diff --git a/src/arch/alpha/boot/Config b/src/arch/alpha/boot/Config
new file mode 100644
index 0000000000..178df54605
--- /dev/null
+++ b/src/arch/alpha/boot/Config
@@ -0,0 +1 @@
+object boot.o
diff --git a/src/arch/alpha/boot/boot.c b/src/arch/alpha/boot/boot.c
new file mode 100644
index 0000000000..367fe6bc75
--- /dev/null
+++ b/src/arch/alpha/boot/boot.c
@@ -0,0 +1,197 @@
+#include <boot/uniform_boot.h>
+#include <boot/elf.h>
+#include <arch/boot/hwrpb.h>
+
+/* FIXME remove the hardcodes
+ * MAX_ASM, CMD_LINE, SYSNAME, CYCLE_FREQ, CPU, SYS_VARIATION, SYS_TYPE
+ */
+
+#ifndef CMD_LINE
+#define CMD_LINE ""
+#endif
+
+/* Magic values.
+ * Hard code these for the DS10 for now.
+ */
+#define HWRPB_SIGNATURE 	0x4857525042000000
+#define HWRPB_SYS_TYPE		ST_DEC_TSUNAMI
+#define HWRPB_SYS_VARIATION	(7 << 10)
+#define HWRPB_CPU		EV6_CPU
+#define HWRPB_CYCLE_FREQ	462962962
+#define HWRPB_INTR_FREQ		(4096*1024)
+#define HWRPB_SYSNAME		"linuxBIOS DS10"
+#define HWRPB_COMMAND_LINE	CMD_LINE
+#define HWRPB_MAX_ASN		255
+
+#define OFFSET(x, y) ((unsigned long)(((char *)&(x)) - ((char *)&(y))))
+#define ELEMENTS(x) (sizeof(x)/sizeof((x)[0]))
+#define ADDR(x) ((unsigned long)&(x))
+
+struct sysname_struct {
+	unsigned long length;
+	unsigned char name[30];
+};
+
+static struct boot_data {
+	struct hwrpb_struct hwrpb;
+	struct dsr_struct dsr;
+	struct sysname_struct sysname;
+	struct percpu_struct cpu[1];
+	struct memdesc_struct mem;
+	struct memclust_struct mem_cluster[2];
+} boot_data = 
+{
+	.hwrpb = {
+		.phys_addr = ADDR(boot_data.hwrpb),
+		.id = HWRPB_SIGNATURE,
+		.revision = 6,
+		.size = sizeof(boot_data.hwrpb),
+		.cpuid = 0,
+		.pagesize = 8192,
+		.pa_bits = 13,
+		.max_asn = HWRPB_MAX_ASN,
+		.ssn = "MILO-LinuxBIOS",
+		.sys_type = HWRPB_SYS_TYPE, 
+		.sys_variation = HWRPB_SYS_VARIATION,
+		.sys_revision = 0,
+		.intr_freq = HWRPB_INTR_FREQ,
+		.cycle_freq = HWRPB_CYCLE_FREQ,
+		.vptb = 0,
+		.res1 = 0,
+		.tbhb_offset = 0,
+		.nr_processors = ELEMENTS(boot_data.cpu),
+		.processor_size = sizeof(boot_data.cpu[0]),
+		.processor_offset = OFFSET(boot_data.cpu, boot_data.hwrpb),
+		.ctb_nr = 0,
+		.ctb_size = 0,
+		.ctbt_offset = 0,
+		.mddt_offset = OFFSET(boot_data.mem, boot_data.hwrpb),
+		.cdb_offset = 0,
+		.frut_offset = 0,
+		.save_terminal = 0,
+		.save_terminal_data = 0,
+		.restore_terminal = 0,
+		.restore_terminal_data = 0,
+		.CPU_restart = 0,
+		.CPU_restart_data = 0,
+		.res2 = 0,
+		.res3 = 0,
+		.chksum = 0,
+		.rxrdy = 0,
+		.txrdy = 0,
+		.dsr_offset = OFFSET(boot_data.dsr, boot_data.hwrpb),
+	},
+	.dsr = {
+		.smm = 0,
+		.lurt_off = 0,
+		.sysname_off = OFFSET(boot_data.sysname, boot_data.dsr),
+	},
+	.sysname = {
+		.length = sizeof(HWRPB_SYSNAME) -1,
+		.name = HWRPB_SYSNAME,
+	},
+
+	.cpu = {
+		{
+			.serial_no = { 
+				0x73695f78756e694c,
+				0x002174616572475f,
+			},
+			.type = HWRPB_CPU,
+		}
+	},
+	.mem = {
+		.chksum = 0,
+		.optional_pa = ADDR(boot_data.mem), 
+		.numclusters = ELEMENTS(boot_data.mem_cluster),
+	},
+	.mem_cluster = {
+		{
+			.start_pfn = 0,
+			.numpages = 128,
+			.numtested = 0,
+			.bitmap_va = 0,
+			.bitmap_pa = 0,
+			.bitmap_chksum = 0,
+			.usage = 1, /* console/PALcode reserved */
+		},
+		{
+			.start_pfn = 128,
+			.numpages = 0,
+			.numtested = 0,
+			.bitmap_va = 0,
+			.bitmap_pa = 0,
+			.bitmap_chksum = 0,
+			.usage = 0,
+		},
+	},
+};
+
+
+static struct {
+	struct uniform_boot_header header;
+	struct {
+		struct ube_hwrpb hwrpb;
+	} env;
+	unsigned char command_line[1024];
+} ube_all = {
+	.header = {
+		.header_bytes = sizeof(ube_all.header),
+		.header_checksum = 0,
+		.arg = (unsigned long)&ube_all.command_line,
+		.arg_bytes = sizeof(ube_all.command_line),
+		.env = (unsigned long)&ube_all.env,
+		.env_bytes = sizeof(ube_all.env),
+	},
+	.env = {
+		.hwrpb = {
+			.tag = UBE_TAG_HWRPB,
+			.size = sizeof(ube_all.env.hwrpb),
+			.hwrpb = (unsigned long)&boot_data.hwrpb,
+		},
+	},
+	.command_line = HWRPB_COMMAND_LINE,
+};
+
+static unsigned long hwrpb_compute_checksum(struct hwrpb_struct *hwrpb)
+{
+	unsigned long sum = 0, *l;
+        for (l = (unsigned long *)hwrpb; l <= (unsigned long *) &hwrpb->chksum; ++l)
+                sum += *l;
+	return sum;
+
+}
+
+
+void *get_ube_pointer(unsigned long totalram)
+{
+	/* Set the amount of RAM I have */
+	boot_data.mem_cluster[1].numpages = (totalram >> 3) -
+		boot_data.mem_cluster[1].start_pfn;
+	boot_data.hwrpb.chksum = 0;
+	boot_data.hwrpb.chksum = hwrpb_compute_checksum(&boot_data.hwrpb);
+	ube_all.header.header_checksum = 0;
+	ube_all.header.header_checksum = 
+		uniform_boot_compute_header_checksum(&ube_all.header);
+	return &ube_all;
+}
+
+
+int elf_check_arch(Elf_ehdr *ehdr)
+{
+	return (
+		(ehdr->e_machine == EM_ALPHA) &&
+		(ehdr->e_ident[EI_CLASS] == ELFCLASS64) &&
+		(ehdr->e_ident[EI_DATA] == ELFDATA2LSB)
+		);
+}
+
+void jmp_to_elf_entry(void *entry, void *ptr)
+{
+	void (*kernel_entry)(void *ptr);
+	kernel_entry = entry;
+
+	/* Jump to kernel */
+	kernel_entry(ptr);
+}
+
diff --git a/src/arch/alpha/lib/Config b/src/arch/alpha/lib/Config
new file mode 100644
index 0000000000..02e6d884cd
--- /dev/null
+++ b/src/arch/alpha/lib/Config
@@ -0,0 +1,9 @@
+object hardwaremain.o
+
+makedefine OBJECTS+= __divqu.o __remqu.o __divlu.o __remlu.o
+
+makerule __divqu.o: $(TOP)/src/arch/alpha/lib/divide.S; $(CC) $(CFLAGS) -DDIV -c -o __divqu.o $<
+makerule __remqu.o: $(TOP)/src/arch/alpha/lib/divide.S; $(CC) $(CFLAGS) -DREM -c -o __remqu.o $<
+makerule __divlu.o: $(TOP)/src/arch/alpha/lib/divide.S; $(CC) $(CFLAGS) -DDIV -DINTSIZE -c -o __divlu.o $<
+makerule __remlu.o: $(TOP)/src/arch/alpha/lib/divide.S; $(CC) $(CFLAGS) -DREM -DINTSIZE -c -o __remlu.o $<
+
diff --git a/src/arch/alpha/lib/divide.S b/src/arch/alpha/lib/divide.S
new file mode 100644
index 0000000000..a4bc1f751e
--- /dev/null
+++ b/src/arch/alpha/lib/divide.S
@@ -0,0 +1,195 @@
+/*
+ * cpu/ev6/divide.S
+ *
+ * (C) 1995 Linus Torvalds
+ *
+ * Alpha division..
+ */
+
+/*
+ * The alpha chip doesn't provide hardware division, so we have to do it
+ * by hand.  The compiler expects the functions
+ *
+ *	__divqu: 64-bit unsigned long divide
+ *	__remqu: 64-bit unsigned long remainder
+ *	__divqs/__remqs: signed 64-bit
+ *	__divlu/__remlu: unsigned 32-bit
+ *	__divls/__remls: signed 32-bit
+ *
+ * These are not normal C functions: instead of the normal
+ * calling sequence, these expect their arguments in registers
+ * $24 and $25, and return the result in $27. Register $28 may
+ * be clobbered (assembly temporary), anything else must be saved. 
+ *
+ * In short: painful.
+ *
+ * This is a rather simple bit-at-a-time algorithm: it's very good
+ * at dividing random 64-bit numbers, but the more usual case where
+ * the divisor is small is handled better by the DEC algorithm
+ * using lookup tables. This uses much less memory, though, and is
+ * nicer on the cache.. Besides, I don't know the copyright status
+ * of the DEC code.
+ */
+
+/*
+ * My temporaries:
+ *	$0 - current bit
+ *	$1 - shifted divisor
+ *	$2 - modulus/quotient
+ *
+ *	$23 - return address
+ *	$24 - dividend
+ *	$25 - divisor
+ *
+ *	$27 - quotient/modulus
+ *	$28 - compare status
+ */
+
+#define halt .long 0
+
+/*
+ * Select function type and registers
+ */
+#define mask	$0
+#define divisor	$1
+#define compare $28
+#define tmp1	$3
+#define tmp2	$4
+
+#ifdef DIV
+#define DIV_ONLY(x,y...) x,##y
+#define MOD_ONLY(x,y...)
+#define func(x) __div##x
+#define modulus $2
+#define quotient $27
+#define GETSIGN(x) xor $24,$25,x
+#define STACK 48
+#else
+#define DIV_ONLY(x,y...)
+#define MOD_ONLY(x,y...) x,##y
+#define func(x) __rem##x
+#define modulus $27
+#define quotient $2
+#define GETSIGN(x) bis $24,$24,x
+#define STACK 32
+#endif
+
+/*
+ * For 32-bit operations, we need to extend to 64-bit
+ */
+#ifdef INTSIZE
+#define ufunction func(lu)
+#define sfunction func(l)
+#define LONGIFY(x) zapnot x,15,x
+#define SLONGIFY(x) addl x,0,x
+#else
+#define ufunction func(qu)
+#define sfunction func(q)
+#define LONGIFY(x)
+#define SLONGIFY(x)
+#endif
+
+.set noat
+.align	3
+.globl	ufunction
+.ent	ufunction
+ufunction:
+	subq	$30,STACK,$30
+	.frame	$30,STACK,$23
+	.prologue 0
+
+7:	stq	$1, 0($30)
+	bis	$25,$25,divisor
+	stq	$2, 8($30)
+	bis	$24,$24,modulus
+	stq	$0,16($30)
+	bis	$31,$31,quotient
+	LONGIFY(divisor)
+	stq	tmp1,24($30)
+	LONGIFY(modulus)
+	bis	$31,1,mask
+	DIV_ONLY(stq tmp2,32($30))
+	beq	divisor, 9f			/* div by zero */
+
+#ifdef INTSIZE
+	/*
+	 * shift divisor left, using 3-bit shifts for
+	 * 32-bit divides as we can't overflow. Three-bit
+	 * shifts will result in looping three times less
+	 * here, but can result in two loops more later.
+	 * Thus using a large shift isn't worth it (and
+	 * s8add pairs better than a sll..)
+	 */
+1:	cmpult	divisor,modulus,compare
+	s8addq	divisor,$31,divisor
+	s8addq	mask,$31,mask
+	bne	compare,1b
+#else
+1:	cmpult	divisor,modulus,compare
+	blt     divisor, 2f
+	addq	divisor,divisor,divisor
+	addq	mask,mask,mask
+	bne	compare,1b
+	unop
+#endif
+
+	/* ok, start to go right again.. */
+2:	DIV_ONLY(addq quotient,mask,tmp2)
+	srl	mask,1,mask
+	cmpule	divisor,modulus,compare
+	subq	modulus,divisor,tmp1
+	DIV_ONLY(cmovne compare,tmp2,quotient)
+	srl	divisor,1,divisor
+	cmovne	compare,tmp1,modulus
+	bne	mask,2b
+
+9:	ldq	$1, 0($30)
+	ldq	$2, 8($30)
+	ldq	$0,16($30)
+	ldq	tmp1,24($30)
+	DIV_ONLY(ldq tmp2,32($30))
+	addq	$30,STACK,$30
+	ret	$31,($23),1
+	.end	ufunction
+
+/*
+ * Uhh.. Ugly signed division. I'd rather not have it at all, but
+ * it's needed in some circumstances. There are different ways to
+ * handle this, really. This does:
+ * 	-a / b = a / -b = -(a / b)
+ *	-a % b = -(a % b)
+ *	a % -b = a % b
+ * which is probably not the best solution, but at least should
+ * have the property that (x/y)*y + (x%y) = x.
+ */
+.align 3
+.globl	sfunction
+.ent	sfunction
+sfunction:
+	subq	$30,STACK,$30
+	.frame	$30,STACK,$23
+	.prologue 0
+	bis	$24,$25,$28
+	SLONGIFY($28)
+	bge	$28,7b
+	stq	$24,0($30)
+	subq	$31,$24,$28
+	stq	$25,8($30)
+	cmovlt	$24,$28,$24	/* abs($24) */
+	stq	$23,16($30)
+	subq	$31,$25,$28
+	stq	tmp1,24($30)
+	cmovlt	$25,$28,$25	/* abs($25) */
+	unop
+	bsr	$23,ufunction
+	ldq	$24,0($30)
+	ldq	$25,8($30)
+	GETSIGN($28)
+	subq	$31,$27,tmp1
+	SLONGIFY($28)
+	ldq	$23,16($30)
+	cmovlt	$28,tmp1,$27
+	ldq	tmp1,24($30)
+	addq	$30,STACK,$30
+	ret	$31,($23),1
+	.end	sfunction
diff --git a/src/arch/alpha/lib/hardwaremain.c b/src/arch/alpha/lib/hardwaremain.c
new file mode 100644
index 0000000000..aaf634911f
--- /dev/null
+++ b/src/arch/alpha/lib/hardwaremain.c
@@ -0,0 +1,29 @@
+#include <stdlib.h>
+#include <boot/elf.h>
+#include <printk.h>
+#include <subr.h>
+
+void hardwaremain(unsigned long signature, unsigned long memsize, 
+	unsigned long cpu_speed)
+{
+	early_mainboard_init();
+	displayinit();
+
+	printk("\n\nsignature=0x%016lx memsize=0x%016lx cpu_speed=0x%016lx\n",
+		signature, memsize, cpu_speed);
+	elfboot(memsize >> 10 /* In kilobytes */);
+	printk("\n after elfboot\n");
+}
+
+void fatal_error(unsigned long exception_handler, unsigned long exception_at, 
+	unsigned long return_address)
+{
+	/* Just in case we are totally messed up */
+	early_mainboard_init();
+	displayinit();
+	printk("\n\nFault: 0x%0lx\n", exception_handler);
+	printk("PC: 0x%0lx\n", exception_at);
+	printk("RA: 0x%0lx\n", return_address);
+	/* Now spin forever */
+	while(1) {;}
+}