drm/nv40/pm: write nv40-specific reclocking routines

Not 100% perfect yet, but a good start towards what it'll look like in the
end.

Actually seems stable on a NV44 I have here, as much as running around OA
for a fair amount of time constantly switching between performance levels
can prove..

My NV49 isn't quite so happy, and semaphores mess up somehow (sometimes) as
a result of the memory reclocking.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drivers/gpu/drm/nouveau/Makefile

@@ -31,7 +31,7 @@ nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \
              nv04_fbcon.o nv50_fbcon.o nvc0_fbcon.o \
              nv10_gpio.o nv50_gpio.o \
 	     nv50_calc.o \
-	     nv04_pm.o nv50_pm.o nva3_pm.o nvc0_pm.o \
+	     nv04_pm.o nv40_pm.o nv50_pm.o nva3_pm.o nvc0_pm.o \
 	     nv50_vram.o nvc0_vram.o \
 	     nv50_vm.o nvc0_vm.o
 

drivers/gpu/drm/nouveau/nouveau_pm.h

@@ -52,6 +52,11 @@ void *nv04_pm_clock_pre(struct drm_device *, struct nouveau_pm_level *,
 			u32 id, int khz);
 void nv04_pm_clock_set(struct drm_device *, void *);
 
+/* nv40_pm.c */
+int nv40_pm_clocks_get(struct drm_device *, struct nouveau_pm_level *);
+void *nv40_pm_clocks_pre(struct drm_device *, struct nouveau_pm_level *);
+void nv40_pm_clocks_set(struct drm_device *, void *);
+
 /* nv50_pm.c */
 int nv50_pm_clock_get(struct drm_device *, u32 id);
 void *nv50_pm_clock_pre(struct drm_device *, struct nouveau_pm_level *,

drivers/gpu/drm/nouveau/nouveau_state.c

@@ -286,9 +286,9 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
 		engine->gpio.get		= nv10_gpio_get;
 		engine->gpio.set		= nv10_gpio_set;
 		engine->gpio.irq_enable		= NULL;
-		engine->pm.clock_get		= nv04_pm_clock_get;
-		engine->pm.clock_pre		= nv04_pm_clock_pre;
-		engine->pm.clock_set		= nv04_pm_clock_set;
+		engine->pm.clocks_get		= nv40_pm_clocks_get;
+		engine->pm.clocks_pre		= nv40_pm_clocks_pre;
+		engine->pm.clocks_set		= nv40_pm_clocks_set;
 		engine->pm.voltage_get		= nouveau_voltage_gpio_get;
 		engine->pm.voltage_set		= nouveau_voltage_gpio_set;
 		engine->pm.temp_get		= nv40_temp_get;

drivers/gpu/drm/nouveau/nv40_pm.c

@@ -0,0 +1,333 @@
+/*
+ * Copyright 2011 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Ben Skeggs
+ */
+
+#include "drmP.h"
+#include "nouveau_drv.h"
+#include "nouveau_bios.h"
+#include "nouveau_pm.h"
+#include "nouveau_hw.h"
+
+#define min2(a,b) ((a) < (b) ? (a) : (b))
+
+static u32
+read_pll_1(struct drm_device *dev, u32 reg)
+{
+	u32 ctrl = nv_rd32(dev, reg + 0x00);
+	int P = (ctrl & 0x00070000) >> 16;
+	int N = (ctrl & 0x0000ff00) >> 8;
+	int M = (ctrl & 0x000000ff) >> 0;
+	u32 ref = 27000, clk = 0;
+
+	if (ctrl & 0x80000000)
+		clk = ref * N / M;
+
+	return clk >> P;
+}
+
+static u32
+read_pll_2(struct drm_device *dev, u32 reg)
+{
+	u32 ctrl = nv_rd32(dev, reg + 0x00);
+	u32 coef = nv_rd32(dev, reg + 0x04);
+	int N2 = (coef & 0xff000000) >> 24;
+	int M2 = (coef & 0x00ff0000) >> 16;
+	int N1 = (coef & 0x0000ff00) >> 8;
+	int M1 = (coef & 0x000000ff) >> 0;
+	int P = (ctrl & 0x00070000) >> 16;
+	u32 ref = 27000, clk = 0;
+
+	if (ctrl & 0x80000000)
+		clk = ref * N1 / M1;
+
+	if (!(ctrl & 0x00000100)) {
+		if (ctrl & 0x40000000)
+			clk = clk * N2 / M2;
+	}
+
+	return clk >> P;
+}
+
+static u32
+read_clk(struct drm_device *dev, u32 src)
+{
+	switch (src) {
+	case 3:
+		return read_pll_2(dev, 0x004000);
+	case 2:
+		return read_pll_1(dev, 0x004008);
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+int
+nv40_pm_clocks_get(struct drm_device *dev, struct nouveau_pm_level *perflvl)
+{
+	u32 ctrl = nv_rd32(dev, 0x00c040);
+
+	perflvl->core   = read_clk(dev, (ctrl & 0x00000003) >> 0);
+	perflvl->shader = read_clk(dev, (ctrl & 0x00000030) >> 4);
+	perflvl->memory = read_pll_2(dev, 0x4020);
+	return 0;
+}
+
+struct nv40_pm_state {
+	u32 ctrl;
+	u32 npll_ctrl;
+	u32 npll_coef;
+	u32 spll;
+	u32 mpll_ctrl;
+	u32 mpll_coef;
+};
+
+static int
+nv40_calc_pll(struct drm_device *dev, u32 reg, struct pll_lims *pll,
+	      u32 clk, int *N1, int *M1, int *N2, int *M2, int *log2P)
+{
+	struct nouveau_pll_vals coef;
+	int ret;
+
+	ret = get_pll_limits(dev, reg, pll);
+	if (ret)
+		return ret;
+
+	if (clk < pll->vco1.maxfreq)
+		pll->vco2.maxfreq = 0;
+
+	ret = nouveau_calc_pll_mnp(dev, pll, clk, &coef);
+	if (ret == 0)
+		return -ERANGE;
+
+	*N1 = coef.N1;
+	*M1 = coef.M1;
+	if (N2 && M2) {
+		if (pll->vco2.maxfreq) {
+			*N2 = coef.N2;
+			*M2 = coef.M2;
+		} else {
+			*N2 = 1;
+			*M2 = 1;
+		}
+	}
+	*log2P = coef.log2P;
+	return 0;
+}
+
+void *
+nv40_pm_clocks_pre(struct drm_device *dev, struct nouveau_pm_level *perflvl)
+{
+	struct nv40_pm_state *info;
+	struct pll_lims pll;
+	int N1, N2, M1, M2, log2P;
+	int ret;
+
+	info = kmalloc(sizeof(*info), GFP_KERNEL);
+	if (!info)
+		return ERR_PTR(-ENOMEM);
+
+	/* core/geometric clock */
+	ret = nv40_calc_pll(dev, 0x004000, &pll, perflvl->core,
+			    &N1, &M1, &N2, &M2, &log2P);
+	if (ret < 0)
+		goto out;
+
+	if (N2 == M2) {
+		info->npll_ctrl = 0x80000100 | (log2P << 16);
+		info->npll_coef = (N1 << 8) | M1;
+	} else {
+		info->npll_ctrl = 0xc0000000 | (log2P << 16);
+		info->npll_coef = (N2 << 24) | (M2 << 16) | (N1 << 8) | M1;
+	}
+
+	/* use the second PLL for shader/rop clock, if it differs from core */
+	if (perflvl->shader && perflvl->shader != perflvl->core) {
+		ret = nv40_calc_pll(dev, 0x004008, &pll, perflvl->shader,
+				    &N1, &M1, NULL, NULL, &log2P);
+		if (ret < 0)
+			goto out;
+
+		info->spll = 0xc0000000 | (log2P << 16) | (N1 << 8) | M1;
+		info->ctrl = 0x00000223;
+	} else {
+		info->spll = 0x00000000;
+		info->ctrl = 0x00000333;
+	}
+
+	/* memory clock */
+	ret = nv40_calc_pll(dev, 0x004020, &pll, perflvl->memory,
+			    &N1, &M1, &N2, &M2, &log2P);
+	if (ret < 0)
+		goto out;
+
+	info->mpll_ctrl  = 0x80000000 | (log2P << 16);
+	info->mpll_ctrl |= min2(pll.log2p_bias + log2P, pll.max_log2p) << 20;
+	if (N2 == M2) {
+		info->mpll_ctrl |= 0x00000100;
+		info->mpll_coef  = (N1 << 8) | M1;
+	} else {
+		info->mpll_ctrl |= 0x40000000;
+		info->mpll_coef  = (N2 << 24) | (M2 << 16) | (N1 << 8) | M1;
+	}
+
+out:
+	if (ret < 0) {
+		kfree(info);
+		info = ERR_PTR(ret);
+	}
+	return info;
+}
+
+static bool
+nv40_pm_gr_idle(void *data)
+{
+	struct drm_device *dev = data;
+
+	if ((nv_rd32(dev, 0x400760) & 0x000000f0) >> 4 !=
+	    (nv_rd32(dev, 0x400760) & 0x0000000f))
+		return false;
+
+	if (nv_rd32(dev, 0x400700))
+		return false;
+
+	return true;
+}
+
+void
+nv40_pm_clocks_set(struct drm_device *dev, void *pre_state)
+{
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	struct nv40_pm_state *info = pre_state;
+	unsigned long flags;
+	u32 crtc_mask = 0;
+	u8 sr1[2];
+	int i;
+
+	/* determine which CRTCs are active, fetch VGA_SR1 for each */
+	for (i = 0; i < 2; i++) {
+		u32 vbl = nv_rd32(dev, 0x600808 + (i * 0x2000));
+		u32 cnt = 0;
+		do {
+			if (vbl != nv_rd32(dev, 0x600808 + (i * 0x2000))) {
+				nv_wr08(dev, 0x0c03c4 + (i * 0x2000), 0x01);
+				sr1[i] = nv_rd08(dev, 0x0c03c5 + (i * 0x2000));
+				if (!(sr1[i] & 0x20))
+					crtc_mask |= (1 << i);
+				break;
+			}
+			udelay(1);
+		} while (cnt++ < 32);
+	}
+
+	/* halt and idle engines */
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+	nv_mask(dev, 0x002500, 0x00000001, 0x00000000);
+	if (!nv_wait(dev, 0x002500, 0x00000010, 0x00000000))
+		goto resume;
+	nv_mask(dev, 0x003220, 0x00000001, 0x00000000);
+	if (!nv_wait(dev, 0x003220, 0x00000010, 0x00000000))
+		goto resume;
+	nv_mask(dev, 0x003200, 0x00000001, 0x00000000);
+	nv04_fifo_cache_pull(dev, false);
+
+	if (!nv_wait_cb(dev, nv40_pm_gr_idle, dev))
+		goto resume;
+
+	/* set engine clocks */
+	nv_mask(dev, 0x00c040, 0x00000333, 0x00000000);
+	nv_wr32(dev, 0x004004, info->npll_coef);
+	nv_mask(dev, 0x004000, 0xc0070100, info->npll_ctrl);
+	nv_mask(dev, 0x004008, 0xc007ffff, info->spll);
+	mdelay(5);
+	nv_mask(dev, 0x00c040, 0x00000333, info->ctrl);
+
+	/* wait for vblank start on active crtcs, disable memory access */
+	for (i = 0; i < 2; i++) {
+		if (!(crtc_mask & (1 << i)))
+			continue;
+		nv_wait(dev, 0x600808 + (i * 0x2000), 0x00010000, 0x00000000);
+		nv_wait(dev, 0x600808 + (i * 0x2000), 0x00010000, 0x00010000);
+		nv_wr08(dev, 0x0c03c4 + (i * 0x2000), 0x01);
+		nv_wr08(dev, 0x0c03c5 + (i * 0x2000), sr1[i] | 0x20);
+	}
+
+	/* prepare ram for reclocking */
+	nv_wr32(dev, 0x1002d4, 0x00000001); /* precharge */
+	nv_wr32(dev, 0x1002d0, 0x00000001); /* refresh */
+	nv_wr32(dev, 0x1002d0, 0x00000001); /* refresh */
+	nv_mask(dev, 0x100210, 0x80000000, 0x00000000); /* no auto refresh */
+	nv_wr32(dev, 0x1002dc, 0x00000001); /* enable self-refresh */
+
+	/* change the PLL of each memory partition */
+	nv_mask(dev, 0x00c040, 0x0000c000, 0x00000000);
+	switch (dev_priv->chipset) {
+	case 0x40:
+	case 0x45:
+	case 0x41:
+	case 0x42:
+	case 0x47:
+		nv_mask(dev, 0x004044, 0xc0771100, info->mpll_ctrl);
+		nv_mask(dev, 0x00402c, 0xc0771100, info->mpll_ctrl);
+		nv_wr32(dev, 0x004048, info->mpll_coef);
+		nv_wr32(dev, 0x004030, info->mpll_coef);
+	case 0x43:
+	case 0x49:
+	case 0x4b:
+		nv_mask(dev, 0x004038, 0xc0771100, info->mpll_ctrl);
+		nv_wr32(dev, 0x00403c, info->mpll_coef);
+	default:
+		nv_mask(dev, 0x004020, 0xc0771100, info->mpll_ctrl);
+		nv_wr32(dev, 0x004024, info->mpll_coef);
+		break;
+	}
+	udelay(100);
+	nv_mask(dev, 0x00c040, 0x0000c000, 0x0000c000);
+
+	/* re-enable normal operation of memory controller */
+	nv_wr32(dev, 0x1002dc, 0x00000000);
+	nv_mask(dev, 0x100210, 0x80000000, 0x80000000);
+	udelay(100);
+
+	/* make sure we're in vblank (hopefully the same one as before), and
+	 * then re-enable crtc memory access
+	 */
+	for (i = 0; i < 2; i++) {
+		if (!(crtc_mask & (1 << i)))
+			continue;
+		nv_wait(dev, 0x600808 + (i * 0x2000), 0x00010000, 0x00010000);
+		nv_wr08(dev, 0x0c03c4 + (i * 0x2000), 0x01);
+		nv_wr08(dev, 0x0c03c5 + (i * 0x2000), sr1[i]);
+	}
+
+	/* resume engines */
+resume:
+	nv_wr32(dev, 0x003250, 0x00000001);
+	nv_mask(dev, 0x003220, 0x00000001, 0x00000001);
+	nv_wr32(dev, 0x003200, 0x00000001);
+	nv_wr32(dev, 0x002500, 0x00000001);
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
+
+	kfree(info);
+}