diff --git a/.gitignore b/.gitignore
deleted file mode 100644
index 485dee6..0000000
--- a/.gitignore
+++ /dev/null
@@ -1 +0,0 @@
-.idea
diff --git a/kernel.c b/kernel.c
deleted file mode 100644
index 6ac43f6..0000000
--- a/kernel.c
+++ /dev/null
@@ -1,34 +0,0 @@
-void kprintstr(char* str);
-void khang();
-
-/* HACK. kmain has to come first because of flat binary */
-
-void kmain(void)
-{
-    char* k_msg = "Hello world from 32-bit C kernel!";
-    kprintstr(k_msg);
-    khang();
-}
-
-void kprintstr(char* str)
-{
-    /* last line of  80x25 */
-    char* ch = (char*)0xb8000 + 2 * 80 * 22;
-    while(*str)
-    {
-        *ch = *str;
-        ch++;
-        *ch = 0xf; /* white character */
-        ch++;
-        str++;
-    }
-    return;
-}
-
-void khang()
-{
-    while(1)
-    {
-
-    }
-}
diff --git a/src/addr.inc b/src/addr.inc
new file mode 100644
index 0000000..d4db610
--- /dev/null
+++ b/src/addr.inc
@@ -0,0 +1,11 @@
+STAGE2_BASE equ 0x8000
+STAGE2_OFFSET equ 0x0
+STAGE2_FLAT equ (STAGE2_BASE * 0x10) + STAGE2_OFFSET
+
+STAGE3_BASE equ 0x1000
+STAGE3_OFFSET equ 0
+STAGE3_FLAT equ (STAGE3_BASE * 0x10) + STAGE3_OFFSET
+
+KERNEL_BASE equ 0x4000
+KERNEL_OFFSET equ 0
+KERNEL_FLAT equ (KERNEL_BASE * 0x10) + KERNEL_OFFSET
diff --git a/bochsrc b/src/bochsrc
similarity index 100%
rename from bochsrc
rename to src/bochsrc
diff --git a/fat12_drv.lib b/src/fat12_drv.lib
similarity index 100%
rename from fat12_drv.lib
rename to src/fat12_drv.lib
diff --git a/fat12_hdr.inc b/src/fat12_hdr.inc
similarity index 100%
rename from fat12_hdr.inc
rename to src/fat12_hdr.inc
diff --git a/src/kernel.c b/src/kernel.c
new file mode 100644
index 0000000..a847d6b
--- /dev/null
+++ b/src/kernel.c
@@ -0,0 +1,54 @@
+void kprintstr(char* str);
+void khang();
+void kdelay(int mills);
+/* HACK. kmain has to come first because of flat binary */
+
+void kmain(void)
+{
+    char* k_msg = "Hello world from 32-bit C kernel!"; /*33 characters*/
+    kprintstr(k_msg);
+}
+
+void kdelay(int millis)
+{
+    for(int i = 0; i < millis; i++)
+    {
+        for(int j = 0; j < millis; j++)
+        {
+
+        }
+    }
+}
+
+void kprintstr(char* str)
+{
+    /* 80x25 */
+    char* ch = (char*)0xb8000 + 2 * 80 * 21;
+    char color = 0xF;
+    char swap[] = {'|','\\', '/', '\\', '/', '|', '|'};
+    while(*str)
+    {
+        *ch = *str;
+        ch++;
+        *ch = color; /* white character */
+        ch++;
+        str++;
+    }
+    ch = (char*)0xb8000 + 2 * 80 * 22;
+    while(1)
+    {
+        int i = 0;
+        while(1)
+        {
+            /*display 10 characters doing*/
+            for(int j = 0; j < 80 * 3; j++)
+            {
+                *(ch + j * 2) = swap[(i * 1337 + j * 33 ) % 3];
+                *(ch + j * 2+ 1) = color;
+                color = (color + 1) % 0xF;
+            }
+            kdelay(5000);
+            i++;
+        }
+    }
+}
\ No newline at end of file
diff --git a/makefile b/src/makefile
similarity index 95%
rename from makefile
rename to src/makefile
index 037d73f..f44d9af 100644
--- a/makefile
+++ b/src/makefile
@@ -6,7 +6,7 @@ ASM_FLAGS=-f bin -o
 CFLAGS=-ffreestanding -m32 -target i386-pc-none-elf -c -Wall -Wextra -Werror -mno-red-zone -std=c17 -o
 
 STAGE1_TARGET=stage1.bin
-ASM_TARGET=stage2.bin stage1.bin
+ASM_TARGET=stage2.bin stage1.bin stage3.bin
 TARGET_DIR=out
 
 all : createdir $(ASM_TARGET) kernel.bin image
diff --git a/print16.lib b/src/print16.lib
similarity index 100%
rename from print16.lib
rename to src/print16.lib
diff --git a/src/print32.lib b/src/print32.lib
new file mode 100644
index 0000000..337bea0
--- /dev/null
+++ b/src/print32.lib
@@ -0,0 +1,122 @@
+; global position for vram
+vb_pos: dd 0
+
+;;;;;;;;;;;;;;;;;;;;;
+;print a 32-bit integer
+; input: push u32
+;;;;;;;;;;;;;;;;;;;;;;
+print_u32:
+push ebp
+mov ebp, esp
+
+push eax
+push ebx
+push ecx
+
+; loop 4 times
+mov ebx, 8
+mov ecx, 32 - 4
+.lp:
+; eax = thing to print
+mov eax, dword [ebp + 2 * 4]
+shr eax, cl
+and eax, 0xF
+
+call al_to_char
+call print_char
+
+sub ecx, 4
+dec ebx
+test ebx, ebx
+jne .lp
+
+pop ecx
+pop ebx
+pop eax
+pop ebp
+ret
+
+;;;;;;;;;;;;;;;;;;;;;
+;print a char in
+;;;;;;;;;;;;;;;;;;;;;;
+print_char:
+; input: al = char
+push edi
+push eax
+push esi
+push ecx
+
+; esi points to current frame buffer
+mov esi, 0xb8000
+; edi points to offset
+mov edi, dword [vb_pos]
+
+; handle return character
+cmp al, 0xA
+jne .char
+; round up to 80 * 2
+mov eax, edi
+add ax, 2 * 80 - 1
+mov cl, 2 * 80
+div cl
+mul cl
+mov edi, eax
+jmp .end
+.char:
+mov byte [esi + edi], al
+mov word [esi + edi + 1], 0xf
+add edi, 2
+
+.end:
+mov dword [vb_pos], edi
+
+pop ecx
+pop esi
+pop eax
+pop edi
+ret
+
+;;;;;;;;;;;;;;;;;;;;;
+;print a null-terminated string
+;;;;;;;;;;;;;;;;;;;;;;
+print_str:
+;offset
+push ebp
+mov ebp, esp
+
+push esi
+push eax
+push edi
+
+mov esi, dword [ebp + 2 * 4]
+
+mov edi, 0xb8000
+
+.loop_each_char:
+mov al, byte [esi]
+test al, al
+je .end
+call print_char
+inc esi
+jmp .loop_each_char
+
+.end:
+pop edi
+pop eax
+pop esi
+pop ebp
+ret
+
+;;;;;;;;;;;;;;;;;;;;;
+; helper function converting 4 bit val in al to its character
+;;;;;;;;;;;;;;;;;;;;;;
+al_to_char:
+; no parameters
+cmp al,09h
+jna .l2
+add al,37h
+jmp .end
+.l2:
+add al,30h
+.end:
+ret
\ No newline at end of file
diff --git a/stage1.asm b/src/stage1.asm
similarity index 94%
rename from stage1.asm
rename to src/stage1.asm
index 911d1bc..832632d 100644
--- a/stage1.asm
+++ b/src/stage1.asm
@@ -1,4 +1,5 @@
 org 0x7c00
+%include "addr.inc"
 ;;;;;;;;;;;;;;;;;;;;;;;
 ; Constant definition
 ;;;;;;;;;;;;;;;;;;;;;;;
@@ -40,7 +41,6 @@ jmp $
 fn_stage2: db 'STAGE2  BIN',0
 
 %include "fat12_drv.lib"
-%include "stage2.inc"
 
 times 510-($ - $$) db 0
 dw 0xaa55
\ No newline at end of file
diff --git a/stage2.asm b/src/stage2.asm
similarity index 75%
rename from stage2.asm
rename to src/stage2.asm
index 10cb6a1..83ab013 100644
--- a/stage2.asm
+++ b/src/stage2.asm
@@ -1,4 +1,4 @@
-%include "stage2.inc"
+%include "addr.inc"
 
 org STAGE2_OFFSET
 jmp start16
@@ -16,22 +16,36 @@ mov ax, 0xA
 call print_char
 %endmacro
 
+%macro print_addr 2
+push %1
+call print_u16
+add sp, 2
+
+mov ax, 0x3A ; colon
+call print_char
+
+push %2
+call print_u16
+add sp, 2
+%endmacro
+
 [section .data]
 [bits 16]
 fn_kernel: db 'KERNEL  BIN',0
+fn_stage3: db 'STAGE3  BIN',0
 
-msg_ok: db 'OK', 0
-msg_err: db 'ERR', 0
-msg_stage2_hello: db 'Stage2 loader at ', 0
-msg_det_mem_pre: db 'Detecting memory... ', 0
-msg_det_mem_title: db "Memory Information:", 0
-msg_det_mem_tbl_hdr: db 'Base                Length              Type',0 ; each field is of length 20
+msg_ok: db 'OK ', 0
+msg_err: db 'ERR ', 0
+
+msg_stage2_hello: db 'Stage2 loader loaded at ', 0
+msg_det_mem: db 'Detecting memory... ', 0
+msg_det_mem_title: db "Memory Information: ", 0
+msg_det_mem_tbl_hdr: db 'Base                Length              Type ',0 ; each field is of length 20
 msg_det_mem_tbl_pad: db '    ',0 ; 4 x spaces
 
-msg_ld_krnl_pre: db 'Loading kernel.bin... ', 0
-msg_ld_krnl_post: db 'kernel.bin loaded to ', 0
-
-msg_pm_pre: db 'Initializing protected mode and jumping to kernel... ', 0
+msg_ld_stage3: db 'Loading stage3.bin... ', 0
+msg_ld_krnl: db 'Loading kernel.bin... ', 0
+msg_init_pm: db 'Initializing protected mode and jumping to stage3 loader... ', 0
 
 
 %include "fat12_hdr.inc"
@@ -52,37 +66,37 @@ mov al, 0x03
 int 0x10
 
 print_msg msg_stage2_hello
-
-push word STAGE2_BASE
-call print_u16
-add sp, 2
-
-mov ax, 0x3A ; colon
-call print_char
-
-push word STAGE2_OFFSET
-call print_u16
-add sp, 2
-
+print_addr STAGE2_BASE, STAGE2_OFFSET
 print_rtn
-print_msg msg_det_mem_pre
 
+print_msg msg_det_mem
 call detect_mem
-
 cmp ax, 1
 je .err
 
 print_msg msg_ok
 print_rtn
+
 print_msg msg_det_mem_title
 print_rtn
 
 call print_mem_layout
-
 print_rtn
 
-print_msg msg_ld_krnl_pre
+print_msg msg_ld_stage3
+push word fn_stage3
+push word STAGE3_OFFSET
+push word STAGE3_BASE
+call fat12_read_file
+add esp, 6 
+cmp ax, 1
+je .err
 
+print_msg msg_ok
+print_addr STAGE3_BASE, STAGE3_OFFSET
+print_rtn
+
+print_msg msg_ld_krnl
 push word fn_kernel
 push word KERNEL_OFFSET
 push word KERNEL_BASE
@@ -92,25 +106,10 @@ cmp ax, 1
 je .err
 
 print_msg msg_ok 
+print_addr KERNEL_BASE, KERNEL_OFFSET
 print_rtn
 
-print_msg msg_ld_krnl_post
-
-push word KERNEL_BASE
-call print_u16
-add sp, 2
-
-mov ax, 0x3A ; colon
-call print_char
-
-push word KERNEL_OFFSET
-call print_u16
-add sp, 2
-
-print_rtn
-
-print_msg msg_pm_pre
-
+print_msg msg_init_pm
 
 ;load gdt
 lgdt [gdtr32]
@@ -125,6 +124,10 @@ mov eax,cr0
 or eax,1
 mov cr0,eax
 
+; pass the frame buffer
+xor ebx, ebx
+mov bx, word [vb_pos]
+
 jmp dword gdt32.sel_code:(STAGE2_FLAT + start32)
 
 start16.err:
@@ -235,23 +238,20 @@ ret
 [section .text]
 [bits 32]
 start32:
+; ebx holds -> vram_buffer
 mov ax,gdt32.sel_data
 mov ss,ax
 mov ds,ax
 mov es,ax
 mov gs,ax
+; adjust 
 
-mov esp, STAGE2_FLAT + stack32
-
-;Jump to The Kernel
-mov eax, KERNEL_FLAT
+;Jump to stage 3
+mov eax, STAGE3_FLAT
 jmp eax
 
 [section .data]
 [bits 32]
-times 1024 db 0
-stack32:
-
 gdt32:
 dd 0
 dd 0
diff --git a/src/stage3.asm b/src/stage3.asm
new file mode 100644
index 0000000..a09a00a
--- /dev/null
+++ b/src/stage3.asm
@@ -0,0 +1,141 @@
+org 0x10000 ; 64k this is terrible but IDC
+%include "addr.inc"
+%macro print_msg 1
+mov eax, %1
+push eax
+call print_str
+add esp, 4
+%endmacro
+
+%macro print_rtn 0
+mov eax, 0xA
+call print_char
+%endmacro
+
+%macro print_addr 2
+push %1
+call print_u32
+add esp, 4
+
+mov eax, 0x3A ; colon
+call print_char
+
+push %2
+call print_u32
+add esp, 4
+%endmacro
+
+[SECTION .text]
+[BITS 32]
+jmp start
+%include "print32.lib"
+start:
+mov dword [vb_pos], ebx ; ebx holds framebuffer pointer
+
+print_msg msg_ok
+print_rtn
+
+print_msg msg_switch_gdt_pre
+; switch stack and GDT
+lgdt [gdtr32]
+jmp dword gdt32.sel_code:.reload
+.reload:
+mov ax, gdt32.sel_data
+mov ds, ax
+mov ss, ax
+mov es, ax
+mov gs, ax
+mov fs, ax
+mov esp, stack
+
+print_msg msg_ok
+print_rtn
+
+print_msg msg_init_idt
+
+; write offset to idt
+mov eax, dummy_handler
+mov word [idt32], ax
+mov cl, 16
+shr eax, cl
+mov word [idt32 + 6], ax
+
+lidt [idtr32]
+print_msg msg_ok
+print_rtn
+
+print_msg msg_test_div
+xor ax, ax
+div ax
+next_op:
+print_msg msg_ok
+print_rtn
+
+print_msg msg_init_krnl
+mov eax, KERNEL_FLAT
+jmp eax
+
+err:
+print_msg msg_err
+jmp $
+
+
+dummy_handler:
+print_msg msg_div
+call print_u32  ; top of stack is the address
+print_msg msg_space
+mov dword [esp], next_op ; overwrite return address
+iretd
+
+[SECTION .data]
+[BITS 32]
+msg_div: db '#DIV at ', 0
+msg_space: db ' ', 0
+msg_ok: db 'OK ', 0
+msg_err: db 'ERR ', 0
+msg_switch_gdt_pre: db 'Reloading GDT and stack... ', 0
+msg_init_idt: db 'Initializing IDT... ', 0
+msg_test_div: db 'Testing #DIV exception... ', 0
+msg_init_krnl: db 'Jumping to kernel... ', 0
+
+align 4096
+times 4096 db 0
+stack:
+
+align 16
+idt32:
+dw 0  ; offset low 16 bits
+dw gdt32.sel_code ; selector
+db 0
+db 0b10001111 ; Present | DPL 0 | Trap gate
+dw 0
+.size equ $ - idt32
+
+idtr32: 
+dw idt32.size - 1
+dd idt32
+
+align 16
+gdt32:
+dd 0
+dd 0
+.sel_code: equ $ - gdt32 
+dw 0xFFFF ; LIMIT = 0xFFFF
+dw 0 ; BASE LOW = 0
+db 0 ; BASE MID = 0
+db 10011010b; PRESENT | RW | EXE
+db 11001111b; Granularity = 4k | 32 bits
+db 0 ; BASE_HIGH = 0
+.sel_data: equ $ - gdt32
+dw 0xFFFF ; LIMIT = 0xFFFF
+dw 0 ; BASE LOW = 0
+db 0 ; BASE MID = 0
+db 10010010b; PRESENT | RW | EXE
+db 11001111b; Granularity = 4k | 32 bits
+db 0 ; BASE_HIGH = 0
+.size: equ $ - gdt32
+
+align 16
+gdtr32: 
+dw gdt32.size - 1
+dd gdt32
\ No newline at end of file
diff --git a/stage2.inc b/stage2.inc
deleted file mode 100644
index e03a42e..0000000
--- a/stage2.inc
+++ /dev/null
@@ -1,7 +0,0 @@
-STAGE2_BASE equ 0x8000
-STAGE2_OFFSET equ 0x0
-STAGE2_FLAT equ (STAGE2_BASE * 0x10) + STAGE2_OFFSET
-
-KERNEL_BASE equ 0x4000
-KERNEL_OFFSET equ 0x0
-KERNEL_FLAT equ (KERNEL_BASE * 0x10) + KERNEL_OFFSET
\ No newline at end of file