Merge branch 'master' of git://git.denx.de/u-boot-blackfin

11 files changed, 209 insertions, 497 deletions

diff --git a/cpu/blackfin/Makefile b/cpu/blackfin/Makefile
index 8fed4b4..b4049ff 100644
--- a/cpu/blackfin/Makefile
+++ b/cpu/blackfin/Makefile
@@ -17,14 +17,15 @@ EXTRA    :=
 CEXTRA   := initcode.o
 SEXTRA   := start.o
 SOBJS    := interrupt.o cache.o
-COBJS    := cpu.o traps.o interrupts.o reset.o serial.o i2c.o watchdog.o
+COBJS-y  := cpu.o traps.o interrupts.o reset.o serial.o watchdog.o
+COBJS-$(CONFIG_JTAG_CONSOLE) += jtag-console.o
 
 ifeq ($(CONFIG_BFIN_BOOT_MODE),BFIN_BOOT_BYPASS)
-COBJS    += initcode.o
+COBJS-y  += initcode.o
 endif
 
-SRCS     := $(SEXTRA:.o=.S) $(SOBJS:.o=.S) $(COBJS:.o=.c)
-OBJS     := $(addprefix $(obj),$(COBJS) $(SOBJS))
+SRCS     := $(SEXTRA:.o=.S) $(SOBJS:.o=.S) $(COBJS-y:.o=.c)
+OBJS     := $(addprefix $(obj),$(COBJS-y) $(SOBJS))
 EXTRA    := $(addprefix $(obj),$(EXTRA))
 CEXTRA   := $(addprefix $(obj),$(CEXTRA))
 SEXTRA   := $(addprefix $(obj),$(SEXTRA))
diff --git a/cpu/blackfin/cache.S b/cpu/blackfin/cache.S
index 51bdb30..9facadf 100644
--- a/cpu/blackfin/cache.S
+++ b/cpu/blackfin/cache.S
@@ -39,7 +39,7 @@ ENTRY(_blackfin_dcache_flush_range)
 	RTS;
 ENDPROC(_blackfin_dcache_flush_range)
 
-ENTRY(_blackfin_dcache_invalidate_range)
+ENTRY(_blackfin_dcache_flush_invalidate_range)
 	R2 = -32;
 	R2 = R0 & R2;
 	P0 = R2;
@@ -58,4 +58,4 @@ ENTRY(_blackfin_dcache_invalidate_range)
 	FLUSHINV[P0];
 	SSYNC;
 	RTS;
-ENDPROC(_blackfin_dcache_invalidate_range)
+ENDPROC(_blackfin_dcache_flush_invalidate_range)
diff --git a/cpu/blackfin/cpu.c b/cpu/blackfin/cpu.c
index 9efd88e..30c214b 100644
--- a/cpu/blackfin/cpu.c
+++ b/cpu/blackfin/cpu.c
@@ -14,11 +14,14 @@
 #include <asm/blackfin.h>
 #include <asm/cplb.h>
 #include <asm/mach-common/bits/core.h>
+#include <asm/mach-common/bits/ebiu.h>
 #include <asm/mach-common/bits/trace.h>
 
 #include "cpu.h"
 #include "serial.h"
 
+ulong bfin_poweron_retx;
+
 __attribute__ ((__noreturn__))
 void cpu_init_f(ulong bootflag, ulong loaded_from_ldr)
 {
@@ -48,6 +51,9 @@ void cpu_init_f(ulong bootflag, ulong loaded_from_ldr)
 		bfin_write_EBIU_AMGCTL(CONFIG_EBIU_AMGCTL_VAL);
 #endif
 
+	/* Save RETX so we can pass it while booting Linux */
+	bfin_poweron_retx = bootflag;
+
 #ifdef CONFIG_DEBUG_DUMP
 	/* Turn on hardware trace buffer */
 	bfin_write_TBUFCTL(TBUFPWR | TBUFEN);
diff --git a/cpu/blackfin/i2c.c b/cpu/blackfin/i2c.c
deleted file mode 100644
index 2a3e223..0000000
--- a/cpu/blackfin/i2c.c
+++ /dev/null
@@ -1,428 +0,0 @@
-/*
- * i2c.c - driver for Blackfin on-chip TWI/I2C
- *
- * Copyright (c) 2006-2008 Analog Devices Inc.
- *
- * Licensed under the GPL-2 or later.
- */
-
-#include <common.h>
-
-#ifdef CONFIG_HARD_I2C
-
-#include <asm/blackfin.h>
-#include <i2c.h>
-#include <asm/io.h>
-#include <asm/mach-common/bits/twi.h>
-
-/* Two-Wire Interface		(0xFFC01400 - 0xFFC014FF) */
-#ifdef TWI0_CLKDIV
-#define bfin_read_TWI_CLKDIV()               bfin_read_TWI0_CLKDIV()
-#define bfin_write_TWI_CLKDIV(val)           bfin_write_TWI0_CLKDIV(val)
-#define bfin_read_TWI_CONTROL()              bfin_read_TWI0_CONTROL()
-#define bfin_write_TWI_CONTROL(val)          bfin_write_TWI0_CONTROL(val)
-#define bfin_read_TWI_SLAVE_CTL()            bfin_read_TWI0_SLAVE_CTL()
-#define bfin_write_TWI_SLAVE_CTL(val)        bfin_write_TWI0_SLAVE_CTL(val)
-#define bfin_read_TWI_SLAVE_STAT()           bfin_read_TWI0_SLAVE_STAT()
-#define bfin_write_TWI_SLAVE_STAT(val)       bfin_write_TWI0_SLAVE_STAT(val)
-#define bfin_read_TWI_SLAVE_ADDR()           bfin_read_TWI0_SLAVE_ADDR()
-#define bfin_write_TWI_SLAVE_ADDR(val)       bfin_write_TWI0_SLAVE_ADDR(val)
-#define bfin_read_TWI_MASTER_CTL()           bfin_read_TWI0_MASTER_CTL()
-#define bfin_write_TWI_MASTER_CTL(val)       bfin_write_TWI0_MASTER_CTL(val)
-#define bfin_read_TWI_MASTER_STAT()          bfin_read_TWI0_MASTER_STAT()
-#define bfin_write_TWI_MASTER_STAT(val)      bfin_write_TWI0_MASTER_STAT(val)
-#define bfin_read_TWI_MASTER_ADDR()          bfin_read_TWI0_MASTER_ADDR()
-#define bfin_write_TWI_MASTER_ADDR(val)      bfin_write_TWI0_MASTER_ADDR(val)
-#define bfin_read_TWI_INT_STAT()             bfin_read_TWI0_INT_STAT()
-#define bfin_write_TWI_INT_STAT(val)         bfin_write_TWI0_INT_STAT(val)
-#define bfin_read_TWI_INT_MASK()             bfin_read_TWI0_INT_MASK()
-#define bfin_write_TWI_INT_MASK(val)         bfin_write_TWI0_INT_MASK(val)
-#define bfin_read_TWI_FIFO_CTL()             bfin_read_TWI0_FIFO_CTL()
-#define bfin_write_TWI_FIFO_CTL(val)         bfin_write_TWI0_FIFO_CTL(val)
-#define bfin_read_TWI_FIFO_STAT()            bfin_read_TWI0_FIFO_STAT()
-#define bfin_write_TWI_FIFO_STAT(val)        bfin_write_TWI0_FIFO_STAT(val)
-#define bfin_read_TWI_XMT_DATA8()            bfin_read_TWI0_XMT_DATA8()
-#define bfin_write_TWI_XMT_DATA8(val)        bfin_write_TWI0_XMT_DATA8(val)
-#define bfin_read_TWI_XMT_DATA_16()          bfin_read_TWI0_XMT_DATA16()
-#define bfin_write_TWI_XMT_DATA16(val)       bfin_write_TWI0_XMT_DATA16(val)
-#define bfin_read_TWI_RCV_DATA8()            bfin_read_TWI0_RCV_DATA8()
-#define bfin_write_TWI_RCV_DATA8(val)        bfin_write_TWI0_RCV_DATA8(val)
-#define bfin_read_TWI_RCV_DATA16()           bfin_read_TWI0_RCV_DATA16()
-#define bfin_write_TWI_RCV_DATA16(val)       bfin_write_TWI0_RCV_DATA16(val)
-#endif
-
-#ifdef DEBUG_I2C
-#define PRINTD(fmt,args...)	do {	\
-	DECLARE_GLOBAL_DATA_PTR;	\
-	if (gd->have_console)		\
-		printf(fmt ,##args);	\
-	} while (0)
-#else
-#define PRINTD(fmt,args...)
-#endif
-
-#ifndef CONFIG_TWICLK_KHZ
-#define CONFIG_TWICLK_KHZ	50
-#endif
-
-/* All transfers are described by this data structure */
-struct i2c_msg {
-	u16 addr;		/* slave address */
-	u16 flags;
-#define I2C_M_STOP		0x2
-#define I2C_M_RD		0x1
-	u16 len;		/* msg length */
-	u8 *buf;		/* pointer to msg data */
-};
-
-/**
- * i2c_reset: - reset the host controller
- */
-static void i2c_reset(void)
-{
-	/* Disable TWI */
-	bfin_write_TWI_CONTROL(0);
-	SSYNC();
-
-	/* Set TWI internal clock as 10MHz */
-	bfin_write_TWI_CONTROL(((get_sclk() / 1024 / 1024 + 5) / 10) & 0x7F);
-
-	/* Set Twi interface clock as specified */
-	if (CONFIG_TWICLK_KHZ > 400)
-		bfin_write_TWI_CLKDIV(((5 * 1024 / 400) << 8) | ((5 * 1024 /
-						400) & 0xFF));
-	else
-		bfin_write_TWI_CLKDIV(((5 * 1024 /
-					CONFIG_TWICLK_KHZ) << 8) | ((5 * 1024 /
-						CONFIG_TWICLK_KHZ)
-						& 0xFF));
-
-	/* Enable TWI */
-	bfin_write_TWI_CONTROL(bfin_read_TWI_CONTROL() | TWI_ENA);
-	SSYNC();
-}
-
-int wait_for_completion(struct i2c_msg *msg, int timeout_count)
-{
-	unsigned short twi_int_stat;
-	unsigned short mast_stat;
-	int i;
-
-	for (i = 0; i < timeout_count; i++) {
-		twi_int_stat = bfin_read_TWI_INT_STAT();
-		mast_stat = bfin_read_TWI_MASTER_STAT();
-
-		if (XMTSERV & twi_int_stat) {
-			/* Transmit next data */
-			if (msg->len > 0) {
-				bfin_write_TWI_XMT_DATA8(*(msg->buf++));
-				msg->len--;
-			} else if (msg->flags & I2C_M_STOP)
-				bfin_write_TWI_MASTER_CTL
-				    (bfin_read_TWI_MASTER_CTL() | STOP);
-			SSYNC();
-			/* Clear status */
-			bfin_write_TWI_INT_STAT(XMTSERV);
-			SSYNC();
-			i = 0;
-		}
-		if (RCVSERV & twi_int_stat) {
-			if (msg->len > 0) {
-				/* Receive next data */
-				*(msg->buf++) = bfin_read_TWI_RCV_DATA8();
-				msg->len--;
-			} else if (msg->flags & I2C_M_STOP) {
-				bfin_write_TWI_MASTER_CTL
-				    (bfin_read_TWI_MASTER_CTL() | STOP);
-				SSYNC();
-			}
-			/* Clear interrupt source */
-			bfin_write_TWI_INT_STAT(RCVSERV);
-			SSYNC();
-			i = 0;
-		}
-		if (MERR & twi_int_stat) {
-			bfin_write_TWI_INT_STAT(MERR);
-			bfin_write_TWI_INT_MASK(0);
-			bfin_write_TWI_MASTER_STAT(0x3e);
-			bfin_write_TWI_MASTER_CTL(0);
-			SSYNC();
-			/*
-			 * if both err and complete int stats are set,
-			 * return proper results.
-			 */
-			if (MCOMP & twi_int_stat) {
-				bfin_write_TWI_INT_STAT(MCOMP);
-				bfin_write_TWI_INT_MASK(0);
-				bfin_write_TWI_MASTER_CTL(0);
-				SSYNC();
-				/*
-				 * If it is a quick transfer,
-				 * only address bug no data, not an err.
-				 */
-				if (msg->len == 0 && mast_stat & BUFRDERR)
-					return 0;
-				/*
-				 * If address not acknowledged return -3,
-				 * else return 0.
-				 */
-				else if (!(mast_stat & ANAK))
-					return 0;
-				else
-					return -3;
-			}
-			return -1;
-		}
-		if (MCOMP & twi_int_stat) {
-			bfin_write_TWI_INT_STAT(MCOMP);
-			SSYNC();
-			bfin_write_TWI_INT_MASK(0);
-			bfin_write_TWI_MASTER_CTL(0);
-			SSYNC();
-			return 0;
-		}
-	}
-	if (msg->flags & I2C_M_RD)
-		return -4;
-	else
-		return -2;
-}
-
-/**
- * i2c_transfer: - Transfer one byte over the i2c bus
- *
- * This function can tranfer a byte over the i2c bus in both directions.
- * It is used by the public API functions.
- *
- * @return:	 0: transfer successful
- *		-1: transfer fail
- *		-2: transmit timeout
- *		-3: ACK missing
- *		-4: receive timeout
- *		-5: controller not ready
- */
-int i2c_transfer(struct i2c_msg *msg)
-{
-	int ret = 0;
-	int timeout_count = 10000;
-	int len = msg->len;
-
-	if (!(bfin_read_TWI_CONTROL() & TWI_ENA)) {
-		ret = -5;
-		goto transfer_error;
-	}
-
-	while (bfin_read_TWI_MASTER_STAT() & BUSBUSY)
-		continue;
-
-	/* Set Transmit device address */
-	bfin_write_TWI_MASTER_ADDR(msg->addr);
-
-	/*
-	 * FIFO Initiation.
-	 * Data in FIFO should be discarded before start a new operation.
-	 */
-	bfin_write_TWI_FIFO_CTL(0x3);
-	SSYNC();
-	bfin_write_TWI_FIFO_CTL(0);
-	SSYNC();
-
-	if (!(msg->flags & I2C_M_RD)) {
-		/* Transmit first data */
-		if (msg->len > 0) {
-			PRINTD("1 in i2c_transfer: buf=%d, len=%d\n", *msg->buf,
-			       len);
-			bfin_write_TWI_XMT_DATA8(*(msg->buf++));
-			msg->len--;
-			SSYNC();
-		}
-	}
-
-	/* clear int stat */
-	bfin_write_TWI_INT_STAT(MERR | MCOMP | XMTSERV | RCVSERV);
-
-	/* Interrupt mask . Enable XMT, RCV interrupt */
-	bfin_write_TWI_INT_MASK(MCOMP | MERR |
-			((msg->flags & I2C_M_RD) ? RCVSERV : XMTSERV));
-	SSYNC();
-
-	if (len > 0 && len <= 255)
-		bfin_write_TWI_MASTER_CTL((len << 6));
-	else if (msg->len > 255) {
-		bfin_write_TWI_MASTER_CTL((0xff << 6));
-		msg->flags &= I2C_M_STOP;
-	} else
-		bfin_write_TWI_MASTER_CTL(0);
-
-	/* Master enable */
-	bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN |
-			((msg->flags & I2C_M_RD)
-			 ? MDIR : 0) | ((CONFIG_TWICLK_KHZ >
-					 100) ? FAST : 0));
-	SSYNC();
-
-	ret = wait_for_completion(msg, timeout_count);
-	PRINTD("3 in i2c_transfer: ret=%d\n", ret);
-
- transfer_error:
-	switch (ret) {
-	case 1:
-		PRINTD(("i2c_transfer: error: transfer fail\n"));
-		break;
-	case 2:
-		PRINTD(("i2c_transfer: error: transmit timeout\n"));
-		break;
-	case 3:
-		PRINTD(("i2c_transfer: error: ACK missing\n"));
-		break;
-	case 4:
-		PRINTD(("i2c_transfer: error: receive timeout\n"));
-		break;
-	case 5:
-		PRINTD(("i2c_transfer: error: controller not ready\n"));
-		i2c_reset();
-		break;
-	default:
-		break;
-	}
-	return ret;
-
-}
-
-/* ---------------------------------------------------------------------*/
-/* API Functions							*/
-/* ---------------------------------------------------------------------*/
-
-void i2c_init(int speed, int slaveaddr)
-{
-	i2c_reset();
-}
-
-/**
- * i2c_probe: - Test if a chip answers for a given i2c address
- *
- * @chip:	address of the chip which is searched for
- * @return:	0 if a chip was found, -1 otherwhise
- */
-
-int i2c_probe(uchar chip)
-{
-	struct i2c_msg msg;
-	u8 probebuf;
-
-	i2c_reset();
-
-	probebuf = 0;
-	msg.addr = chip;
-	msg.flags = 0;
-	msg.len = 1;
-	msg.buf = &probebuf;
-	if (i2c_transfer(&msg))
-		return -1;
-
-	msg.addr = chip;
-	msg.flags = I2C_M_RD;
-	msg.len = 1;
-	msg.buf = &probebuf;
-	if (i2c_transfer(&msg))
-		return -1;
-
-	return 0;
-}
-
-/**
- *   i2c_read: - Read multiple bytes from an i2c device
- *
- *   chip:    I2C chip address, range 0..127
- *   addr:    Memory (register) address within the chip
- *   alen:    Number of bytes to use for addr (typically 1, 2 for larger
- *		memories, 0 for register type devices with only one
- *		register)
- *   buffer:  Where to read/write the data
- *   len:     How many bytes to read/write
- *
- *   Returns: 0 on success, not 0 on failure
- */
-
-int i2c_read(uchar chip, uint addr, int alen, uchar * buffer, int len)
-{
-	struct i2c_msg msg;
-	u8 addr_bytes[3];	/* lowest...highest byte of data address */
-
-	PRINTD("i2c_read: chip=0x%x, addr=0x%x, alen=0x%x, len=0x%x\n", chip,
-			addr, alen, len);
-
-	if (alen > 0) {
-		addr_bytes[0] = (u8) ((addr >> 0) & 0x000000FF);
-		addr_bytes[1] = (u8) ((addr >> 8) & 0x000000FF);
-		addr_bytes[2] = (u8) ((addr >> 16) & 0x000000FF);
-		msg.addr = chip;
-		msg.flags = 0;
-		msg.len = alen;
-		msg.buf = addr_bytes;
-		if (i2c_transfer(&msg))
-			return -1;
-	}
-
-	/* start read sequence */
-	PRINTD(("i2c_read: start read sequence\n"));
-	msg.addr = chip;
-	msg.flags = I2C_M_RD;
-	msg.len = len;
-	msg.buf = buffer;
-	if (i2c_transfer(&msg))
-		return -1;
-
-	return 0;
-}
-
-/**
- *   i2c_write: -  Write multiple bytes to an i2c device
- *
- *   chip:    I2C chip address, range 0..127
- *   addr:    Memory (register) address within the chip
- *   alen:    Number of bytes to use for addr (typically 1, 2 for larger
- *		memories, 0 for register type devices with only one
- *		register)
- *   buffer:  Where to read/write the data
- *   len:     How many bytes to read/write
- *
- *   Returns: 0 on success, not 0 on failure
- */
-
-int i2c_write(uchar chip, uint addr, int alen, uchar * buffer, int len)
-{
-	struct i2c_msg msg;
-	u8 addr_bytes[3];	/* lowest...highest byte of data address */
-
-	PRINTD
-		("i2c_write: chip=0x%x, addr=0x%x, alen=0x%x, len=0x%x, buf0=0x%x\n",
-		 chip, addr, alen, len, buffer[0]);
-
-	/* chip address write */
-	if (alen > 0) {
-		addr_bytes[0] = (u8) ((addr >> 0) & 0x000000FF);
-		addr_bytes[1] = (u8) ((addr >> 8) & 0x000000FF);
-		addr_bytes[2] = (u8) ((addr >> 16) & 0x000000FF);
-		msg.addr = chip;
-		msg.flags = 0;
-		msg.len = alen;
-		msg.buf = addr_bytes;
-		if (i2c_transfer(&msg))
-			return -1;
-	}
-
-	/* start read sequence */
-	PRINTD(("i2c_write: start write sequence\n"));
-	msg.addr = chip;
-	msg.flags = 0;
-	msg.len = len;
-	msg.buf = buffer;
-	if (i2c_transfer(&msg))
-		return -1;
-
-	return 0;
-
-}
-
-#endif /* CONFIG_HARD_I2C */
diff --git a/cpu/blackfin/initcode.c b/cpu/blackfin/initcode.c
index ffc8420..e733dd2 100644
--- a/cpu/blackfin/initcode.c
+++ b/cpu/blackfin/initcode.c
@@ -158,7 +158,7 @@ static inline void serial_putc(char c)
 #endif
 
 #ifndef CONFIG_PLL_CTL_VAL
-# define CONFIG_PLL_CTL_VAL (SPORT_HYST | (CONFIG_VCO_MULT << 9))
+# define CONFIG_PLL_CTL_VAL (SPORT_HYST | (CONFIG_VCO_MULT << 9) | CONFIG_CLKIN_HALF)
 #endif
 
 #ifndef CONFIG_EBIU_RSTCTL_VAL
diff --git a/cpu/blackfin/jtag-console.c b/cpu/blackfin/jtag-console.c
new file mode 100644
index 0000000..44c0a83
--- /dev/null
+++ b/cpu/blackfin/jtag-console.c
@@ -0,0 +1,125 @@
+/*
+ * jtag-console.c - console driver over Blackfin JTAG
+ *
+ * Copyright (c) 2008 Analog Devices Inc.
+ *
+ * Licensed under the GPL-2 or later.
+ */
+
+#include <common.h>
+#include <devices.h>
+#include <asm/blackfin.h>
+
+#ifndef CONFIG_JTAG_CONSOLE_TIMEOUT
+# define CONFIG_JTAG_CONSOLE_TIMEOUT 100
+#endif
+
+/* The Blackfin tends to be much much faster than the JTAG hardware. */
+static void jtag_write_emudat(uint32_t emudat)
+{
+	static bool overflowed = false;
+	ulong timeout = get_timer(0) + CONFIG_JTAG_CONSOLE_TIMEOUT;
+	while (bfin_read_DBGSTAT() & 0x1) {
+		if (overflowed)
+			return;
+		if (timeout < get_timer(0))
+			overflowed = true;
+	}
+	overflowed = false;
+	__asm__ __volatile__("emudat = %0;" : : "d"(emudat));
+}
+/* Transmit a buffer.  The format is:
+ * [32bit length][actual data]
+ */
+static void jtag_send(const char *c, uint32_t len)
+{
+	uint32_t i;
+
+	if (len == 0)
+		return;
+
+	/* First send the length */
+	jtag_write_emudat(len);
+
+	/* Then send the data */
+	for (i = 0; i < len; i += 4)
+		jtag_write_emudat((c[i] << 0) | (c[i+1] << 8) | (c[i+2] << 16) | (c[i+3] << 24));
+}
+static void jtag_putc(const char c)
+{
+	jtag_send(&c, 1);
+}
+static void jtag_puts(const char *s)
+{
+	jtag_send(s, strlen(s));
+}
+
+static int jtag_tstc(void)
+{
+	return (bfin_read_DBGSTAT() & 0x2);
+}
+
+/* Receive a buffer.  The format is:
+ * [32bit length][actual data]
+ */
+static size_t inbound_len;
+static int leftovers_len;
+static uint32_t leftovers;
+static int jtag_getc(void)
+{
+	int ret;
+	uint32_t emudat;
+
+	/* see if any data is left over */
+	if (leftovers_len) {
+		--leftovers_len;
+		ret = leftovers & 0xff;
+		leftovers >>= 8;
+		return ret;
+	}
+
+	/* wait for new data ! */
+	while (!jtag_tstc())
+		continue;
+	__asm__("%0 = emudat;" : "=d"(emudat));
+
+	if (inbound_len == 0) {
+		/* grab the length */
+		inbound_len = emudat;
+	} else {
+		/* store the bytes */
+		leftovers_len = min(4, inbound_len);
+		inbound_len -= leftovers_len;
+		leftovers = emudat;
+	}
+
+	return jtag_getc();
+}
+
+int drv_jtag_console_init(void)
+{
+	device_t dev;
+	int ret;
+
+	memset(&dev, 0x00, sizeof(dev));
+	strcpy(dev.name, "jtag");
+	dev.flags = DEV_FLAGS_OUTPUT | DEV_FLAGS_INPUT | DEV_FLAGS_SYSTEM;
+	dev.putc = jtag_putc;
+	dev.puts = jtag_puts;
+	dev.tstc = jtag_tstc;
+	dev.getc = jtag_getc;
+
+	ret = device_register(&dev);
+	return (ret == 0 ? 1 : ret);
+}
+
+#ifdef CONFIG_UART_CONSOLE_IS_JTAG
+/* Since the JTAG is always available (at power on), allow it to fake a UART */
+void serial_set_baud(uint32_t baud) {}
+void serial_setbrg(void)            {}
+int serial_init(void)               { return 0; }
+void serial_putc(const char c)      __attribute__((alias("jtag_putc")));
+void serial_puts(const char *s)     __attribute__((alias("jtag_puts")));
+int serial_tstc(void)               __attribute__((alias("jtag_tstc")));
+int serial_getc(void)               __attribute__((alias("jtag_getc")));
+#endif
diff --git a/cpu/blackfin/reset.c b/cpu/blackfin/reset.c
index d1e34b3..284cea5 100644
--- a/cpu/blackfin/reset.c
+++ b/cpu/blackfin/reset.c
@@ -29,26 +29,35 @@ void bfin_reset(void)
 	 */
 	__builtin_bfin_ssync();
 
-	while (1) {
+	/* The bootrom checks to see how it was reset and will
+	 * automatically perform a software reset for us when
+	 * it starts executing after the core reset.
+	 */
+	if (ANOMALY_05000353 || ANOMALY_05000386) {
 		/* Initiate System software reset. */
 		bfin_write_SWRST(0x7);
 
 		/* Due to the way reset is handled in the hardware, we need
-		 * to delay for 7 SCLKS.  The only reliable way to do this is
-		 * to calculate the CCLK/SCLK ratio and multiply 7.  For now,
+		 * to delay for 10 SCLKS.  The only reliable way to do this is
+		 * to calculate the CCLK/SCLK ratio and multiply 10.  For now,
 		 * we'll assume worse case which is a 1:15 ratio.
 		 */
 		asm(
 			"LSETUP (1f, 1f) LC0 = %0\n"
 			"1: nop;"
 			:
-			: "a" (15 * 7)
+			: "a" (15 * 10)
 			: "LC0", "LB0", "LT0"
 		);
 
 		/* Clear System software reset */
 		bfin_write_SWRST(0);
 
+		/* The BF526 ROM will crash during reset */
+#if defined(__ADSPBF522__) || defined(__ADSPBF524__) || defined(__ADSPBF526__)
+		bfin_read_SWRST();
+#endif
+
 		/* Wait for the SWRST write to complete.  Cannot rely on SSYNC
 		 * though as the System state is all reset now.
 		 */
@@ -59,10 +68,11 @@ void bfin_reset(void)
 			: "a" (15 * 1)
 			: "LC1", "LB1", "LT1"
 		);
+	}
 
+	while (1)
 		/* Issue core reset */
 		asm("raise 1");
-	}
 }
 
 /* We need to trampoline ourselves up into L1 since our linker
diff --git a/cpu/blackfin/serial.c b/cpu/blackfin/serial.c
index 0d6f377..42534bd 100644
--- a/cpu/blackfin/serial.c
+++ b/cpu/blackfin/serial.c
@@ -29,6 +29,8 @@
 #include <asm/blackfin.h>
 #include <asm/mach-common/bits/uart.h>
 
+#ifdef CONFIG_UART_CONSOLE
+
 #if defined(UART_LSR) && (CONFIG_UART_CONSOLE != 0)
 # error CONFIG_UART_CONSOLE must be 0 on parts with only one UART
 #endif
@@ -170,3 +172,5 @@ void serial_puts(const char *s)
 	while (*s)
 		serial_putc(*s++);
 }
+
+#endif
diff --git a/cpu/blackfin/serial.h b/cpu/blackfin/serial.h
index ec40c26..f671096 100644
--- a/cpu/blackfin/serial.h
+++ b/cpu/blackfin/serial.h
@@ -14,6 +14,10 @@
 #include <asm/blackfin.h>
 #include <asm/mach-common/bits/uart.h>
 
+#ifndef CONFIG_UART_CONSOLE
+# define CONFIG_UART_CONSOLE 0
+#endif
+
 #ifdef CONFIG_DEBUG_EARLY_SERIAL
 # define BFIN_DEBUG_EARLY_SERIAL 1
 #else
@@ -95,7 +99,16 @@
 __attribute__((always_inline))
 static inline void serial_do_portmux(void)
 {
-#ifdef __ADSPBF52x__
+#if defined(__ADSPBF51x__)
+# define DO_MUX(port, mux_tx, mux_rx, tx, rx) \
+	bfin_write_PORT##port##_MUX((bfin_read_PORT##port##_MUX() & ~(PORT_x_MUX_##mux_tx##_MASK | PORT_x_MUX_##mux_rx##_MASK)) | PORT_x_MUX_##mux_tx##_FUNC_2 | PORT_x_MUX_##mux_rx##_FUNC_2); \
+	bfin_write_PORT##port##_FER(bfin_read_PORT##port##_FER() | P##port##tx | P##port##rx);
+	switch (CONFIG_UART_CONSOLE) {
+	case 0: DO_MUX(G, 5, 5, 9, 10);  break;	/* Port G; mux 5; PG9 and PG10 */
+	case 1: DO_MUX(F, 2, 3, 14, 15); break;	/* Port H; mux 2/3; PH14 and PH15 */
+	}
+	SSYNC();
+#elif defined(__ADSPBF52x__)
 # define DO_MUX(port, mux, tx, rx) \
 	bfin_write_PORT##port##_MUX((bfin_read_PORT##port##_MUX() & ~PORT_x_MUX_##mux##_MASK) | PORT_x_MUX_##mux##_FUNC_3); \
 	bfin_write_PORT##port##_FER(bfin_read_PORT##port##_FER() | P##port##tx | P##port##rx);
diff --git a/cpu/blackfin/start.S b/cpu/blackfin/start.S
index 9975a0c..6c8def4 100644
--- a/cpu/blackfin/start.S
+++ b/cpu/blackfin/start.S
@@ -125,8 +125,11 @@ ENTRY(_start)
 	 */
 	r6 = 1 (x);
 
-	/* Relocate from wherever are (FLASH/RAM/etc...) to the
-	 * hardcoded monitor location in the end of RAM.
+	/* Relocate from wherever we are (FLASH/RAM/etc...) to the hardcoded
+	 * monitor location in the end of RAM.  We know that memcpy() only
+	 * uses registers, so it is safe to call here.  Note that this only
+	 * copies to external memory ... we do not start executing out of
+	 * it yet (see "lower to 15" below).
 	 */
 	serial_early_puts("Relocate");
 	call _get_pc;
@@ -135,27 +138,16 @@ ENTRY(_start)
 	r2.h = .Loffset;
 	r3.l = _start;
 	r3.h = _start;
-	r1 = r2 - r3;
-
-	r0 = r0 - r1;
-
-	cc = r0 == r3;
+	r2 = r2 - r3;
+	r1 = r0 - r2;
+	cc = r1 == r3;
 	if cc jump .Lnorelocate;
-
 	r6 = 0 (x);
-	p1 = r0;
-
-	p2.l = LO(CONFIG_SYS_MONITOR_BASE);
-	p2.h = HI(CONFIG_SYS_MONITOR_BASE);
 
-	p3 = 0x04;
-	p4.l = LO(CONFIG_SYS_MONITOR_BASE + CONFIG_SYS_MONITOR_LEN);
-	p4.h = HI(CONFIG_SYS_MONITOR_BASE + CONFIG_SYS_MONITOR_LEN);
-.Lloop1:
-	r1 = [p1 ++ p3];
-	[p2 ++ p3] = r1;
-	cc=p2==p4;
-	if !cc jump .Lloop1;
+	r0 = r3;
+	r2.l = LO(CONFIG_SYS_MONITOR_LEN);
+	r2.h = HI(CONFIG_SYS_MONITOR_LEN);
+	call _memcpy_ASM;
 
 	/* Initialize BSS section ... we know that memset() does not
 	 * use the BSS, so it is safe to call here.  The bootrom LDR
@@ -173,9 +165,8 @@ ENTRY(_start)
 .Lnorelocate:
 
 	/* Setup the actual stack in external memory */
-	r0.h = HI(CONFIG_STACKBASE);
-	r0.l = LO(CONFIG_STACKBASE);
-	sp = r0;
+	sp.h = HI(CONFIG_STACKBASE);
+	sp.l = LO(CONFIG_STACKBASE);
 	fp = sp;
 
 	/* Now lower ourselves from the highest interrupt level to
@@ -183,7 +174,9 @@ ENTRY(_start)
 	 * setting the 15 handler to ".Lenable_nested", raising the 15
 	 * interrupt, and then returning from the highest interrupt
 	 * level to the dummy "jump" until the interrupt controller
-	 * services the pending 15 interrupt.
+	 * services the pending 15 interrupt.  If executing out of
+	 * flash, these steps also changes the code flow from flash
+	 * to external memory.
 	 */
 	serial_early_puts("Lower to 15");
 	r0 = r7;
diff --git a/cpu/blackfin/traps.c b/cpu/blackfin/traps.c
index d17c0a1..a2c6f1e 100644
--- a/cpu/blackfin/traps.c
+++ b/cpu/blackfin/traps.c
@@ -111,23 +111,12 @@ void trap_c(struct pt_regs *regs)
 		}
 		if (i == ARRAY_SIZE(bfin_memory_map)) {
 			printf("%cCPLB exception outside of memory map at 0x%p\n",
-				(data ? 'D' : 'I'), new_cplb_addr);
+				(data ? 'D' : 'I'), (void *)new_cplb_addr);
 			bfin_panic(regs);
 		} else
 			debug("CPLB addr %p matches map 0x%p - 0x%p\n", new_cplb_addr, bfin_memory_map[i].start, bfin_memory_map[i].end);
 		new_cplb_data = (data ? bfin_memory_map[i].data_flags : bfin_memory_map[i].inst_flags);
 
-		/* Turn the cache off */
-		SSYNC();
-		if (data) {
-			asm(" .align 8; ");
-			*pDMEM_CONTROL &= ~ENDCPLB;
-		} else {
-			asm(" .align 8; ");
-			*pIMEM_CONTROL &= ~ENICPLB;
-		}
-		SSYNC();
-
 		if (data) {
 			CPLB_ADDR_BASE = (uint32_t *)DCPLB_ADDR0;
 			CPLB_DATA_BASE = (uint32_t *)DCPLB_DATA0;
@@ -149,8 +138,17 @@ void trap_c(struct pt_regs *regs)
 
 		debug("evicting entry %i: 0x%p 0x%08X\n", i, *CPLB_ADDR, *CPLB_DATA);
 		last_evicted = i + 1;
+
+		/* need to turn off cplbs whenever we muck with the cplb table */
+#if ENDCPLB != ENICPLB
+# error cplb enable bit violates my sanity
+#endif
+		uint32_t mem_control = (data ? DMEM_CONTROL : IMEM_CONTROL);
+		bfin_write32(mem_control, bfin_read32(mem_control) & ~ENDCPLB);
 		*CPLB_ADDR = new_cplb_addr;
 		*CPLB_DATA = new_cplb_data;
+		bfin_write32(mem_control, bfin_read32(mem_control) | ENDCPLB);
+		SSYNC();
 
 		/* dump current table for debugging purposes */
 		CPLB_ADDR = CPLB_ADDR_BASE;
@@ -158,17 +156,6 @@ void trap_c(struct pt_regs *regs)
 		for (i = 0; i < 16; ++i)
 			debug("%2i 0x%p 0x%08X\n", i, *CPLB_ADDR++, *CPLB_DATA++);
 
-		/* Turn the cache back on */
-		SSYNC();
-		if (data) {
-			asm(" .align 8; ");
-			*pDMEM_CONTROL |= ENDCPLB;
-		} else {
-			asm(" .align 8; ");
-			*pIMEM_CONTROL |= ENICPLB;
-		}
-		SSYNC();
-
 		break;
 	}
 
@@ -220,20 +207,21 @@ static const char *symbol_lookup(unsigned long addr, unsigned long *caddr)
 static void decode_address(char *buf, unsigned long address)
 {
 	unsigned long sym_addr;
+	void *paddr = (void *)address;
 	const char *sym = symbol_lookup(address, &sym_addr);
 
 	if (sym) {
-		sprintf(buf, "<0x%p> { %s + 0x%x }", address, sym, address - sym_addr);
+		sprintf(buf, "<0x%p> { %s + 0x%lx }", paddr, sym, address - sym_addr);
 		return;
 	}
 
 	if (!address)
-		sprintf(buf, "<0x%p> /* Maybe null pointer? */", address);
+		sprintf(buf, "<0x%p> /* Maybe null pointer? */", paddr);
 	else if (address >= CONFIG_SYS_MONITOR_BASE &&
 	         address < CONFIG_SYS_MONITOR_BASE + CONFIG_SYS_MONITOR_LEN)
-		sprintf(buf, "<0x%p> /* somewhere in u-boot */", address);
+		sprintf(buf, "<0x%p> /* somewhere in u-boot */", paddr);
 	else
-		sprintf(buf, "<0x%p> /* unknown address */", address);
+		sprintf(buf, "<0x%p> /* unknown address */", paddr);
 }
 
 static char *strhwerrcause(uint16_t hwerrcause)
@@ -273,7 +261,7 @@ static char *strexcause(uint16_t excause)
 void dump(struct pt_regs *fp)
 {
 	char buf[150];
-	size_t i;
+	int i;
 	uint16_t hwerrcause, excause;
 
 	if (!ENABLE_DUMP)
@@ -288,8 +276,8 @@ void dump(struct pt_regs *fp)
 	printf("SEQUENCER STATUS:\n");
 	printf(" SEQSTAT: %08lx  IPEND: %04lx  SYSCFG: %04lx\n",
 		fp->seqstat, fp->ipend, fp->syscfg);
-	printf("  HWERRCAUSE: 0x%lx: %s\n", hwerrcause, strhwerrcause(hwerrcause));
-	printf("  EXCAUSE   : 0x%lx: %s\n", excause, strexcause(excause));
+	printf("  HWERRCAUSE: 0x%x: %s\n", hwerrcause, strhwerrcause(hwerrcause));
+	printf("  EXCAUSE   : 0x%x: %s\n", excause, strexcause(excause));
 	for (i = 6; i <= 15; ++i) {
 		if (fp->ipend & (1 << i)) {
 			decode_address(buf, bfin_read32(EVT0 + 4*i));
@@ -323,7 +311,7 @@ void dump(struct pt_regs *fp)
 	printf(" P0 : %08lx    P1 : %08lx    P2 : %08lx    P3 : %08lx\n",
 		fp->p0, fp->p1, fp->p2, fp->p3);
 	printf(" P4 : %08lx    P5 : %08lx    FP : %08lx    SP : %08lx\n",
-		fp->p4, fp->p5, fp->fp, fp);
+		fp->p4, fp->p5, fp->fp, (unsigned long)fp);
 	printf(" LB0: %08lx    LT0: %08lx    LC0: %08lx\n",
 		fp->lb0, fp->lt0, fp->lc0);
 	printf(" LB1: %08lx    LT1: %08lx    LC1: %08lx\n",
@@ -349,7 +337,7 @@ void dump_bfin_trace_buffer(void)
 {
 	char buf[150];
 	unsigned long tflags;
-	size_t i = 0;
+	int i = 0;
 
 	if (!ENABLE_DUMP)
 		return;