[PATCH v2 2/2] phy: qcom: edp: Add X1E80100 eDP and DP PHYs
Johan Hovold
johan at kernel.org
Mon Dec 4 07:29:53 PST 2023
On Mon, Dec 04, 2023 at 05:01:37PM +0200, Abel Vesa wrote:
> The Qualcomm X1E80100 platform has a number of eDP and DP PHY instances,
> add support for these.
>
> Co-developed-by: Abhinav Kumar <quic_abhinavk at quicinc.com>
> Signed-off-by: Abhinav Kumar <quic_abhinavk at quicinc.com>
> Signed-off-by: Abel Vesa <abel.vesa at linaro.org>
> ---
> /* Turn on BIAS current for PHY/PLL */
> - writel(0x17, edp->pll + QSERDES_V4_COM_BIAS_EN_CLKBUFLR_EN);
> + if (cfg->qserdes_version == QSERDES_V6)
> + writel(0x1f, edp->pll + QSERDES_V6_COM_PLL_BIAS_EN_CLK_BUFLR_EN);
> + else
> + writel(0x17, edp->pll + QSERDES_V4_COM_BIAS_EN_CLKBUFLR_EN);
Sprinkling version conditionals like this throughout the driver look a
bit sloppy and makes the code harder to read (and maintain long term).
Not as easy to clean up when you're using magic numbers instead of
proper value defines, but perhaps you can just fix that up as well as
you should access to the documentation.
> writel(DP_PHY_PD_CTL_PSR_PWRDN, edp->edp + DP_PHY_PD_CTL);
> msleep(20);
> @@ -261,7 +290,10 @@ static int qcom_edp_set_voltages(struct qcom_edp *edp, const struct phy_configur
> if (swing == 0xff || emph == 0xff)
> return -EINVAL;
>
> - ldo_config = (cfg && cfg->is_dp) ? 0x1 : 0x0;
> + if (cfg->qserdes_version == QSERDES_V6)
> + ldo_config = 0x91;
> + else
> + ldo_config = (cfg && cfg->is_dp) ? 0x1 : 0x0;
Similar here.
> writel(ldo_config, edp->tx0 + TXn_LDO_CONFIG);
> writel(swing, edp->tx0 + TXn_TX_DRV_LVL);
> @@ -288,7 +320,7 @@ static int qcom_edp_phy_configure(struct phy *phy, union phy_configure_opts *opt
> return ret;
> }
>
> -static int qcom_edp_configure_ssc(const struct qcom_edp *edp)
> +static int qcom_edp_configure_ssc_v4(const struct qcom_edp *edp)
> {
> const struct phy_configure_opts_dp *dp_opts = &edp->dp_opts;
> u32 step1;
> @@ -349,6 +415,7 @@ static int qcom_edp_configure_pll(const struct qcom_edp *edp)
> div_frac_start3_mode0 = 0x07;
> lock_cmp1_mode0 = 0x0f;
> lock_cmp2_mode0 = 0x0e;
> +
> break;
>
> case 5400:
> @@ -358,6 +425,7 @@ static int qcom_edp_configure_pll(const struct qcom_edp *edp)
> div_frac_start3_mode0 = 0x0a;
> lock_cmp1_mode0 = 0x1f;
> lock_cmp2_mode0 = 0x1c;
> +
> break;
>
> case 8100:
> @@ -367,6 +435,7 @@ static int qcom_edp_configure_pll(const struct qcom_edp *edp)
> div_frac_start3_mode0 = 0x07;
> lock_cmp1_mode0 = 0x2f;
> lock_cmp2_mode0 = 0x2a;
> +
> break;
>
> default:
This looks like unrelated changes that do not belong in this patch.
> @@ -408,6 +477,103 @@ static int qcom_edp_configure_pll(const struct qcom_edp *edp)
> return 0;
> }
>
> +static int qcom_edp_configure_pll_v6(const struct qcom_edp *edp)
> +{
> + const struct phy_configure_opts_dp *dp_opts = &edp->dp_opts;
> + u32 div_frac_start2_mode0;
> + u32 div_frac_start3_mode0;
> + u32 dec_start_mode0;
> + u32 lock_cmp1_mode0;
> + u32 lock_cmp2_mode0;
> + u32 code1_mode0;
> + u32 code2_mode0;
> + u32 hsclk_sel;
> +
> + switch (dp_opts->link_rate) {
> + case 1620:
> + hsclk_sel = 0x5;
> + dec_start_mode0 = 0x34;
> + div_frac_start2_mode0 = 0xc0;
> + div_frac_start3_mode0 = 0x0b;
> + lock_cmp1_mode0 = 0x37;
> + lock_cmp2_mode0 = 0x04;
> + code1_mode0 = 0x71;
> + code2_mode0 = 0x0C;
> + break;
> +
Drop newline before case throughout? Doesn't seem to help with
readability. But I see now that some other functions do this already.
> + case 2700:
> + hsclk_sel = 0x3;
> + dec_start_mode0 = 0x34;
> + div_frac_start2_mode0 = 0xc0;
> + div_frac_start3_mode0 = 0x0b;
> + lock_cmp1_mode0 = 0x07;
> + lock_cmp2_mode0 = 0x07;
> + code1_mode0 = 0x71;
> + code2_mode0 = 0x0C;
> + break;
> +
> + case 5400:
> static int qcom_edp_set_vco_div(const struct qcom_edp *edp, unsigned long *pixel_freq)
> {
> const struct phy_configure_opts_dp *dp_opts = &edp->dp_opts;
> @@ -462,13 +628,22 @@ static int qcom_edp_phy_power_on(struct phy *phy)
> edp->edp + DP_PHY_PD_CTL);
> writel(0xfc, edp->edp + DP_PHY_MODE);
>
> - timeout = readl_poll_timeout(edp->pll + QSERDES_V4_COM_CMN_STATUS,
> - val, val & BIT(7), 5, 200);
> + if (cfg->qserdes_version == QSERDES_V6) {
> + timeout = readl_poll_timeout(edp->pll + QSERDES_V6_COM_CMN_STATUS,
> + val, val & BIT(7), 5, 200);
> + } else {
> + timeout = readl_poll_timeout(edp->pll + QSERDES_V4_COM_CMN_STATUS,
> + val, val & BIT(7), 5, 200);
> + }
This too looks a bit sloppy. Should you not just abstract the status
register offset?
> +
> if (timeout)
> return timeout;
>
>
> - ldo_config = (cfg && cfg->is_dp) ? 0x1 : 0x0;
> + if (cfg->qserdes_version == QSERDES_V6)
> + ldo_config = 0x91;
> + else
> + ldo_config = (cfg && cfg->is_dp) ? 0x1 : 0x0;
Same construct and comment as above.
>
> writel(ldo_config, edp->tx0 + TXn_LDO_CONFIG);
> writel(ldo_config, edp->tx1 + TXn_LDO_CONFIG);
> @@ -512,10 +695,18 @@ static int qcom_edp_phy_power_on(struct phy *phy)
> writel(0x01, edp->edp + DP_PHY_CFG);
> writel(0x09, edp->edp + DP_PHY_CFG);
>
> - writel(0x20, edp->pll + QSERDES_V4_COM_RESETSM_CNTRL);
> + if (cfg->qserdes_version == QSERDES_V6) {
> + writel(0x20, edp->pll + QSERDES_V6_COM_RESETSM_CNTRL);
> +
> + timeout = readl_poll_timeout(edp->pll + QSERDES_V6_COM_C_READY_STATUS,
> + val, val & BIT(0), 500, 10000);
> + } else {
> + writel(0x20, edp->pll + QSERDES_V4_COM_RESETSM_CNTRL);
> +
> + timeout = readl_poll_timeout(edp->pll + QSERDES_V4_COM_C_READY_STATUS,
> + val, val & BIT(0), 500, 10000);
> + }
Similar here.
> - timeout = readl_poll_timeout(edp->pll + QSERDES_V4_COM_C_READY_STATUS,
> - val, val & BIT(0), 500, 10000);
> if (timeout)
> return timeout;
>
> @@ -744,7 +935,6 @@ static int qcom_edp_clks_register(struct qcom_edp *edp, struct device_node *np)
> data = devm_kzalloc(edp->dev, struct_size(data, hws, 2), GFP_KERNEL);
> if (!data)
> return -ENOMEM;
> - data->num = 2;
>
> snprintf(name, sizeof(name), "%s::link_clk", dev_name(edp->dev));
> init.ops = &qcom_edp_dp_link_clk_ops;
> @@ -764,6 +954,7 @@ static int qcom_edp_clks_register(struct qcom_edp *edp, struct device_node *np)
>
> data->hws[0] = &edp->dp_link_hw;
> data->hws[1] = &edp->dp_pixel_hw;
> + data->num = 2;
Another seemingly unrelated change.
> return devm_of_clk_add_hw_provider(edp->dev, of_clk_hw_onecell_get, data);
> }
> @@ -806,6 +997,7 @@ static int qcom_edp_phy_probe(struct platform_device *pdev)
>
> edp->supplies[0].supply = "vdda-phy";
> edp->supplies[1].supply = "vdda-pll";
> + edp->clks[2].id = "tcsr";
The *clocks* will already have been requested above so this is at best a
noop.
> ret = devm_regulator_bulk_get(dev, ARRAY_SIZE(edp->supplies), edp->supplies);
> if (ret)
> return ret;
Johan
More information about the linux-phy
mailing list