Linux-2.6.12-rc2

Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!

8 files changed, 2430 insertions, 0 deletions

diff --git a/drivers/video/kyro/Makefile b/drivers/video/kyro/Makefile
new file mode 100644
index 000000000000..2fd66f551bae
--- /dev/null
+++ b/drivers/video/kyro/Makefile
@@ -0,0 +1,8 @@
+#
+# Makefile for the Kyro framebuffer driver
+#
+
+obj-$(CONFIG_FB_KYRO)	+= kyrofb.o
+
+kyrofb-objs	:= STG4000Ramdac.o STG4000VTG.o STG4000OverlayDevice.o \
+		   STG4000InitDevice.o fbdev.o
diff --git a/drivers/video/kyro/STG4000InitDevice.c b/drivers/video/kyro/STG4000InitDevice.c
new file mode 100644
index 000000000000..7e33cd307d47
--- /dev/null
+++ b/drivers/video/kyro/STG4000InitDevice.c
@@ -0,0 +1,326 @@
+/*
+ *  linux/drivers/video/kyro/STG4000InitDevice.c
+ *
+ *  Copyright (C) 2000 Imagination Technologies Ltd
+ *  Copyright (C) 2002 STMicroelectronics
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of this archive
+ * for more details.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+
+#include "STG4000Reg.h"
+
+/* SDRAM fixed settings */
+#define SDRAM_CFG_0   0x49A1
+#define SDRAM_CFG_1   0xA732
+#define SDRAM_CFG_2   0x31
+#define SDRAM_ARB_CFG 0xA0
+#define SDRAM_REFRESH 0x20
+
+/* Reset values */
+#define PMX2_SOFTRESET_DAC_RST		0x0001
+#define PMX2_SOFTRESET_C1_RST		0x0004
+#define PMX2_SOFTRESET_C2_RST		0x0008
+#define PMX2_SOFTRESET_3D_RST		0x0010
+#define PMX2_SOFTRESET_VIDIN_RST	0x0020
+#define PMX2_SOFTRESET_TLB_RST		0x0040
+#define PMX2_SOFTRESET_SD_RST		0x0080
+#define PMX2_SOFTRESET_VGA_RST		0x0100
+#define PMX2_SOFTRESET_ROM_RST		0x0200	/* reserved bit, do not reset */
+#define PMX2_SOFTRESET_TA_RST		0x0400
+#define PMX2_SOFTRESET_REG_RST		0x4000
+#define PMX2_SOFTRESET_ALL		0x7fff
+
+/* Core clock freq */
+#define CORE_PLL_FREQ 1000000
+
+/* Reference Clock freq */
+#define REF_FREQ 14318
+
+/* PCI Registers */
+static u16 CorePllControl = 0x70;
+
+#define	PCI_CONFIG_SUBSYS_ID	0x2e
+
+/* Misc */
+#define CORE_PLL_MODE_REG_0_7      3
+#define CORE_PLL_MODE_REG_8_15     2
+#define CORE_PLL_MODE_CONFIG_REG   1
+#define DAC_PLL_CONFIG_REG         0
+
+#define STG_MAX_VCO 500000
+#define STG_MIN_VCO 100000
+
+/* PLL Clock */
+#define    STG4K3_PLL_SCALER      8	/* scale numbers by 2^8 for fixed point calc */
+#define    STG4K3_PLL_MIN_R       2	/* Minimum multiplier */
+#define    STG4K3_PLL_MAX_R       33	/* Max */
+#define    STG4K3_PLL_MIN_F       2	/* Minimum divisor */
+#define    STG4K3_PLL_MAX_F       513	/* Max */
+#define    STG4K3_PLL_MIN_OD      0	/* Min output divider (shift) */
+#define    STG4K3_PLL_MAX_OD      2	/* Max */
+#define    STG4K3_PLL_MIN_VCO_SC  (100000000 >> STG4K3_PLL_SCALER)	/* Min VCO rate */
+#define    STG4K3_PLL_MAX_VCO_SC  (500000000 >> STG4K3_PLL_SCALER)	/* Max VCO rate */
+#define    STG4K3_PLL_MINR_VCO_SC (100000000 >> STG4K3_PLL_SCALER)	/* Min VCO rate (restricted) */
+#define    STG4K3_PLL_MAXR_VCO_SC (500000000 >> STG4K3_PLL_SCALER)	/* Max VCO rate (restricted) */
+#define    STG4K3_PLL_MINR_VCO    100000000	/* Min VCO rate (restricted) */
+#define    STG4K3_PLL_MAX_VCO     500000000	/* Max VCO rate */
+#define    STG4K3_PLL_MAXR_VCO    500000000	/* Max VCO rate (restricted) */
+
+#define OS_DELAY(X) \
+{ \
+volatile u32 i,count=0; \
+    for(i=0;i<X;i++) count++; \
+}
+
+static u32 InitSDRAMRegisters(volatile STG4000REG __iomem *pSTGReg,
+			      u32 dwSubSysID, u32 dwRevID)
+{
+	u32 adwSDRAMArgCfg0[] = { 0xa0, 0x80, 0xa0, 0xa0, 0xa0 };
+	u32 adwSDRAMCfg1[] = { 0x8732, 0x8732, 0xa732, 0xa732, 0x8732 };
+	u32 adwSDRAMCfg2[] = { 0x87d2, 0x87d2, 0xa7d2, 0x87d2, 0xa7d2 };
+	u32 adwSDRAMRsh[] = { 36, 39, 40 };
+	u32 adwChipSpeed[] = { 110, 120, 125 };
+	u32 dwMemTypeIdx;
+	u32 dwChipSpeedIdx;
+
+	/* Get memory tpye and chip speed indexs from the SubSysDevID */
+	dwMemTypeIdx = (dwSubSysID & 0x70) >> 4;
+	dwChipSpeedIdx = (dwSubSysID & 0x180) >> 7;
+
+	if (dwMemTypeIdx > 4 || dwChipSpeedIdx > 2)
+		return 0;
+
+	/* Program SD-RAM interface */
+	STG_WRITE_REG(SDRAMArbiterConf, adwSDRAMArgCfg0[dwMemTypeIdx]);
+	if (dwRevID < 5) {
+		STG_WRITE_REG(SDRAMConf0, 0x49A1);
+		STG_WRITE_REG(SDRAMConf1, adwSDRAMCfg1[dwMemTypeIdx]);
+	} else {
+		STG_WRITE_REG(SDRAMConf0, 0x4DF1);
+		STG_WRITE_REG(SDRAMConf1, adwSDRAMCfg2[dwMemTypeIdx]);
+	}
+
+	STG_WRITE_REG(SDRAMConf2, 0x31);
+	STG_WRITE_REG(SDRAMRefresh, adwSDRAMRsh[dwChipSpeedIdx]);
+
+	return adwChipSpeed[dwChipSpeedIdx] * 10000;
+}
+
+u32 ProgramClock(u32 refClock,
+		   u32 coreClock,
+		   u32 * FOut, u32 * ROut, u32 * POut)
+{
+	u32 R = 0, F = 0, OD = 0, ODIndex = 0;
+	u32 ulBestR = 0, ulBestF = 0, ulBestOD = 0;
+	u32 ulBestVCO = 0, ulBestClk = 0, ulBestScore = 0;
+	u32 ulScore, ulPhaseScore, ulVcoScore;
+	u32 ulTmp = 0, ulVCO;
+	u32 ulScaleClockReq, ulMinClock, ulMaxClock;
+	u32 ODValues[] = { 1, 2, 0 };
+
+	/* Translate clock in Hz */
+	coreClock *= 100;	/* in Hz */
+	refClock *= 1000;	/* in Hz */
+
+	/* Work out acceptable clock
+	 * The method calculates ~ +- 0.4% (1/256)
+	 */
+	ulMinClock = coreClock - (coreClock >> 8);
+	ulMaxClock = coreClock + (coreClock >> 8);
+
+	/* Scale clock required for use in calculations */
+	ulScaleClockReq = coreClock >> STG4K3_PLL_SCALER;
+
+	/* Iterate through post divider values */
+	for (ODIndex = 0; ODIndex < 3; ODIndex++) {
+		OD = ODValues[ODIndex];
+		R = STG4K3_PLL_MIN_R;
+
+		/* loop for pre-divider from min to max  */
+		while (R <= STG4K3_PLL_MAX_R) {
+			/* estimate required feedback multiplier */
+			ulTmp = R * (ulScaleClockReq << OD);
+
+			/* F = ClkRequired * R * (2^OD) / Fref */
+			F = (u32)(ulTmp / (refClock >> STG4K3_PLL_SCALER));
+
+			/* compensate for accuracy */
+			if (F > STG4K3_PLL_MIN_F)
+				F--;
+
+
+			/*
+			 * We should be close to our target frequency (if it's
+			 * achievable with current OD & R) let's iterate
+			 * through F for best fit
+			 */
+			while ((F >= STG4K3_PLL_MIN_F) &&
+			       (F <= STG4K3_PLL_MAX_F)) {
+				/* Calc VCO at full accuracy */
+				ulVCO = refClock / R;
+				ulVCO = F * ulVCO;
+
+				/*
+				 * Check it's within restricted VCO range
+				 * unless of course the desired frequency is
+				 * above the restricted range, then test
+				 * against VCO limit
+				 */
+				if ((ulVCO >= STG4K3_PLL_MINR_VCO) &&
+				    ((ulVCO <= STG4K3_PLL_MAXR_VCO) ||
+				     ((coreClock > STG4K3_PLL_MAXR_VCO)
+				      && (ulVCO <= STG4K3_PLL_MAX_VCO)))) {
+					ulTmp = (ulVCO >> OD);	/* Clock = VCO / (2^OD) */
+
+					/* Is this clock good enough? */
+					if ((ulTmp >= ulMinClock)
+					    && (ulTmp <= ulMaxClock)) {
+						ulPhaseScore = (((refClock / R) - (refClock / STG4K3_PLL_MAX_R))) / ((refClock - (refClock / STG4K3_PLL_MAX_R)) >> 10);
+
+						ulVcoScore = ((ulVCO - STG4K3_PLL_MINR_VCO)) / ((STG4K3_PLL_MAXR_VCO - STG4K3_PLL_MINR_VCO) >> 10);
+						ulScore = ulPhaseScore + ulVcoScore;
+
+						if (!ulBestScore) {
+							ulBestVCO = ulVCO;
+							ulBestOD = OD;
+							ulBestF = F;
+							ulBestR = R;
+							ulBestClk = ulTmp;
+							ulBestScore =
+							    ulScore;
+						}
+						/* is this better, ( aim for highest Score) */
+			/*--------------------------------------------------------------------------
+                             Here we want to use a scoring system which will take account of both the
+                            value at the phase comparater and the VCO output
+                             to do this we will use a cumulative score between the two
+                          The way this ends up is that we choose the first value in the loop anyway
+                          but we shall keep this code in case new restrictions come into play
+                          --------------------------------------------------------------------------*/
+						if ((ulScore >= ulBestScore) && (OD > 0)) {
+							ulBestVCO = ulVCO;
+							ulBestOD = OD;
+							ulBestF = F;
+							ulBestR = R;
+							ulBestClk = ulTmp;
+							ulBestScore =
+							    ulScore;
+						}
+					}
+				}
+				F++;
+			}
+			R++;
+		}
+	}
+
+	/*
+	   did we find anything?
+	   Then return RFOD
+	 */
+	if (ulBestScore) {
+		*ROut = ulBestR;
+		*FOut = ulBestF;
+
+		if ((ulBestOD == 2) || (ulBestOD == 3)) {
+			*POut = 3;
+		} else
+			*POut = ulBestOD;
+
+	}
+
+	return (ulBestClk);
+}
+
+int SetCoreClockPLL(volatile STG4000REG __iomem *pSTGReg, struct pci_dev *pDev)
+{
+	u32 F, R, P;
+	u16 core_pll = 0, sub;
+	u32 ulCoreClock;
+	u32 tmp;
+	u32 ulChipSpeed;
+	u8 rev;
+
+	STG_WRITE_REG(IntMask, 0xFFFF);
+
+	/* Disable Primary Core Thread0 */
+	tmp = STG_READ_REG(Thread0Enable);
+	CLEAR_BIT(0);
+	STG_WRITE_REG(Thread0Enable, tmp);
+
+	/* Disable Primary Core Thread1 */
+	tmp = STG_READ_REG(Thread1Enable);
+	CLEAR_BIT(0);
+	STG_WRITE_REG(Thread1Enable, tmp);
+
+	STG_WRITE_REG(SoftwareReset,
+		      PMX2_SOFTRESET_REG_RST | PMX2_SOFTRESET_ROM_RST);
+	STG_WRITE_REG(SoftwareReset,
+		      PMX2_SOFTRESET_REG_RST | PMX2_SOFTRESET_TA_RST |
+		      PMX2_SOFTRESET_ROM_RST);
+
+	/* Need to play around to reset TA */
+	STG_WRITE_REG(TAConfiguration, 0);
+	STG_WRITE_REG(SoftwareReset,
+		      PMX2_SOFTRESET_REG_RST | PMX2_SOFTRESET_ROM_RST);
+	STG_WRITE_REG(SoftwareReset,
+		      PMX2_SOFTRESET_REG_RST | PMX2_SOFTRESET_TA_RST |
+		      PMX2_SOFTRESET_ROM_RST);
+
+	pci_read_config_word(pDev, PCI_CONFIG_SUBSYS_ID, &sub);
+	pci_read_config_byte(pDev, PCI_REVISION_ID, &rev);
+
+	ulChipSpeed = InitSDRAMRegisters(pSTGReg, (u32)sub, (u32)rev);
+
+	if (ulChipSpeed == 0)
+		return -EINVAL;
+
+	ulCoreClock = ProgramClock(REF_FREQ, CORE_PLL_FREQ, &F, &R, &P);
+
+	core_pll |= ((P) | ((F - 2) << 2) | ((R - 2) << 11));
+
+	/* Set Core PLL Control to Core PLL Mode  */
+
+	/* Send bits 0:7 of the Core PLL Mode register */
+	tmp = ((CORE_PLL_MODE_REG_0_7 << 8) | (core_pll & 0x00FF));
+	pci_write_config_word(pDev, CorePllControl, tmp);
+	/* Without some delay between the PCI config writes the clock does
+	   not reliably set when the code is compiled -O3
+	 */
+	OS_DELAY(1000000);
+
+	tmp |= SET_BIT(14);
+	pci_write_config_word(pDev, CorePllControl, tmp);
+	OS_DELAY(1000000);
+
+	/* Send bits 8:15 of the Core PLL Mode register */
+	tmp =
+	    ((CORE_PLL_MODE_REG_8_15 << 8) | ((core_pll & 0xFF00) >> 8));
+	pci_write_config_word(pDev, CorePllControl, tmp);
+	OS_DELAY(1000000);
+
+	tmp |= SET_BIT(14);
+	pci_write_config_word(pDev, CorePllControl, tmp);
+	OS_DELAY(1000000);
+
+	STG_WRITE_REG(SoftwareReset, PMX2_SOFTRESET_ALL);
+
+#if 0
+	/* Enable Primary Core Thread0 */
+	tmp = ((STG_READ_REG(Thread0Enable)) | SET_BIT(0));
+	STG_WRITE_REG(Thread0Enable, tmp);
+
+	/* Enable Primary Core Thread1 */
+	tmp = ((STG_READ_REG(Thread1Enable)) | SET_BIT(0));
+	STG_WRITE_REG(Thread1Enable, tmp);
+#endif
+
+	return 0;
+}
diff --git a/drivers/video/kyro/STG4000Interface.h b/drivers/video/kyro/STG4000Interface.h
new file mode 100644
index 000000000000..e75b3b4a4aa1
--- /dev/null
+++ b/drivers/video/kyro/STG4000Interface.h
@@ -0,0 +1,60 @@
+/*
+ *  linux/drivers/video/kyro/STG4000Interface.h
+ *
+ *  Copyright (C) 2002 STMicroelectronics
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of this archive
+ * for more details.
+ */
+
+#ifndef _STG4000INTERFACE_H
+#define _STG4000INTERFACE_H
+
+struct pci_dev;
+
+/*
+ * Ramdac Setup
+ */
+extern int InitialiseRamdac(volatile STG4000REG __iomem *pSTGReg, u32 displayDepth,
+			    u32 displayWidth, u32 displayHeight,
+			    s32 HSyncPolarity, s32 VSyncPolarity,
+			    u32 *pixelClock);
+
+extern void DisableRamdacOutput(volatile STG4000REG __iomem *pSTGReg);
+extern void EnableRamdacOutput(volatile STG4000REG __iomem *pSTGReg);
+
+/*
+ * Timing generator setup
+ */
+extern void DisableVGA(volatile STG4000REG __iomem *pSTGReg);
+extern void StopVTG(volatile STG4000REG __iomem *pSTGReg);
+extern void StartVTG(volatile STG4000REG __iomem *pSTGReg);
+extern void SetupVTG(volatile STG4000REG __iomem *pSTGReg,
+		     const struct kyrofb_info * pTiming);
+
+extern u32 ProgramClock(u32 refClock, u32 coreClock, u32 *FOut, u32 *ROut, u32 *POut);
+extern int SetCoreClockPLL(volatile STG4000REG __iomem *pSTGReg, struct pci_dev *pDev);
+
+/*
+ * Overlay setup
+ */
+extern void ResetOverlayRegisters(volatile STG4000REG __iomem *pSTGReg);
+
+extern int CreateOverlaySurface(volatile STG4000REG __iomem *pSTGReg,
+				u32 ulWidth, u32 ulHeight,
+				int bLinear,
+				u32 ulOverlayOffset,
+				u32 * retStride, u32 * retUVStride);
+
+extern int SetOverlayBlendMode(volatile STG4000REG __iomem *pSTGReg,
+			       OVRL_BLEND_MODE mode,
+			       u32 ulAlpha, u32 ulColorKey);
+
+extern int SetOverlayViewPort(volatile STG4000REG __iomem *pSTGReg,
+			      u32 left, u32 top,
+			      u32 right, u32 bottom);
+
+extern void EnableOverlayPlane(volatile STG4000REG __iomem *pSTGReg);
+
+#endif /* _STG4000INTERFACE_H */
diff --git a/drivers/video/kyro/STG4000OverlayDevice.c b/drivers/video/kyro/STG4000OverlayDevice.c
new file mode 100644
index 000000000000..2ae9bafacdd0
--- /dev/null
+++ b/drivers/video/kyro/STG4000OverlayDevice.c
@@ -0,0 +1,600 @@
+/*
+ *  linux/drivers/video/kyro/STG4000OverlayDevice.c
+ *
+ *  Copyright (C) 2000 Imagination Technologies Ltd
+ *  Copyright (C) 2002 STMicroelectronics
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of this archive
+ * for more details.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+
+#include "STG4000Reg.h"
+
+/* HW Defines */
+
+#define STG4000_NO_SCALING    0x800
+#define STG4000_NO_DECIMATION 0xFFFFFFFF
+
+/* Primary surface */
+#define STG4000_PRIM_NUM_PIX   5
+#define STG4000_PRIM_ALIGN     4
+#define STG4000_PRIM_ADDR_BITS 20
+
+#define STG4000_PRIM_MIN_WIDTH  640
+#define STG4000_PRIM_MAX_WIDTH  1600
+#define STG4000_PRIM_MIN_HEIGHT 480
+#define STG4000_PRIM_MAX_HEIGHT 1200
+
+/* Overlay surface */
+#define STG4000_OVRL_NUM_PIX   4
+#define STG4000_OVRL_ALIGN     2
+#define STG4000_OVRL_ADDR_BITS 20
+#define STG4000_OVRL_NUM_MODES 5
+
+#define STG4000_OVRL_MIN_WIDTH  0
+#define STG4000_OVRL_MAX_WIDTH  720
+#define STG4000_OVRL_MIN_HEIGHT 0
+#define STG4000_OVRL_MAX_HEIGHT 576
+
+/* Decimation and Scaling */
+static u32 adwDecim8[33] = {
+	    0xffffffff, 0xfffeffff, 0xffdffbff, 0xfefefeff, 0xfdf7efbf,
+	    0xfbdf7bdf, 0xf7bbddef, 0xeeeeeeef, 0xeeddbb77, 0xedb76db7,
+	    0xdb6db6db, 0xdb5b5b5b, 0xdab5ad6b, 0xd5ab55ab, 0xd555aaab,
+	    0xaaaaaaab, 0xaaaa5555, 0xaa952a55, 0xa94a5295, 0xa5252525,
+	    0xa4924925, 0x92491249, 0x91224489, 0x91111111, 0x90884211,
+	    0x88410821, 0x88102041, 0x81010101, 0x80800801, 0x80010001,
+	    0x80000001, 0x00000001, 0x00000000
+};
+
+typedef struct _OVRL_SRC_DEST {
+	/*clipped on-screen pixel position of overlay */
+	u32 ulDstX1;
+	u32 ulDstY1;
+	u32 ulDstX2;
+	u32 ulDstY2;
+
+	/*clipped pixel pos of source data within buffer thses need to be 128 bit word aligned */
+	u32 ulSrcX1;
+	u32 ulSrcY1;
+	u32 ulSrcX2;
+	u32 ulSrcY2;
+
+	/* on-screen pixel position of overlay */
+	s32 lDstX1;
+	s32 lDstY1;
+	s32 lDstX2;
+	s32 lDstY2;
+} OVRL_SRC_DEST;
+
+static u32 ovlWidth, ovlHeight, ovlStride;
+static int ovlLinear;
+
+void ResetOverlayRegisters(volatile STG4000REG __iomem *pSTGReg)
+{
+	u32 tmp;
+
+	/* Set Overlay address to default */
+	tmp = STG_READ_REG(DACOverlayAddr);
+	CLEAR_BITS_FRM_TO(0, 20);
+	CLEAR_BIT(31);
+	STG_WRITE_REG(DACOverlayAddr, tmp);
+
+	/* Set Overlay U address */
+	tmp = STG_READ_REG(DACOverlayUAddr);
+	CLEAR_BITS_FRM_TO(0, 20);
+	STG_WRITE_REG(DACOverlayUAddr, tmp);
+
+	/* Set Overlay V address */
+	tmp = STG_READ_REG(DACOverlayVAddr);
+	CLEAR_BITS_FRM_TO(0, 20);
+	STG_WRITE_REG(DACOverlayVAddr, tmp);
+
+	/* Set Overlay Size */
+	tmp = STG_READ_REG(DACOverlaySize);
+	CLEAR_BITS_FRM_TO(0, 10);
+	CLEAR_BITS_FRM_TO(12, 31);
+	STG_WRITE_REG(DACOverlaySize, tmp);
+
+	/* Set Overlay Vt Decimation */
+	tmp = STG4000_NO_DECIMATION;
+	STG_WRITE_REG(DACOverlayVtDec, tmp);
+
+	/* Set Overlay format to default value */
+	tmp = STG_READ_REG(DACPixelFormat);
+	CLEAR_BITS_FRM_TO(4, 7);
+	CLEAR_BITS_FRM_TO(16, 22);
+	STG_WRITE_REG(DACPixelFormat, tmp);
+
+	/* Set Vertical scaling to default */
+	tmp = STG_READ_REG(DACVerticalScal);
+	CLEAR_BITS_FRM_TO(0, 11);
+	CLEAR_BITS_FRM_TO(16, 22);
+	tmp |= STG4000_NO_SCALING;	/* Set to no scaling */
+	STG_WRITE_REG(DACVerticalScal, tmp);
+
+	/* Set Horizontal Scaling to default */
+	tmp = STG_READ_REG(DACHorizontalScal);
+	CLEAR_BITS_FRM_TO(0, 11);
+	CLEAR_BITS_FRM_TO(16, 17);
+	tmp |= STG4000_NO_SCALING;	/* Set to no scaling */
+	STG_WRITE_REG(DACHorizontalScal, tmp);
+
+	/* Set Blend mode to Alpha Blend */
+	/* ????? SG 08/11/2001 Surely this isn't the alpha blend mode,
+	   hopefully its overwrite
+	 */
+	tmp = STG_READ_REG(DACBlendCtrl);
+	CLEAR_BITS_FRM_TO(0, 30);
+	tmp = (GRAPHICS_MODE << 28);
+	STG_WRITE_REG(DACBlendCtrl, tmp);
+
+}
+
+int CreateOverlaySurface(volatile STG4000REG __iomem *pSTGReg,
+			 u32 inWidth,
+			 u32 inHeight,
+			 int bLinear,
+			 u32 ulOverlayOffset,
+			 u32 * retStride, u32 * retUVStride)
+{
+	u32 tmp;
+	u32 ulStride;
+
+	if (inWidth > STG4000_OVRL_MAX_WIDTH ||
+	    inHeight > STG4000_OVRL_MAX_HEIGHT) {
+		return -EINVAL;
+	}
+
+	/* Stride in 16 byte words - 16Bpp */
+	if (bLinear) {
+		/* Format is 16bits so num 16 byte words is width/8 */
+		if ((inWidth & 0x7) == 0) {	/* inWidth % 8 */
+			ulStride = (inWidth / 8);
+		} else {
+			/* Round up to next 16byte boundary */
+			ulStride = ((inWidth + 8) / 8);
+		}
+	} else {
+		/* Y component is 8bits so num 16 byte words is width/16 */
+		if ((inWidth & 0xf) == 0) {	/* inWidth % 16 */
+			ulStride = (inWidth / 16);
+		} else {
+			/* Round up to next 16byte boundary */
+			ulStride = ((inWidth + 16) / 16);
+		}
+	}
+
+
+	/* Set Overlay address and Format mode */
+	tmp = STG_READ_REG(DACOverlayAddr);
+	CLEAR_BITS_FRM_TO(0, 20);
+	if (bLinear) {
+		CLEAR_BIT(31);	/* Overlay format to Linear */
+	} else {
+		tmp |= SET_BIT(31);	/* Overlay format to Planer */
+	}
+
+	/* Only bits 24:4 of the Overlay address */
+	tmp |= (ulOverlayOffset >> 4);
+	STG_WRITE_REG(DACOverlayAddr, tmp);
+
+	if (!bLinear) {
+		u32 uvSize =
+		    (inWidth & 0x1) ? (inWidth + 1 / 2) : (inWidth / 2);
+		u32 uvStride;
+		u32 ulOffset;
+		/* Y component is 8bits so num 32 byte words is width/32 */
+		if ((uvSize & 0xf) == 0) {	/* inWidth % 16 */
+			uvStride = (uvSize / 16);
+		} else {
+			/* Round up to next 32byte boundary */
+			uvStride = ((uvSize + 16) / 16);
+		}
+
+		ulOffset = ulOverlayOffset + (inHeight * (ulStride * 16));
+		/* Align U,V data to 32byte boundary */
+		if ((ulOffset & 0x1f) != 0)
+			ulOffset = (ulOffset + 32L) & 0xffffffE0L;
+
+		tmp = STG_READ_REG(DACOverlayUAddr);
+		CLEAR_BITS_FRM_TO(0, 20);
+		tmp |= (ulOffset >> 4);
+		STG_WRITE_REG(DACOverlayUAddr, tmp);
+
+		ulOffset += (inHeight / 2) * (uvStride * 16);
+		/* Align U,V data to 32byte boundary */
+		if ((ulOffset & 0x1f) != 0)
+			ulOffset = (ulOffset + 32L) & 0xffffffE0L;
+
+		tmp = STG_READ_REG(DACOverlayVAddr);
+		CLEAR_BITS_FRM_TO(0, 20);
+		tmp |= (ulOffset >> 4);
+		STG_WRITE_REG(DACOverlayVAddr, tmp);
+
+		*retUVStride = uvStride * 16;
+	}
+
+
+	/* Set Overlay YUV pixel format
+	 * Make sure that LUT not used - ??????
+	 */
+	tmp = STG_READ_REG(DACPixelFormat);
+	/* Only support Planer or UYVY linear formats */
+	CLEAR_BITS_FRM_TO(4, 9);
+	STG_WRITE_REG(DACPixelFormat, tmp);
+
+	ovlWidth = inWidth;
+	ovlHeight = inHeight;
+	ovlStride = ulStride;
+	ovlLinear = bLinear;
+	*retStride = ulStride << 4;	/* In bytes */
+
+	return 0;
+}
+
+int SetOverlayBlendMode(volatile STG4000REG __iomem *pSTGReg,
+			OVRL_BLEND_MODE mode,
+			u32 ulAlpha, u32 ulColorKey)
+{
+	u32 tmp;
+
+	tmp = STG_READ_REG(DACBlendCtrl);
+	CLEAR_BITS_FRM_TO(28, 30);
+	tmp |= (mode << 28);
+
+	switch (mode) {
+	case COLOR_KEY:
+		CLEAR_BITS_FRM_TO(0, 23);
+		tmp |= (ulColorKey & 0x00FFFFFF);
+		break;
+
+	case GLOBAL_ALPHA:
+		CLEAR_BITS_FRM_TO(24, 27);
+		tmp |= ((ulAlpha & 0xF) << 24);
+		break;
+
+	case CK_PIXEL_ALPHA:
+		CLEAR_BITS_FRM_TO(0, 23);
+		tmp |= (ulColorKey & 0x00FFFFFF);
+		break;
+
+	case CK_GLOBAL_ALPHA:
+		CLEAR_BITS_FRM_TO(0, 23);
+		tmp |= (ulColorKey & 0x00FFFFFF);
+		CLEAR_BITS_FRM_TO(24, 27);
+		tmp |= ((ulAlpha & 0xF) << 24);
+		break;
+
+	case GRAPHICS_MODE:
+	case PER_PIXEL_ALPHA:
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	STG_WRITE_REG(DACBlendCtrl, tmp);
+
+	return 0;
+}
+
+void EnableOverlayPlane(volatile STG4000REG __iomem *pSTGReg)
+{
+	u32 tmp;
+	/* Enable Overlay */
+	tmp = STG_READ_REG(DACPixelFormat);
+	tmp |= SET_BIT(7);
+	STG_WRITE_REG(DACPixelFormat, tmp);
+
+	/* Set video stream control */
+	tmp = STG_READ_REG(DACStreamCtrl);
+	tmp |= SET_BIT(1);	/* video stream */
+	STG_WRITE_REG(DACStreamCtrl, tmp);
+}
+
+static u32 Overlap(u32 ulBits, u32 ulPattern)
+{
+	u32 ulCount = 0;
+
+	while (ulBits) {
+		if (!(ulPattern & 1))
+			ulCount++;
+		ulBits--;
+		ulPattern = ulPattern >> 1;
+	}
+
+	return ulCount;
+
+}
+
+int SetOverlayViewPort(volatile STG4000REG __iomem *pSTGReg,
+		       u32 left, u32 top,
+		       u32 right, u32 bottom)
+{
+	OVRL_SRC_DEST srcDest;
+
+	u32 ulSrcTop, ulSrcBottom;
+	u32 ulSrc, ulDest;
+	u32 ulFxScale, ulFxOffset;
+	u32 ulHeight, ulWidth;
+	u32 ulPattern;
+	u32 ulDecimate, ulDecimated;
+	u32 ulApplied;
+	u32 ulDacXScale, ulDacYScale;
+	u32 ulScale;
+	u32 ulLeft, ulRight;
+	u32 ulSrcLeft, ulSrcRight;
+	u32 ulScaleLeft, ulScaleRight;
+	u32 ulhDecim;
+	u32 ulsVal;
+	u32 ulVertDecFactor;
+	int bResult;
+	u32 ulClipOff = 0;
+	u32 ulBits = 0;
+	u32 ulsAdd = 0;
+	u32 tmp, ulStride;
+	u32 ulExcessPixels, ulClip, ulExtraLines;
+
+
+	srcDest.ulSrcX1 = 0;
+	srcDest.ulSrcY1 = 0;
+	srcDest.ulSrcX2 = ovlWidth - 1;
+	srcDest.ulSrcY2 = ovlHeight - 1;
+
+	srcDest.ulDstX1 = left;
+	srcDest.ulDstY1 = top;
+	srcDest.ulDstX2 = right;
+	srcDest.ulDstY2 = bottom;
+
+	srcDest.lDstX1 = srcDest.ulDstX1;
+	srcDest.lDstY1 = srcDest.ulDstY1;
+	srcDest.lDstX2 = srcDest.ulDstX2;
+	srcDest.lDstY2 = srcDest.ulDstY2;
+
+    /************* Vertical decimation/scaling ******************/
+
+	/* Get Src Top and Bottom */
+	ulSrcTop = srcDest.ulSrcY1;
+	ulSrcBottom = srcDest.ulSrcY2;
+
+	ulSrc = ulSrcBottom - ulSrcTop;
+	ulDest = srcDest.lDstY2 - srcDest.lDstY1;	/* on-screen overlay */
+
+	if (ulSrc <= 1)
+		return -EINVAL;
+
+	/* First work out the position we are to display as offset from the
+	 * source of the buffer
+	 */
+	ulFxScale = (ulDest << 11) / ulSrc;	/* fixed point scale factor */
+	ulFxOffset = (srcDest.lDstY2 - srcDest.ulDstY2) << 11;
+
+	ulSrcBottom = ulSrcBottom - (ulFxOffset / ulFxScale);
+	ulSrc = ulSrcBottom - ulSrcTop;
+	ulHeight = ulSrc;
+
+	ulDest = srcDest.ulDstY2 - (srcDest.ulDstY1 - 1);
+	ulPattern = adwDecim8[ulBits];
+
+	/* At this point ulSrc represents the input decimator */
+	if (ulSrc > ulDest) {
+		ulDecimate = ulSrc - ulDest;
+		ulBits = 0;
+		ulApplied = ulSrc / 32;
+
+		while (((ulBits * ulApplied) +
+			Overlap((ulSrc % 32),
+				adwDecim8[ulBits])) < ulDecimate)
+			ulBits++;
+
+		ulPattern = adwDecim8[ulBits];
+		ulDecimated =
+		    (ulBits * ulApplied) + Overlap((ulSrc % 32),
+						   ulPattern);
+		ulSrc = ulSrc - ulDecimated;	/* the number number of lines that will go into the scaler */
+	}
+
+	if (ulBits && (ulBits != 32)) {
+		ulVertDecFactor = (63 - ulBits) / (32 - ulBits);	/* vertical decimation factor scaled up to nearest integer */
+	} else {
+		ulVertDecFactor = 1;
+	}
+
+	ulDacYScale = ((ulSrc - 1) * 2048) / (ulDest + 1);
+
+	tmp = STG_READ_REG(DACOverlayVtDec);	/* Decimation */
+	CLEAR_BITS_FRM_TO(0, 31);
+	tmp = ulPattern;
+	STG_WRITE_REG(DACOverlayVtDec, tmp);
+
+	/***************** Horizontal decimation/scaling ***************************/
+
+	/*
+	 * Now we handle the horizontal case, this is a simplified verison of
+	 * the vertical case in that we decimate by factors of 2.  as we are
+	 * working in words we should always be able to decimate by these
+	 * factors.  as we always have to have a buffer which is aligned to a
+	 * whole number of 128 bit words, we must align the left side to the
+	 * lowest to the next lowest 128 bit boundary, and the right hand edge
+	 * to the next largets boundary, (in a similar way to how we didi it in
+	 * PMX1) as the left and right hand edges are aligned to these
+	 * boundaries normally this only becomes an issue when we are chopping
+	 * of one of the sides We shall work out vertical stuff first
+	 */
+	ulSrc = srcDest.ulSrcX2 - srcDest.ulSrcX1;
+	ulDest = srcDest.lDstX2 - srcDest.lDstX1;
+#ifdef _OLDCODE
+	ulLeft = srcDest.ulDstX1;
+	ulRight = srcDest.ulDstX2;
+#else
+	if (srcDest.ulDstX1 > 2) {
+		ulLeft = srcDest.ulDstX1 + 2;
+		ulRight = srcDest.ulDstX2 + 1;
+	} else {
+		ulLeft = srcDest.ulDstX1;
+		ulRight = srcDest.ulDstX2 + 1;
+	}
+#endif
+	/* first work out the position we are to display as offset from the source of the buffer */
+	bResult = 1;
+
+	do {
+		if (ulDest == 0)
+			return -EINVAL;
+
+		/* source pixels per dest pixel <<11 */
+		ulFxScale = ((ulSrc - 1) << 11) / (ulDest);
+
+		/* then number of destination pixels out we are */
+		ulFxOffset = ulFxScale * ((srcDest.ulDstX1 - srcDest.lDstX1) + ulClipOff);
+		ulFxOffset >>= 11;
+
+		/* this replaces the code which was making a decision as to use either ulFxOffset or ulSrcX1 */
+		ulSrcLeft = srcDest.ulSrcX1 + ulFxOffset;
+
+		/* then number of destination pixels out we are */
+		ulFxOffset = ulFxScale * (srcDest.lDstX2 - srcDest.ulDstX2);
+		ulFxOffset >>= 11;
+
+		ulSrcRight = srcDest.ulSrcX2 - ulFxOffset;
+
+		/*
+		 * we must align these to our 128 bit boundaries. we shall
+		 * round down the pixel pos to the nearest 8 pixels.
+		 */
+		ulScaleLeft = ulSrcLeft;
+		ulScaleRight = ulSrcRight;
+
+		/* shift fxscale until it is in the range of the scaler */
+		ulhDecim = 0;
+		ulScale = (((ulSrcRight - ulSrcLeft) - 1) << (11 - ulhDecim)) / (ulRight - ulLeft + 2);
+
+		while (ulScale > 0x800) {
+			ulhDecim++;
+			ulScale = (((ulSrcRight - ulSrcLeft) - 1) << (11 - ulhDecim)) / (ulRight - ulLeft + 2);
+		}
+
+		/*
+		 * to try and get the best values We first try and use
+		 * src/dwdest for the scale factor, then we move onto src-1
+		 *
+		 * we want to check to see if we will need to clip data, if so
+		 * then we should clip our source so that we don't need to
+		 */
+		if (!ovlLinear) {
+			ulSrcLeft &= ~0x1f;
+
+			/*
+			 * we must align the right hand edge to the next 32
+			 * pixel` boundary, must be on a 256 boundary so u, and
+			 * v are 128 bit aligned
+			 */
+			ulSrcRight = (ulSrcRight + 0x1f) & ~0x1f;
+		} else {
+			ulSrcLeft &= ~0x7;
+
+			/*
+			 * we must align the right hand edge to the next
+			 * 8pixel` boundary
+			 */
+			ulSrcRight = (ulSrcRight + 0x7) & ~0x7;
+		}
+
+		/* this is the input size line store needs to cope with */
+		ulWidth = ulSrcRight - ulSrcLeft;
+
+		/*
+		 * use unclipped value to work out scale factror this is the
+		 * scale factor we want we shall now work out the horizonal
+		 * decimation and scaling
+		 */
+		ulsVal = ((ulWidth / 8) >> ulhDecim);
+
+		if ((ulWidth != (ulsVal << ulhDecim) * 8))
+			ulsAdd = 1;
+
+		/* input pixels to scaler; */
+		ulSrc = ulWidth >> ulhDecim;
+
+		if (ulSrc <= 2)
+			return -EINVAL;
+
+		ulExcessPixels = ((((ulScaleLeft - ulSrcLeft)) << (11 - ulhDecim)) / ulScale);
+
+		ulClip = (ulSrc << 11) / ulScale;
+		ulClip -= (ulRight - ulLeft);
+		ulClip += ulExcessPixels;
+
+		if (ulClip)
+			ulClip--;
+
+		/* We may need to do more here if we really have a HW rev < 5 */
+	} while (!bResult);
+
+	ulExtraLines = (1 << ulhDecim) * ulVertDecFactor;
+	ulExtraLines += 64;
+	ulHeight += ulExtraLines;
+
+	ulDacXScale = ulScale;
+
+
+	tmp = STG_READ_REG(DACVerticalScal);
+	CLEAR_BITS_FRM_TO(0, 11);
+	CLEAR_BITS_FRM_TO(16, 22);	/* Vertical Scaling */
+
+	/* Calculate new output line stride, this is always the number of 422
+	   words in the line buffer, so it doesn't matter if the
+	   mode is 420. Then set the vertical scale register.
+	 */
+	ulStride = (ulWidth >> (ulhDecim + 3)) + ulsAdd;
+	tmp |= ((ulStride << 16) | (ulDacYScale));	/* DAC_LS_CTRL = stride */
+	STG_WRITE_REG(DACVerticalScal, tmp);
+
+	/* Now set up the overlay size using the modified width and height
+	   from decimate and scaling calculations
+	 */
+	tmp = STG_READ_REG(DACOverlaySize);
+	CLEAR_BITS_FRM_TO(0, 10);
+	CLEAR_BITS_FRM_TO(12, 31);
+
+	if (ovlLinear) {
+		tmp |=
+		    (ovlStride | ((ulHeight + 1) << 12) |
+		     (((ulWidth / 8) - 1) << 23));
+	} else {
+		tmp |=
+		    (ovlStride | ((ulHeight + 1) << 12) |
+		     (((ulWidth / 32) - 1) << 23));
+	}
+
+	STG_WRITE_REG(DACOverlaySize, tmp);
+
+	/* Set Video Window Start */
+	tmp = ((ulLeft << 16)) | (srcDest.ulDstY1);
+	STG_WRITE_REG(DACVidWinStart, tmp);
+
+	/* Set Video Window End */
+	tmp = ((ulRight) << 16) | (srcDest.ulDstY2);
+	STG_WRITE_REG(DACVidWinEnd, tmp);
+
+	/* Finally set up the rest of the overlay regs in the order
+	   done in the IMG driver
+	 */
+	tmp = STG_READ_REG(DACPixelFormat);
+	tmp = ((ulExcessPixels << 16) | tmp) & 0x7fffffff;
+	STG_WRITE_REG(DACPixelFormat, tmp);
+
+	tmp = STG_READ_REG(DACHorizontalScal);
+	CLEAR_BITS_FRM_TO(0, 11);
+	CLEAR_BITS_FRM_TO(16, 17);
+	tmp |= ((ulhDecim << 16) | (ulDacXScale));
+	STG_WRITE_REG(DACHorizontalScal, tmp);
+
+	return 0;
+}
diff --git a/drivers/video/kyro/STG4000Ramdac.c b/drivers/video/kyro/STG4000Ramdac.c
new file mode 100644
index 000000000000..e6ad037e4396
--- /dev/null
+++ b/drivers/video/kyro/STG4000Ramdac.c
@@ -0,0 +1,163 @@
+/*
+ *  linux/drivers/video/kyro/STG4000Ramdac.c
+ *
+ *  Copyright (C) 2002 STMicroelectronics
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file COPYING in the main directory of this archive
+ * for more details.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <video/kyro.h>
+
+#include "STG4000Reg.h"
+#include "STG4000Interface.h"
+
+static u32 STG_PIXEL_BUS_WIDTH = 128;	/* 128 bit bus width      */
+static u32 REF_CLOCK = 14318;
+
+int InitialiseRamdac(volatile STG4000REG __iomem * pSTGReg,
+		     u32 displayDepth,
+		     u32 displayWidth,
+		     u32 displayHeight,
+		     s32 HSyncPolarity,
+		     s32 VSyncPolarity, u32 * pixelClock)
+{
+	u32 tmp = 0;
+	u32 F = 0, R = 0, P = 0;
+	u32 stride = 0;
+	u32 ulPdiv = 0;
+	u32 physicalPixelDepth = 0;
+	/* Make sure DAC is in Reset */
+	tmp = STG_READ_REG(SoftwareReset);
+
+	if (tmp & 0x1) {
+		CLEAR_BIT(1);
+		STG_WRITE_REG(SoftwareReset, tmp);
+	}
+
+	/* Set Pixel Format */
+	tmp = STG_READ_REG(DACPixelFormat);
+	CLEAR_BITS_FRM_TO(0, 2);
+
+	/* Set LUT not used from 16bpp to 32 bpp ??? */
+	CLEAR_BITS_FRM_TO(8, 9);
+
+	switch (displayDepth) {
+	case 16:
+		{
+			physicalPixelDepth = 16;
+			tmp |= _16BPP;<