ARM: keystone: move SoC sources to mach-keystone

Move
arch/arm/cpu/armv7/keystone/* -> arch/arm/mach-keystone/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>

14 files changed, 2189 insertions, 0 deletions

diff --git a/arch/arm/mach-keystone/Kconfig b/arch/arm/mach-keystone/Kconfig
new file mode 100644
index 0000000000..134ae87fe1
--- /dev/null
+++ b/arch/arm/mach-keystone/Kconfig
@@ -0,0 +1,22 @@
+if ARCH_KEYSTONE
+
+choice
+	prompt "TI Keystone board select"
+
+config TARGET_K2HK_EVM
+	bool "TI Keystone 2 Kepler/Hawking EVM"
+
+config TARGET_K2E_EVM
+	bool "TI Keystone 2 Edison EVM"
+
+config TARGET_K2L_EVM
+	bool "TI Keystone 2 Lamar EVM"
+
+endchoice
+
+config SYS_SOC
+	default "keystone"
+
+source "board/ti/ks2_evm/Kconfig"
+
+endif
diff --git a/arch/arm/mach-keystone/Makefile b/arch/arm/mach-keystone/Makefile
new file mode 100644
index 0000000000..ed030db2c8
--- /dev/null
+++ b/arch/arm/mach-keystone/Makefile
@@ -0,0 +1,18 @@
+#
+# (C) Copyright 2012-2014
+#     Texas Instruments Incorporated, <www.ti.com>
+#
+# SPDX-License-Identifier:     GPL-2.0+
+#
+
+obj-y	+= init.o
+obj-y	+= psc.o
+obj-y	+= clock.o
+obj-$(CONFIG_SOC_K2HK) += clock-k2hk.o
+obj-$(CONFIG_SOC_K2E) += clock-k2e.o
+obj-$(CONFIG_SOC_K2L) += clock-k2l.o
+obj-y	+= cmd_clock.o
+obj-y	+= cmd_mon.o
+obj-y	+= msmc.o
+obj-y	+= ddr3.o cmd_ddr3.o
+obj-y	+= keystone.o
diff --git a/arch/arm/mach-keystone/clock-k2e.c b/arch/arm/mach-keystone/clock-k2e.c
new file mode 100644
index 0000000000..31f66613ef
--- /dev/null
+++ b/arch/arm/mach-keystone/clock-k2e.c
@@ -0,0 +1,117 @@
+/*
+ * Keystone2: get clk rate for K2E
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/clock_defs.h>
+
+const struct keystone_pll_regs keystone_pll_regs[] = {
+	[CORE_PLL] = {KS2_MAINPLLCTL0, KS2_MAINPLLCTL1},
+	[PASS_PLL] = {KS2_PASSPLLCTL0, KS2_PASSPLLCTL1},
+	[DDR3_PLL] = {KS2_DDR3APLLCTL0, KS2_DDR3APLLCTL1},
+};
+
+int dev_speeds[] = {
+	SPD800,
+	SPD850,
+	SPD1000,
+	SPD1250,
+	SPD1350,
+	SPD1400,
+	SPD1500,
+	SPD1400,
+	SPD1350,
+	SPD1250,
+	SPD1000,
+	SPD850,
+	SPD800
+};
+
+/**
+ * pll_freq_get - get pll frequency
+ * Fout = Fref * NF(mult) / NR(prediv) / OD
+ * @pll:	pll identifier
+ */
+static unsigned long pll_freq_get(int pll)
+{
+	unsigned long mult = 1, prediv = 1, output_div = 2;
+	unsigned long ret;
+	u32 tmp, reg;
+
+	if (pll == CORE_PLL) {
+		ret = external_clk[sys_clk];
+		if (pllctl_reg_read(pll, ctl) & PLLCTL_PLLEN) {
+			/* PLL mode */
+			tmp = __raw_readl(KS2_MAINPLLCTL0);
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = (((tmp & PLLM_MULT_HI_SMASK) >> 6) |
+				(pllctl_reg_read(pll, mult) &
+				PLLM_MULT_LO_MASK)) + 1;
+			output_div = ((pllctl_reg_read(pll, secctl) >>
+				       PLL_CLKOD_SHIFT) & PLL_CLKOD_MASK) + 1;
+
+			ret = ret / prediv / output_div * mult;
+		}
+	} else {
+		switch (pll) {
+		case PASS_PLL:
+			ret = external_clk[pa_clk];
+			reg = KS2_PASSPLLCTL0;
+			break;
+		case DDR3_PLL:
+			ret = external_clk[ddr3_clk];
+			reg = KS2_DDR3APLLCTL0;
+			break;
+		default:
+			return 0;
+		}
+
+		tmp = __raw_readl(reg);
+
+		if (!(tmp & PLLCTL_BYPASS)) {
+			/* Bypass disabled */
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = ((tmp >> PLL_MULT_SHIFT) & PLL_MULT_MASK) + 1;
+			output_div = ((tmp >> PLL_CLKOD_SHIFT) &
+				      PLL_CLKOD_MASK) + 1;
+			ret = ((ret / prediv) * mult) / output_div;
+		}
+	}
+
+	return ret;
+}
+
+unsigned long clk_get_rate(unsigned int clk)
+{
+	switch (clk) {
+	case core_pll_clk:      return pll_freq_get(CORE_PLL);
+	case pass_pll_clk:      return pll_freq_get(PASS_PLL);
+	case ddr3_pll_clk:      return pll_freq_get(DDR3_PLL);
+	case sys_clk0_1_clk:
+	case sys_clk0_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(1);
+	case sys_clk1_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(2);
+	case sys_clk2_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(3);
+	case sys_clk3_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(4);
+	case sys_clk0_2_clk:    return clk_get_rate(sys_clk0_clk) / 2;
+	case sys_clk0_3_clk:    return clk_get_rate(sys_clk0_clk) / 3;
+	case sys_clk0_4_clk:    return clk_get_rate(sys_clk0_clk) / 4;
+	case sys_clk0_6_clk:    return clk_get_rate(sys_clk0_clk) / 6;
+	case sys_clk0_8_clk:    return clk_get_rate(sys_clk0_clk) / 8;
+	case sys_clk0_12_clk:   return clk_get_rate(sys_clk0_clk) / 12;
+	case sys_clk0_24_clk:   return clk_get_rate(sys_clk0_clk) / 24;
+	case sys_clk1_3_clk:    return clk_get_rate(sys_clk1_clk) / 3;
+	case sys_clk1_4_clk:    return clk_get_rate(sys_clk1_clk) / 4;
+	case sys_clk1_6_clk:    return clk_get_rate(sys_clk1_clk) / 6;
+	case sys_clk1_12_clk:   return clk_get_rate(sys_clk1_clk) / 12;
+	default:
+		break;
+	}
+
+	return 0;
+}
diff --git a/arch/arm/mach-keystone/clock-k2hk.c b/arch/arm/mach-keystone/clock-k2hk.c
new file mode 100644
index 0000000000..1591960795
--- /dev/null
+++ b/arch/arm/mach-keystone/clock-k2hk.c
@@ -0,0 +1,145 @@
+/*
+ * Keystone2: get clk rate for K2HK
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/clock_defs.h>
+
+const struct keystone_pll_regs keystone_pll_regs[] = {
+	[CORE_PLL]	= {KS2_MAINPLLCTL0, KS2_MAINPLLCTL1},
+	[PASS_PLL]	= {KS2_PASSPLLCTL0, KS2_PASSPLLCTL1},
+	[TETRIS_PLL]	= {KS2_ARMPLLCTL0, KS2_ARMPLLCTL1},
+	[DDR3A_PLL]	= {KS2_DDR3APLLCTL0, KS2_DDR3APLLCTL1},
+	[DDR3B_PLL]	= {KS2_DDR3BPLLCTL0, KS2_DDR3BPLLCTL1},
+};
+
+int dev_speeds[] = {
+	SPD800,
+	SPD1000,
+	SPD1200,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD1200,
+	SPD1000,
+	SPD800,
+	SPD800,
+	SPD800,
+};
+
+int arm_speeds[] = {
+	SPD800,
+	SPD1000,
+	SPD1200,
+	SPD1350,
+	SPD1400,
+	SPD800,
+	SPD1400,
+	SPD1350,
+	SPD1200,
+	SPD1000,
+	SPD800,
+	SPD800,
+	SPD800,
+};
+
+/**
+ * pll_freq_get - get pll frequency
+ * Fout = Fref * NF(mult) / NR(prediv) / OD
+ * @pll:	pll identifier
+ */
+static unsigned long pll_freq_get(int pll)
+{
+	unsigned long mult = 1, prediv = 1, output_div = 2;
+	unsigned long ret;
+	u32 tmp, reg;
+
+	if (pll == CORE_PLL) {
+		ret = external_clk[sys_clk];
+		if (pllctl_reg_read(pll, ctl) & PLLCTL_PLLEN) {
+			/* PLL mode */
+			tmp = __raw_readl(KS2_MAINPLLCTL0);
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = (((tmp & PLLM_MULT_HI_SMASK) >> 6) |
+				(pllctl_reg_read(pll, mult) &
+				 PLLM_MULT_LO_MASK)) + 1;
+			output_div = ((pllctl_reg_read(pll, secctl) >>
+				       PLL_CLKOD_SHIFT) & PLL_CLKOD_MASK) + 1;
+
+			ret = ret / prediv / output_div * mult;
+		}
+	} else {
+		switch (pll) {
+		case PASS_PLL:
+			ret = external_clk[pa_clk];
+			reg = KS2_PASSPLLCTL0;
+			break;
+		case TETRIS_PLL:
+			ret = external_clk[tetris_clk];
+			reg = KS2_ARMPLLCTL0;
+			break;
+		case DDR3A_PLL:
+			ret = external_clk[ddr3a_clk];
+			reg = KS2_DDR3APLLCTL0;
+			break;
+		case DDR3B_PLL:
+			ret = external_clk[ddr3b_clk];
+			reg = KS2_DDR3BPLLCTL0;
+			break;
+		default:
+			return 0;
+		}
+
+		tmp = __raw_readl(reg);
+
+		if (!(tmp & PLLCTL_BYPASS)) {
+			/* Bypass disabled */
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = ((tmp >> PLL_MULT_SHIFT) & PLL_MULT_MASK) + 1;
+			output_div = ((tmp >> PLL_CLKOD_SHIFT) &
+				      PLL_CLKOD_MASK) + 1;
+			ret = ((ret / prediv) * mult) / output_div;
+		}
+	}
+
+	return ret;
+}
+
+unsigned long clk_get_rate(unsigned int clk)
+{
+	switch (clk) {
+	case core_pll_clk:	return pll_freq_get(CORE_PLL);
+	case pass_pll_clk:	return pll_freq_get(PASS_PLL);
+	case tetris_pll_clk:	return pll_freq_get(TETRIS_PLL);
+	case ddr3a_pll_clk:	return pll_freq_get(DDR3A_PLL);
+	case ddr3b_pll_clk:	return pll_freq_get(DDR3B_PLL);
+	case sys_clk0_1_clk:
+	case sys_clk0_clk:	return pll_freq_get(CORE_PLL) / pll0div_read(1);
+	case sys_clk1_clk:	return pll_freq_get(CORE_PLL) / pll0div_read(2);
+	case sys_clk2_clk:	return pll_freq_get(CORE_PLL) / pll0div_read(3);
+	case sys_clk3_clk:	return pll_freq_get(CORE_PLL) / pll0div_read(4);
+	case sys_clk0_2_clk:	return clk_get_rate(sys_clk0_clk) / 2;
+	case sys_clk0_3_clk:	return clk_get_rate(sys_clk0_clk) / 3;
+	case sys_clk0_4_clk:	return clk_get_rate(sys_clk0_clk) / 4;
+	case sys_clk0_6_clk:	return clk_get_rate(sys_clk0_clk) / 6;
+	case sys_clk0_8_clk:	return clk_get_rate(sys_clk0_clk) / 8;
+	case sys_clk0_12_clk:	return clk_get_rate(sys_clk0_clk) / 12;
+	case sys_clk0_24_clk:	return clk_get_rate(sys_clk0_clk) / 24;
+	case sys_clk1_3_clk:	return clk_get_rate(sys_clk1_clk) / 3;
+	case sys_clk1_4_clk:	return clk_get_rate(sys_clk1_clk) / 4;
+	case sys_clk1_6_clk:	return clk_get_rate(sys_clk1_clk) / 6;
+	case sys_clk1_12_clk:	return clk_get_rate(sys_clk1_clk) / 12;
+	default:
+		break;
+	}
+
+	return 0;
+}
diff --git a/arch/arm/mach-keystone/clock-k2l.c b/arch/arm/mach-keystone/clock-k2l.c
new file mode 100644
index 0000000000..1c5e4d54d8
--- /dev/null
+++ b/arch/arm/mach-keystone/clock-k2l.c
@@ -0,0 +1,138 @@
+/*
+ * Keystone2: get clk rate for K2L
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/clock_defs.h>
+
+const struct keystone_pll_regs keystone_pll_regs[] = {
+	[CORE_PLL] = {KS2_MAINPLLCTL0, KS2_MAINPLLCTL1},
+	[PASS_PLL] = {KS2_PASSPLLCTL0, KS2_PASSPLLCTL1},
+	[TETRIS_PLL] = {KS2_ARMPLLCTL0,  KS2_ARMPLLCTL1},
+	[DDR3_PLL] = {KS2_DDR3APLLCTL0, KS2_DDR3APLLCTL1},
+};
+
+int dev_speeds[] = {
+	SPD800,
+	SPD1000,
+	SPD1200,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD800,
+	SPD1200,
+	SPD1000,
+	SPD800,
+	SPD800,
+	SPD800,
+};
+
+int arm_speeds[] = {
+	SPD800,
+	SPD1000,
+	SPD1200,
+	SPD1350,
+	SPD1400,
+	SPD800,
+	SPD1400,
+	SPD1350,
+	SPD1200,
+	SPD1000,
+	SPD800,
+	SPD800,
+	SPD800,
+};
+
+/**
+ * pll_freq_get - get pll frequency
+ * Fout = Fref * NF(mult) / NR(prediv) / OD
+ * @pll:	pll identifier
+ */
+static unsigned long pll_freq_get(int pll)
+{
+	unsigned long mult = 1, prediv = 1, output_div = 2;
+	unsigned long ret;
+	u32 tmp, reg;
+
+	if (pll == CORE_PLL) {
+		ret = external_clk[sys_clk];
+		if (pllctl_reg_read(pll, ctl) & PLLCTL_PLLEN) {
+			/* PLL mode */
+			tmp = __raw_readl(KS2_MAINPLLCTL0);
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = (((tmp & PLLM_MULT_HI_SMASK) >> 6) |
+				(pllctl_reg_read(pll, mult) &
+				PLLM_MULT_LO_MASK)) + 1;
+			output_div = ((pllctl_reg_read(pll, secctl) >>
+					PLL_CLKOD_SHIFT) & PLL_CLKOD_MASK) + 1;
+
+			ret = ret / prediv / output_div * mult;
+		}
+	} else {
+		switch (pll) {
+		case PASS_PLL:
+			ret = external_clk[pa_clk];
+			reg = KS2_PASSPLLCTL0;
+			break;
+		case TETRIS_PLL:
+			ret = external_clk[tetris_clk];
+			reg = KS2_ARMPLLCTL0;
+			break;
+		case DDR3_PLL:
+			ret = external_clk[ddr3_clk];
+			reg = KS2_DDR3APLLCTL0;
+			break;
+		default:
+			return 0;
+		}
+
+		tmp = __raw_readl(reg);
+		if (!(tmp & PLLCTL_BYPASS)) {
+			/* Bypass disabled */
+			prediv = (tmp & PLL_DIV_MASK) + 1;
+			mult = ((tmp >> PLL_MULT_SHIFT) & PLL_MULT_MASK) + 1;
+			output_div = ((tmp >> PLL_CLKOD_SHIFT) &
+				      PLL_CLKOD_MASK) + 1;
+			ret = ((ret / prediv) * mult) / output_div;
+		}
+	}
+
+	return ret;
+}
+
+unsigned long clk_get_rate(unsigned int clk)
+{
+	switch (clk) {
+	case core_pll_clk:      return pll_freq_get(CORE_PLL);
+	case pass_pll_clk:      return pll_freq_get(PASS_PLL);
+	case tetris_pll_clk:    return pll_freq_get(TETRIS_PLL);
+	case ddr3_pll_clk:      return pll_freq_get(DDR3_PLL);
+	case sys_clk0_1_clk:
+	case sys_clk0_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(1);
+	case sys_clk1_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(2);
+	case sys_clk2_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(3);
+	case sys_clk3_clk:      return pll_freq_get(CORE_PLL) / pll0div_read(4);
+	case sys_clk0_2_clk:    return clk_get_rate(sys_clk0_clk) / 2;
+	case sys_clk0_3_clk:    return clk_get_rate(sys_clk0_clk) / 3;
+	case sys_clk0_4_clk:    return clk_get_rate(sys_clk0_clk) / 4;
+	case sys_clk0_6_clk:    return clk_get_rate(sys_clk0_clk) / 6;
+	case sys_clk0_8_clk:    return clk_get_rate(sys_clk0_clk) / 8;
+	case sys_clk0_12_clk:   return clk_get_rate(sys_clk0_clk) / 12;
+	case sys_clk0_24_clk:   return clk_get_rate(sys_clk0_clk) / 24;
+	case sys_clk1_3_clk:    return clk_get_rate(sys_clk1_clk) / 3;
+	case sys_clk1_4_clk:    return clk_get_rate(sys_clk1_clk) / 4;
+	case sys_clk1_6_clk:    return clk_get_rate(sys_clk1_clk) / 6;
+	case sys_clk1_12_clk:   return clk_get_rate(sys_clk1_clk) / 12;
+	default:
+		break;
+	}
+
+	return 0;
+}
diff --git a/arch/arm/mach-keystone/clock.c b/arch/arm/mach-keystone/clock.c
new file mode 100644
index 0000000000..d13fbc1a4b
--- /dev/null
+++ b/arch/arm/mach-keystone/clock.c
@@ -0,0 +1,272 @@
+/*
+ * Keystone2: pll initialization
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/clock_defs.h>
+
+#define MAX_SPEEDS		13
+
+static void wait_for_completion(const struct pll_init_data *data)
+{
+	int i;
+	for (i = 0; i < 100; i++) {
+		sdelay(450);
+		if ((pllctl_reg_read(data->pll, stat) & PLLSTAT_GO) == 0)
+			break;
+	}
+}
+
+void init_pll(const struct pll_init_data *data)
+{
+	u32 tmp, tmp_ctl, pllm, plld, pllod, bwadj;
+
+	pllm = data->pll_m - 1;
+	plld = (data->pll_d - 1) & PLL_DIV_MASK;
+	pllod = (data->pll_od - 1) & PLL_CLKOD_MASK;
+
+	if (data->pll == MAIN_PLL) {
+		/* The requered delay before main PLL configuration */
+		sdelay(210000);
+
+		tmp = pllctl_reg_read(data->pll, secctl);
+
+		if (tmp & (PLLCTL_BYPASS)) {
+			setbits_le32(keystone_pll_regs[data->pll].reg1,
+				     BIT(MAIN_ENSAT_OFFSET));
+
+			pllctl_reg_clrbits(data->pll, ctl, PLLCTL_PLLEN |
+					   PLLCTL_PLLENSRC);
+			sdelay(340);
+
+			pllctl_reg_setbits(data->pll, secctl, PLLCTL_BYPASS);
+			pllctl_reg_setbits(data->pll, ctl, PLLCTL_PLLPWRDN);
+			sdelay(21000);
+
+			pllctl_reg_clrbits(data->pll, ctl, PLLCTL_PLLPWRDN);
+		} else {
+			pllctl_reg_clrbits(data->pll, ctl, PLLCTL_PLLEN |
+					   PLLCTL_PLLENSRC);
+			sdelay(340);
+		}
+
+		pllctl_reg_write(data->pll, mult, pllm & PLLM_MULT_LO_MASK);
+
+		clrsetbits_le32(keystone_pll_regs[data->pll].reg0,
+				PLLM_MULT_HI_SMASK, (pllm << 6));
+
+		/* Set the BWADJ     (12 bit field)  */
+		tmp_ctl = pllm >> 1; /* Divide the pllm by 2 */
+		clrsetbits_le32(keystone_pll_regs[data->pll].reg0,
+				PLL_BWADJ_LO_SMASK,
+				(tmp_ctl << PLL_BWADJ_LO_SHIFT));
+		clrsetbits_le32(keystone_pll_regs[data->pll].reg1,
+				PLL_BWADJ_HI_MASK,
+				(tmp_ctl >> 8));
+
+		/*
+		 * Set the pll divider (6 bit field) *
+		 * PLLD[5:0] is located in MAINPLLCTL0
+		 */
+		clrsetbits_le32(keystone_pll_regs[data->pll].reg0,
+				PLL_DIV_MASK, plld);
+
+		/* Set the OUTPUT DIVIDE (4 bit field) in SECCTL */
+		pllctl_reg_rmw(data->pll, secctl, PLL_CLKOD_SMASK,
+			       (pllod << PLL_CLKOD_SHIFT));
+		wait_for_completion(data);
+
+		pllctl_reg_write(data->pll, div1, PLLM_RATIO_DIV1);
+		pllctl_reg_write(data->pll, div2, PLLM_RATIO_DIV2);
+		pllctl_reg_write(data->pll, div3, PLLM_RATIO_DIV3);
+		pllctl_reg_write(data->pll, div4, PLLM_RATIO_DIV4);
+		pllctl_reg_write(data->pll, div5, PLLM_RATIO_DIV5);
+
+		pllctl_reg_setbits(data->pll, alnctl, 0x1f);
+
+		/*
+		 * Set GOSET bit in PLLCMD to initiate the GO operation
+		 * to change the divide
+		 */
+		pllctl_reg_setbits(data->pll, cmd, PLLSTAT_GO);
+		sdelay(1500); /* wait for the phase adj */
+		wait_for_completion(data);
+
+		/* Reset PLL */
+		pllctl_reg_setbits(data->pll, ctl, PLLCTL_PLLRST);
+		sdelay(21000);	/* Wait for a minimum of 7 us*/
+		pllctl_reg_clrbits(data->pll, ctl, PLLCTL_PLLRST);
+		sdelay(105000);	/* Wait for PLL Lock time (min 50 us) */
+
+		pllctl_reg_clrbits(data->pll, secctl, PLLCTL_BYPASS);
+
+		tmp = pllctl_reg_setbits(data->pll, ctl, PLLCTL_PLLEN);
+
+#ifndef CONFIG_SOC_K2E
+	} else if (data->pll == TETRIS_PLL) {
+		bwadj = pllm >> 1;
+		/* 1.5 Set PLLCTL0[BYPASS] =1 (enable bypass), */
+		setbits_le32(keystone_pll_regs[data->pll].reg0,  PLLCTL_BYPASS);
+		/*
+		 * Set CHIPMISCCTL1[13] = 0 (enable glitchfree bypass)
+		 * only applicable for Kepler
+		 */
+		clrbits_le32(KS2_MISC_CTRL, KS2_ARM_PLL_EN);
+		/* 2 In PLLCTL1, write PLLRST = 1 (PLL is reset) */
+		setbits_le32(keystone_pll_regs[data->pll].reg1 ,
+			     PLL_PLLRST | PLLCTL_ENSAT);
+
+		/*
+		 * 3 Program PLLM and PLLD in PLLCTL0 register
+		 * 4 Program BWADJ[7:0] in PLLCTL0 and BWADJ[11:8] in
+		 * PLLCTL1 register. BWADJ value must be set
+		 * to ((PLLM + 1) >> 1) – 1)
+		 */
+		tmp = ((bwadj & PLL_BWADJ_LO_MASK) << PLL_BWADJ_LO_SHIFT) |
+			(pllm << 6) |
+			(plld & PLL_DIV_MASK) |
+			(pllod << PLL_CLKOD_SHIFT) | PLLCTL_BYPASS;
+		__raw_writel(tmp, keystone_pll_regs[data->pll].reg0);
+
+		/* Set BWADJ[11:8] bits */
+		tmp = __raw_readl(keystone_pll_regs[data->pll].reg1);
+		tmp &= ~(PLL_BWADJ_HI_MASK);
+		tmp |= ((bwadj>>8) & PLL_BWADJ_HI_MASK);
+		__raw_writel(tmp, keystone_pll_regs[data->pll].reg1);
+		/*
+		 * 5 Wait for at least 5 us based on the reference
+		 * clock (PLL reset time)
+		 */
+		sdelay(21000);	/* Wait for a minimum of 7 us*/
+
+		/* 6 In PLLCTL1, write PLLRST = 0 (PLL reset is released) */
+		clrbits_le32(keystone_pll_regs[data->pll].reg1, PLL_PLLRST);
+		/*
+		 * 7 Wait for at least 500 * REFCLK cycles * (PLLD + 1)
+		 * (PLL lock time)
+		 */
+		sdelay(105000);
+		/* 8 disable bypass */
+		clrbits_le32(keystone_pll_regs[data->pll].reg0, PLLCTL_BYPASS);
+		/*
+		 * 9 Set CHIPMISCCTL1[13] = 1 (disable glitchfree bypass)
+		 * only applicable for Kepler
+		 */
+		setbits_le32(KS2_MISC_CTRL, KS2_ARM_PLL_EN);
+#endif
+	} else {
+		setbits_le32(keystone_pll_regs[data->pll].reg1, PLLCTL_ENSAT);
+		/*
+		 * process keeps state of Bypass bit while programming
+		 * all other DDR PLL settings
+		 */
+		tmp = __raw_readl(keystone_pll_regs[data->pll].reg0);
+		tmp &= PLLCTL_BYPASS;	/* clear everything except Bypass */
+
+		/*
+		 * Set the BWADJ[7:0], PLLD[5:0] and PLLM to PLLCTL0,
+		 * bypass disabled
+		 */
+		bwadj = pllm >> 1;
+		tmp |= ((bwadj & PLL_BWADJ_LO_MASK) << PLL_BWADJ_LO_SHIFT) |
+			(pllm << PLL_MULT_SHIFT) |
+			(plld & PLL_DIV_MASK) |
+			(pllod << PLL_CLKOD_SHIFT);
+		__raw_writel(tmp, keystone_pll_regs[data->pll].reg0);
+
+		/* Set BWADJ[11:8] bits */
+		tmp = __raw_readl(keystone_pll_regs[data->pll].reg1);
+		tmp &= ~(PLL_BWADJ_HI_MASK);
+		tmp |= ((bwadj >> 8) & PLL_BWADJ_HI_MASK);
+
+		__raw_writel(tmp, keystone_pll_regs[data->pll].reg1);
+
+		/* Reset bit: bit 14 for both DDR3 & PASS PLL */
+		tmp = PLL_PLLRST;
+		/* Set RESET bit = 1 */
+		setbits_le32(keystone_pll_regs[data->pll].reg1, tmp);
+		/* Wait for a minimum of 7 us*/
+		sdelay(21000);
+		/* Clear RESET bit */
+		clrbits_le32(keystone_pll_regs[data->pll].reg1, tmp);
+		sdelay(105000);
+
+		/* clear BYPASS (Enable PLL Mode) */
+		clrbits_le32(keystone_pll_regs[data->pll].reg0, PLLCTL_BYPASS);
+		sdelay(21000);	/* Wait for a minimum of 7 us*/
+	}
+
+	/*
+	 * This is required to provide a delay between multiple
+	 * consequent PPL configurations
+	 */
+	sdelay(210000);
+}
+
+void init_plls(int num_pll, struct pll_init_data *config)
+{
+	int i;
+
+	for (i = 0; i < num_pll; i++)
+		init_pll(&config[i]);
+}
+
+static int get_max_speed(u32 val, int *speeds)
+{
+	int j;
+
+	if (!val)
+		return speeds[0];
+
+	for (j = 1; j < MAX_SPEEDS; j++) {
+		if (val == 1)
+			return speeds[j];
+		val >>= 1;
+	}
+
+	return SPD800;
+}
+
+#ifdef CONFIG_SOC_K2HK
+static u32 read_efuse_bootrom(void)
+{
+	return (cpu_revision() > 1) ? __raw_readl(KS2_EFUSE_BOOTROM) :
+		__raw_readl(KS2_REV1_DEVSPEED);
+}
+#else
+static inline u32 read_efuse_bootrom(void)
+{
+	return __raw_readl(KS2_EFUSE_BOOTROM);
+}
+#endif
+
+inline int get_max_dev_speed(void)
+{
+	return get_max_speed(read_efuse_bootrom() & 0xffff, dev_speeds);
+}
+
+#ifndef CONFIG_SOC_K2E
+inline int get_max_arm_speed(void)
+{
+	return get_max_speed((read_efuse_bootrom() >> 16) & 0xffff, arm_speeds);
+}
+#endif
+
+void pass_pll_pa_clk_enable(void)
+{
+	u32 reg;
+
+	reg = readl(keystone_pll_regs[PASS_PLL].reg1);
+
+	reg |= PLLCTL_PAPLL;
+	writel(reg, keystone_pll_regs[PASS_PLL].reg1);
+
+	/* wait till clock is enabled */
+	sdelay(15000);
+}
diff --git a/arch/arm/mach-keystone/cmd_clock.c b/arch/arm/mach-keystone/cmd_clock.c
new file mode 100644
index 0000000000..af1b701e82
--- /dev/null
+++ b/arch/arm/mach-keystone/cmd_clock.c
@@ -0,0 +1,135 @@
+/*
+ * keystone2: commands for clocks
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <command.h>
+#include <asm/arch/hardware.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/psc_defs.h>
+
+struct pll_init_data cmd_pll_data = {
+	.pll = MAIN_PLL,
+	.pll_m = 16,
+	.pll_d = 1,
+	.pll_od = 2,
+};
+
+int do_pll_cmd(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
+{
+	if (argc != 5)
+		goto pll_cmd_usage;
+
+	if (strncmp(argv[1], "pa", 2) == 0)
+		cmd_pll_data.pll = PASS_PLL;
+#ifndef CONFIG_SOC_K2E
+	else if (strncmp(argv[1], "arm", 3) == 0)
+		cmd_pll_data.pll = TETRIS_PLL;
+#endif
+#ifdef CONFIG_SOC_K2HK
+	else if (strncmp(argv[1], "ddr3a", 5) == 0)
+		cmd_pll_data.pll = DDR3A_PLL;
+	else if (strncmp(argv[1], "ddr3b", 5) == 0)
+		cmd_pll_data.pll = DDR3B_PLL;
+#else
+	else if (strncmp(argv[1], "ddr3", 4) == 0)
+		cmd_pll_data.pll = DDR3_PLL;
+#endif
+	else
+		goto pll_cmd_usage;
+
+	cmd_pll_data.pll_m   = simple_strtoul(argv[2], NULL, 10);
+	cmd_pll_data.pll_d   = simple_strtoul(argv[3], NULL, 10);
+	cmd_pll_data.pll_od  = simple_strtoul(argv[4], NULL, 10);
+
+	printf("Trying to set pll %d; mult %d; div %d; OD %d\n",
+	       cmd_pll_data.pll, cmd_pll_data.pll_m,
+	       cmd_pll_data.pll_d, cmd_pll_data.pll_od);
+	init_pll(&cmd_pll_data);
+
+	return 0;
+
+pll_cmd_usage:
+	return cmd_usage(cmdtp);
+}
+
+U_BOOT_CMD(
+	pllset, 5,      0,      do_pll_cmd,
+	"set pll multiplier and pre divider",
+	PLLSET_CMD_LIST " <mult> <div> <OD>\n"
+);
+
+int do_getclk_cmd(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
+{
+	unsigned int clk;
+	unsigned int freq;
+
+	if (argc != 2)
+		goto getclk_cmd_usage;
+
+	clk = simple_strtoul(argv[1], NULL, 10);
+
+	freq = clk_get_rate(clk);
+	printf("clock index [%d] - frequency %u\n", clk, freq);
+	return 0;
+
+getclk_cmd_usage:
+	return cmd_usage(cmdtp);
+}
+
+U_BOOT_CMD(
+	getclk,	2,	0,	do_getclk_cmd,
+	"get clock rate",
+	"<clk index>\n"
+	"The indexes for clocks:\n"
+	CLOCK_INDEXES_LIST
+);
+
+int do_psc_cmd(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
+{
+	int	psc_module;
+	int	res;
+
+	if (argc != 3)
+		goto psc_cmd_usage;
+
+	psc_module = simple_strtoul(argv[1], NULL, 10);
+	if (strcmp(argv[2], "en") == 0) {
+		res = psc_enable_module(psc_module);
+		printf("psc_enable_module(%d) - %s\n", psc_module,
+		       (res) ? "ERROR" : "OK");
+		return 0;
+	}
+
+	if (strcmp(argv[2], "di") == 0) {
+		res = psc_disable_module(psc_module);
+		printf("psc_disable_module(%d) - %s\n", psc_module,
+		       (res) ? "ERROR" : "OK");
+		return 0;
+	}
+
+	if (strcmp(argv[2], "domain") == 0) {
+		res = psc_disable_domain(psc_module);
+		printf("psc_disable_domain(%d) - %s\n", psc_module,
+		       (res) ? "ERROR" : "OK");
+		return 0;
+	}
+
+psc_cmd_usage:
+	return cmd_usage(cmdtp);
+}
+
+U_BOOT_CMD(
+	psc,	3,	0,	do_psc_cmd,
+	"<enable/disable psc module os disable domain>",
+	"<mod/domain index> <en|di|domain>\n"
+	"Intended to control Power and Sleep Controller (PSC) domains and\n"
+	"modules. The module or domain index exectly corresponds to ones\n"
+	"listed in official TRM. For instance, to enable MSMC RAM clock\n"
+	"domain use command: psc 14 en.\n"
+);
diff --git a/arch/arm/mach-keystone/cmd_ddr3.c b/arch/arm/mach-keystone/cmd_ddr3.c
new file mode 100644
index 0000000000..ea78ad8fd5
--- /dev/null
+++ b/arch/arm/mach-keystone/cmd_ddr3.c
@@ -0,0 +1,248 @@
+/*
+ * Keystone2: DDR3 test commands
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <asm/arch/hardware.h>
+#include <asm/arch/ddr3.h>
+#include <common.h>
+#include <command.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+#define DDR_MIN_ADDR		CONFIG_SYS_SDRAM_BASE
+
+#define DDR_REMAP_ADDR		0x80000000
+#define ECC_START_ADDR1		((DDR_MIN_ADDR - DDR_REMAP_ADDR) >> 17)
+
+#define ECC_END_ADDR1		(((gd->start_addr_sp - DDR_REMAP_ADDR - \
+				 CONFIG_STACKSIZE) >> 17) - 2)
+
+#define DDR_TEST_BURST_SIZE	1024
+
+static int ddr_memory_test(u32 start_address, u32 end_address, int quick)
+{
+	u32 index_start, value, index;
+
+	index_start = start_address;
+
+	while (1) {
+		/* Write a pattern */
+		for (index = index_start;
+				index < index_start + DDR_TEST_BURST_SIZE;
+				index += 4)
+			__raw_writel(index, index);
+
+		/* Read and check the pattern */
+		for (index = index_start;
+				index < index_start + DDR_TEST_BURST_SIZE;
+				index += 4) {
+			value = __raw_readl(index);
+			if (value != index) {
+				printf("ddr_memory_test: Failed at address index = 0x%x value = 0x%x *(index) = 0x%x\n",
+				       index, value, __raw_readl(index));
+
+				return -1;
+			}
+		}
+
+		index_start += DDR_TEST_BURST_SIZE;
+		if (index_start >= end_address)
+			break;
+
+		if (quick)
+			continue;
+
+		/* Write a pattern for complementary values */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 4)
+			__raw_writel((u32)~index, index);
+
+		/* Read and check the pattern */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 4) {
+			value = __raw_readl(index);
+			if (value != ~index) {
+				printf("ddr_memory_test: Failed at address index = 0x%x value = 0x%x *(index) = 0x%x\n",
+				       index, value, __raw_readl(index));
+
+				return -1;
+			}
+		}
+
+		index_start += DDR_TEST_BURST_SIZE;
+		if (index_start >= end_address)
+			break;
+
+		/* Write a pattern */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 2)
+			__raw_writew((u16)index, index);
+
+		/* Read and check the pattern */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 2) {
+			value = __raw_readw(index);
+			if (value != (u16)index) {
+				printf("ddr_memory_test: Failed at address index = 0x%x value = 0x%x *(index) = 0x%x\n",
+				       index, value, __raw_readw(index));
+
+				return -1;
+			}
+		}
+
+		index_start += DDR_TEST_BURST_SIZE;
+		if (index_start >= end_address)
+			break;
+
+		/* Write a pattern */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 1)
+			__raw_writeb((u8)index, index);
+
+		/* Read and check the pattern */
+		for (index = index_start;
+		     index < index_start + DDR_TEST_BURST_SIZE;
+		     index += 1) {
+			value = __raw_readb(index);
+			if (value != (u8)index) {
+				printf("ddr_memory_test: Failed at address index = 0x%x value = 0x%x *(index) = 0x%x\n",
+				       index, value, __raw_readb(index));
+
+				return -1;
+			}
+		}
+
+		index_start += DDR_TEST_BURST_SIZE;
+		if (index_start >= end_address)
+			break;
+	}
+
+	puts("ddr memory test PASSED!\n");
+	return 0;
+}
+
+static int ddr_memory_compare(u32 address1, u32 address2, u32 size)
+{
+	u32 index, value, index2, value2;
+
+	for (index = address1, index2 = address2;
+	     index < address1 + size;
+	     index += 4, index2 += 4) {
+		value = __raw_readl(index);
+		value2 = __raw_readl(index2);
+
+		if (value != value2) {
+			printf("ddr_memory_test: Compare failed at address = 0x%x value = 0x%x, address2 = 0x%x value2 = 0x%x\n",
+			       index, value, index2, value2);
+
+			return -1;
+		}
+	}
+
+	puts("ddr memory compare PASSED!\n");
+	return 0;
+}
+
+static int ddr_memory_ecc_err(u32 base, u32 address, u32 ecc_err)
+{
+	u32 value1, value2, value3;
+
+	puts("Disabling DDR ECC ...\n");
+	ddr3_disable_ecc(base);
+
+	value1 = __raw_readl(address);
+	value2 = value1 ^ ecc_err;
+	__raw_writel(value2, address);
+
+	value3 = __raw_readl(address);
+	printf("ECC err test, addr 0x%x, read data 0x%x, wrote data 0x%x, err pattern: 0x%x, read after write data 0x%x\n",
+	       address, value1, value2, ecc_err, value3);
+
+	__raw_writel(ECC_START_ADDR1 | (ECC_END_ADDR1 << 16),
+		     base + KS2_DDR3_ECC_ADDR_RANGE1_OFFSET);
+
+	puts("Enabling DDR ECC ...\n");
+	ddr3_enable_ecc(base, 1);
+
+	value1 = __raw_readl(address);
+	printf("ECC err test, addr 0x%x, read data 0x%x\n", address, value1);
+
+	ddr3_check_ecc_int(base);
+	return 0;
+}
+
+static int do_ddr_test(cmd_tbl_t *cmdtp,
+		       int flag, int argc, char * const argv[])
+{
+	u32 start_addr, end_addr, size, ecc_err;
+
+	if ((argc == 4) && (strncmp(argv[1], "ecc_err", 8) == 0)) {
+		if (!ddr3_ecc_support_rmw(KS2_DDR3A_EMIF_CTRL_BASE)) {
+			puts("ECC RMW isn't supported for this SOC\n");
+			return 1;
+		}
+
+		start_addr = simple_strtoul(argv[2], NULL, 16);
+		ecc_err = simple_strtoul(argv[3], NULL, 16);
+
+		if ((start_addr < CONFIG_SYS_SDRAM_BASE) ||
+		    (start_addr > (CONFIG_SYS_SDRAM_BASE +
+		     CONFIG_MAX_RAM_BANK_SIZE - 1))) {
+			puts("Invalid address!\n");
+			return cmd_usage(cmdtp);
+		}
+
+		ddr_memory_ecc_err(KS2_DDR3A_EMIF_CTRL_BASE,
+				   start_addr, ecc_err);
+		return 0;
+	}
+
+	if (!(((argc == 4) && (strncmp(argv[1], "test", 5) == 0)) ||
+	      ((argc == 5) && (strncmp(argv[1], "compare", 8) == 0))))
+		return cmd_usage(cmdtp);
+
+	start_addr = simple_strtoul(argv[2], NULL, 16);
+	end_addr = simple_strtoul(argv[3], NULL, 16);
+
+	if ((start_addr < CONFIG_SYS_SDRAM_BASE) ||
+	    (start_addr > (CONFIG_SYS_SDRAM_BASE +
+	     CONFIG_MAX_RAM_BANK_SIZE - 1)) ||
+	    (end_addr < CONFIG_SYS_SDRAM_BASE) ||
+	    (end_addr > (CONFIG_SYS_SDRAM_BASE +
+	     CONFIG_MAX_RAM_BANK_SIZE - 1)) || (start_addr >= end_addr)) {
+		puts("Invalid start or end address!\n");
+		return cmd_usage(cmdtp);
+	}
+
+	puts("Please wait ...\n");
+	if (argc == 5) {
+		size = simple_strtoul(argv[4], NULL, 16);
+		ddr_memory_compare(start_addr, end_addr, size);
+	} else {
+		ddr_memory_test(start_addr, end_addr, 0);
+	}
+
+	return 0;
+}
+
+U_BOOT_CMD(ddr,	5, 1, do_ddr_test,
+	   "DDR3 test",
+	   "test <start_addr in hex> <end_addr in hex> - test DDR from start\n"
+	   "	address to end address\n"
+	   "ddr compare <start_addr in hex> <end_addr in hex> <size in hex> -\n"
+	   "	compare DDR data of (size) bytes from start address to end\n"
+	   "	address\n"
+	   "ddr ecc_err <addr in hex> <bit_err in hex> - generate bit errors\n"
+	   "	in DDR data at <addr>, the command will read a 32-bit data\n"
+	   "	from <addr>, and write (data ^ bit_err) back to <addr>\n"
+);
diff --git a/arch/arm/mach-keystone/cmd_mon.c b/arch/arm/mach-keystone/cmd_mon.c
new file mode 100644
index 0000000000..f9f58a37df
--- /dev/null
+++ b/arch/arm/mach-keystone/cmd_mon.c
@@ -0,0 +1,131 @@
+/*
+ * K2HK: secure kernel command file
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <command.h>
+asm(".arch_extension sec\n\t");
+
+static int mon_install(u32 addr, u32 dpsc, u32 freq)
+{
+	int result;
+
+	__asm__ __volatile__ (
+		"stmfd r13!, {lr}\n"
+		"mov r0, %1\n"
+		"mov r1, %2\n"
+		"mov r2, %3\n"
+		"blx r0\n"
+		"ldmfd r13!, {lr}\n"
+		: "=&r" (result)
+		: "r" (addr), "r" (dpsc), "r" (freq)
+		: "cc", "r0", "r1", "r2", "memory");
+	return result;
+}
+
+static int do_mon_install(cmd_tbl_t *cmdtp, int flag, int argc,
+			  char * const argv[])
+{
+	u32 addr, dpsc_base = 0x1E80000, freq;
+	int     rcode = 0;
+
+	if (argc < 2)
+		return CMD_RET_USAGE;
+
+	freq = clk_get_rate(sys_clk0_6_clk);
+
+	addr = simple_strtoul(argv[1], NULL, 16);
+
+	rcode = mon_install(addr, dpsc_base, freq);
+	printf("## installed monitor, freq [%d], status %d\n",
+	       freq, rcode);
+
+	return 0;
+}
+
+U_BOOT_CMD(mon_install, 2, 0, do_mon_install,
+	   "Install boot kernel at 'addr'",
+	   ""
+);
+
+static void core_spin(void)
+{
+	while (1)
+		; /* forever */;
+}
+
+int mon_power_on(int core_id, void *ep)
+{
+	int result;
+
+	asm volatile (
+		"stmfd  r13!, {lr}\n"
+		"mov r1, %1\n"
+		"mov r2, %2\n"
+		"mov r0, #0\n"
+		"smc	#0\n"
+		"ldmfd  r13!, {lr}\n"
+		: "=&r" (result)
+		: "r" (core_id), "r" (ep)
+		: "cc", "r0", "r1", "r2", "memory");
+	return  result;
+}
+
+int mon_power_off(int core_id)
+{
+	int result;
+
+	asm volatile (
+		"stmfd  r13!, {lr}\n"
+		"mov r1, %1\n"
+		"mov r0, #1\n"
+		"smc	#1\n"
+		"ldmfd  r13!, {lr}\n"
+		: "=&r" (result)
+		: "r" (core_id)
+		: "cc", "r0", "r1", "memory");
+	return  result;
+}
+
+int do_mon_power(cmd_tbl_t *cmdtp, int flag, int argc,
+			char * const argv[])
+{
+	int     rcode = 0, core_id, on;
+	void (*fn)(void);
+
+	fn = core_spin;
+
+	if (argc < 3)
+		return CMD_RET_USAGE;
+
+	core_id = simple_strtoul(argv[1], NULL, 16);
+	on = simple_strtoul(argv[2], NULL, 16);
+
+	if (on)
+		rcode = mon_power_on(core_id, fn);
+	else
+		rcode = mon_power_off(core_id);
+
+	if (on) {
+		if (!rcode)
+			printf("core %d powered on successfully\n", core_id);
+		else
+			printf("core %d power on failure\n", core_id);
+	} else {
+		printf("core %d powered off successfully\n", core_id);
+	}
+
+	return 0;
+}
+
+U_BOOT_CMD(mon_power, 3, 0, do_mon_power,
+	   "Power On/Off secondary core",
+	   "mon_power <coreid> <oper>\n"
+	   "- coreid (1-3) and oper (1 - ON, 0 - OFF)\n"
+	   ""
+);
diff --git a/arch/arm/mach-keystone/ddr3.c b/arch/arm/mach-keystone/ddr3.c
new file mode 100644
index 0000000000..dfb27b5ba2
--- /dev/null
+++ b/arch/arm/mach-keystone/ddr3.c
@@ -0,0 +1,404 @@
+/*
+ * Keystone2: DDR3 initialization
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <asm/io.h>
+#include <common.h>
+#include <asm/arch/msmc.h>
+#include <asm/arch/ddr3.h>
+#include <asm/arch/psc_defs.h>
+
+#include <asm/ti-common/ti-edma3.h>
+
+#define DDR3_EDMA_BLK_SIZE_SHIFT	10
+#define DDR3_EDMA_BLK_SIZE		(1 << DDR3_EDMA_BLK_SIZE_SHIFT)
+#define DDR3_EDMA_BCNT			0x8000
+#define DDR3_EDMA_CCNT			1
+#define DDR3_EDMA_XF_SIZE		(DDR3_EDMA_BLK_SIZE * DDR3_EDMA_BCNT)
+#define DDR3_EDMA_SLOT_NUM		1
+
+void ddr3_init_ddrphy(u32 base, struct ddr3_phy_config *phy_cfg)
+{
+	unsigned int tmp;
+
+	while ((__raw_readl(base + KS2_DDRPHY_PGSR0_OFFSET)
+		 & 0x00000001) != 0x00000001)
+		;
+
+	__raw_writel(phy_cfg->pllcr, base + KS2_DDRPHY_PLLCR_OFFSET);
+
+	tmp = __raw_readl(base + KS2_DDRPHY_PGCR1_OFFSET);
+	tmp &= ~(phy_cfg->pgcr1_mask);
+	tmp |= phy_cfg->pgcr1_val;
+	__raw_writel(tmp, base + KS2_DDRPHY_PGCR1_OFFSET);
+
+	__raw_writel(phy_cfg->ptr0,   base + KS2_DDRPHY_PTR0_OFFSET);
+	__raw_writel(phy_cfg->ptr1,   base + KS2_DDRPHY_PTR1_OFFSET);
+	__raw_writel(phy_cfg->ptr3,  base + KS2_DDRPHY_PTR3_OFFSET);
+	__raw_writel(phy_cfg->ptr4,  base + KS2_DDRPHY_PTR4_OFFSET);
+
+	tmp =  __raw_readl(base + KS2_DDRPHY_DCR_OFFSET);
+	tmp &= ~(phy_cfg->dcr_mask);
+	tmp |= phy_cfg->dcr_val;
+	__raw_writel(tmp, base + KS2_DDRPHY_DCR_OFFSET);
+
+	__raw_writel(phy_cfg->dtpr0, base + KS2_DDRPHY_DTPR0_OFFSET);
+	__raw_writel(phy_cfg->dtpr1, base + KS2_DDRPHY_DTPR1_OFFSET);
+	__raw_writel(phy_cfg->dtpr2, base + KS2_DDRPHY_DTPR2_OFFSET);
+	__raw_writel(phy_cfg->mr0,   base + KS2_DDRPHY_MR0_OFFSET);
+	__raw_writel(phy_cfg->mr1,   base + KS2_DDRPHY_MR1_OFFSET);
+	__raw_writel(phy_cfg->mr2,   base + KS2_DDRPHY_MR2_OFFSET);
+	__raw_writel(phy_cfg->dtcr,  base + KS2_DDRPHY_DTCR_OFFSET);
+	__raw_writel(phy_cfg->pgcr2, base + KS2_DDRPHY_PGCR2_OFFSET);
+
+	__raw_writel(phy_cfg->zq0cr1, base + KS2_DDRPHY_ZQ0CR1_OFFSET);
+	__raw_writel(phy_cfg->zq1cr1, base + KS2_DDRPHY_ZQ1CR1_OFFSET);
+	__raw_writel(phy_cfg->zq2cr1, base + KS2_DDRPHY_ZQ2CR1_OFFSET);
+
+	__raw_writel(phy_cfg->pir_v1, base + KS2_DDRPHY_PIR_OFFSET);
+	while ((__raw_readl(base + KS2_DDRPHY_PGSR0_OFFSET) & 0x1) != 0x1)
+		;
+
+	__raw_writel(phy_cfg->pir_v2, base + KS2_DDRPHY_PIR_OFFSET);
+	while ((__raw_readl(base + KS2_DDRPHY_PGSR0_OFFSET) & 0x1) != 0x1)
+		;
+}
+
+void ddr3_init_ddremif(u32 base, struct ddr3_emif_config *emif_cfg)
+{
+	__raw_writel(emif_cfg->sdcfg,  base + KS2_DDR3_SDCFG_OFFSET);
+	__raw_writel(emif_cfg->sdtim1, base + KS2_DDR3_SDTIM1_OFFSET);
+	__raw_writel(emif_cfg->sdtim2, base + KS2_DDR3_SDTIM2_OFFSET);
+	__raw_writel(emif_cfg->sdtim3, base + KS2_DDR3_SDTIM3_OFFSET);
+	__raw_writel(emif_cfg->sdtim4, base + KS2_DDR3_SDTIM4_OFFSET);
+	__raw_writel(emif_cfg->zqcfg,  base + KS2_DDR3_ZQCFG_OFFSET);
+	__raw_writel(emif_cfg->sdrfc,  base + KS2_DDR3_SDRFC_OFFSET);
+}
+
+int ddr3_ecc_support_rmw(u32 base)
+{
+	u32 value = __raw_readl(base + KS2_DDR3_MIDR_OFFSET);
+
+	/* Check the DDR3 controller ID reg if the controllers
+	   supports ECC RMW or not */
+	if (value == 0x40461C02)
+		return 1;
+
+	return 0;
+}
+
+static void ddr3_ecc_config(u32 base, u32 value)
+{
+	u32 data;
+
+	__raw_writel(value,  base + KS2_DDR3_ECC_CTRL_OFFSET);
+	udelay(100000); /* delay required to synchronize across clock domains */
+
+	if (value & KS2_DDR3_ECC_EN) {
+		/* Clear the 1-bit error count */
+		data = __raw_readl(base + KS2_DDR3_ONE_BIT_ECC_ERR_CNT_OFFSET);
+		__raw_writel(data, base + KS2_DDR3_ONE_BIT_ECC_ERR_CNT_OFFSET);
+
+		/* enable the ECC interrupt */
+		__raw_writel(KS2_DDR3_1B_ECC_ERR_SYS | KS2_DDR3_2B_ECC_ERR_SYS |
+			     KS2_DDR3_WR_ECC_ERR_SYS,
+			     base + KS2_DDR3_ECC_INT_ENABLE_SET_SYS_OFFSET);
+
+		/* Clear the ECC error interrupt status */
+		__raw_writel(KS2_DDR3_1B_ECC_ERR_SYS | KS2_DDR3_2B_ECC_ERR_SYS |
+			     KS2_DDR3_WR_ECC_ERR_SYS,
+			     base + KS2_DDR3_ECC_INT_STATUS_OFFSET);
+	}
+}
+
+static void ddr3_reset_data(u32 base, u32 ddr3_size)
+{
+	u32 mpax[2];
+	u32 seg_num;
+	u32 seg, blks, dst, edma_blks;
+	struct edma3_slot_config slot;
+	struct edma3_channel_config edma_channel;
+	u32 edma_src[DDR3_EDMA_BLK_SIZE/4] __aligned(16) = {0, };
+
+	/* Setup an edma to copy the 1k block to the entire DDR */
+	puts("\nClear entire DDR3 memory to enable ECC\n");
+
+	/* save the SES MPAX regs */
+	msmc_get_ses_mpax(8, 0, mpax);
+
+	/* setup edma slot 1 configuration */
+	slot.opt = EDMA3_SLOPT_TRANS_COMP_INT_ENB |
+		   EDMA3_SLOPT_COMP_CODE(0) |
+		   EDMA3_SLOPT_STATIC | EDMA3_SLOPT_AB_SYNC;
+	slot.bcnt = DDR3_EDMA_BCNT;
+	slot.acnt = DDR3_EDMA_BLK_SIZE;
+	slot.ccnt = DDR3_EDMA_CCNT;
+	slot.src_bidx = 0;
+	slot.dst_bidx = DDR3_EDMA_BLK_SIZE;
+	slot.src_cidx = 0;
+	slot.dst_cidx = 0;
+	slot.link = EDMA3_PARSET_NULL_LINK;
+	slot.bcntrld = 0;
+	edma3_slot_configure(KS2_EDMA0_BASE, DDR3_EDMA_SLOT_NUM, &slot);
+
+	/* configure quik edma channel */
+	edma_channel.slot = DDR3_EDMA_SLOT_NUM;
+	edma_channel.chnum = 0;
+	edma_channel.complete_code = 0;
+	/* event trigger after dst update */
+	edma_channel.trigger_slot_word = EDMA3_TWORD(dst);
+	qedma3_start(KS2_EDMA0_BASE, &edma_channel);
+
+	/* DDR3 size in segments (4KB seg size) */
+	seg_num = ddr3_size << (30 - KS2_MSMC_SEG_SIZE_SHIFT);
+
+	for (seg = 0; seg < seg_num; seg += KS2_MSMC_MAP_SEG_NUM) {
+		/* map 2GB 36-bit DDR address to 32-bit DDR address in EMIF
+		   access slave interface so that edma driver can access */
+		msmc_map_ses_segment(8, 0, base >> KS2_MSMC_SEG_SIZE_SHIFT,
+				     KS2_MSMC_DST_SEG_BASE + seg, MPAX_SEG_2G);
+
+		if ((seg_num - seg) > KS2_MSMC_MAP_SEG_NUM)
+			edma_blks = KS2_MSMC_MAP_SEG_NUM <<
+					(KS2_MSMC_SEG_SIZE_SHIFT
+					- DDR3_EDMA_BLK_SIZE_SHIFT);
+		else
+			edma_blks = (seg_num - seg) << (KS2_MSMC_SEG_SIZE_SHIFT
+					- DDR3_EDMA_BLK_SIZE_SHIFT);
+
+		/* Use edma driver to scrub 2GB DDR memory */
+		for (dst = base, blks = 0; blks < edma_blks;
+		     blks += DDR3_EDMA_BCNT, dst += DDR3_EDMA_XF_SIZE) {
+			edma3_set_src_addr(KS2_EDMA0_BASE,
+					   edma_channel.slot, (u32)edma_src);
+			edma3_set_dest_addr(KS2_EDMA0_BASE,
+					    edma_channel.slot, (u32)dst);
+
+			while (edma3_check_for_transfer(KS2_EDMA0_BASE,
+							&edma_channel))
+				udelay(10);
+		}
+	}
+
+	qedma3_stop(KS2_EDMA0_BASE, &edma_channel);
+
+	/* restore the SES MPAX regs */
+	msmc_set_ses_mpax(8, 0, mpax);
+}
+
+static void ddr3_ecc_init_range(u32 base)
+{
+	u32 ecc_val = KS2_DDR3_ECC_EN;
+	u32 rmw = ddr3_ecc_support_rmw(base);
+
+	if (rmw)
+		ecc_val |= KS2_DDR3_ECC_RMW_EN;
+
+	__raw_writel(0, base + KS2_DDR3_ECC_ADDR_RANGE1_OFFSET);
+
+	ddr3_ecc_config(base, ecc_val);
+}
+
+void ddr3_enable_ecc(u32 base, int test)
+{
+	u32 ecc_val = KS2_DDR3_ECC_ENABLE;
+	u32 rmw = ddr3_ecc_support_rmw(base);
+
+	if (test)
+		ecc_val |= KS2_DDR3_ECC_ADDR_RNG_1_EN;
+
+	if (!rmw) {
+		if (!test)
+			/* by default, disable ecc when rmw = 0 and no
+			   ecc test */
+			ecc_val = 0;
+	} else {
+		ecc_val |= KS2_DDR3_ECC_RMW_EN;
+	}
+
+	ddr3_ecc_config(base, ecc_val);
+}
+
+void ddr3_disable_ecc(u32 base)
+{
+	ddr3_ecc_config(base, 0);
+}
+
+#if defined(CONFIG_SOC_K2HK) || defined(CONFIG_SOC_K2L)
+static void cic_init(u32 base)
+{
+	/* Disable CIC global interrupts */
+	__raw_writel(0, base + KS2_CIC_GLOBAL_ENABLE);
+
+	/* Set to normal mode, no nesting, no priority hold */
+	__raw_writel(0, base + KS2_CIC_CTRL);
+	__raw_writel(0, base + KS2_CIC_HOST_CTRL);
+
+	/* Enable CIC global interrupts */
+	__raw_writel(1, base + KS2_CIC_GLOBAL_ENABLE);
+}
+
+static void cic_map_cic_to_gic(u32 base, u32 chan_num, u32 irq_num)
+{
+	/* Map the system interrupt to a CIC channel */
+	__raw_writeb(chan_num, base + KS2_CIC_CHAN_MAP(0) + irq_num);
+
+	/* Enable CIC system interrupt */
+	__raw_writel(irq_num, base + KS2_CIC_SYS_ENABLE_IDX_SET);
+
+	/* Enable CIC Host interrupt */
+	__raw_writel(chan_num, base + KS2_CIC_HOST_ENABLE_IDX_SET);
+}
+
+static void ddr3_map_ecc_cic2_irq(u32 base)
+{
+	cic_init(base);
+	cic_map_cic_to_gic(base, KS2_CIC2_DDR3_ECC_CHAN_NUM,
+			   KS2_CIC2_DDR3_ECC_IRQ_NUM);
+}
+#endif
+
+void ddr3_init_ecc(u32 base, u32 ddr3_size)
+{
+	if (!ddr3_ecc_support_rmw(base)) {
+		ddr3_disable_ecc(base);
+		return;
+	}
+
+	ddr3_ecc_init_range(base);
+	ddr3_reset_data(CONFIG_SYS_SDRAM_BASE, ddr3_size);
+
+	/* mapping DDR3 ECC system interrupt from CIC2 to GIC */
+#if defined(CONFIG_SOC_K2HK) || defined(CONFIG_SOC_K2L)
+	ddr3_map_ecc_cic2_irq(KS2_CIC2_BASE);
+#endif
+	ddr3_enable_ecc(base, 0);
+}
+
+void ddr3_check_ecc_int(u32 base)
+{
+	char *env;
+	int ecc_test = 0;
+	u32 value = __raw_readl(base + KS2_DDR3_ECC_INT_STATUS_OFFSET);
+
+	env = getenv("ecc_test");
+	if (env)
+		ecc_test = simple_strtol(env, NULL, 0);
+
+	if (value & KS2_DDR3_WR_ECC_ERR_SYS)
+		puts("DDR3 ECC write error interrupted\n");
+
+	if (value & KS2_DDR3_2B_ECC_ERR_SYS) {
+		puts("DDR3 ECC 2-bit error interrupted\n");
+
+		if (!ecc_test) {
+			puts("Reseting the device ...\n");
+			reset_cpu(0);
+		}
+	}
+
+	value = __raw_readl(base + KS2_DDR3_ONE_BIT_ECC_ERR_CNT_OFFSET);
+	if (value) {
+		printf("1-bit ECC err count: 0x%x\n", value);
+		value = __raw_readl(base +
+				    KS2_DDR3_ONE_BIT_ECC_ERR_ADDR_LOG_OFFSET);
+		printf("1-bit ECC err address log: 0x%x\n", value);
+	}
+}
+
+void ddr3_reset_ddrphy(void)
+{
+	u32 tmp;
+
+	/* Assert DDR3A  PHY reset */
+	tmp = readl(KS2_DDR3APLLCTL1);
+	tmp |= KS2_DDR3_PLLCTRL_PHY_RESET;
+	writel(tmp, KS2_DDR3APLLCTL1);
+
+	/* wait 10us to catch the reset */
+	udelay(10);
+
+	/* Release DDR3A PHY reset */
+	tmp = readl(KS2_DDR3APLLCTL1);
+	tmp &= ~KS2_DDR3_PLLCTRL_PHY_RESET;
+	__raw_writel(tmp, KS2_DDR3APLLCTL1);
+}
+
+#ifdef CONFIG_SOC_K2HK
+/**
+ * ddr3_reset_workaround - reset workaround in case if leveling error
+ * detected for PG 1.0 and 1.1 k2hk SoCs
+ */
+void ddr3_err_reset_workaround(void)
+{
+	unsigned int tmp;
+	unsigned int tmp_a;
+	unsigned int tmp_b;
+
+	/*
+	 * Check for PGSR0 error bits of DDR3 PHY.
+	 * Check for WLERR, QSGERR, WLAERR,
+	 * RDERR, WDERR, REERR, WEERR error to see if they are set or not
+	 */
+	tmp_a = __raw_readl(KS2_DDR3A_DDRPHYC + KS2_DDRPHY_PGSR0_OFFSET);
+	tmp_b = __raw_readl(KS2_DDR3B_DDRPHYC + KS2_DDRPHY_PGSR0_OFFSET);
+
+	if (((tmp_a & 0x0FE00000) != 0) || ((tmp_b & 0x0FE00000) != 0)) {
+		printf("DDR Leveling Error Detected!\n");
+		printf("DDR3A PGSR0 = 0x%x\n", tmp_a);
+		printf("DDR3B PGSR0 = 0x%x\n", tmp_b);
+
+		/*
+		 * Write Keys to KICK registers to enable writes to registers
+		 * in boot config space
+		 */
+		__raw_writel(KS2_KICK0_MAGIC, KS2_KICK0);
+		__raw_writel(KS2_KICK1_MAGIC, KS2_KICK1);
+
+		/*
+		 * Move DDR3A Module out of reset isolation by setting
+		 * MDCTL23[12] = 0
+		 */
+		tmp_a = __raw_readl(KS2_PSC_BASE +
+				    PSC_REG_MDCTL(KS2_LPSC_EMIF4F_DDR3A));
+
+		tmp_a = PSC_REG_MDCTL_SET_RESET_ISO(tmp_a, 0);
+		__raw_writel(tmp_a, KS2_PSC_BASE +
+			     PSC_REG_MDCTL(KS2_LPSC_EMIF4F_DDR3A));
+
+		/*
+		 * Move DDR3B Module out of reset isolation by setting
+		 * MDCTL24[12] = 0
+		 */
+		tmp_b = __raw_readl(KS2_PSC_BASE +
+				    PSC_REG_MDCTL(KS2_LPSC_EMIF4F_DDR3B));
+		tmp_b = PSC_REG_MDCTL_SET_RESET_ISO(tmp_b, 0);
+		__raw_writel(tmp_b, KS2_PSC_BASE +
+			     PSC_REG_MDCTL(KS2_LPSC_EMIF4F_DDR3B));
+
+		/*
+		 * Write 0x5A69 Key to RSTCTRL[15:0] to unlock writes
+		 * to RSTCTRL and RSTCFG
+		 */
+		tmp = __raw_readl(KS2_RSTCTRL);
+		tmp &= KS2_RSTCTRL_MASK;
+		tmp |= KS2_RSTCTRL_KEY;
+		__raw_writel(tmp, KS2_RSTCTRL);
+
+		/*
+		 * Set PLL Controller to drive hard reset on SW trigger by
+		 * setting RSTCFG[13] = 0
+		 */
+		tmp = __raw_readl(KS2_RSTCTRL_RSCFG);
+		tmp &= ~KS2_RSTYPE_PLL_SOFT;
+		__raw_writel(tmp, KS2_RSTCTRL_RSCFG);
+
+		reset_cpu(0);
+	}
+}
+#endif
diff --git a/arch/arm/mach-keystone/init.c b/arch/arm/mach-keystone/init.c
new file mode 100644
index 0000000000..c96845c4e2
--- /dev/null
+++ b/arch/arm/mach-keystone/init.c
@@ -0,0 +1,151 @@
+/*
+ * Keystone2: Architecture initialization
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <ns16550.h>
+#include <asm/io.h>
+#include <asm/arch/msmc.h>
+#include <asm/arch/clock.h>
+#include <asm/arch/hardware.h>
+#include <asm/arch/psc_defs.h>
+
+#define MAX_PCI_PORTS		2
+enum pci_mode	{
+	ENDPOINT,
+	LEGACY_ENDPOINT,
+	ROOTCOMPLEX,
+};
+
+#define DEVCFG_MODE_MASK		(BIT(2) | BIT(1))
+#define DEVCFG_MODE_SHIFT		1
+
+void chip_configuration_unlock(void)
+{
+	__raw_writel(KS2_KICK0_MAGIC, KS2_KICK0);
+	__raw_writel(KS2_KICK1_MAGIC, KS2_KICK1);
+}
+
+#ifdef CONFIG_SOC_K2L
+void osr_init(void)
+{
+	u32 i;
+	u32 j;
+	u32 val;
+	u32 base = KS2_OSR_CFG_BASE;
+	u32 ecc_ctrl[KS2_OSR_NUM_RAM_BANKS];
+
+	/* Enable the OSR clock domain */
+	psc_enable_module(KS2_LPSC_OSR);
+
+	/* Disable OSR ECC check for all the ram banks */
+	for (i = 0; i < KS2_OSR_NUM_RAM_BANKS; i++) {
+		val = i | KS2_OSR_ECC_VEC_TRIG_RD |
+			(KS2_OSR_ECC_CTRL << KS2_OSR_ECC_VEC_RD_ADDR_SH);
+
+		writel(val , base + KS2_OSR_ECC_VEC);
+
+		/**
+		 * wait till read is done.
+		 * Print should be added after earlyprintk support is added.
+		 */
+		for (j = 0; j < 10000; j++) {
+			val = readl(base + KS2_OSR_ECC_VEC);
+			if (val & KS2_OSR_ECC_VEC_RD_DONE)
+				break;
+		}
+
+		ecc_ctrl[i] = readl(base + KS2_OSR_ECC_CTRL) ^
+						KS2_OSR_ECC_CTRL_CHK;
+
+		writel(ecc_ctrl[i], KS2_MSMC_DATA_BASE + i * 4);
+		writel(ecc_ctrl[i], base + KS2_OSR_ECC_CTRL);
+	}
+
+	/* Reset OSR memory to all zeros */
+	for (i = 0; i < KS2_OSR_SIZE; i += 4)
+		writel(0, KS2_OSR_DATA_BASE + i);
+
+	/* Enable OSR ECC check for all the ram banks */
+	for (i = 0; i < KS2_OSR_NUM_RAM_BANKS; i++)
+		writel(ecc_ctrl[i] |
+		       KS2_OSR_ECC_CTRL_CHK, base + KS2_OSR_ECC_CTRL);
+}
+#endif
+
+/* Function to set up PCIe mode */
+static void config_pcie_mode(int pcie_port,  enum pci_mode mode)
+{
+	u32 val = __raw_readl(KS2_DEVCFG);
+
+	if (pcie_port >= MAX_PCI_PORTS)
+		return;
+
+	/**
+	 * each pci port has two bits for mode and it starts at
+	 * bit 1. So use port number to get the right bit position.
+	 */
+	pcie_port <<= 1;
+	val &= ~(DEVCFG_MODE_MASK << pcie_port);
+	val |= ((mode << DEVCFG_MODE_SHIFT) << pcie_port);
+	__raw_writel(val, KS2_DEVCFG);
+}
+
+int arch_cpu_init(void)
+{
+	chip_configuration_unlock();
+	icache_enable();
+
+	msmc_share_all_segments(KS2_MSMC_SEGMENT_TETRIS);
+	msmc_share_all_segments(KS2_MSMC_SEGMENT_NETCP);
+	msmc_share_all_segments(KS2_MSMC_SEGMENT_QM_PDSP);
+	msmc_share_all_segments(KS2_MSMC_SEGMENT_PCIE0);
+
+	/* Initialize the PCIe-0 to work as Root Complex */
+	config_pcie_mode(0, ROOTCOMPLEX);
+#if defined(CONFIG_SOC_K2E) || defined(CONFIG_SOC_K2L)
+	msmc_share_all_segments(KS2_MSMC_SEGMENT_PCIE1);
+	/* Initialize the PCIe-1 to work as Root Complex */
+	config_pcie_mode(1, ROOTCOMPLEX);
+#endif
+#ifdef CONFIG_SOC_K2L
+	osr_init();
+#endif
+
+	/*
+	 * just initialise the COM2 port so that TI specific
+	 * UART register PWREMU_MGMT is initialized. Linux UART
+	 * driver doesn't handle this.
+	 */
+	NS16550_init((NS16550_t)(CONFIG_SYS_NS16550_COM2),
+		     CONFIG_SYS_NS16550_CLK / 16 / CONFIG_BAUDRATE);
+
+	return 0;
+}
+
+void reset_cpu(ulong addr)
+{
+	volatile u32 *rstctrl = (volatile u32 *)(KS2_RSTCTRL);
+	u32 tmp;
+
+	tmp = *rstctrl & KS2_RSTCTRL_MASK;
+	*rstctrl = tmp | KS2_RSTCTRL_KEY;
+
+	*rstctrl &= KS2_RSTCTRL_SWRST;
+
+	for (;;)
+		;
+}
+
+void enable_caches(void)
+{
+#ifndef CONFIG_SYS_DCACHE_OFF
+	/* Enable D-cache. I-cache is already enabled in start.S */
+	dcache_enable();
+#endif
+}
diff --git a/arch/arm/mach-keystone/keystone.c b/arch/arm/mach-keystone/keystone.c
new file mode 100644
index 0000000000..11a9357db4
--- /dev/null
+++ b/arch/arm/mach-keystone/keystone.c
@@ -0,0 +1,87 @@
+/*
+ * Keystone EVM : Board initialization
+ *
+ * (C) Copyright 2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/io.h>
+#include <asm/arch/mon.h>
+#include <asm/arch/psc_defs.h>
+#include <asm/arch/hardware.h>
+#include <asm/arch/hardware.h>
+
+/**
+ * cpu_to_bus - swap bytes of the 32-bit data if the device is BE
+ * @ptr - array of data
+ * @length - lenght of data array
+ */
+int cpu_to_bus(u32 *ptr, u32 length)
+{
+	u32 i;
+
+	if (!(readl(KS2_DEVSTAT) & 0x1))
+		for (i = 0; i < length; i++, ptr++)
+			*ptr = cpu_to_be32(*ptr);
+
+	return 0;
+}
+
+static int turn_off_myself(void)
+{
+	printf("Turning off ourselves\r\n");
+	mon_power_off(0);
+
+	psc_disable_module(KS2_LPSC_TETRIS);
+	psc_disable_domain(KS2_TETRIS_PWR_DOMAIN);
+
+	asm volatile ("isb\n"
+		      "dsb\n"
+		      "wfi\n");
+
+	printf("What! Should not see that\n");
+	return 0;
+}
+
+static void turn_off_all_dsps(int num_dsps)
+{
+	int i;
+
+	for (i = 0; i < num_dsps; i++) {
+		if (psc_disable_module(i + KS2_LPSC_GEM_0))
+			printf("Cannot disable module for #%d DSP", i);
+
+		if (psc_disable_domain(i + 8))
+			printf("Cannot disable domain for #%d DSP", i);
+	}
+}
+
+int do_killme_cmd(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
+{
+	return turn_off_myself();
+}
+
+U_BOOT_CMD(
+	killme, 1,      0,      do_killme_cmd,
+	"turn off main ARM core",
+	"turn off main ARM core. Should not live after that :(\n"
+);
+
+int misc_init_r(void)
+{
+	char *env;
+	long ks2_debug = 0;
+
+	env = getenv("ks2_debug");
+
+	if (env)
+		ks2_debug = simple_strtol(env, NULL, 0);
+
+	if ((ks2_debug & DBG_LEAVE_DSPS_ON) == 0)
+		turn_off_all_dsps(KS2_NUM_DSPS);
+
+	return 0;
+}
diff --git a/arch/arm/mach-keystone/msmc.c b/arch/arm/mach-keystone/msmc.c
new file mode 100644
index 0000000000..7899141d54
--- /dev/null
+++ b/arch/arm/mach-keystone/msmc.c
@@ -0,0 +1,94 @@
+/*
+ * MSMC controller utilities
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm/arch/msmc.h>
+
+struct mpax {
+	u32	mpaxl;
+	u32	mpaxh;
+};
+
+struct msms_regs {
+	u32	pid;
+	u32	_res_04;
+	u32	smcerrar;
+	u32	smcerrxr;
+	u32	smedcc;
+	u32	smcea;
+	u32	smsecc;
+	u32	smpfar;
+	u32	smpfxr;
+	u32	smpfr;
+	u32	smpfcr;
+	u32	_res_2c;
+	u32	sbndc[8];
+	u32	sbndm;
+	u32	sbnde;
+	u32	_res_58;
+	u32	cfglck;
+	u32	cfgulck;
+	u32	cfglckstat;
+	u32	sms_mpax_lck;
+	u32	sms_mpax_ulck;
+	u32	sms_mpax_lckstat;
+	u32	ses_mpax_lck;
+	u32	ses_mpax_ulck;
+	u32	ses_mpax_lckstat;
+	u32	smestat;
+	u32	smirstat;
+	u32	smirc;
+	u32	smiestat;
+	u32	smiec;
+	u32	_res_94_c0[12];
+	u32	smncerrar;
+	u32	smncerrxr;
+	u32	smncea;
+	u32	_res_d0_1fc[76];
+	struct mpax sms[16][8];
+	struct mpax ses[16][8];
+};
+
+
+void msmc_share_all_segments(int priv_id)
+{
+	struct msms_regs *msmc = (struct msms_regs *)KS2_MSMC_CTRL_BASE;
+	int j;
+
+	for (j = 0; j < 8; j++) {
+		msmc->sms[priv_id][j].mpaxh &= 0xffffff7ful;
+		msmc->ses[priv_id][j].mpaxh &= 0xffffff7ful;
+	}
+}
+
+void msmc_map_ses_segment(int priv_id, int ses_pair,
+			  u32 src_pfn, u32 dst_pfn, enum mpax_seg_size size)
+{
+	struct msms_regs *msmc = (struct msms_regs *)KS2_MSMC_CTRL_BASE;
+
+	msmc->ses[priv_id][ses_pair].mpaxh = src_pfn << 12 |
+					     (size & 0x1f) | 0x80;
+	msmc->ses[priv_id][ses_pair].mpaxl = dst_pfn << 8 | 0x3f;
+}
+
+void msmc_get_ses_mpax(int priv_id, int ses_pair, u32 *mpax)
+{
+	struct msms_regs *msmc = (struct msms_regs *)KS2_MSMC_CTRL_BASE;
+
+	*mpax++ = msmc->ses[priv_id][ses_pair].mpaxl;
+	*mpax = msmc->ses[priv_id][ses_pair].mpaxh;
+}
+
+void msmc_set_ses_mpax(int priv_id, int ses_pair, u32 *mpax)
+{
+	struct msms_regs *msmc = (struct msms_regs *)KS2_MSMC_CTRL_BASE;
+
+	msmc->ses[priv_id][ses_pair].mpaxl = *mpax++;
+	msmc->ses[priv_id][ses_pair].mpaxh = *mpax;
+}
diff --git a/arch/arm/mach-keystone/psc.c b/arch/arm/mach-keystone/psc.c
new file mode 100644
index 0000000000..237e776e87
--- /dev/null
+++ b/arch/arm/mach-keystone/psc.c
@@ -0,0 +1,227 @@
+/*
+ * Keystone: PSC configuration module
+ *
+ * (C) Copyright 2012-2014
+ *     Texas Instruments Incorporated, <www.ti.com>
+ *
+ * SPDX-License-Identifier:     GPL-2.0+
+ */
+
+#include <common.h>
+#include <asm-generic/errno.h>
+#include <asm/io.h>
+#include <asm/processor.h>
+#include <asm/arch/psc_defs.h>
+
+int psc_delay(void)
+{
+	udelay(10);
+	return 10;
+}
+
+/*
+ * FUNCTION PURPOSE: Wait for end of transitional state
+ *
+ * DESCRIPTION: Polls pstat for the selected domain and waits for transitions
+ *              to be complete.
+ *
+ *              Since this is boot loader code it is *ASSUMED* that interrupts
+ *              are disabled and no other core is mucking around with the psc
+ *              at the same time.
+ *
+ *              Returns 0 when the domain is free. Returns -1 if a timeout
+ *              occurred waiting for the completion.
+ */
+int psc_wait(u32 domain_num)
+{
+	u32 retry;
+	u32 ptstat;
+
+	/*
+	 * Do nothing if the power domain is in transition. This should never
+	 * happen since the boot code is the only software accesses psc.
+	 * It's still remotely possible that the hardware state machines
+	 * initiate transitions.
+	 * Don't trap if the domain (or a module in this domain) is
+	 * stuck in transition.
+	 */
+	retry = 0;
+
+	do {
+		ptstat = __raw_readl(KS2_PSC_BASE + PSC_REG_PSTAT);
+		ptstat = ptstat & (1 << domain_num);
+	} while ((ptstat != 0) && ((retry += psc_delay()) <
+		 PSC_PTSTAT_TIMEOUT_LIMIT));
+
+	if (retry >= PSC_PTSTAT_TIMEOUT_LIMIT)
+		return -1;
+
+	return 0;
+}
+
+u32 psc_get_domain_num(u32 mod_num)
+{
+	u32 domain_num;
+
+	/* Get the power domain associated with the module number */
+	domain_num = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCFG(mod_num));
+	domain_num = PSC_REG_MDCFG_GET_PD(domain_num);
+
+	return domain_num;
+}
+
+/*
+ * FUNCTION PURPOSE: Power up/down a module
+ *
+ * DESCRIPTION: Powers up/down the requested module and the associated power
+ *		domain if required. No action is taken it the module is
+ *		already powered up/down.
+ *
+ *              This only controls modules. The domain in which the module
+ *              resides will be left in the power on state. Multiple modules
+ *              can exist in a power domain, so powering down the domain based
+ *              on a single module is not done.
+ *
+ *              Returns 0 on success, -1 if the module can't be powered up, or
+ *              if there is a timeout waiting for the transition.
+ */
+int psc_set_state(u32 mod_num, u32 state)
+{
+	u32 domain_num;
+	u32 pdctl;
+	u32 mdctl;
+	u32 ptcmd;
+	u32 reset_iso;
+	u32 v;
+
+	/*
+	 * Get the power domain associated with the module number, and reset
+	 * isolation functionality
+	 */
+	v = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCFG(mod_num));
+	domain_num = PSC_REG_MDCFG_GET_PD(v);
+	reset_iso  = PSC_REG_MDCFG_GET_RESET_ISO(v);
+
+	/* Wait for the status of the domain/module to be non-transitional */
+	if (psc_wait(domain_num) != 0)
+		return -1;
+
+	/*
+	 * Perform configuration even if the current status matches the
+	 * existing state
+	 *
+	 * Set the next state of the power domain to on. It's OK if the domain
+	 * is always on. This code will not ever power down a domain, so no
+	 * change is made if the new state is power down.
+	 */
+	if (state == PSC_REG_VAL_MDCTL_NEXT_ON) {
+		pdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_PDCTL(domain_num));
+		pdctl = PSC_REG_PDCTL_SET_NEXT(pdctl,
+					       PSC_REG_VAL_PDCTL_NEXT_ON);
+		__raw_writel(pdctl, KS2_PSC_BASE + PSC_REG_PDCTL(domain_num));
+	}
+
+	/* Set the next state for the module to enabled/disabled */
+	mdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+	mdctl = PSC_REG_MDCTL_SET_NEXT(mdctl, state);
+	mdctl = PSC_REG_MDCTL_SET_RESET_ISO(mdctl, reset_iso);
+	__raw_writel(mdctl, KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+
+	/* Trigger the enable */
+	ptcmd = __raw_readl(KS2_PSC_BASE + PSC_REG_PTCMD);
+	ptcmd |= (u32)(1<<domain_num);
+	__raw_writel(ptcmd, KS2_PSC_BASE + PSC_REG_PTCMD);
+
+	/* Wait on the complete */
+	return psc_wait(domain_num);
+}
+
+/*
+ * FUNCTION PURPOSE: Power up a module
+ *
+ * DESCRIPTION: Powers up the requested module and the associated power domain
+ *              if required. No action is taken it the module is already
+ *              powered up.
+ *
+ *              Returns 0 on success, -1 if the module can't be powered up, or
+ *              if there is a timeout waiting for the transition.
+ */
+int psc_enable_module(u32 mod_num)
+{
+	u32 mdctl;
+
+	/* Set the bit to apply reset */
+	mdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+	if ((mdctl & 0x3f) == PSC_REG_VAL_MDSTAT_STATE_ON)
+		return 0;
+
+	return psc_set_state(mod_num, PSC_REG_VAL_MDCTL_NEXT_ON);
+}
+
+/*
+ * FUNCTION PURPOSE: Power down a module
+ *
+ * DESCRIPTION: Powers down the requested module.
+ *
+ *              Returns 0 on success, -1 on failure or timeout.
+ */
+int psc_disable_module(u32 mod_num)
+{
+	u32 mdctl;
+
+	/* Set the bit to apply reset */
+	mdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+	if ((mdctl & 0x3f) == 0)
+		return 0;
+	mdctl = PSC_REG_MDCTL_SET_LRSTZ(mdctl, 0);
+	__raw_writel(mdctl, KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+
+	return psc_set_state(mod_num, PSC_REG_VAL_MDCTL_NEXT_SWRSTDISABLE);
+}
+
+/*
+ * FUNCTION PURPOSE: Set the reset isolation bit in mdctl
+ *
+ * DESCRIPTION: The reset isolation enable bit is set. The state of the module
+ *              is not changed. Returns 0 if the module config showed that
+ *              reset isolation is supported. Returns 1 otherwise. This is not
+ *              an error, but setting the bit in mdctl has no effect.
+ */
+int psc_set_reset_iso(u32 mod_num)
+{
+	u32 v;
+	u32 mdctl;
+
+	/* Set the reset isolation bit */
+	mdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+	mdctl = PSC_REG_MDCTL_SET_RESET_ISO(mdctl, 1);
+	__raw_writel(mdctl, KS2_PSC_BASE + PSC_REG_MDCTL(mod_num));
+
+	v = __raw_readl(KS2_PSC_BASE + PSC_REG_MDCFG(mod_num));
+	if (PSC_REG_MDCFG_GET_RESET_ISO(v) == 1)
+		return 0;
+
+	return 1;
+}
+
+/*
+ * FUNCTION PURPOSE: Disable a power domain
+ *
+ * DESCRIPTION: The power domain is disabled
+ */
+int psc_disable_domain(u32 domain_num)
+{
+	u32 pdctl;
+	u32 ptcmd;
+
+	pdctl = __raw_readl(KS2_PSC_BASE + PSC_REG_PDCTL(domain_num));
+	pdctl = PSC_REG_PDCTL_SET_NEXT(pdctl, PSC_REG_VAL_PDCTL_NEXT_OFF);
+	pdctl = PSC_REG_PDCTL_SET_PDMODE(pdctl, PSC_REG_VAL_PDCTL_PDMODE_SLEEP);
+	__raw_writel(pdctl, KS2_PSC_BASE + PSC_REG_PDCTL(domain_num));
+
+	ptcmd = __raw_readl(KS2_PSC_BASE + PSC_REG_PTCMD);
+	ptcmd |= (u32)(1 << domain_num);
+	__raw_writel(ptcmd, KS2_PSC_BASE + PSC_REG_PTCMD);
+
+	return psc_wait(domain_num);
+}