[3/3] phy: rockchip-typec: Do the calibration more correctly

Commit Message

Doug Anderson Sept. 6, 2017, 8:52 p.m. UTC

Calculate the calibration code as per the docs.  The docs talk about
reading and averaging the pullup and pulldown calibration codes.  They
also talk about adding in some adjustment codes.  Let's do what the
docs say.

In practice this doesn't seem to matter a whole lot.  On a device I
tested the pullup and pulldown codes were nearly the same (0x23 and
0x24) and the adjustment codes were 0.

Reviewed-by: Chris Zhong <[e-mail]>
Signed-off-by: Douglas Anderson <[e-mail]>
---

 drivers/phy/rockchip/phy-rockchip-typec.c | 28 +++++++++++++++++++---------
 1 file changed, 19 insertions(+), 9 deletions(-)

Comments

Doug Anderson Sept. 6, 2017, 9:24 p.m. UTC | #1

Hi,

On Wed, Sep 6, 2017 at 1:52 PM, Douglas Anderson <[e-mail]> wrote:
> Calculate the calibration code as per the docs.  The docs talk about
> reading and averaging the pullup and pulldown calibration codes.  They
> also talk about adding in some adjustment codes.  Let's do what the
> docs say.
>
> In practice this doesn't seem to matter a whole lot.  On a device I
> tested the pullup and pulldown codes were nearly the same (0x23 and
> 0x24) and the adjustment codes were 0.
>
> Reviewed-by: Chris Zhong <[e-mail]>
> Signed-off-by: Douglas Anderson <[e-mail]>
> ---
>
>  drivers/phy/rockchip/phy-rockchip-typec.c | 28 +++++++++++++++++++---------
>  1 file changed, 19 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/phy/rockchip/phy-rockchip-typec.c b/drivers/phy/rockchip/phy-rockchip-typec.c
> index b8696a7c45fb..e59983a82dcc 100644
> --- a/drivers/phy/rockchip/phy-rockchip-typec.c
> +++ b/drivers/phy/rockchip/phy-rockchip-typec.c
> @@ -194,6 +194,7 @@
>  /* Use this for "n" in macros like "_MULT_XXX" to target the aux channel */
>  #define AUX_CH_LANE                    8
>
> +

Not sure how I missed that this was adding a blank line here.  :(  If
there are no other issues I'm happy to re-post or happy if this blank
line is removed when the patch is applied.  Please let me know which
you'd prefer.

-Doug

Patch

diff --git a/drivers/phy/rockchip/phy-rockchip-typec.c b/drivers/phy/rockchip/phy-rockchip-typec.c
index b8696a7c45fb..e59983a82dcc 100644
--- a/drivers/phy/rockchip/phy-rockchip-typec.c
+++ b/drivers/phy/rockchip/phy-rockchip-typec.c
@@ -194,6 +194,7 @@ 
 /* Use this for "n" in macros like "_MULT_XXX" to target the aux channel */
 #define AUX_CH_LANE			8
 
+
 #define TX_ANA_CTRL_REG_1		(0x5020 << 2)
 
 #define TXDA_DP_AUX_EN			BIT(15)
@@ -560,24 +561,33 @@  static void tcphy_dp_aux_calibration(struct rockchip_typec_phy *tcphy)
 	u16 val;
 	u16 tx_ana_ctrl_reg_1;
 	u16 tx_ana_ctrl_reg_2;
-	s32 pu_calib_code;
-
-	/* disable txda_cal_latch_en for rewrite the calibration values */
-	tx_ana_ctrl_reg_1 = readl(tcphy->base + TX_ANA_CTRL_REG_1);
-	tx_ana_ctrl_reg_1 &= ~TXDA_CAL_LATCH_EN;
-	writel(tx_ana_ctrl_reg_1, tcphy->base + TX_ANA_CTRL_REG_1);
+	s32 pu_calib_code, pd_calib_code;
+	s32 pu_adj, pd_adj;
+	u16 calib;
 
 	/*
-	 * read a resistor calibration code from CMN_TXPUCAL_CTRL[5:0] and
-	 * write it to TX_DIG_CTRL_REG_2[5:0].
+	 * Calculate calibration code as per docs: use an average of the
+	 * pull down and pull up.  Then add in adjustments.
 	 */
 	val = readl(tcphy->base + CMN_TXPUCAL_CTRL);
 	pu_calib_code = CMN_CALIB_CODE_POS(val);
+	val = readl(tcphy->base + CMN_TXPDCAL_CTRL);
+	pd_calib_code = CMN_CALIB_CODE_POS(val);
+	val = readl(tcphy->base + CMN_TXPU_ADJ_CTRL);
+	pu_adj = CMN_CALIB_CODE(val);
+	val = readl(tcphy->base + CMN_TXPD_ADJ_CTRL);
+	pd_adj = CMN_CALIB_CODE(val);
+	calib = (pu_calib_code + pd_calib_code) / 2 + pu_adj + pd_adj;
+
+	/* disable txda_cal_latch_en for rewrite the calibration values */
+	tx_ana_ctrl_reg_1 = readl(tcphy->base + TX_ANA_CTRL_REG_1);
+	tx_ana_ctrl_reg_1 &= ~TXDA_CAL_LATCH_EN;
+	writel(tx_ana_ctrl_reg_1, tcphy->base + TX_ANA_CTRL_REG_1);
 
 	/* write the calibration, then delay 10 ms as sample in docs */
 	val = readl(tcphy->base + TX_DIG_CTRL_REG_2);
 	val &= ~(TX_RESCAL_CODE_MASK << TX_RESCAL_CODE_OFFSET);
-	val |= pu_calib_code << TX_RESCAL_CODE_OFFSET;
+	val |= calib << TX_RESCAL_CODE_OFFSET;
 	writel(val, tcphy->base + TX_DIG_CTRL_REG_2);
 	usleep_range(10000, 10050);