[PATCH v2] ahci: disable DEVSLP for Intel Valleyview

[PATCH v2] ahci: disable DEVSLP for Intel Valleyview

[Jacob Pan]

On Intel Valleyview SoC, SATA device sleep is not reliable. When
DEVSLP is attempted on certain SSDs, port_devslp write would fail
and result in malfunction of AHCI controller. AHCI controller may
be not shown in PCI enumeration after reset. Complete power source
removal may be required to recover from this failure. So we blacklist
this device and override host device reported capabilities such that
device LPM will only attempt slumber but not DEVSLP.

Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Acked-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/ata/ahci.c    | 15 +++++++++++++++
 drivers/ata/ahci.h    |  1 +
 drivers/ata/libahci.c |  7 +++++++
 3 files changed, 23 insertions(+)

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index 5a0bf8e..e090d5f 100644
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -1115,6 +1115,17 @@ static bool ahci_broken_online(struct pci_dev *pdev)
 	return pdev->bus->number == (val >> 8) && pdev->devfn == (val & 0xff);
 }
 
+static bool ahci_broken_devslp(struct pci_dev *pdev)
+{
+	/* device with broken DEVSLP but still showing SDS capability */
+	static const struct pci_device_id ids[] = {
+		{ PCI_VDEVICE(INTEL, 0x0f23)}, /* Valleyview SoC */
+		{}
+	};
+
+	return pci_match_id(ids, pdev);
+}
+
 #ifdef CONFIG_ATA_ACPI
 static void ahci_gtf_filter_workaround(struct ata_host *host)
 {
@@ -1357,6 +1368,10 @@ static int ahci_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	hpriv->mmio = pcim_iomap_table(pdev)[ahci_pci_bar];
 
+	/* must set flag prior to save config in order to take effect */
+	if (ahci_broken_devslp(pdev))
+		hpriv->flags |= AHCI_HFLAG_NO_DEVSLP;
+
 	/* save initial config */
 	ahci_pci_save_initial_config(pdev, hpriv);
 
diff --git a/drivers/ata/ahci.h b/drivers/ata/ahci.h
index 0b670c0..630089a 100644
--- a/drivers/ata/ahci.h
+++ b/drivers/ata/ahci.h
@@ -235,6 +235,7 @@ enum {
 						        port start (wait until
 						        error-handling stage) */
 	AHCI_HFLAG_MULTI_MSI		= (1 << 16), /* multiple PCI MSIs */
+	AHCI_HFLAG_NO_DEVSLP		= (1 << 17), /* no device sleep */
 
 	/* ap->flags bits */
 
diff --git a/drivers/ata/libahci.c b/drivers/ata/libahci.c
index 8cd0184..32fd860 100644
--- a/drivers/ata/libahci.c
+++ b/drivers/ata/libahci.c
@@ -653,6 +653,13 @@ void ahci_save_initial_config(struct device *dev,
 		cap &= ~HOST_CAP_SNTF;
 	}
 
+	if ((cap2 & HOST_CAP2_SDS) && (hpriv->flags & AHCI_HFLAG_NO_DEVSLP)) {
+		dev_info(dev,
+			 "controller can't do DEVSLP, turning off\n");
+		cap2 &= ~HOST_CAP2_SDS;
+		cap2 &= ~HOST_CAP2_SADM;
+	}
+
 	if (!(cap & HOST_CAP_FBS) && (hpriv->flags & AHCI_HFLAG_YES_FBS)) {
 		dev_info(dev, "controller can do FBS, turning on CAP_FBS\n");
 		cap |= HOST_CAP_FBS;
-- 
1.8.1.2

Re: [PATCH v2] ahci: disable DEVSLP for Intel Valleyview

[Tejun Heo]

On Tue, Apr 15, 2014 at 10:27:11PM -0700, Jacob Pan wrote:
> On Intel Valleyview SoC, SATA device sleep is not reliable. When
> DEVSLP is attempted on certain SSDs, port_devslp write would fail
> and result in malfunction of AHCI controller. AHCI controller may
> be not shown in PCI enumeration after reset. Complete power source
> removal may be required to recover from this failure. So we blacklist
> this device and override host device reported capabilities such that
> device LPM will only attempt slumber but not DEVSLP.
> 
> Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
> Acked-by: Dan Williams <dan.j.williams@intel.com>

Applied to libata/for-3.15-fixes.

Thanks.

-- 
tejun