drm/msm/dsi: Add DSI PLL for 28nm 8960 PHY
authorArchit Taneja <architt@codeaurora.org>
Wed, 14 Oct 2015 06:33:44 +0000 (12:03 +0530)
committerRob Clark <robdclark@gmail.com>
Mon, 14 Dec 2015 15:40:12 +0000 (10:40 -0500)
Add DSI PLL common clock framework clocks for 8960 PHY.

The PLL here is different from the ones found in B family msm chips. As
before, the DSI provides two clocks to the outside world. dsixpll and
dsixpllbyte (x = 1, 2). dsixpll is a regular clock divider, but
dsixpllbyte is modelled as a custom clock divider.

dsixpllbyte is the starting point of the PLL configuration. It is the
one that sets up the VCO clock rate. We need the VCO clock rate in the
form: F * byteclk, where F is a multiplication factor that varies on
the byte clock the DSI driver is trying to set. We use the custom
clk_ops for dsixpllbyte to ensure that the parent (VCO) is set at this
rate.

An additional divider (POSTDIV1) generates the bitclk. Since bit clock
can be derived from byteclock, we calculate it internally, and don't
expose it as a clock.

Cc: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
drivers/gpu/drm/msm/Makefile
drivers/gpu/drm/msm/dsi/pll/dsi_pll.c
drivers/gpu/drm/msm/dsi/pll/dsi_pll.h
drivers/gpu/drm/msm/dsi/pll/dsi_pll_28nm_8960.c [new file with mode: 0644]

index 1888fd34d51fe48258cd1505e74da8ee14044401..065ad4138799927f35904e4f3d4cae8560a8688a 100644 (file)
@@ -68,6 +68,7 @@ msm-$(CONFIG_DRM_MSM_DSI_28NM_8960_PHY) += dsi/phy/dsi_phy_28nm_8960.o
 ifeq ($(CONFIG_DRM_MSM_DSI_PLL),y)
 msm-y += dsi/pll/dsi_pll.o
 msm-$(CONFIG_DRM_MSM_DSI_28NM_PHY) += dsi/pll/dsi_pll_28nm.o
+msm-$(CONFIG_DRM_MSM_DSI_28NM_8960_PHY) += dsi/pll/dsi_pll_28nm_8960.o
 endif
 
 obj-$(CONFIG_DRM_MSM)  += msm.o
index 5104fc9f9a5398e71d4869264f49f3a74c801b84..5cd438f91afea11c8dc9f756d8028741b0fde73f 100644 (file)
@@ -151,6 +151,9 @@ struct msm_dsi_pll *msm_dsi_pll_init(struct platform_device *pdev,
        case MSM_DSI_PHY_28NM_LP:
                pll = msm_dsi_pll_28nm_init(pdev, type, id);
                break;
+       case MSM_DSI_PHY_28NM_8960:
+               pll = msm_dsi_pll_28nm_8960_init(pdev, id);
+               break;
        default:
                pll = ERR_PTR(-ENXIO);
                break;
index 063caa2c5740fe9462c6646f171d4f9422831eb4..80b6038334a601342d6637a50a6ecbd8140de008 100644 (file)
@@ -93,6 +93,16 @@ static inline struct msm_dsi_pll *msm_dsi_pll_28nm_init(
        return ERR_PTR(-ENODEV);
 }
 #endif
+#ifdef CONFIG_DRM_MSM_DSI_28NM_8960_PHY
+struct msm_dsi_pll *msm_dsi_pll_28nm_8960_init(struct platform_device *pdev,
+                                              int id);
+#else
+struct msm_dsi_pll *msm_dsi_pll_28nm_8960_init(struct platform_device *pdev,
+                                              int id)
+{
+       return ERR_PTR(-ENODEV);
+}
+#endif
 
 #endif /* __DSI_PLL_H__ */
 
diff --git a/drivers/gpu/drm/msm/dsi/pll/dsi_pll_28nm_8960.c b/drivers/gpu/drm/msm/dsi/pll/dsi_pll_28nm_8960.c
new file mode 100644 (file)
index 0000000..38c90e1
--- /dev/null
@@ -0,0 +1,533 @@
+/*
+ * Copyright (c) 2012-2015, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/clk-provider.h>
+
+#include "dsi_pll.h"
+#include "dsi.xml.h"
+
+/*
+ * DSI PLL 28nm (8960/A family) - clock diagram (eg: DSI1):
+ *
+ *
+ *                        +------+
+ *  dsi1vco_clk ----o-----| DIV1 |---dsi1pllbit (not exposed as clock)
+ *  F * byte_clk    |     +------+
+ *                  | bit clock divider (F / 8)
+ *                  |
+ *                  |     +------+
+ *                  o-----| DIV2 |---dsi0pllbyte---o---> To byte RCG
+ *                  |     +------+                 | (sets parent rate)
+ *                  | byte clock divider (F)       |
+ *                  |                              |
+ *                  |                              o---> To esc RCG
+ *                  |                                (doesn't set parent rate)
+ *                  |
+ *                  |     +------+
+ *                  o-----| DIV3 |----dsi0pll------o---> To dsi RCG
+ *                        +------+                 | (sets parent rate)
+ *                  dsi clock divider (F * magic)  |
+ *                                                 |
+ *                                                 o---> To pixel rcg
+ *                                                  (doesn't set parent rate)
+ */
+
+#define POLL_MAX_READS         8000
+#define POLL_TIMEOUT_US                1
+
+#define NUM_PROVIDED_CLKS      2
+
+#define VCO_REF_CLK_RATE       27000000
+#define VCO_MIN_RATE           600000000
+#define VCO_MAX_RATE           1200000000
+
+#define DSI_BYTE_PLL_CLK       0
+#define DSI_PIXEL_PLL_CLK      1
+
+#define VCO_PREF_DIV_RATIO     27
+
+struct pll_28nm_cached_state {
+       unsigned long vco_rate;
+       u8 postdiv3;
+       u8 postdiv2;
+       u8 postdiv1;
+};
+
+struct clk_bytediv {
+       struct clk_hw hw;
+       void __iomem *reg;
+};
+
+struct dsi_pll_28nm {
+       struct msm_dsi_pll base;
+
+       int id;
+       struct platform_device *pdev;
+       void __iomem *mmio;
+
+       /* custom byte clock divider */
+       struct clk_bytediv *bytediv;
+
+       /* private clocks: */
+       struct clk *clks[NUM_DSI_CLOCKS_MAX];
+       u32 num_clks;
+
+       /* clock-provider: */
+       struct clk *provided_clks[NUM_PROVIDED_CLKS];
+       struct clk_onecell_data clk_data;
+
+       struct pll_28nm_cached_state cached_state;
+};
+
+#define to_pll_28nm(x) container_of(x, struct dsi_pll_28nm, base)
+
+static bool pll_28nm_poll_for_ready(struct dsi_pll_28nm *pll_28nm,
+                                   int nb_tries, int timeout_us)
+{
+       bool pll_locked = false;
+       u32 val;
+
+       while (nb_tries--) {
+               val = pll_read(pll_28nm->mmio + REG_DSI_28nm_8960_PHY_PLL_RDY);
+               pll_locked = !!(val & DSI_28nm_8960_PHY_PLL_RDY_PLL_RDY);
+
+               if (pll_locked)
+                       break;
+
+               udelay(timeout_us);
+       }
+       DBG("DSI PLL is %slocked", pll_locked ? "" : "*not* ");
+
+       return pll_locked;
+}
+
+/*
+ * Clock Callbacks
+ */
+static int dsi_pll_28nm_clk_set_rate(struct clk_hw *hw, unsigned long rate,
+                                    unsigned long parent_rate)
+{
+       struct msm_dsi_pll *pll = hw_clk_to_pll(hw);
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+       void __iomem *base = pll_28nm->mmio;
+       u32 val, temp, fb_divider;
+
+       DBG("rate=%lu, parent's=%lu", rate, parent_rate);
+
+       temp = rate / 10;
+       val = VCO_REF_CLK_RATE / 10;
+       fb_divider = (temp * VCO_PREF_DIV_RATIO) / val;
+       fb_divider = fb_divider / 2 - 1;
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_1,
+                       fb_divider & 0xff);
+
+       val = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_2);
+
+       val |= (fb_divider >> 8) & 0x07;
+
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_2,
+                       val);
+
+       val = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_3);
+
+       val |= (VCO_PREF_DIV_RATIO - 1) & 0x3f;
+
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_3,
+                       val);
+
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_6,
+                       0xf);
+
+       val = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8);
+       val |= 0x7 << 4;
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8,
+                       val);
+
+       return 0;
+}
+
+static int dsi_pll_28nm_clk_is_enabled(struct clk_hw *hw)
+{
+       struct msm_dsi_pll *pll = hw_clk_to_pll(hw);
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+
+       return pll_28nm_poll_for_ready(pll_28nm, POLL_MAX_READS,
+                                       POLL_TIMEOUT_US);
+}
+
+static unsigned long dsi_pll_28nm_clk_recalc_rate(struct clk_hw *hw,
+                                                 unsigned long parent_rate)
+{
+       struct msm_dsi_pll *pll = hw_clk_to_pll(hw);
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+       void __iomem *base = pll_28nm->mmio;
+       unsigned long vco_rate;
+       u32 status, fb_divider, temp, ref_divider;
+
+       VERB("parent_rate=%lu", parent_rate);
+
+       status = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_0);
+
+       if (status & DSI_28nm_8960_PHY_PLL_CTRL_0_ENABLE) {
+               fb_divider = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_1);
+               fb_divider &= 0xff;
+               temp = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_2) & 0x07;
+               fb_divider = (temp << 8) | fb_divider;
+               fb_divider += 1;
+
+               ref_divider = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_3);
+               ref_divider &= 0x3f;
+               ref_divider += 1;
+
+               /* multiply by 2 */
+               vco_rate = (parent_rate / ref_divider) * fb_divider * 2;
+       } else {
+               vco_rate = 0;
+       }
+
+       DBG("returning vco rate = %lu", vco_rate);
+
+       return vco_rate;
+}
+
+static const struct clk_ops clk_ops_dsi_pll_28nm_vco = {
+       .round_rate = msm_dsi_pll_helper_clk_round_rate,
+       .set_rate = dsi_pll_28nm_clk_set_rate,
+       .recalc_rate = dsi_pll_28nm_clk_recalc_rate,
+       .prepare = msm_dsi_pll_helper_clk_prepare,
+       .unprepare = msm_dsi_pll_helper_clk_unprepare,
+       .is_enabled = dsi_pll_28nm_clk_is_enabled,
+};
+
+/*
+ * Custom byte clock divier clk_ops
+ *
+ * This clock is the entry point to configuring the PLL. The user (dsi host)
+ * will set this clock's rate to the desired byte clock rate. The VCO lock
+ * frequency is a multiple of the byte clock rate. The multiplication factor
+ * (shown as F in the diagram above) is a function of the byte clock rate.
+ *
+ * This custom divider clock ensures that its parent (VCO) is set to the
+ * desired rate, and that the byte clock postdivider (POSTDIV2) is configured
+ * accordingly
+ */
+#define to_clk_bytediv(_hw) container_of(_hw, struct clk_bytediv, hw)
+
+static unsigned long clk_bytediv_recalc_rate(struct clk_hw *hw,
+               unsigned long parent_rate)
+{
+       struct clk_bytediv *bytediv = to_clk_bytediv(hw);
+       unsigned int div;
+
+       div = pll_read(bytediv->reg) & 0xff;
+
+       return parent_rate / (div + 1);
+}
+
+/* find multiplication factor(wrt byte clock) at which the VCO should be set */
+static unsigned int get_vco_mul_factor(unsigned long byte_clk_rate)
+{
+       unsigned long bit_mhz;
+
+       /* convert to bit clock in Mhz */
+       bit_mhz = (byte_clk_rate * 8) / 1000000;
+
+       if (bit_mhz < 125)
+               return 64;
+       else if (bit_mhz < 250)
+               return 32;
+       else if (bit_mhz < 600)
+               return 16;
+       else
+               return 8;
+}
+
+static long clk_bytediv_round_rate(struct clk_hw *hw, unsigned long rate,
+                                  unsigned long *prate)
+{
+       unsigned long best_parent;
+       unsigned int factor;
+
+       factor = get_vco_mul_factor(rate);
+
+       best_parent = rate * factor;
+       *prate = clk_hw_round_rate(clk_hw_get_parent(hw), best_parent);
+
+       return *prate / factor;
+}
+
+static int clk_bytediv_set_rate(struct clk_hw *hw, unsigned long rate,
+                               unsigned long parent_rate)
+{
+       struct clk_bytediv *bytediv = to_clk_bytediv(hw);
+       u32 val;
+       unsigned int factor;
+
+       factor = get_vco_mul_factor(rate);
+
+       val = pll_read(bytediv->reg);
+       val |= (factor - 1) & 0xff;
+       pll_write(bytediv->reg, val);
+
+       return 0;
+}
+
+/* Our special byte clock divider ops */
+static const struct clk_ops clk_bytediv_ops = {
+       .round_rate = clk_bytediv_round_rate,
+       .set_rate = clk_bytediv_set_rate,
+       .recalc_rate = clk_bytediv_recalc_rate,
+};
+
+/*
+ * PLL Callbacks
+ */
+static int dsi_pll_28nm_enable_seq(struct msm_dsi_pll *pll)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+       struct device *dev = &pll_28nm->pdev->dev;
+       void __iomem *base = pll_28nm->mmio;
+       bool locked;
+       unsigned int bit_div, byte_div;
+       int max_reads = 1000, timeout_us = 100;
+       u32 val;
+
+       DBG("id=%d", pll_28nm->id);
+
+       /*
+        * before enabling the PLL, configure the bit clock divider since we
+        * don't expose it as a clock to the outside world
+        * 1: read back the byte clock divider that should already be set
+        * 2: divide by 8 to get bit clock divider
+        * 3: write it to POSTDIV1
+        */
+       val = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_9);
+       byte_div = val + 1;
+       bit_div = byte_div / 8;
+
+       val = pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8);
+       val &= ~0xf;
+       val |= (bit_div - 1);
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8, val);
+
+       /* enable the PLL */
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_0,
+                       DSI_28nm_8960_PHY_PLL_CTRL_0_ENABLE);
+
+       locked = pll_28nm_poll_for_ready(pll_28nm, max_reads, timeout_us);
+
+       if (unlikely(!locked))
+               dev_err(dev, "DSI PLL lock failed\n");
+       else
+               DBG("DSI PLL lock success");
+
+       return locked ? 0 : -EINVAL;
+}
+
+static void dsi_pll_28nm_disable_seq(struct msm_dsi_pll *pll)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+
+       DBG("id=%d", pll_28nm->id);
+       pll_write(pll_28nm->mmio + REG_DSI_28nm_8960_PHY_PLL_CTRL_0, 0x00);
+}
+
+static void dsi_pll_28nm_save_state(struct msm_dsi_pll *pll)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+       struct pll_28nm_cached_state *cached_state = &pll_28nm->cached_state;
+       void __iomem *base = pll_28nm->mmio;
+
+       cached_state->postdiv3 =
+                       pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_10);
+       cached_state->postdiv2 =
+                       pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_9);
+       cached_state->postdiv1 =
+                       pll_read(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8);
+
+       cached_state->vco_rate = clk_hw_get_rate(&pll->clk_hw);
+}
+
+static int dsi_pll_28nm_restore_state(struct msm_dsi_pll *pll)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+       struct pll_28nm_cached_state *cached_state = &pll_28nm->cached_state;
+       void __iomem *base = pll_28nm->mmio;
+       int ret;
+
+       ret = dsi_pll_28nm_clk_set_rate(&pll->clk_hw,
+                                       cached_state->vco_rate, 0);
+       if (ret) {
+               dev_err(&pll_28nm->pdev->dev,
+                       "restore vco rate failed. ret=%d\n", ret);
+               return ret;
+       }
+
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_10,
+                       cached_state->postdiv3);
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_9,
+                       cached_state->postdiv2);
+       pll_write(base + REG_DSI_28nm_8960_PHY_PLL_CTRL_8,
+                       cached_state->postdiv1);
+
+       return 0;
+}
+
+static int dsi_pll_28nm_get_provider(struct msm_dsi_pll *pll,
+                               struct clk **byte_clk_provider,
+                               struct clk **pixel_clk_provider)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+
+       if (byte_clk_provider)
+               *byte_clk_provider = pll_28nm->provided_clks[DSI_BYTE_PLL_CLK];
+       if (pixel_clk_provider)
+               *pixel_clk_provider =
+                               pll_28nm->provided_clks[DSI_PIXEL_PLL_CLK];
+
+       return 0;
+}
+
+static void dsi_pll_28nm_destroy(struct msm_dsi_pll *pll)
+{
+       struct dsi_pll_28nm *pll_28nm = to_pll_28nm(pll);
+
+       msm_dsi_pll_helper_unregister_clks(pll_28nm->pdev,
+                                       pll_28nm->clks, pll_28nm->num_clks);
+}
+
+static int pll_28nm_register(struct dsi_pll_28nm *pll_28nm)
+{
+       char *clk_name, *parent_name, *vco_name;
+       struct clk_init_data vco_init = {
+               .parent_names = (const char *[]){ "pxo" },
+               .num_parents = 1,
+               .ops = &clk_ops_dsi_pll_28nm_vco,
+       };
+       struct device *dev = &pll_28nm->pdev->dev;
+       struct clk **clks = pll_28nm->clks;
+       struct clk **provided_clks = pll_28nm->provided_clks;
+       struct clk_bytediv *bytediv;
+       struct clk_init_data bytediv_init = { };
+       int ret, num = 0;
+
+       DBG("%d", pll_28nm->id);
+
+       bytediv = devm_kzalloc(dev, sizeof(*bytediv), GFP_KERNEL);
+       if (!bytediv)
+               return -ENOMEM;
+
+       vco_name = devm_kzalloc(dev, 32, GFP_KERNEL);
+       if (!vco_name)
+               return -ENOMEM;
+
+       parent_name = devm_kzalloc(dev, 32, GFP_KERNEL);
+       if (!parent_name)
+               return -ENOMEM;
+
+       clk_name = devm_kzalloc(dev, 32, GFP_KERNEL);
+       if (!clk_name)
+               return -ENOMEM;
+
+       pll_28nm->bytediv = bytediv;
+
+       snprintf(vco_name, 32, "dsi%dvco_clk", pll_28nm->id);
+       vco_init.name = vco_name;
+
+       pll_28nm->base.clk_hw.init = &vco_init;
+
+       clks[num++] = clk_register(dev, &pll_28nm->base.clk_hw);
+
+       /* prepare and register bytediv */
+       bytediv->hw.init = &bytediv_init;
+       bytediv->reg = pll_28nm->mmio + REG_DSI_28nm_8960_PHY_PLL_CTRL_9;
+
+       snprintf(parent_name, 32, "dsi%dvco_clk", pll_28nm->id);
+       snprintf(clk_name, 32, "dsi%dpllbyte", pll_28nm->id);
+
+       bytediv_init.name = clk_name;
+       bytediv_init.ops = &clk_bytediv_ops;
+       bytediv_init.flags = CLK_SET_RATE_PARENT;
+       bytediv_init.parent_names = (const char * const *) &parent_name;
+       bytediv_init.num_parents = 1;
+
+       /* DIV2 */
+       clks[num++] = provided_clks[DSI_BYTE_PLL_CLK] =
+                       clk_register(dev, &bytediv->hw);
+
+       snprintf(clk_name, 32, "dsi%dpll", pll_28nm->id);
+       /* DIV3 */
+       clks[num++] = provided_clks[DSI_PIXEL_PLL_CLK] =
+                       clk_register_divider(dev, clk_name,
+                               parent_name, 0, pll_28nm->mmio +
+                               REG_DSI_28nm_8960_PHY_PLL_CTRL_10,
+                               0, 8, 0, NULL);
+
+       pll_28nm->num_clks = num;
+
+       pll_28nm->clk_data.clk_num = NUM_PROVIDED_CLKS;
+       pll_28nm->clk_data.clks = provided_clks;
+
+       ret = of_clk_add_provider(dev->of_node,
+                       of_clk_src_onecell_get, &pll_28nm->clk_data);
+       if (ret) {
+               dev_err(dev, "failed to register clk provider: %d\n", ret);
+               return ret;
+       }
+
+       return 0;
+}
+
+struct msm_dsi_pll *msm_dsi_pll_28nm_8960_init(struct platform_device *pdev,
+                                              int id)
+{
+       struct dsi_pll_28nm *pll_28nm;
+       struct msm_dsi_pll *pll;
+       int ret;
+
+       if (!pdev)
+               return ERR_PTR(-ENODEV);
+
+       pll_28nm = devm_kzalloc(&pdev->dev, sizeof(*pll_28nm), GFP_KERNEL);
+       if (!pll_28nm)
+               return ERR_PTR(-ENOMEM);
+
+       pll_28nm->pdev = pdev;
+       pll_28nm->id = id + 1;
+
+       pll_28nm->mmio = msm_ioremap(pdev, "dsi_pll", "DSI_PLL");
+       if (IS_ERR_OR_NULL(pll_28nm->mmio)) {
+               dev_err(&pdev->dev, "%s: failed to map pll base\n", __func__);
+               return ERR_PTR(-ENOMEM);
+       }
+
+       pll = &pll_28nm->base;
+       pll->min_rate = VCO_MIN_RATE;
+       pll->max_rate = VCO_MAX_RATE;
+       pll->get_provider = dsi_pll_28nm_get_provider;
+       pll->destroy = dsi_pll_28nm_destroy;
+       pll->disable_seq = dsi_pll_28nm_disable_seq;
+       pll->save_state = dsi_pll_28nm_save_state;
+       pll->restore_state = dsi_pll_28nm_restore_state;
+
+       pll->en_seq_cnt = 1;
+       pll->enable_seqs[0] = dsi_pll_28nm_enable_seq;
+
+       ret = pll_28nm_register(pll_28nm);
+       if (ret) {
+               dev_err(&pdev->dev, "failed to register PLL: %d\n", ret);
+               return ERR_PTR(ret);
+       }
+
+       return pll;
+}