diff --git a/src/asm/Makefile b/src/asm/Makefile
index b574de2..f20064d 100644
--- a/src/asm/Makefile
+++ b/src/asm/Makefile
@@ -1,14 +1,20 @@
 TARGET=r3000
 ROM_WORDADDR_WIDTH=12
 RAM_WORDADDR_WIDTH=11
+BUILD_DIR:=./build
 
 AR=mips-elf-ar
 CC=mips-elf-gcc
 LD=mips-elf-ld
 OBJCOPY=mips-elf-objcopy
+OBJDUMP=mips-elf-objdump
 ROMGEN=romgen
 
-CFLAGS=-march=$(TARGET)
+CFLAGS := -march=$(TARGET) -L. -T link/link.ld -Wl,--gc-sections -G 0
+CFLAGS += -fno-exceptions -ffunction-sections -fdata-sections -fstrict-volatile-bitfields -nostartfiles
+LIBS := -lsys -lc
+
+all: testbench.elf
 
 libsys.a:
 	$(CC) $(CFLAGS) -c libsys/startup.s -o startup.o
@@ -19,13 +25,17 @@ libsys.a:
 
 
 testbench.elf: libsys.a
-	$(CC) $(CFLAGS) -c testbench.c -o testbench.o
-	$(LD) $(CFLAGS) -lsys -o $@
-	$(OBJCOPY) -j .stext -j .ktext -j .text $@ -O binary $@.ROM.bin
-	$(OBJCOPY) -j .data $@.elf -O binary $@.RAM.bin
-	$(ROMGEN) $@.ROM.bin $ROM_WORDADDR_WIDTH
-	$(ROMGEN) $@.RAM.bin $RAM_WORDADDR_WIDTH
+	$(CC) $(CFLAGS) -o $@ testbench.c $(LIBS)
+	$(OBJDUMP) -d $@ > $@.dis
+	$(OBJCOPY) $@ -O binary $@.ROM.bin
+	$(OBJCOPY) -j .data $@ -O binary $@.RAM.bin
+	$(ROMGEN) $@.ROM.bin $(ROM_WORDADDR_WIDTH)
+	$(ROMGEN) $@.RAM.bin $(RAM_WORDADDR_WIDTH)
 
 #xtclsh.exe $TARGET.RAM.tcl
 #xtclsh.exe $TARGET.ROM.tcl
 
+clean:
+	rm -rf libsys.a
+	rm -rf testbench.elf
+	
diff --git a/src/asm/libsys/init.s b/src/asm/libsys/init.s
index 41a35da..0d4e38f 100644
--- a/src/asm/libsys/init.s
+++ b/src/asm/libsys/init.s
@@ -1,9 +1,10 @@
 		.file	"init.s"
 	
-		.text
-		.align	2
+		.section .init,"ax"
+		.align	4
 		.globl _init
 		.extern end
+		.extern main
 	
 _init:
 		.set	noreorder
diff --git a/src/asm/libsys/startup.s b/src/asm/libsys/startup.s
index a53059e..610a3fc 100644
--- a/src/asm/libsys/startup.s
+++ b/src/asm/libsys/startup.s
@@ -1,8 +1,8 @@
 		.file	"startup.s"
 	
-		.section .stext,"ax"
+		.section .init,"ax"
 		.extern _init
-		.align	2
+		.align	4
 	
 _start:
 		.set	noreorder
diff --git a/src/asm/link/link.ld b/src/asm/link/link.ld
index e175fa8..e22c0c9 100644
--- a/src/asm/link/link.ld
+++ b/src/asm/link/link.ld
@@ -1,7 +1,10 @@
+/* Specify the memory areas */
 MEMORY
 {
-  imem       : ORIGIN = 0x00400000, LENGTH = 0x00004000  /*   8K    */
-  dmem       : ORIGIN = 0x00000000, LENGTH = 0x00002000  /*   8K    */
+  rom          : ORIGIN = 0xBFC00000, LENGTH = 0x00002000  /*   8K    */
+  ram          : ORIGIN = 0x40000000, LENGTH = 0x04000000  /*  64M    */
+  flash        : ORIGIN = 0x00000000, LENGTH = 0x00800000  /*   8M    */
+  io           : ORIGIN = 0xA0000000, LENGTH = 0x00800000  /*   8M    */
 }
 
 end = 0x7FFFEFFC;
@@ -13,35 +16,86 @@ sys_uart_data = 0x80000004;
 sys_timer_sec = 0x80000014;
 sys_timer_usec = 0x80000010;
 
+STARTUP(startup.o)
+INPUT(kernel.o)
+
+DEBUGGER_VAR_SIZE = 0x10000;
 SECTIONS
 {
-   .stext 0x00400000 :
-   {
-     *(.stext)
-   }  > imem
-   
-   .ktext 0x00400180 :
-   {
-     *(.ktext)
-   }  > imem
+    .start  : 
+    {
+    	. = ALIGN(4);
+        start = .;
+        entry = .;
+        _entry = .;
+        __entry = .;
+        KEEP (*(.start))
+    } >ram
 
-   .text . :
-   {
-     *(.text)
-   }  > imem
-
-   .data 0x00000000 : ALIGN(2)
-   {
-      *(.rodata*) *(.data) *(.sdata) *(.sbss) *(.kdata)
-   }  > dmem
-
-   .bss . :
-   {
-      *(.bss)
-   }  > dmem
-
-   .eh_frame . :
-   {
-      *(.eh_frame)
-   }  > dmem
+    .exc_vector ORIGIN(ram) + 0x80 :
+    {
+        KEEP (*(.exc_vect))
+    } >ram
+    .init           :
+    {
+       KEEP (*(.init))
+    } >ram
+    .text           :
+    {
+        *(.text)
+        *(.text.*)
+    } >ram
+    .fini           :
+    {
+      KEEP (*(.fini))
+    } >ram
+    .rodata         : { *(.rodata .rodata.*) } >ram
+    .rodata1        : { *(.rodata1) } >ram
+    .sdata2         :
+    {
+        *(.sdata2 .sdata2.*)
+    } >ram
+    .sbss2          : { *(.sbss2 .sbss2.*) } >ram
+    .data           :
+    {
+        *(.data .data.*)
+    } >ram
+    .data1          : { *(.data1) } >ram
+    .got.plt        : { *(.got.plt) } >ram
+    . = ALIGN(4);
+    _gp = ALIGN(16) + 0x7ff0;
+    .got            : { *(.got) }
+    /* We want the small data sections together, so single-instruction offsets
+       can access them all, and initialized data all before uninitialized, so
+       we can shorten the on-disk segment size.  */
+    .sdata          :
+    {
+        *(.sdata .sdata.*)
+    } >ram
+    __bss_start = ALIGN(4);
+    .sbss           :
+    {
+        *(.sbss .sbss.*)
+        *(.scommon)
+    } >ram
+    .bss            :
+    {
+        *(.bss .bss.*)
+        *(COMMON)
+        /* Align here to ensure that the .bss section occupies space up to
+           _end.  Align after .bss to ensure correct alignment even if the
+           .bss section disappears because there are no input sections.
+           FIXME: Why do we need it? When there is no .bss section, we don't
+           pad the .data section.  */
+        . = ALIGN(. != 0 ? 32 / 8 : 1);
+    } >ram
+    . = ALIGN(32 / 8);
+    _end = .; PROVIDE (end = .);
+    . = (ORIGIN(ram) + LENGTH(ram) - DEBUGGER_VAR_SIZE - 0x10);
+    stack_ptr = . | ~(LENGTH(ram)-1) & 0x7fffffff;
+    . = . + 0x10;
+    .dbg_vars . (NOLOAD)  :
+    {
+        *(.dbg_vars)
+    }
 }