diff --git a/Makefile b/Makefile
index 7d6f408..53dbdd3 100644
--- a/Makefile
+++ b/Makefile
@@ -381,6 +381,15 @@
 ALL-$(CONFIG_SPL) += $(obj)spl/u-boot-spl.bin
 ALL-$(CONFIG_OF_SEPARATE) += $(obj)u-boot.dtb $(obj)u-boot-dtb.bin
 
+# enable combined SPL/u-boot/dtb rules for tegra
+ifeq ($(SOC),tegra20)
+ifeq ($(CONFIG_OF_SEPARATE),y)
+ALL-y += $(obj)u-boot-dtb-tegra.bin
+else
+ALL-y += $(obj)u-boot-nodtb-tegra.bin
+endif
+endif
+
 all:		$(ALL-y) $(SUBDIR_EXAMPLES)
 
 $(obj)u-boot.dtb:	$(obj)u-boot
diff --git a/arch/arm/cpu/armv7/Makefile b/arch/arm/cpu/armv7/Makefile
index 788eada..4fdbee4 100644
--- a/arch/arm/cpu/armv7/Makefile
+++ b/arch/arm/cpu/armv7/Makefile
@@ -32,7 +32,7 @@
 COBJS	+= cpu.o
 COBJS	+= syslib.o
 
-ifneq ($(CONFIG_AM33XX)$(CONFIG_OMAP44XX)$(CONFIG_OMAP54XX),)
+ifneq ($(CONFIG_AM33XX)$(CONFIG_OMAP44XX)$(CONFIG_OMAP54XX)$(CONFIG_TEGRA20),)
 SOBJS	+= lowlevel_init.o
 endif
 
diff --git a/arch/arm/cpu/armv7/start.S b/arch/arm/cpu/armv7/start.S
index 38cce93..b06bafd 100644
--- a/arch/arm/cpu/armv7/start.S
+++ b/arch/arm/cpu/armv7/start.S
@@ -133,7 +133,6 @@
 	orr	r0, r0, #0xd3
 	msr	cpsr,r0
 
-#if !defined(CONFIG_TEGRA20)
 /*
  * Setup vector:
  * (OMAP4 spl TEXT_BASE is not 32 byte aligned.
@@ -149,7 +148,6 @@
 	ldr	r0, =_start
 	mcr	p15, 0, r0, c12, c0, 0	@Set VBAR
 #endif
-#endif	/* !Tegra20 */
 
 	/* the mask ROM code should have PLL and others stable */
 #ifndef CONFIG_SKIP_LOWLEVEL_INIT
diff --git a/arch/arm/cpu/tegra20-common/ap20.c b/arch/arm/cpu/tegra20-common/ap20.c
index 6ff71e0..00588da 100644
--- a/arch/arm/cpu/tegra20-common/ap20.c
+++ b/arch/arm/cpu/tegra20-common/ap20.c
@@ -20,16 +20,11 @@
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
 * MA 02111-1307 USA
 */
-
 #include <asm/io.h>
-#include <asm/arch/tegra20.h>
 #include <asm/arch/ap20.h>
-#include <asm/arch/clk_rst.h>
-#include <asm/arch/clock.h>
 #include <asm/arch/fuse.h>
 #include <asm/arch/gp_padctrl.h>
 #include <asm/arch/pmc.h>
-#include <asm/arch/pinmux.h>
 #include <asm/arch/scu.h>
 #include <asm/arch/warmboot.h>
 #include <common.h>
@@ -68,235 +63,7 @@
 	return TEGRA_SOC_UNKNOWN;
 }
 
-/* Returns 1 if the current CPU executing is a Cortex-A9, else 0 */
-static int ap20_cpu_is_cortexa9(void)
-{
-	u32 id = readb(NV_PA_PG_UP_BASE + PG_UP_TAG_0);
-	return id == (PG_UP_TAG_0_PID_CPU & 0xff);
-}
-
-void init_pllx(void)
-{
-	struct clk_rst_ctlr *clkrst =
-			(struct clk_rst_ctlr *)NV_PA_CLK_RST_BASE;
-	struct clk_pll_simple *pll =
-		&clkrst->crc_pll_simple[CLOCK_ID_XCPU - CLOCK_ID_FIRST_SIMPLE];
-	u32 reg;
-
-	/* If PLLX is already enabled, just return */
-	if (readl(&pll->pll_base) & PLL_ENABLE_MASK)
-		return;
-
-	/* Set PLLX_MISC */
-	writel(1 << PLL_CPCON_SHIFT, &pll->pll_misc);
-
-	/* Use 12MHz clock here */
-	reg = PLL_BYPASS_MASK | (12 << PLL_DIVM_SHIFT);
-	reg |= 1000 << PLL_DIVN_SHIFT;
-	writel(reg, &pll->pll_base);
-
-	reg |= PLL_ENABLE_MASK;
-	writel(reg, &pll->pll_base);
-
-	reg &= ~PLL_BYPASS_MASK;
-	writel(reg, &pll->pll_base);
-}
-
-static void enable_cpu_clock(int enable)
-{
-	struct clk_rst_ctlr *clkrst = (struct clk_rst_ctlr *)NV_PA_CLK_RST_BASE;
-	u32 clk;
-
-	/*
-	 * NOTE:
-	 * Regardless of whether the request is to enable or disable the CPU
-	 * clock, every processor in the CPU complex except the master (CPU 0)
-	 * will have it's clock stopped because the AVP only talks to the
-	 * master. The AVP does not know (nor does it need to know) that there
-	 * are multiple processors in the CPU complex.
-	 */
-
-	if (enable) {
-		/* Initialize PLLX */
-		init_pllx();
-
-		/* Wait until all clocks are stable */
-		udelay(PLL_STABILIZATION_DELAY);
-
-		writel(CCLK_BURST_POLICY, &clkrst->crc_cclk_brst_pol);
-		writel(SUPER_CCLK_DIVIDER, &clkrst->crc_super_cclk_div);
-	}
-
-	/*
-	 * Read the register containing the individual CPU clock enables and
-	 * always stop the clock to CPU 1.
-	 */
-	clk = readl(&clkrst->crc_clk_cpu_cmplx);
-	clk |= 1 << CPU1_CLK_STP_SHIFT;
-
-	/* Stop/Unstop the CPU clock */
-	clk &= ~CPU0_CLK_STP_MASK;
-	clk |= !enable << CPU0_CLK_STP_SHIFT;
-	writel(clk, &clkrst->crc_clk_cpu_cmplx);
-
-	clock_enable(PERIPH_ID_CPU);
-}
-
-static int is_cpu_powered(void)
-{
-	struct pmc_ctlr *pmc = (struct pmc_ctlr *)TEGRA20_PMC_BASE;
-
-	return (readl(&pmc->pmc_pwrgate_status) & CPU_PWRED) ? 1 : 0;
-}
-
-static void remove_cpu_io_clamps(void)
-{
-	struct pmc_ctlr *pmc = (struct pmc_ctlr *)TEGRA20_PMC_BASE;
-	u32 reg;
-
-	/* Remove the clamps on the CPU I/O signals */
-	reg = readl(&pmc->pmc_remove_clamping);
-	reg |= CPU_CLMP;
-	writel(reg, &pmc->pmc_remove_clamping);
-
-	/* Give I/O signals time to stabilize */
-	udelay(IO_STABILIZATION_DELAY);
-}
-
-static void powerup_cpu(void)
-{
-	struct pmc_ctlr *pmc = (struct pmc_ctlr *)TEGRA20_PMC_BASE;
-	u32 reg;
-	int timeout = IO_STABILIZATION_DELAY;
-
-	if (!is_cpu_powered()) {
-		/* Toggle the CPU power state (OFF -> ON) */
-		reg = readl(&pmc->pmc_pwrgate_toggle);
-		reg &= PARTID_CP;
-		reg |= START_CP;
-		writel(reg, &pmc->pmc_pwrgate_toggle);
-
-		/* Wait for the power to come up */
-		while (!is_cpu_powered()) {
-			if (timeout-- == 0)
-				printf("CPU failed to power up!\n");
-			else
-				udelay(10);
-		}
-
-		/*
-		 * Remove the I/O clamps from CPU power partition.
-		 * Recommended only on a Warm boot, if the CPU partition gets
-		 * power gated. Shouldn't cause any harm when called after a
-		 * cold boot according to HW, probably just redundant.
-		 */
-		remove_cpu_io_clamps();
-	}
-}
-
-static void enable_cpu_power_rail(void)
-{
-	struct pmc_ctlr *pmc = (struct pmc_ctlr *)TEGRA20_PMC_BASE;
-	u32 reg;
-
-	reg = readl(&pmc->pmc_cntrl);
-	reg |= CPUPWRREQ_OE;
-	writel(reg, &pmc->pmc_cntrl);
-
-	/*
-	 * The TI PMU65861C needs a 3.75ms delay between enabling
-	 * the power rail and enabling the CPU clock.  This delay
-	 * between SM1EN and SM1 is for switching time + the ramp
-	 * up of the voltage to the CPU (VDD_CPU from PMU).
-	 */
-	udelay(3750);
-}
-
-static void reset_A9_cpu(int reset)
-{
-	/*
-	* NOTE:  Regardless of whether the request is to hold the CPU in reset
-	*        or take it out of reset, every processor in the CPU complex
-	*        except the master (CPU 0) will be held in reset because the
-	*        AVP only talks to the master. The AVP does not know that there
-	*        are multiple processors in the CPU complex.
-	*/
-
-	/* Hold CPU 1 in reset, and CPU 0 if asked */
-	reset_cmplx_set_enable(1, crc_rst_cpu | crc_rst_de | crc_rst_debug, 1);
-	reset_cmplx_set_enable(0, crc_rst_cpu | crc_rst_de | crc_rst_debug,
-			       reset);
-
-	/* Enable/Disable master CPU reset */
-	reset_set_enable(PERIPH_ID_CPU, reset);
-}
-
-static void clock_enable_coresight(int enable)
-{
-	u32 rst, src;
-
-	clock_set_enable(PERIPH_ID_CORESIGHT, enable);
-	reset_set_enable(PERIPH_ID_CORESIGHT, !enable);
-
-	if (enable) {
-		/*
-		 * Put CoreSight on PLLP_OUT0 (216 MHz) and divide it down by
-		 *  1.5, giving an effective frequency of 144MHz.
-		 * Set PLLP_OUT0 [bits31:30 = 00], and use a 7.1 divisor
-		 *  (bits 7:0), so 00000001b == 1.5 (n+1 + .5)
-		 */
-		src = CLK_DIVIDER(NVBL_PLLP_KHZ, 144000);
-		clock_ll_set_source_divisor(PERIPH_ID_CSI, 0, src);
-
-		/* Unlock the CPU CoreSight interfaces */
-		rst = 0xC5ACCE55;
-		writel(rst, CSITE_CPU_DBG0_LAR);
-		writel(rst, CSITE_CPU_DBG1_LAR);
-	}
-}
-
-void start_cpu(u32 reset_vector)
-{
-	/* Enable VDD_CPU */
-	enable_cpu_power_rail();
-
-	/* Hold the CPUs in reset */
-	reset_A9_cpu(1);
-
-	/* Disable the CPU clock */
-	enable_cpu_clock(0);
-
-	/* Enable CoreSight */
-	clock_enable_coresight(1);
-
-	/*
-	 * Set the entry point for CPU execution from reset,
-	 *  if it's a non-zero value.
-	 */
-	if (reset_vector)
-		writel(reset_vector, EXCEP_VECTOR_CPU_RESET_VECTOR);
-
-	/* Enable the CPU clock */
-	enable_cpu_clock(1);
-
-	/* If the CPU doesn't already have power, power it up */
-	powerup_cpu();
-
-	/* Take the CPU out of reset */
-	reset_A9_cpu(0);
-}
-
-
-void halt_avp(void)
-{
-	for (;;) {
-		writel((HALT_COP_EVENT_JTAG | HALT_COP_EVENT_IRQ_1 \
-			| HALT_COP_EVENT_FIQ_1 | (FLOW_MODE_STOP<<29)),
-			FLOW_CTLR_HALT_COP_EVENTS);
-	}
-}
-
-void enable_scu(void)
+static void enable_scu(void)
 {
 	struct scu_ctlr *scu = (struct scu_ctlr *)NV_PA_ARM_PERIPHBASE;
 	u32 reg;
@@ -332,7 +99,7 @@
 	return odmdata;
 }
 
-void init_pmc_scratch(void)
+static void init_pmc_scratch(void)
 {
 	struct pmc_ctlr *const pmc = (struct pmc_ctlr *)TEGRA20_PMC_BASE;
 	u32 odmdata;
@@ -347,27 +114,8 @@
 	writel(odmdata, &pmc->pmc_scratch20);
 }
 
-void tegra20_start(void)
+void s_init(void)
 {
-	struct pmux_tri_ctlr *pmt = (struct pmux_tri_ctlr *)NV_PA_APB_MISC_BASE;
-
-	/* If we are the AVP, start up the first Cortex-A9 */
-	if (!ap20_cpu_is_cortexa9()) {
-		/* enable JTAG */
-		writel(0xC0, &pmt->pmt_cfg_ctl);
-
-		/*
-		 * If we are ARM7 - give it a different stack. We are about to
-		 * start up the A9 which will want to use this one.
-		 */
-		asm volatile("mov	sp, %0\n"
-			: : "r"(AVP_EARLY_BOOT_STACK_LIMIT));
-
-		start_cpu((u32)_start);
-		halt_avp();
-		/* not reached */
-	}
-
 	/* Init PMC scratch memory */
 	init_pmc_scratch();
 
diff --git a/arch/arm/cpu/tegra20-common/board.c b/arch/arm/cpu/tegra20-common/board.c
index e595ff9..598023a 100644
--- a/arch/arm/cpu/tegra20-common/board.c
+++ b/arch/arm/cpu/tegra20-common/board.c
@@ -23,12 +23,12 @@
 
 #include <common.h>
 #include <asm/io.h>
-#include <asm/arch/ap20.h>
 #include <asm/arch/clock.h>
 #include <asm/arch/funcmux.h>
 #include <asm/arch/pmc.h>
 #include <asm/arch/sys_proto.h>
 #include <asm/arch/tegra20.h>
+#include <asm/arch/warmboot.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -80,27 +80,6 @@
 }
 #endif	/* CONFIG_DISPLAY_BOARDINFO */
 
-#ifdef CONFIG_ARCH_CPU_INIT
-/*
- * Note this function is executed by the ARM7TDMI AVP. It does not return
- * in this case. It is also called once the A9 starts up, but does nothing in
- * that case.
- */
-int arch_cpu_init(void)
-{
-	/* Fire up the Cortex A9 */
-	tegra20_start();
-
-	/* We didn't do this init in start.S, so do it now */
-	cpu_init_cp15();
-
-	/* Initialize essential common plls */
-	clock_early_init();
-
-	return 0;
-}
-#endif
-
 static int uart_configs[] = {
 #if defined(CONFIG_TEGRA20_UARTA_UAA_UAB)
 	FUNCMUX_UART1_UAA_UAB,
diff --git a/boards.cfg b/boards.cfg
index 64b0a30..3319298 100644
--- a/boards.cfg
+++ b/boards.cfg
@@ -265,10 +265,10 @@
 smdk5250		     arm	 armv7	     smdk5250		 samsung	exynos
 smdkv310		     arm	 armv7	     smdkv310		 samsung	exynos
 trats                        arm         armv7       trats               samsung        exynos
-harmony                      arm         armv7       harmony             nvidia         tegra20
-seaboard                     arm         armv7       seaboard            nvidia         tegra20
-ventana                      arm         armv7       ventana             nvidia         tegra20
-whistler                     arm         armv7       whistler            nvidia         tegra20
+harmony                      arm         armv7:arm720t harmony           nvidia         tegra20
+seaboard                     arm         armv7:arm720t seaboard          nvidia         tegra20
+ventana                      arm         armv7:arm720t ventana           nvidia         tegra20
+whistler                     arm         armv7:arm720t whistler          nvidia         tegra20
 u8500_href                   arm         armv7       u8500               st-ericsson    u8500
 snowball                     arm         armv7       snowball               st-ericsson    u8500
 actux1_4_16                  arm         ixp         actux1              -              -           actux1:FLASH2X2
@@ -295,11 +295,11 @@
 zipitz2                      arm         pxa
 colibri_pxa270               arm         pxa         -                   toradex
 jornada                      arm         sa1100
-plutux                       arm         armv7       plutux              avionic-design tegra20
-medcom                       arm         armv7       medcom              avionic-design tegra20
-tec                          arm         armv7       tec                 avionic-design tegra20
-paz00                        arm         armv7       paz00               compal         tegra20
-trimslice                    arm         armv7       trimslice           compulab       tegra20
+plutux                       arm         armv7:arm720t plutux            avionic-design tegra20
+medcom                       arm         armv7:arm720t medcom            avionic-design tegra20
+tec                          arm         armv7:arm720t tec               avionic-design tegra20
+paz00                        arm         armv7:arm720t paz00             compal         tegra20
+trimslice                    arm         armv7:arm720t trimslice         compulab       tegra20
 atngw100                     avr32       at32ap      -                   atmel          at32ap700x
 atstk1002                    avr32       at32ap      atstk1000           atmel          at32ap700x
 atstk1003                    avr32       at32ap      atstk1000           atmel          at32ap700x
diff --git a/include/configs/tegra20-common-post.h b/include/configs/tegra20-common-post.h
index e8badb9..42f270f 100644
--- a/include/configs/tegra20-common-post.h
+++ b/include/configs/tegra20-common-post.h
@@ -146,4 +146,69 @@
 	"fdt_high=01100000\0" \
 	BOOTCMDS_COMMON
 
+/* overrides for SPL build here */
+#ifdef CONFIG_SPL_BUILD
+
+/* remove devicetree support */
+#ifdef CONFIG_OF_CONTROL
+#undef CONFIG_OF_CONTROL
+#endif
+
+/* remove SERIAL_MULTI */
+#ifdef CONFIG_SERIAL_MULTI
+#undef CONFIG_SERIAL_MULTI
+#endif
+
+/* remove I2C support */
+#ifdef CONFIG_TEGRA_I2C
+#undef CONFIG_TEGRA_I2C
+#endif
+#ifdef CONFIG_CMD_I2C
+#undef CONFIG_CMD_I2C
+#endif
+
+/* remove MMC support */
+#ifdef CONFIG_MMC
+#undef CONFIG_MMC
+#endif
+#ifdef CONFIG_GENERIC_MMC
+#undef CONFIG_GENERIC_MMC
+#endif
+#ifdef CONFIG_TEGRA20_MMC
+#undef CONFIG_TEGRA20_MMC
+#endif
+#ifdef CONFIG_CMD_MMC
+#undef CONFIG_CMD_MMC
+#endif
+
+/* remove partitions/filesystems */
+#ifdef CONFIG_DOS_PARTITION
+#undef CONFIG_DOS_PARTITION
+#endif
+#ifdef CONFIG_EFI_PARTITION
+#undef CONFIG_EFI_PARTITION
+#endif
+#ifdef CONFIG_CMD_EXT2
+#undef CONFIG_CMD_EXT2
+#endif
+#ifdef CONFIG_CMD_FAT
+#undef CONFIG_CMD_FAT
+#endif
+
+/* remove USB */
+#ifdef CONFIG_USB_EHCI
+#undef CONFIG_USB_EHCI
+#endif
+#ifdef CONFIG_USB_EHCI_TEGRA
+#undef CONFIG_USB_EHCI_TEGRA
+#endif
+#ifdef CONFIG_USB_STORAGE
+#undef CONFIG_USB_STORAGE
+#endif
+#ifdef CONFIG_CMD_USB
+#undef CONFIG_CMD_USB
+#endif
+
+#endif /* CONFIG_SPL_BUILD */
+
 #endif /* __TEGRA20_COMMON_POST_H */
diff --git a/include/configs/tegra20-common.h b/include/configs/tegra20-common.h
index 769728b..1c0d235 100644
--- a/include/configs/tegra20-common.h
+++ b/include/configs/tegra20-common.h
@@ -43,8 +43,6 @@
 
 #define CONFIG_SYS_CACHELINE_SIZE	32
 
-#define CONFIG_ARCH_CPU_INIT		/* Fire up the A9 core */
-
 #include <asm/arch/tegra20.h>		/* get chip and board defs */
 
 /*
@@ -53,8 +51,6 @@
 #define CONFIG_DISPLAY_CPUINFO
 #define CONFIG_DISPLAY_BOARDINFO
 
-#define CONFIG_SKIP_LOWLEVEL_INIT
-
 #define CONFIG_CMDLINE_TAG		/* enable passing of ATAGs */
 #define CONFIG_OF_LIBFDT		/* enable passing of devicetree */
 
@@ -182,7 +178,7 @@
 #define PHYS_SDRAM_1		TEGRA20_SDRC_CS0
 #define PHYS_SDRAM_1_SIZE	0x20000000	/* 512M */
 
-#define CONFIG_SYS_TEXT_BASE	0x00108000
+#define CONFIG_SYS_TEXT_BASE	0x0010c000
 #define CONFIG_SYS_SDRAM_BASE	PHYS_SDRAM_1
 
 #define CONFIG_SYS_INIT_RAM_ADDR	CONFIG_STACKBASE
@@ -195,4 +191,20 @@
 #define CONFIG_CMD_GPIO
 #define CONFIG_CMD_ENTERRCM
 #define CONFIG_CMD_BOOTZ
+
+/* Defines for SPL */
+#define CONFIG_SPL
+#define CONFIG_SPL_NAND_SIMPLE
+#define CONFIG_SPL_TEXT_BASE		0x00108000
+#define CONFIG_SPL_MAX_SIZE		0x00004000
+#define CONFIG_SYS_SPL_MALLOC_START	0x00090000
+#define CONFIG_SYS_SPL_MALLOC_SIZE	0x00010000
+#define CONFIG_SPL_STACK		0x000ffffc
+
+#define CONFIG_SPL_LIBCOMMON_SUPPORT
+#define CONFIG_SPL_LIBGENERIC_SUPPORT
+#define CONFIG_SPL_SERIAL_SUPPORT
+#define CONFIG_SPL_GPIO_SUPPORT
+#define CONFIG_SPL_LDSCRIPT		"$(CPUDIR)/tegra20/u-boot-spl.lds"
+
 #endif /* __TEGRA20_COMMON_H */