From: Eric Anholt <eric@anholt.net>
Date: Mon, 8 May 2017 12:33:48 -0700
Subject: drm/pl111: Register the clock divider and use it.
Git-commit: 032838f9cb4014af8a974374db9e2ce6f3aa8d3b
Patch-mainline: v4.13-rc1
References: FATE#326289 FATE#326079 FATE#326049 FATE#322398 FATE#326166

This is required for the panel to work on bcm911360, where CLCDCLK is
the fixed 200Mhz AXI41 clock.  The rate set is still passed up to the
CLCDCLK, for platforms that have a settable rate on that one.

v2: Set SET_RATE_PARENT (caught by Linus Walleij), depend on
    COMMON_CLK.
v3: Mark the clk_ops static (caught by Stephen).

Signed-off-by: Eric Anholt <eric@anholt.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170508193348.30236-1-eric@anholt.net
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Petr Tesarik <ptesarik@suse.com>
---
 drivers/gpu/drm/pl111/Kconfig         |    1 
 drivers/gpu/drm/pl111/pl111_display.c |  162 ++++++++++++++++++++++++++++++----
 drivers/gpu/drm/pl111/pl111_drm.h     |    8 +
 drivers/gpu/drm/pl111/pl111_drv.c     |   11 --
 include/linux/amba/clcd-regs.h        |    5 +
 5 files changed, 163 insertions(+), 24 deletions(-)

--- a/drivers/gpu/drm/pl111/Kconfig
+++ b/drivers/gpu/drm/pl111/Kconfig
@@ -2,6 +2,7 @@ config DRM_PL111
 	tristate "DRM Support for PL111 CLCD Controller"
 	depends on DRM
 	depends on ARM || ARM64 || COMPILE_TEST
+	depends on COMMON_CLK
 	select DRM_KMS_HELPER
 	select DRM_KMS_CMA_HELPER
 	select DRM_GEM_CMA_HELPER
--- a/drivers/gpu/drm/pl111/pl111_display.c
+++ b/drivers/gpu/drm/pl111/pl111_display.c
@@ -108,7 +108,7 @@ static void pl111_display_enable(struct
 	u32 cntl;
 	u32 ppl, hsw, hfp, hbp;
 	u32 lpp, vsw, vfp, vbp;
-	u32 cpl;
+	u32 cpl, tim2;
 	int ret;
 
 	ret = clk_set_rate(priv->clk, mode->clock * 1000);
@@ -142,20 +142,28 @@ static void pl111_display_enable(struct
 	       (vfp << 16) |
 	       (vbp << 24),
 	       priv->regs + CLCD_TIM1);
-	/* XXX: We currently always use CLCDCLK with no divisor.  We
-	 * could probably reduce power consumption by using HCLK
-	 * (apb_pclk) with a divisor when it gets us near our target
-	 * pixel clock.
-	 */
-	writel(((mode->flags & DRM_MODE_FLAG_NHSYNC) ? TIM2_IHS : 0) |
-	       ((mode->flags & DRM_MODE_FLAG_NVSYNC) ? TIM2_IVS : 0) |
-	       ((connector->display_info.bus_flags &
-		 DRM_BUS_FLAG_DE_LOW) ? TIM2_IOE : 0) |
-	       ((connector->display_info.bus_flags &
-		 DRM_BUS_FLAG_PIXDATA_NEGEDGE) ? TIM2_IPC : 0) |
-	       TIM2_BCD |
-	       (cpl << 16),
-	       priv->regs + CLCD_TIM2);
+
+	spin_lock(&priv->tim2_lock);
+
+	tim2 = readl(priv->regs + CLCD_TIM2);
+	tim2 &= (TIM2_BCD | TIM2_PCD_LO_MASK | TIM2_PCD_HI_MASK);
+
+	if (mode->flags & DRM_MODE_FLAG_NHSYNC)
+		tim2 |= TIM2_IHS;
+
+	if (mode->flags & DRM_MODE_FLAG_NVSYNC)
+		tim2 |= TIM2_IVS;
+
+	if (connector->display_info.bus_flags & DRM_BUS_FLAG_DE_LOW)
+		tim2 |= TIM2_IOE;
+
+	if (connector->display_info.bus_flags & DRM_BUS_FLAG_PIXDATA_NEGEDGE)
+		tim2 |= TIM2_IPC;
+
+	tim2 |= cpl << 16;
+	writel(tim2, priv->regs + CLCD_TIM2);
+	spin_unlock(&priv->tim2_lock);
+
 	writel(0, priv->regs + CLCD_TIM3);
 
 	drm_panel_prepare(priv->connector.panel);
@@ -288,6 +296,126 @@ const struct drm_simple_display_pipe_fun
 	.prepare_fb = pl111_display_prepare_fb,
 };
 
+static int pl111_clk_div_choose_div(struct clk_hw *hw, unsigned long rate,
+				    unsigned long *prate, bool set_parent)
+{
+	int best_div = 1, div;
+	struct clk_hw *parent = clk_hw_get_parent(hw);
+	unsigned long best_prate = 0;
+	unsigned long best_diff = ~0ul;
+	int max_div = (1 << (TIM2_PCD_LO_BITS + TIM2_PCD_HI_BITS)) - 1;
+
+	for (div = 1; div < max_div; div++) {
+		unsigned long this_prate, div_rate, diff;
+
+		if (set_parent)
+			this_prate = clk_hw_round_rate(parent, rate * div);
+		else
+			this_prate = *prate;
+		div_rate = DIV_ROUND_UP_ULL(this_prate, div);
+		diff = abs(rate - div_rate);
+
+		if (diff < best_diff) {
+			best_div = div;
+			best_diff = diff;
+			best_prate = this_prate;
+		}
+	}
+
+	*prate = best_prate;
+	return best_div;
+}
+
+static long pl111_clk_div_round_rate(struct clk_hw *hw, unsigned long rate,
+				     unsigned long *prate)
+{
+	int div = pl111_clk_div_choose_div(hw, rate, prate, true);
+
+	return DIV_ROUND_UP_ULL(*prate, div);
+}
+
+static unsigned long pl111_clk_div_recalc_rate(struct clk_hw *hw,
+					       unsigned long prate)
+{
+	struct pl111_drm_dev_private *priv =
+		container_of(hw, struct pl111_drm_dev_private, clk_div);
+	u32 tim2 = readl(priv->regs + CLCD_TIM2);
+	int div;
+
+	if (tim2 & TIM2_BCD)
+		return prate;
+
+	div = tim2 & TIM2_PCD_LO_MASK;
+	div |= (tim2 & TIM2_PCD_HI_MASK) >>
+		(TIM2_PCD_HI_SHIFT - TIM2_PCD_LO_BITS);
+	div += 2;
+
+	return DIV_ROUND_UP_ULL(prate, div);
+}
+
+static int pl111_clk_div_set_rate(struct clk_hw *hw, unsigned long rate,
+				  unsigned long prate)
+{
+	struct pl111_drm_dev_private *priv =
+		container_of(hw, struct pl111_drm_dev_private, clk_div);
+	int div = pl111_clk_div_choose_div(hw, rate, &prate, false);
+	u32 tim2;
+
+	spin_lock(&priv->tim2_lock);
+	tim2 = readl(priv->regs + CLCD_TIM2);
+	tim2 &= ~(TIM2_BCD | TIM2_PCD_LO_MASK | TIM2_PCD_HI_MASK);
+
+	if (div == 1) {
+		tim2 |= TIM2_BCD;
+	} else {
+		div -= 2;
+		tim2 |= div & TIM2_PCD_LO_MASK;
+		tim2 |= (div >> TIM2_PCD_LO_BITS) << TIM2_PCD_HI_SHIFT;
+	}
+
+	writel(tim2, priv->regs + CLCD_TIM2);
+	spin_unlock(&priv->tim2_lock);
+
+	return 0;
+}
+
+static const struct clk_ops pl111_clk_div_ops = {
+	.recalc_rate = pl111_clk_div_recalc_rate,
+	.round_rate = pl111_clk_div_round_rate,
+	.set_rate = pl111_clk_div_set_rate,
+};
+
+static int
+pl111_init_clock_divider(struct drm_device *drm)
+{
+	struct pl111_drm_dev_private *priv = drm->dev_private;
+	struct clk *parent = devm_clk_get(drm->dev, "clcdclk");
+	struct clk_hw *div = &priv->clk_div;
+	const char *parent_name;
+	struct clk_init_data init = {
+		.name = "pl111_div",
+		.ops = &pl111_clk_div_ops,
+		.parent_names = &parent_name,
+		.num_parents = 1,
+		.flags = CLK_SET_RATE_PARENT,
+	};
+	int ret;
+
+	if (IS_ERR(parent)) {
+		dev_err(drm->dev, "CLCD: unable to get clcdclk.\n");
+		return PTR_ERR(parent);
+	}
+	parent_name = __clk_get_name(parent);
+
+	spin_lock_init(&priv->tim2_lock);
+	div->init = &init;
+
+	ret = devm_clk_hw_register(drm->dev, div);
+
+	priv->clk = div->clk;
+	return ret;
+}
+
 int pl111_display_init(struct drm_device *drm)
 {
 	struct pl111_drm_dev_private *priv = drm->dev_private;
@@ -333,6 +461,10 @@ int pl111_display_init(struct drm_device
 		return -EINVAL;
 	}
 
+	ret = pl111_init_clock_divider(drm);
+	if (ret)
+		return ret;
+
 	ret = drm_simple_display_pipe_init(drm, &priv->pipe,
 					   &pl111_display_funcs,
 					   formats, ARRAY_SIZE(formats),
--- a/drivers/gpu/drm/pl111/pl111_drm.h
+++ b/drivers/gpu/drm/pl111/pl111_drm.h
@@ -21,6 +21,7 @@
 
 #include <drm/drm_gem.h>
 #include <drm/drm_simple_kms_helper.h>
+#include <linux/clk-provider.h>
 
 #define CLCD_IRQ_NEXTBASE_UPDATE BIT(2)
 
@@ -37,7 +38,14 @@ struct pl111_drm_dev_private {
 	struct drm_fbdev_cma *fbdev;
 
 	void *regs;
+	/* The pixel clock (a reference to our clock divider off of CLCDCLK). */
 	struct clk *clk;
+	/* pl111's internal clock divider. */
+	struct clk_hw clk_div;
+	/* Lock to sync access to CLCD_TIM2 between the common clock
+	 * subsystem and pl111_display_enable().
+	 */
+	spinlock_t tim2_lock;
 };
 
 #define to_pl111_connector(x) \
--- a/drivers/gpu/drm/pl111/pl111_drv.c
+++ b/drivers/gpu/drm/pl111/pl111_drv.c
@@ -50,8 +50,8 @@
  * - Read back hardware state at boot to skip reprogramming the
  *   hardware when doing a no-op modeset.
  *
- * - Use the internal clock divisor to reduce power consumption by
- *   using HCLK (apb_pclk) when appropriate.
+ * - Use the CLKSEL bit to support switching between the two external
+ *   clock parents.
  */
 
 #include <linux/amba/bus.h>
@@ -195,13 +195,6 @@ static int pl111_amba_probe(struct amba_
 	priv->drm = drm;
 	drm->dev_private = priv;
 
-	priv->clk = devm_clk_get(dev, "clcdclk");
-	if (IS_ERR(priv->clk)) {
-		dev_err(dev, "CLCD: unable to get clk.\n");
-		ret = PTR_ERR(priv->clk);
-		goto dev_unref;
-	}
-
 	priv->regs = devm_ioremap_resource(dev, &amba_dev->res);
 	if (!priv->regs) {
 		dev_err(dev, "%s failed mmio\n", __func__);
--- a/include/linux/amba/clcd-regs.h
+++ b/include/linux/amba/clcd-regs.h
@@ -39,12 +39,17 @@
 #define CLCD_PALL 		0x00000200
 #define CLCD_PALETTE		0x00000200
 
+#define TIM2_PCD_LO_MASK	GENMASK(4, 0)
+#define TIM2_PCD_LO_BITS	5
 #define TIM2_CLKSEL		(1 << 5)
 #define TIM2_IVS		(1 << 11)
 #define TIM2_IHS		(1 << 12)
 #define TIM2_IPC		(1 << 13)
 #define TIM2_IOE		(1 << 14)
 #define TIM2_BCD		(1 << 26)
+#define TIM2_PCD_HI_MASK	GENMASK(31, 27)
+#define TIM2_PCD_HI_BITS	5
+#define TIM2_PCD_HI_SHIFT	27
 
 #define CNTL_LCDEN		(1 << 0)
 #define CNTL_LCDBPP1		(0 << 1)