Skip to contentSkip to conversation

sata_mv: use new sata phy register settings for new devices

PATCH In review Martin Michlmayr 5 participants 8 messages Apr 11, 2009 – May 11, 2009 libata

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sata_mv: use new sata phy register settings for new devices

From: Saeed Bishara <[email protected]>

Marvell's new SoC (65 nano) needs different modification for its SATA PHY registers.

---
This fixes the SATA problems I saw on the Kirkwood-based QNAP TS-219.

drivers/ata/sata_mv.c +67 -2
index 37ae5dc..f19f2e3 100644
@@ -293,6 +293,10 @@ enum {
 	FISCFG_WAIT_DEV_ERR	= (1 << 8),	/* wait for host on DevErr */
 	FISCFG_SINGLE_SYNC	= (1 << 16),	/* SYNC on DMA activation */
 
+	PHY_MODE9_GEN2		= 0x398,
+	PHY_MODE9_GEN1		= 0x39c,
+	PHYCFG_OFS		= 0x3a0,	/* only in 65n devices */
+
 	MV5_PHY_MODE		= 0x74,
 	MV5_LTMODE		= 0x30,
 	MV5_PHY_CTL		= 0x0C,
@@ -609,6 +613,8 @@ static int mv_soc_reset_hc(struct mv_host_priv *hpriv,
 static void mv_soc_reset_flash(struct mv_host_priv *hpriv,
 				      void __iomem *mmio);
 static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio);
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port);
 static void mv_reset_pci_bus(struct ata_host *host, void __iomem *mmio);
 static void mv_reset_channel(struct mv_host_priv *hpriv, void __iomem *mmio,
 			     unsigned int port_no);
@@ -807,6 +813,14 @@ static const struct mv_hw_ops mv_soc_ops = {
 	.reset_bus		= mv_soc_reset_bus,
 };
 
+static const struct mv_hw_ops mv_soc_65n_ops = {
+	.phy_errata		= mv_soc_65n_phy_errata,
+	.enable_leds		= mv_soc_enable_leds,
+	.reset_hc		= mv_soc_reset_hc,
+	.reset_flash		= mv_soc_reset_flash,
+	.reset_bus		= mv_soc_reset_bus,
+};
+
 /*
  * Functions
  */
@@ -3359,6 +3373,53 @@ static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio)
 	return;
 }
 
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port)
+{
+	void __iomem *port_mmio = mv_port_base(mmio, port);
+	u32	reg;
+
+	reg = readl(port_mmio + PHY_MODE3);
+	reg &= ~(0x3 << 27);	/* bits 28:27 to 0x1 */
+	reg |= (0x1 << 27);
+	reg &= ~(0x3 << 29);	/* bits 30:29 to 0x1 */
+	reg |= (0x1 << 29);
+	writel(reg, port_mmio + PHY_MODE3);
+
+	reg = readl(port_mmio + PHY_MODE4);
+	reg &= ~0x1;	/* bit 0 to 0x0, bit 16 must be set */
+	reg |= (0x1 << 16);
+	writel(reg, port_mmio + PHY_MODE4);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN2);
+	reg &= ~0xf;	/* bits 3:0 to 0x8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* bit 14 to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN2);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN1);
+	reg &= ~0xf;	/* bits 3:0 to 0x8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* bit 14 to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN1);
+}
+
+/**
+ *	soc_is_65 - check if the soc is 65 nano device
+ *
+ *	Detect the type of the SoC, this is done by reading the PHYCFG_OFS
+ *	register, this register should contain non-zero value and it exists only
+ *	in the 65 nano devices, when reading it from older devices we get 0.
+ */
+static bool soc_is_65n(struct mv_host_priv *hpriv)
+{
+	void __iomem *port0_mmio = mv_port_base(hpriv->base, 0);
+
+	if (readl(port0_mmio + PHYCFG_OFS))
+		return true;
+	return false;
+}
+
 static void mv_setup_ifcfg(void __iomem *port_mmio, int want_gen2i)
 {
 	u32 ifcfg = readl(port_mmio + SATA_IFCFG);
@@ -3699,7 +3760,10 @@ static int mv_chip_id(struct ata_host *host, unsigned int board_idx)
 		}
 		break;
 	case chip_soc:
-		hpriv->ops = &mv_soc_ops;
+		if (soc_is_65n(hpriv))
+			hpriv->ops = &mv_soc_65n_ops;
+		else
+			hpriv->ops = &mv_soc_ops;
 		hp_flags |= MV_HP_FLAG_SOC | MV_HP_GEN_IIE |
 			MV_HP_ERRATA_60X1C0;
 		break;
@@ -3762,7 +3826,8 @@ static int mv_init_host(struct ata_host *host, unsigned int board_idx)
 	n_hc = mv_get_hc_count(host->ports[0]->flags);
 
 	for (port = 0; port < host->n_ports; port++)
-		hpriv->ops->read_preamp(hpriv, port, mmio);
+		if (hpriv->ops->read_preamp)
+			hpriv->ops->read_preamp(hpriv, port, mmio);
 
 	rc = hpriv->ops->reset_hc(hpriv, mmio, n_hc);
 	if (rc)
Signature · 2 lines
Martin Michlmayr
http://www.cyrius.com/
#1 lore raw
Quoted 56 lines · Martin Michlmayr

From: Saeed Bishara <[email protected]>

Marvell's new SoC (65 nano) needs different modification for its SATA PHY registers.

---
This fixes the SATA problems I saw on the Kirkwood-based QNAP TS-219.

drivers/ata/sata_mv.c +0 -0
index 37ae5dc..f19f2e3 100644
@@ -293,6 +293,10 @@ enum {

       FISCFG_WAIT_DEV_ERR     = (1 << 8),     /* wait for host on DevErr */ FISCFG_SINGLE_SYNC      = (1 << 16),    /* SYNC on DMA activation */

+       PHY_MODE9_GEN2          = 0x398,
+       PHY_MODE9_GEN1          = 0x39c,
+       PHYCFG_OFS              = 0x3a0,        /* only in 65n devices */ +
       MV5_PHY_MODE            = 0x74,
       MV5_LTMODE              = 0x30,
       MV5_PHY_CTL             = 0x0C,

drivers/ata/sata_mv.c +11 -0
@@ -609,6 +613,8 @@ static int mv_soc_reset_hc(struct mv_host_priv *hpriv,
  static void mv_soc_reset_flash(struct mv_host_priv *hpriv,
                                      void __iomem *mmio);
  static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio);
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+                                 void __iomem *mmio, unsigned int port);
  static void mv_reset_pci_bus(struct ata_host *host, void __iomem *mmio);
  static void mv_reset_channel(struct mv_host_priv *hpriv, void __iomem *mmio,
                             unsigned int port_no);
@@ -807,6 +813,14 @@ static const struct mv_hw_ops mv_soc_ops = {
        .reset_bus              = mv_soc_reset_bus,
  };
 
+static const struct mv_hw_ops mv_soc_65n_ops = {
+       .phy_errata             = mv_soc_65n_phy_errata,
+       .enable_leds            = mv_soc_enable_leds,
+       .reset_hc               = mv_soc_reset_hc,
+       .reset_flash            = mv_soc_reset_flash,
+       .reset_bus              = mv_soc_reset_bus,
+};
+
  /*
  * Functions
  */
@@ -3359,6 +3373,53 @@ static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio)
        return;
  }
 
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,

drivers/ata/sata_mv.c static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio) · line 3386

Quoted 10 lines
drivers/ata/sata_mv.c +10 -0
+                                 void __iomem *mmio, unsigned int port)
+{
+       void __iomem *port_mmio = mv_port_base(mmio, port);
+       u32     reg;
+
+       reg = readl(port_mmio + PHY_MODE3);
+       reg &= ~(0x3 << 27);    /* bits 28:27 to 0x1 */
+       reg |= (0x1 << 27);
+       reg &= ~(0x3 << 29);    /* bits 30:29 to 0x1 */
+       reg |= (0x1 << 29);

Saeed,
Any chance these magic numbers can be documented?

No objection to this patch and I would prefer a follow on patch to document the bits if that's possible.

thanks,
grant

Quoted 70 lines
drivers/ata/sata_mv.c +36 -0
+       writel(reg, port_mmio + PHY_MODE3);
+
+       reg = readl(port_mmio + PHY_MODE4);
+       reg &= ~0x1;    /* bit 0 to 0x0, bit 16 must be set */
+       reg |= (0x1 << 16);
+       writel(reg, port_mmio + PHY_MODE4);
+
+       reg = readl(port_mmio + PHY_MODE9_GEN2);
+       reg &= ~0xf;    /* bits 3:0 to 0x8 */
+       reg |= 0x8;
+       reg &= ~(0x1 << 14);    /* bit 14 to 0 */
+       writel(reg, port_mmio + PHY_MODE9_GEN2);
+
+       reg = readl(port_mmio + PHY_MODE9_GEN1);
+       reg &= ~0xf;    /* bits 3:0 to 0x8 */
+       reg |= 0x8;
+       reg &= ~(0x1 << 14);    /* bit 14 to 0 */
+       writel(reg, port_mmio + PHY_MODE9_GEN1);
+}
+
+/**
+ *     soc_is_65 - check if the soc is 65 nano device
+ *
+ *     Detect the type of the SoC, this is done by reading the PHYCFG_OFS
+ *     register, this register should contain non-zero value and it exists only
+ *     in the 65 nano devices, when reading it from older devices we get 0.
+ */
+static bool soc_is_65n(struct mv_host_priv *hpriv)
+{
+       void __iomem *port0_mmio = mv_port_base(hpriv->base, 0);
+
+       if (readl(port0_mmio + PHYCFG_OFS))
+               return true;
+       return false;
+}
+
  static void mv_setup_ifcfg(void __iomem *port_mmio, int want_gen2i)
  {
        u32 ifcfg = readl(port_mmio + SATA_IFCFG);
@@ -3699,7 +3760,10 @@ static int mv_chip_id(struct ata_host *host, unsigned int board_idx)
                }
                break;

       case chip_soc:
-               hpriv->ops = &mv_soc_ops;
+               if (soc_is_65n(hpriv))
+                       hpriv->ops = &mv_soc_65n_ops;
+               else
+                       hpriv->ops = &mv_soc_ops; hp_flags |= MV_HP_FLAG_SOC | MV_HP_GEN_IIE | MV_HP_ERRATA_60X1C0;
               break;

drivers/ata/sata_mv.c +2 -2
@@ -3762,7 +3826,8 @@ static int mv_init_host(struct ata_host *host, unsigned int board_idx)
        n_hc = mv_get_hc_count(host->ports[0]->flags);
 
        for (port = 0; port < host->n_ports; port++)
-               hpriv->ops->read_preamp(hpriv, port, mmio);
+               if (hpriv->ops->read_preamp)
+                       hpriv->ops->read_preamp(hpriv, port, mmio);
 
        rc = hpriv->ops->reset_hc(hpriv, mmio, n_hc);
        if (rc)
 
--

Martin Michlmayr
http://www.cyrius.com/
--
To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

#2 lore raw
Quoted 4 lines

Any chance these magic numbers can be documented?

No objection to this patch and I would prefer a follow on patch to document the bits if that's possible.

Grant,
those bits control analog/electrical settings of the sata phy
module, the functional spec uses acronyms to name it, and without much documentation for it.

saeed

#3 lore raw
Quoted 8 lines · saeed bishara
Quoted 4 lines

Any chance these magic numbers can be documented?

No objection to this patch and I would prefer a follow on patch to document the bits if that's possible.

Grant,
those bits control analog/electrical settings of the sata phy
module, the functional spec uses acronyms to name it, and without much documentation for it.

Named constants based on acronyms are preferable to numeric constants. A little information and disambiguation is better than none at all.

Jeff

Tested-by Martin Michlmayr
#4 lore raw

Here the patch with names constants:
To: Mark Lord <[email protected]>
Cc: Martin Michlmayr <[email protected]>, [email protected]
Subject: [PATCH] sata_mv: use new sata phy register settings for new devices

Marvell's new SoC (65 nano) needs different settings for its SATA PHY registers.

drivers/ata/sata_mv.c +14 -0
@@ -293,6 +293,10 @@ enum {
 	FISCFG_WAIT_DEV_ERR	= (1 << 8),	/* wait for host on DevErr */
 	FISCFG_SINGLE_SYNC	= (1 << 16),	/* SYNC on DMA activation */
 
+	PHY_MODE9_GEN2		= 0x398,
+	PHY_MODE9_GEN1		= 0x39c,
+	PHYCFG_OFS		= 0x3a0,	/* only in 65n devices */
+
 	MV5_PHY_MODE		= 0x74,
 	MV5_LTMODE		= 0x30,
 	MV5_PHY_CTL		= 0x0C,
@@ -609,6 +613,8 @@ static int mv_soc_reset_hc(struct mv_host_priv *hpriv,
 static void mv_soc_reset_flash(struct mv_host_priv *hpriv,
 				      void __iomem *mmio);
 static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio);
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port);
 static void mv_reset_pci_bus(struct ata_host *host, void __iomem *mmio);
 static void mv_reset_channel(struct mv_host_priv *hpriv, void __iomem *mmio,
 			     unsigned int port_no);
@@ -807,6 +813,14 @@ static const struct mv_hw_ops mv_soc_ops = {
 	.reset_bus		= mv_soc_reset_bus,
 };
 
+static const struct mv_hw_ops mv_soc_65n_ops = {
+	.phy_errata		= mv_soc_65n_phy_errata,
+	.enable_leds		= mv_soc_enable_leds,
+	.reset_hc		= mv_soc_reset_hc,
+	.reset_flash		= mv_soc_reset_flash,
+	.reset_bus		= mv_soc_reset_bus,
+};
+
 /*
  * Functions
  */
@@ -3359,6 +3373,53 @@ static void mv_soc_reset_bus(struct ata_host

*host, void __iomem *mmio)
return;
}

+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+ void __iomem *mmio, unsigned int port)
+{
+ void __iomem *port_mmio = mv_port_base(mmio, port);
+ u32 reg;
+
+ reg = readl(port_mmio + PHY_MODE3);
+ reg &= ~(0x3 << 27); /* SELMUPF (bits 28:27) to 1 */
+ reg |= (0x1 << 27);
+ reg &= ~(0x3 << 29); /* SELMUPI (bits 30:29) to 1 */
+ reg |= (0x1 << 29);
+ writel(reg, port_mmio + PHY_MODE3);
+
+ reg = readl(port_mmio + PHY_MODE4);
+ reg &= ~0x1; /* SATU_OD8 (bit 0) to 0, reserved bit 16 must be set */
+ reg |= (0x1 << 16);
+ writel(reg, port_mmio + PHY_MODE4);
+
+ reg = readl(port_mmio + PHY_MODE9_GEN2);
+ reg &= ~0xf; /* TXAMP[3:0] (bits 3:0) to 8 */
+ reg |= 0x8;
+ reg &= ~(0x1 << 14); /* TXAMP[4] (bit 14) to 0 */
+ writel(reg, port_mmio + PHY_MODE9_GEN2);
+
+ reg = readl(port_mmio + PHY_MODE9_GEN1);
+ reg &= ~0xf; /* TXAMP[3:0] (bits 3:0) to 8 */
+ reg |= 0x8;
+ reg &= ~(0x1 << 14); /* TXAMP[4] (bit 14) to 0 */
+ writel(reg, port_mmio + PHY_MODE9_GEN1);
+}
+
+/**
+ * soc_is_65 - check if the soc is 65 nano device
+ *
+ * Detect the type of the SoC, this is done by reading the PHYCFG_OFS
+ * register, this register should contain non-zero value and it exists only
+ * in the 65 nano devices, when reading it from older devices we get 0.
+ */
+static bool soc_is_65n(struct mv_host_priv *hpriv)
+{
+ void __iomem *port0_mmio = mv_port_base(hpriv->base, 0);
+
+ if (readl(port0_mmio + PHYCFG_OFS))
+ return true;
+ return false;
+}
+
static void mv_setup_ifcfg(void __iomem *port_mmio, int want_gen2i) {
u32 ifcfg = readl(port_mmio + SATA_IFCFG);
@@ -3699,7 +3760,10 @@ static int mv_chip_id(struct ata_host *host, unsigned int board_idx)
}
break;
case chip_soc:
- hpriv->ops = &mv_soc_ops;
+ if (soc_is_65n(hpriv))
+ hpriv->ops = &mv_soc_65n_ops;
+ else
+ hpriv->ops = &mv_soc_ops;
hp_flags |= MV_HP_FLAG_SOC | MV_HP_GEN_IIE |
MV_HP_ERRATA_60X1C0;
break;
@@ -3762,7 +3826,8 @@ static int mv_init_host(struct ata_host *host, unsigned int board_idx)
n_hc = mv_get_hc_count(host->ports[0]->flags);

for (port = 0; port < host->n_ports; port++)
- hpriv->ops->read_preamp(hpriv, port, mmio);
+ if (hpriv->ops->read_preamp)
+ hpriv->ops->read_preamp(hpriv, port, mmio);

rc = hpriv->ops->reset_hc(hpriv, mmio, n_hc); if (rc)

author Tested-by
#5 lore raw

Mark, can this patch go in now? It's needed to make SATA operational on my Kirkwood-based QNAP TS-219 with a Maxtor drive.

Subject: [PATCH] sata_mv: use new sata phy register settings for new devices
From: Saeed Bishara <[email protected]>

Marvell's new SoC (65 nano) needs different settings for its SATA PHY registers.

drivers/ata/sata_mv.c +67 -2
@@ -293,6 +293,10 @@ enum {
 	FISCFG_WAIT_DEV_ERR	= (1 << 8),	/* wait for host on DevErr */
 	FISCFG_SINGLE_SYNC	= (1 << 16),	/* SYNC on DMA activation */
 
+	PHY_MODE9_GEN2		= 0x398,
+	PHY_MODE9_GEN1		= 0x39c,
+	PHYCFG_OFS		= 0x3a0,	/* only in 65n devices */
+
 	MV5_PHY_MODE		= 0x74,
 	MV5_LTMODE		= 0x30,
 	MV5_PHY_CTL		= 0x0C,
@@ -609,6 +613,8 @@ static int mv_soc_reset_hc(struct mv_host_priv *hpriv,
 static void mv_soc_reset_flash(struct mv_host_priv *hpriv,
 				      void __iomem *mmio);
 static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio);
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port);
 static void mv_reset_pci_bus(struct ata_host *host, void __iomem *mmio);
 static void mv_reset_channel(struct mv_host_priv *hpriv, void __iomem *mmio,
 			     unsigned int port_no);
@@ -807,6 +813,14 @@ static const struct mv_hw_ops mv_soc_ops = {
 	.reset_bus		= mv_soc_reset_bus,
 };
 
+static const struct mv_hw_ops mv_soc_65n_ops = {
+	.phy_errata		= mv_soc_65n_phy_errata,
+	.enable_leds		= mv_soc_enable_leds,
+	.reset_hc		= mv_soc_reset_hc,
+	.reset_flash		= mv_soc_reset_flash,
+	.reset_bus		= mv_soc_reset_bus,
+};
+
 /*
  * Functions
  */
@@ -3397,6 +3411,53 @@ static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio)
 	return;
 }
 
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port)
+{
+	void __iomem *port_mmio = mv_port_base(mmio, port);
+	u32	reg;
+
+	reg = readl(port_mmio + PHY_MODE3);
+	reg &= ~(0x3 << 27);	/* SELMUPF (bits 28:27) to 1 */
+	reg |= (0x1 << 27);
+	reg &= ~(0x3 << 29);	/* SELMUPI (bits 30:29) to 1 */
+	reg |= (0x1 << 29);
+	writel(reg, port_mmio + PHY_MODE3);
+
+	reg = readl(port_mmio + PHY_MODE4);
+	reg &= ~0x1;	/* SATU_OD8 (bit 0) to 0, reserved bit 16 must be set */
+	reg |= (0x1 << 16);
+	writel(reg, port_mmio + PHY_MODE4);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN2);
+	reg &= ~0xf;	/* TXAMP[3:0] (bits 3:0) to 8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* TXAMP[4] (bit 14) to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN2);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN1);
+	reg &= ~0xf;	/* TXAMP[3:0] (bits 3:0) to 8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* TXAMP[4] (bit 14) to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN1);
+}
+
+/**
+ *	soc_is_65 - check if the soc is 65 nano device
+ *
+ *	Detect the type of the SoC, this is done by reading the PHYCFG_OFS
+ *	register, this register should contain non-zero value and it exists only
+ *	in the 65 nano devices, when reading it from older devices we get 0.
+ */
+static bool soc_is_65n(struct mv_host_priv *hpriv)
+{
+	void __iomem *port0_mmio = mv_port_base(hpriv->base, 0);
+
+	if (readl(port0_mmio + PHYCFG_OFS))
+		return true;
+	return false;
+}
+
 static void mv_setup_ifcfg(void __iomem *port_mmio, int want_gen2i)
 {
 	u32 ifcfg = readl(port_mmio + SATA_IFCFG);
@@ -3737,7 +3798,10 @@ static int mv_chip_id(struct ata_host *host, unsigned int board_idx)
 		}
 		break;
 	case chip_soc:
-		hpriv->ops = &mv_soc_ops;
+		if (soc_is_65n(hpriv))
+			hpriv->ops = &mv_soc_65n_ops;
+		else
+			hpriv->ops = &mv_soc_ops;
 		hp_flags |= MV_HP_FLAG_SOC | MV_HP_GEN_IIE |
 			MV_HP_ERRATA_60X1C0;
 		break;
@@ -3800,7 +3864,8 @@ static int mv_init_host(struct ata_host *host, unsigned int board_idx)
 	n_hc = mv_get_hc_count(host->ports[0]->flags);
 
 	for (port = 0; port < host->n_ports; port++)
-		hpriv->ops->read_preamp(hpriv, port, mmio);
+		if (hpriv->ops->read_preamp)
+			hpriv->ops->read_preamp(hpriv, port, mmio);
 
 	rc = hpriv->ops->reset_hc(hpriv, mmio, n_hc);
 	if (rc)
Signature · 2 lines
Martin Michlmayr
http://www.cyrius.com/
#6 lore raw
Quoted 2 lines · Martin Michlmayr

Mark, can this patch go in now? It's needed to make SATA operational on my Kirkwood-based QNAP TS-219 with a Maxtor drive.

..

No comment from me on this one. I don't know if this patch has been tested on more than a single device type or not. Eg. does it break CF or ATAPI ?

It *looks* harmless on the non-SOC chips though, so even if it's not perfect it shouldn't do much harm.

Jeff: pick it up if you feel comfortable with the submitter's track record.

Cheers
-ml

..

Quoted 123 lines

Subject: [PATCH] sata_mv: use new sata phy register settings for new devices
From: Saeed Bishara <[email protected]>

Marvell's new SoC (65 nano) needs different settings for its SATA PHY registers.

drivers/ata/sata_mv.c +67 -2
@@ -293,6 +293,10 @@ enum {
 	FISCFG_WAIT_DEV_ERR	= (1 << 8),	/* wait for host on DevErr */
 	FISCFG_SINGLE_SYNC	= (1 << 16),	/* SYNC on DMA activation */
 
+	PHY_MODE9_GEN2		= 0x398,
+	PHY_MODE9_GEN1		= 0x39c,
+	PHYCFG_OFS		= 0x3a0,	/* only in 65n devices */
+
 	MV5_PHY_MODE		= 0x74,
 	MV5_LTMODE		= 0x30,
 	MV5_PHY_CTL		= 0x0C,
@@ -609,6 +613,8 @@ static int mv_soc_reset_hc(struct mv_host_priv *hpriv,
 static void mv_soc_reset_flash(struct mv_host_priv *hpriv,
 				      void __iomem *mmio);
 static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio);
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port);
 static void mv_reset_pci_bus(struct ata_host *host, void __iomem *mmio);
 static void mv_reset_channel(struct mv_host_priv *hpriv, void __iomem *mmio,
 			     unsigned int port_no);
@@ -807,6 +813,14 @@ static const struct mv_hw_ops mv_soc_ops = {
 	.reset_bus		= mv_soc_reset_bus,
 };
 
+static const struct mv_hw_ops mv_soc_65n_ops = {
+	.phy_errata		= mv_soc_65n_phy_errata,
+	.enable_leds		= mv_soc_enable_leds,
+	.reset_hc		= mv_soc_reset_hc,
+	.reset_flash		= mv_soc_reset_flash,
+	.reset_bus		= mv_soc_reset_bus,
+};
+
 /*
  * Functions
  */
@@ -3397,6 +3411,53 @@ static void mv_soc_reset_bus(struct ata_host *host, void __iomem *mmio)
 	return;
 }
 
+static void mv_soc_65n_phy_errata(struct mv_host_priv *hpriv,
+				  void __iomem *mmio, unsigned int port)
+{
+	void __iomem *port_mmio = mv_port_base(mmio, port);
+	u32	reg;
+
+	reg = readl(port_mmio + PHY_MODE3);
+	reg &= ~(0x3 << 27);	/* SELMUPF (bits 28:27) to 1 */
+	reg |= (0x1 << 27);
+	reg &= ~(0x3 << 29);	/* SELMUPI (bits 30:29) to 1 */
+	reg |= (0x1 << 29);
+	writel(reg, port_mmio + PHY_MODE3);
+
+	reg = readl(port_mmio + PHY_MODE4);
+	reg &= ~0x1;	/* SATU_OD8 (bit 0) to 0, reserved bit 16 must be set */
+	reg |= (0x1 << 16);
+	writel(reg, port_mmio + PHY_MODE4);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN2);
+	reg &= ~0xf;	/* TXAMP[3:0] (bits 3:0) to 8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* TXAMP[4] (bit 14) to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN2);
+
+	reg = readl(port_mmio + PHY_MODE9_GEN1);
+	reg &= ~0xf;	/* TXAMP[3:0] (bits 3:0) to 8 */
+	reg |= 0x8;
+	reg &= ~(0x1 << 14);	/* TXAMP[4] (bit 14) to 0 */
+	writel(reg, port_mmio + PHY_MODE9_GEN1);
+}
+
+/**
+ *	soc_is_65 - check if the soc is 65 nano device
+ *
+ *	Detect the type of the SoC, this is done by reading the PHYCFG_OFS
+ *	register, this register should contain non-zero value and it exists only
+ *	in the 65 nano devices, when reading it from older devices we get 0.
+ */
+static bool soc_is_65n(struct mv_host_priv *hpriv)
+{
+	void __iomem *port0_mmio = mv_port_base(hpriv->base, 0);
+
+	if (readl(port0_mmio + PHYCFG_OFS))
+		return true;
+	return false;
+}
+
 static void mv_setup_ifcfg(void __iomem *port_mmio, int want_gen2i)
 {
 	u32 ifcfg = readl(port_mmio + SATA_IFCFG);
@@ -3737,7 +3798,10 @@ static int mv_chip_id(struct ata_host *host, unsigned int board_idx)
 		}
 		break;
 	case chip_soc:
-		hpriv->ops = &mv_soc_ops;
+		if (soc_is_65n(hpriv))
+			hpriv->ops = &mv_soc_65n_ops;
+		else
+			hpriv->ops = &mv_soc_ops;
 		hp_flags |= MV_HP_FLAG_SOC | MV_HP_GEN_IIE |
 			MV_HP_ERRATA_60X1C0;
 		break;
@@ -3800,7 +3864,8 @@ static int mv_init_host(struct ata_host *host, unsigned int board_idx)
 	n_hc = mv_get_hc_count(host->ports[0]->flags);
 
 	for (port = 0; port < host->n_ports; port++)
-		hpriv->ops->read_preamp(hpriv, port, mmio);
+		if (hpriv->ops->read_preamp)
+			hpriv->ops->read_preamp(hpriv, port, mmio);
 
 	rc = hpriv->ops->reset_hc(hpriv, mmio, n_hc);
 	if (rc)
#7 lore raw
In reply to Martin Michlmayr
Quoted 4 lines · Martin Michlmayr

Mark, can this patch go in now? It's needed to make SATA operational on my Kirkwood-based QNAP TS-219 with a Maxtor drive.

Quoted 8 lines

Subject: [PATCH] sata_mv: use new sata phy register settings for new devices
From: Saeed Bishara <[email protected]>

Marvell's new SoC (65 nano) needs different settings for its SATA PHY registers.

applied -- though I still await answers to maintainer Mark Lord's questions: was this tested with ATAPI and other features?

(rhetorical question... if not... test it!)