATA_EH_REVALIDATE	= (1 << 0),
 	ATA_EH_SOFTRESET	= (1 << 1),
 	ATA_EH_HARDRESET	= (1 << 2),
-	ATA_EH_SUSPEND		= (1 << 3),
-	ATA_EH_RESUME		= (1 << 4),
+	ATA_EH_SUSPEND		= (1 << 4),
+	ATA_EH_RESUME		= (1 << 5),
+	ATA_EH_PM_FREEZE	= (1 << 6), /* freeze, not actual suspend */
 
 	ATA_EH_RESET_MASK	= ATA_EH_SOFTRESET | ATA_EH_HARDRESET,
-	ATA_EH_PERDEV_MASK	= ATA_EH_REVALIDATE |
-				  ATA_EH_SUSPEND | ATA_EH_RESUME,
+	ATA_EH_PERDEV_MASK	= ATA_EH_REVALIDATE | ATA_EH_SUSPEND |
+				  ATA_EH_RESUME | ATA_EH_PM_FREEZE,
 
 	/* ata_eh_info->flags */
 	ATA_EHI_HOTPLUGGED	= (1 << 0),  /* could have been hotplugged */
@@ -668,8 +669,6 @@ extern int ata_port_online(struct ata_po
 extern int ata_port_offline(struct ata_port *ap);
 extern int ata_scsi_device_resume(struct scsi_device *);
 extern int ata_scsi_device_suspend(struct scsi_device *, pm_message_t state);
-extern int ata_device_resume(struct ata_device *);
-extern int ata_device_suspend(struct ata_device *, pm_message_t state);
 extern int ata_ratelimit(void);
 extern unsigned int ata_busy_sleep(struct ata_port *ap,
 				   unsigned long timeout_pat,
-- 
1.3.2

[PATCH 8/8] sata_sil24: add suspend/sleep support

[Tejun Heo]

Add suspend/sleep support.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/sata_sil24.c |   25 +++++++++++++++++++++++++
 1 files changed, 25 insertions(+), 0 deletions(-)

995017f5cca97018e0b11c7a0ec27791c71c837e
diff --git a/drivers/scsi/sata_sil24.c b/drivers/scsi/sata_sil24.c
index 2e1dc19..8e3deca 100644
--- a/drivers/scsi/sata_sil24.c
+++ b/drivers/scsi/sata_sil24.c
@@ -92,6 +92,7 @@ enum {
 	HOST_CTRL_STOP		= (1 << 18), /* latched PCI STOP */
 	HOST_CTRL_DEVSEL	= (1 << 19), /* latched PCI DEVSEL */
 	HOST_CTRL_REQ64		= (1 << 20), /* latched PCI REQ64 */
+	HOST_CTRL_GLOBAL_RST	= (1 << 31), /* global reset */
 
 	/*
 	 * Port registers
@@ -338,6 +339,7 @@ static int sil24_port_start(struct ata_p
 static void sil24_port_stop(struct ata_port *ap);
 static void sil24_host_stop(struct ata_host_set *host_set);
 static int sil24_init_one(struct pci_dev *pdev, const struct pci_device_id *ent);
+static int sil24_pci_device_resume(struct pci_dev *pdev);
 
 static const struct pci_device_id sil24_pci_tbl[] = {
 	{ 0x1095, 0x3124, PCI_ANY_ID, PCI_ANY_ID, 0, 0, BID_SIL3124 },
@@ -353,6 +355,8 @@ static struct pci_driver sil24_pci_drive
 	.id_table		= sil24_pci_tbl,
 	.probe			= sil24_init_one,
 	.remove			= ata_pci_remove_one, /* safe? */
+	.suspend		= ata_pci_device_suspend,
+	.resume			= sil24_pci_device_resume,
 };
 
 static struct scsi_host_template sil24_sht = {
@@ -372,6 +376,8 @@ static struct scsi_host_template sil24_s
 	.slave_configure	= ata_scsi_slave_config,
 	.slave_destroy		= ata_scsi_slave_destroy,
 	.bios_param		= ata_std_bios_param,
+	.suspend		= ata_scsi_device_suspend,
+	.resume			= ata_scsi_device_resume,
 };
 
 static const struct ata_port_operations sil24_ops = {
@@ -1179,6 +1185,25 @@ static int sil24_init_one(struct pci_dev
 	return rc;
 }
 
+static int sil24_pci_device_resume(struct pci_dev *pdev)
+{
+	struct ata_host_set *host_set = dev_get_drvdata(&pdev->dev);
+	struct sil24_host_priv *hpriv = host_set->private_data;
+
+	ata_pci_device_do_resume(pdev);
+
+	if (pdev->dev.power.power_state.event == PM_EVENT_SUSPEND)
+		writel(HOST_CTRL_GLOBAL_RST, hpriv->host_base + HOST_CTRL);
+
+	sil24_init_controller(pdev, host_set->n_ports,
+			      host_set->ports[0]->flags,
+			      hpriv->host_base, hpriv->port_base);
+
+	ata_host_set_resume(host_set);
+
+	return 0;
+}
+
 static int __init sil24_init(void)
 {
 	return pci_module_init(&sil24_pci_driver);
-- 
1.3.2

[PATCH 7/8] sata_sil24: separate out sil24_init_controller()

[Tejun Heo]

Separate out controller initialization from sil24_init_one() into
sil24_init_controller().  This will be used by resume.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/sata_sil24.c |  107 ++++++++++++++++++++++++++-------------------
 1 files changed, 62 insertions(+), 45 deletions(-)

fbc7d1ac6173a83309a542bf21a4bb7f644ff5cc
diff --git a/drivers/scsi/sata_sil24.c b/drivers/scsi/sata_sil24.c
index b5f8fa9..2e1dc19 100644
--- a/drivers/scsi/sata_sil24.c
+++ b/drivers/scsi/sata_sil24.c
@@ -988,6 +988,64 @@ static void sil24_host_stop(struct ata_h
 	kfree(hpriv);
 }
 
+static void sil24_init_controller(struct pci_dev *pdev, int n_ports,
+				  unsigned long host_flags,
+				  void __iomem *host_base,
+				  void __iomem *port_base)
+{
+	u32 tmp;
+	int i;
+
+	/* GPIO off */
+	writel(0, host_base + HOST_FLASH_CMD);
+
+	/* clear global reset & mask interrupts during initialization */
+	writel(0, host_base + HOST_CTRL);
+
+	/* init ports */
+	for (i = 0; i < n_ports; i++) {
+		void __iomem *port = port_base + i * PORT_REGS_SIZE;
+
+		/* Initial PHY setting */
+		writel(0x20c, port + PORT_PHY_CFG);
+
+		/* Clear port RST */
+		tmp = readl(port + PORT_CTRL_STAT);
+		if (tmp & PORT_CS_PORT_RST) {
+			writel(PORT_CS_PORT_RST, port + PORT_CTRL_CLR);
+			tmp = ata_wait_register(port + PORT_CTRL_STAT,
+						PORT_CS_PORT_RST,
+						PORT_CS_PORT_RST, 10, 100);
+			if (tmp & PORT_CS_PORT_RST)
+				dev_printk(KERN_ERR, &pdev->dev,
+				           "failed to clear port RST\n");
+		}
+
+		/* Configure IRQ WoC */
+		if (host_flags & SIL24_FLAG_PCIX_IRQ_WOC)
+			writel(PORT_CS_IRQ_WOC, port + PORT_CTRL_STAT);
+		else
+			writel(PORT_CS_IRQ_WOC, port + PORT_CTRL_CLR);
+
+		/* Zero error counters. */
+		writel(0x8000, port + PORT_DECODE_ERR_THRESH);
+		writel(0x8000, port + PORT_CRC_ERR_THRESH);
+		writel(0x8000, port + PORT_HSHK_ERR_THRESH);
+		writel(0x0000, port + PORT_DECODE_ERR_CNT);
+		writel(0x0000, port + PORT_CRC_ERR_CNT);
+		writel(0x0000, port + PORT_HSHK_ERR_CNT);
+
+		/* Always use 64bit activation */
+		writel(PORT_CS_32BIT_ACTV, port + PORT_CTRL_CLR);
+
+		/* Clear port multiplier enable and resume bits */
+		writel(PORT_CS_PM_EN | PORT_CS_RESUME, port + PORT_CTRL_CLR);
+	}
+
+	/* Turn on interrupts */
+	writel(IRQ_STAT_4PORTS, host_base + HOST_CTRL);
+}
+
 static int sil24_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	static int printed_version = 0;
@@ -1076,9 +1134,6 @@ static int sil24_init_one(struct pci_dev
 		}
 	}
 
-	/* GPIO off */
-	writel(0, host_base + HOST_FLASH_CMD);
-
 	/* Apply workaround for completion IRQ loss on PCI-X errata */
 	if (probe_ent->host_flags & SIL24_FLAG_PCIX_IRQ_WOC) {
 		tmp = readl(host_base + HOST_CTRL);
@@ -1090,56 +1145,18 @@ static int sil24_init_one(struct pci_dev
 			probe_ent->host_flags &= ~SIL24_FLAG_PCIX_IRQ_WOC;
 	}
 
-	/* clear global reset & mask interrupts during initialization */
-	writel(0, host_base + HOST_CTRL);
-
 	for (i = 0; i < probe_ent->n_ports; i++) {
-		void __iomem *port = port_base + i * PORT_REGS_SIZE;
-		unsigned long portu = (unsigned long)port;
+		unsigned long portu =
+			(unsigned long)port_base + i * PORT_REGS_SIZE;
 
 		probe_ent->port[i].cmd_addr = portu;
 		probe_ent->port[i].scr_addr = portu + PORT_SCONTROL;
 
 		ata_std_ports(&probe_ent->port[i]);
-
-		/* Initial PHY setting */
-		writel(0x20c, port + PORT_PHY_CFG);
-
-		/* Clear port RST */
-		tmp = readl(port + PORT_CTRL_STAT);
-		if (tmp & PORT_CS_PORT_RST) {
-			writel(PORT_CS_PORT_RST, port + PORT_CTRL_CLR);
-			tmp = ata_wait_register(port + PORT_CTRL_STAT,
-						PORT_CS_PORT_RST,
-						PORT_CS_PORT_RST, 10, 100);
-			if (tmp & PORT_CS_PORT_RST)
-				dev_printk(KERN_ERR, &pdev->dev,
-				           "failed to clear port RST\n");
-		}
-
-		/* Configure IRQ WoC */
-		if (probe_ent->host_flags & SIL24_FLAG_PCIX_IRQ_WOC)
-			writel(PORT_CS_IRQ_WOC, port + PORT_CTRL_STAT);
-		else
-			writel(PORT_CS_IRQ_WOC, port + PORT_CTRL_CLR);
-
-		/* Zero error counters. */
-		writel(0x8000, port + PORT_DECODE_ERR_THRESH);
-		writel(0x8000, port + PORT_CRC_ERR_THRESH);
-		writel(0x8000, port + PORT_HSHK_ERR_THRESH);
-		writel(0x0000, port + PORT_DECODE_ERR_CNT);
-		writel(0x0000, port + PORT_CRC_ERR_CNT);
-		writel(0x0000, port + PORT_HSHK_ERR_CNT);
-
-		/* Always use 64bit activation */
-		writel(PORT_CS_32BIT_ACTV, port + PORT_CTRL_CLR);
-
-		/* Clear port multiplier enable and resume bits */
-		writel(PORT_CS_PM_EN | PORT_CS_RESUME, port + PORT_CTRL_CLR);
 	}
 
-	/* Turn on interrupts */
-	writel(IRQ_STAT_4PORTS, host_base + HOST_CTRL);
+	sil24_init_controller(pdev, probe_ent->n_ports, probe_ent->host_flags,
+			      host_base, port_base);
 
 	pci_set_master(pdev);
 
-- 
1.3.2

[PATCH 6/8] sata_sil: add suspend/sleep support

[Tejun Heo]

Add suspend/sleep support.

Signed-off-by: Tejun Heo <htejun@gmail.com>

---

 drivers/scsi/sata_sil.c |   17 +++++++++++++++++
 1 files changed, 17 insertions(+), 0 deletions(-)

0d50db138ae1cfbfb6e6ce67ce58a0e9215117ac
diff --git a/drivers/scsi/sata_sil.c b/drivers/scsi/sata_sil.c
index 04f2cfe..e943fa8 100644
--- a/drivers/scsi/sata_sil.c
+++ b/drivers/scsi/sata_sil.c
@@ -109,6 +109,7 @@ enum {
 };
 
 static int sil_init_one (struct pci_dev *pdev, const struct pci_device_id *ent);
+static int sil_pci_device_resume(struct pci_dev *pdev);
 static void sil_dev_config(struct ata_port *ap, struct ata_device *dev);
 static u32 sil_scr_read (struct ata_port *ap, unsigned int sc_reg);
 static void sil_scr_write (struct ata_port *ap, unsigned int sc_reg, u32 val);
@@ -160,6 +161,8 @@ static struct pci_driver sil_pci_driver 
 	.id_table		= sil_pci_tbl,
 	.probe			= sil_init_one,
 	.remove			= ata_pci_remove_one,
+	.suspend		= ata_pci_device_suspend,
+	.resume			= sil_pci_device_resume,
 };
 
 static struct scsi_host_template sil_sht = {
@@ -178,6 +181,8 @@ static struct scsi_host_template sil_sht
 	.slave_configure	= ata_scsi_slave_config,
 	.slave_destroy		= ata_scsi_slave_destroy,
 	.bios_param		= ata_std_bios_param,
+	.suspend		= ata_scsi_device_suspend,
+	.resume			= ata_scsi_device_resume,
 };
 
 static const struct ata_port_operations sil_ops = {
@@ -695,6 +700,18 @@ err_out:
 	return rc;
 }
 
+static int sil_pci_device_resume(struct pci_dev *pdev)
+{
+	struct ata_host_set *host_set = dev_get_drvdata(&pdev->dev);
+
+	ata_pci_device_do_resume(pdev);
+	sil_init_controller(pdev, host_set->n_ports, host_set->ports[0]->flags,
+			    host_set->mmio_base);
+	ata_host_set_resume(host_set);
+
+	return 0;
+}
+
 static int __init sil_init(void)
 {
 	return pci_module_init(&sil_pci_driver);
-- 
1.3.2

Re: [git-patches] new Power Management for libata, take 3

[Tejun Heo]

As before, git tree is available.

http://htj.dyndns.org/git/?p=libata-tj.git;a=shortlog;h=new-pm
git://htj.dyndns.org/libata-tj new-pm

Thanks.

-- 
tejun

Re: [PATCHSET] new Power Management for libata, take 3

[Jeff Garzik]

Tejun Heo wrote:
> Hello, all.
> 
> This is the third take of new-power-management patchset.  Changes from
> the last take[1] are...
> 
> * Patches #1-7 are merged to #upstream and thus dropped from this
>   patchset.
> 
> * libata-reimplement-controller-wide-PM patch has been reimplemented
>   to include support for ap->port_suspend() and ap->port_resume().

OK, two minor changes, and I'm ready to pull this.

1) Create ap->dflags or similar, for dynamic port flags.  We shouldn't 
waste too much time worrying about port flag bits, since they apparently 
change all the time anyway :)  Just create a second variable for the 
overflow, and start moving flags that drivers never see into there.

2) Ditch patch #3, I think ata_flush_cache() is more appropriate in 
libata-core, in case other users appear.