# Code for the first few disk sectors that all programs in this directory need:
#   - load sectors past the first (using BIOS primitives) since only the first is available by default
#     - if this fails, print 'D' at top-left of screen and halt
#   - initialize a minimal graphics mode
#   - switch to 32-bit mode (giving up access to BIOS primitives)
#   - set up a handler for keyboard events
#   - jump to start of program

# Code in this file needs to be more deliberate about the SubX facilities it
# uses:
#   - sigils only support 32-bit general-purpose registers, so don't work with segment registers or 16-bit or 8-bit registers
#   - metadata like rm32 and r32 can sometimes misleadingly refer to only the bottom 16 bits of the register; pay attention to the register name
#
# While most of Mu is thoroughly tested, this file is not. I don't yet
# understand hardware interfaces well enough to explain to others.

# Memory map of a Mu computer:
#   code: currently 4 tracks loaded from the primary disk to [0x00007c00, 0x00048600)
#   stack: grows down from 0x02000000 to 0x01000000
#   heap: [0x02000000, 0x08000000)
#     see 120allocate.subx; Qemu initializes with 128MB RAM by default
# Consult https://wiki.osdev.org/Memory_Map_(x86) before modifying any of
# this. And don't forget to keep *stack-debug.subx in sync.

== code

## 16-bit entry point: 0x7c00

# Upon reset, the IBM PC:
#   - loads the first sector (512 bytes)
#     from some bootable image (look for the boot-sector-marker further down this file)
#     to the address range [0x7c00, 0x7e00)
#   - starts executing code at address 0x7c00

  fa/disable-interrupts

  # initialize segment registers
  b8/copy-to-ax 0/imm16
  8e/->seg 3/mod/direct 0/rm32/ax 3/r32/ds
  8e/->seg 3/mod/direct 0/rm32/ax 0/r32/es
  8e/->seg 3/mod/direct 0/rm32/ax 4/r32/fs
  8e/->seg 3/mod/direct 0/rm32/ax 5/r32/gs

  # Temporarily initialize stack to 0x00070000 in real mode.
  # We don't read or write the stack before we get to 32-bit mode, but BIOS
  # calls do. We need to move the stack in case BIOS initializes it to some
  # low address that we want to write code into.
  b8/copy-to-ax 0x7000/imm16
  8e/->seg 3/mod/direct 0/rm32/ax 2/r32/ss
  bc/copy-to-esp 0/imm16

  # undo the A20 hack: https://en.wikipedia.org/wiki/A20_line
  # this is from https://github.com/mit-pdos/xv6-public/blob/master/bootasm.S
  {
    e4/read-port-into-al 0x64/imm8
    a8/test-bits-in-al 0x02/imm8  # set zf if bit 1 (second-least significant) is not set
    75/jump-if-!zero loop/disp8
    b0/copy-to-al 0xd1/imm8
    e6/write-al-into-port 0x64/imm8
  }
  {
    e4/read-port-into-al 0x64/imm8
    a8/test-bits-in-al 0x02/imm8  # set zf if bit 1 (second-least significant) is not set
    75/jump-if-!zero loop/disp8
    b0/copy-to-al 0xdf/imm8
    e6/write-al-into-port 0x64/imm8
  }

  # load remaining sectors from first two tracks of disk into addresses [0x7e00, 0x17800)
  b4/copy-to-ah 2/imm8/read-drive
  # dl comes conveniently initialized at boot time with the index of the device being booted
  b5/copy-to-ch 0/imm8/cylinder
  b6/copy-to-dh 0/imm8/head
  b1/copy-to-cl 2/imm8/sector  # 1-based
  b0/copy-to-al 0x7d/imm8/num-sectors  # 2*63 - 1 = 125
  # address to write sectors to = es:bx = 0x7e00, contiguous with boot segment
  bb/copy-to-bx 0/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 0/r32/es
  bb/copy-to-bx 0x7e00/imm16
  cd/syscall 0x13/imm8/bios-disk-services
  0f 82/jump-if-carry disk_error/disp16

  # load two more tracks of disk into addresses [0x17800, 0x27400)
  b4/copy-to-ah 2/imm8/read-drive
  # dl comes conveniently initialized at boot time with the index of the device being booted
  b5/copy-to-ch 0/imm8/cylinder
  b6/copy-to-dh 2/imm8/head
  b1/copy-to-cl 1/imm8/sector  # 1-based
  b0/copy-to-al 0x7e/imm8/num-sectors  # 2*63 = 126
  # address to write sectors to = es:bx = 0x17800, contiguous with boot segment
  bb/copy-to-bx 0x1780/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 0/r32/es
  bb/copy-to-bx 0/imm16
  cd/syscall 0x13/imm8/bios-disk-services
  0f 82/jump-if-carry disk_error/disp16

  # load two more tracks of disk into addresses [0x27400, 0x37000)
  b4/copy-to-ah 2/imm8/read-drive
  # dl comes conveniently initialized at boot time with the index of the device being booted
  b5/copy-to-ch 0/imm8/cylinder
  b6/copy-to-dh 4/imm8/head
  b1/copy-to-cl 1/imm8/sector  # 1-based
  b0/copy-to-al 0x7e/imm8/num-sectors  # 2*63 = 126
  # address to write sectors to = es:bx = 0x27400, contiguous with boot segment
  bb/copy-to-bx 0x2740/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 0/r32/es
  bb/copy-to-bx 0/imm16
  cd/syscall 0x13/imm8/bios-disk-services
  0f 82/jump-if-carry disk_error/disp16

  # load two more tracks of disk into addresses [0x37000, 0x46c00)
  b4/copy-to-ah 2/imm8/read-drive
  # dl comes conveniently initialized at boot time with the index of the device being booted
  b5/copy-to-ch 0/imm8/cylinder
  b6/copy-to-dh 6/imm8/head
  b1/copy-to-cl 1/imm8/sector  # 1-based
  b0/copy-to-al 0x7e/imm8/num-sectors  # 2*63 = 126
  # address to write sectors to = es:bx = 0x37000, contiguous with boot segment
  bb/copy-to-bx 0x3700/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 0/r32/es
  bb/copy-to-bx 0/imm16
  cd/syscall 0x13/imm8/bios-disk-services
  0f 82/jump-if-carry disk_error/disp16

  # reset es
  bb/copy-to-bx 0/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 0/r32/es

  # adjust video mode
  b4/copy-to-ah 0x4f/imm8  # VBE commands
  b0/copy-to-al 2/imm8  # set video mode
  bb/copy-to-bx 0x4105/imm16  # 0x0105 | 0x4000
                              # 0x0105 = graphics mode 1024x768x256
                              #  (alternative candidate: 0x0101 for 640x480x256)
                              # 0x4000 bit = configure linear frame buffer in Bochs emulator; hopefully this doesn't hurt anything when running natively
  cd/syscall 0x10/imm8/bios-video-services

  # load information for the (hopefully) current video mode
  # mostly just for the address to the linear frame buffer
  b4/copy-to-ah 0x4f/imm8  # VBE commands
  b0/copy-to-al 1/imm8  # get video mode info
  b9/copy-to-cx 0x0105/imm16  # mode we requested
  bf/copy-to-di Video-mode-info/imm16
  cd/syscall 0x10/imm8/bios-video-services

  ## switch to 32-bit mode
  # load global descriptor table
  # We can't refer to the label directly because SubX doesn't do the right
  # thing for lgdt, so rather than make errors worse in most places we instead
  # pin gdt_descriptor below.
  0f 01 2/subop/lgdt 0/mod/indirect 6/rm32/use-disp16 0x7ce0/disp16/gdt_descriptor
  # enable paging
  0f 20/<-cr 3/mod/direct 0/rm32/eax 0/r32/cr0
  66 83 1/subop/or 3/mod/direct 0/rm32/eax 1/imm8  # eax <- or 0x1
  0f 22/->cr 3/mod/direct 0/rm32/eax 0/r32/cr0
  # far jump to initialize_32bit_mode that sets cs to offset 8 in the gdt in the process
  # We can't refer to the label directly because SubX doesn't have syntax for
  # segment selectors. So we instead pin initialize_32bit_mode below.
  ea/jump-far-absolute 0x00087d00/disp32  # address 0x7d00 in offset 8 of the gdt

disk_error:
  # print 'D' to top-left of screen to indicate disk error
  # *0xb8000 <- 0x0f44
  bb/copy-to-bx 0xb800/imm16
  8e/->seg 3/mod/direct 3/rm32/bx 3/r32/ds
  b0/copy-to-al 0x44/imm8/D
  b4/copy-to-ah 0x0f/imm8/white-on-black
  bb/copy-to-bx 0/imm16
  89/<- 0/mod/indirect 7/rm32/bx 0/r32/ax  # *ds:bx <- ax
  # loop forever
  {
    eb/jump loop/disp8
  }

## GDT: 3 records of 8 bytes each
== data 0x7ce0
gdt_descriptor:
  0x17/imm16  # final index of gdt = size of gdt - 1
  gdt_start/imm32/start

gdt_start:
# offset 0: gdt_null:  mandatory null descriptor
  00 00 00 00 00 00 00 00
# offset 8: gdt_code
  ff ff  # limit[0:16]
  00 00 00  # base[0:24]
  9a  # 1/present 00/privilege 1/descriptor type = 1001b
      # 1/code 0/conforming 1/readable 0/accessed = 1010b
  cf  # 1/granularity 1/32-bit 0/64-bit-segment 0/AVL = 1100b
      # limit[16:20] = 1111b
  00  # base[24:32]
# offset 16: gdt_data
  ff ff  # limit[0:16]
  00 00 00  # base[0:24]
  92  # 1/present 00/privilege 1/descriptor type = 1001b
      # 0/data 0/conforming 1/readable 0/accessed = 0010b
  cf  # same as gdt_code
  00  # base[24:32]
# gdt_end:

## 32-bit code from this point

== code 0x7d00
initialize_32bit_mode:
  66 b8/copy-to-ax 0x10/imm16  # offset 16 from gdt_start
  8e/->seg 3/mod/direct 0/rm32/ax 3/r32/ds
  8e/->seg 3/mod/direct 0/rm32/ax 2/r32/ss
  8e/->seg 3/mod/direct 0/rm32/ax 0/r32/es
  8e/->seg 3/mod/direct 0/rm32/ax 4/r32/fs
  8e/->seg 3/mod/direct 0/rm32/ax 5/r32/gs

  bc/copy-to-esp 0x02000000/imm32

  ## load interrupt handlers
  # We can't refer to the label directly because SubX doesn't do the right
  # thing for lidt, so rather than make errors worse in most places we instead
  # pin idt_descriptor below.
  0f 01 3/subop/lidt 0/mod/indirect 5/rm32/use-disp32 0x7e00/disp32/idt_descriptor

  # For now, not bothering reprogramming the IRQ to not conflict with software
  # exceptions.
  #   https://wiki.osdev.org/index.php?title=8259_PIC&oldid=24650#Protected_Mode
  #
  # Interrupt 1 (keyboard) conflicts with debugger faults. We don't use a
  # debugger.
  # Reference:
  #   https://wiki.osdev.org/Exceptions

  # enable timer IRQ0 and keyboard IRQ1
  b0/copy-to-al 0xfc/imm8  # disable mask for IRQ0 and IRQ1
  e6/write-al-into-port 0x21/imm8

  fb/enable-interrupts

  (initialize-mouse)

  ## enable floating point
  db/floating-point-coprocessor e3/initialize
  # eax <- cr4
  0f 20/<-cr 3/mod/direct 0/rm32/eax 4/r32/cr4
  # eax <- or bit 9
  0f ba/bit-test 5/subop/bit-test-and-set 3/mod/direct 0/rm32/eax 9/imm8
  # cr4 <- eax
  0f 22/->cr 3/mod/direct 0/rm32/eax 4/r32/cr4

  e9/jump Entry/disp32

== boot-sector-marker 0x7dfe
# final 2 bytes of boot sector
55 aa

## sector 2 onwards loaded by load_disk, not automatically on boot

== data 0x7e00
idt_descriptor:
  ff 03  # final index of idt = size of idt - 1
  idt_start/imm32/start

# interrupt descriptor table {{{
# 32 entries of 8 bytes each
idt_start:

# entry 0
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00

# By default, BIOS maps IRQ0-7 to interrupt vectors 8-15.
# https://wiki.osdev.org/index.php?title=Interrupts&oldid=25102#Default_PC_Interrupt_Vector_Assignment

# entry 8: https://wiki.osdev.org/Programmable_Interval_Timer
  timer-interrupt-handler/imm16  # target[0:16]
  8/imm16  # segment selector (gdt_code)
  00  # unused
  8e  # 1/p 00/dpl 0 1110/type/32-bit-interrupt-gate
  0/imm16  # target[16:32] -- null-interrupt-handler must be within address 0x10000

# entry 9: keyboard
  keyboard-interrupt-handler/imm16  # target[0:16]
  8/imm16  # segment selector (gdt_code)
  00  # unused
  8e  # 1/p 00/dpl 0 1110/type/32-bit-interrupt-gate
  0/imm16  # target[16:32] -- keyboard-interrupt-handler must be within address 0x10000

00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00
# idt_end:
# }}}

== code

null-interrupt-handler:
  # prologue
  # Don't disable interrupts; the timer has the highest priority anyway,
  # and this interrupt triggers extremely frequently.
  fa/disable-interrupts
  60/push-all-registers
  9c/push-flags
  # acknowledge interrupt
  b0/copy-to-al 0x20/imm8
  e6/write-al-into-port 0x20/imm8
  31/xor %eax 0/r32/eax
$null-interrupt-handler:epilogue:
  # epilogue
  9d/pop-flags
  61/pop-all-registers
  fb/enable-interrupts
  cf/return-from-interrupt

timer-interrupt-handler:
  # prologue
  # Don't disable interrupts; the timer has the highest priority anyway,
  # and this interrupt triggers extremely frequently.
  fa/disable-interrupts
  60/push-all-registers
  9c/push-flags
  # acknowledge interrupt
  b0/copy-to-al 0x20/imm8
  e6/write-al-into-port 0x20/imm8
  31/xor %eax 0/r32/eax
$timer-interrupt-handler:epilogue:
  # epilogue
  9d/pop-flags
  61/pop-all-registers
  fb/enable-interrupts
  cf/return-from-interrupt

keyboard-interrupt-handler:
  # prologue
  fa/disable-interrupts
  60/push-all-registers
  9c/push-flags
  # acknowledge interrupt
  b0/copy-to-al 0x20/imm8
  e6/write-al-into-port 0x20/imm8
  31/xor %eax 0/r32/eax
  # check output buffer of 8042 keyboard controller (https://web.archive.org/web/20040604041507/http://panda.cs.ndsu.nodak.edu/~achapwes/PICmicro/keyboard/atkeyboard.html)
  e4/read-port-into-al 0x64/imm8
  a8/test-bits-in-al 0x01/imm8  # set zf if bit 0 (least significant) is not set
  0f 84/jump-if-not-set $keyboard-interrupt-handler:epilogue/disp32
  # - if keyboard buffer is full, return
  # var dest-addr/ecx: (addr byte) = (keyboard-buffer + *keyboard-buffer:write)
  31/xor %ecx 1/r32/ecx
  8a/byte-> *Keyboard-buffer:write 1/r32/cl
  81 0/subop/add %ecx Keyboard-buffer:data/imm32
  # al = *dest-addr
  8a/byte-> *ecx 0/r32/al
  # if (al != 0) return
  3c/compare-al-and 0/imm8
  0f 85/jump-if-!= $keyboard-interrupt-handler:epilogue/disp32
  # - read keycode
  e4/read-port-into-al 0x60/imm8
  # - key released
  # if (al == 0xaa) shift = false  # left shift is being lifted
  {
    3c/compare-al-and 0xaa/imm8
    75/jump-if-!= break/disp8
    # *shift = 0
    c7 0/subop/copy *Keyboard-shift-pressed? 0/imm32
  }
  # if (al == 0xb6) shift = false  # right shift is being lifted
  {
    3c/compare-al-and 0xb6/imm8
    75/jump-if-!= break/disp8
    # *shift = 0
    c7 0/subop/copy *Keyboard-shift-pressed? 0/imm32
  }
  # if (al == 0x9d) ctrl = false  # ctrl is being lifted
  {
    3c/compare-al-and 0x9d/imm8
    75/jump-if-!= break/disp8
    # *ctrl = 0
    c7 0/subop/copy *Keyboard-ctrl-pressed? 0/imm32
  }
  # if (al & 0x80) a key is being lifted; return
  50/push-eax
  24/and-al-with 0x80/imm8
  3c/compare-al-and 0/imm8
  58/pop-to-eax
  75/jump-if-!= $keyboard-interrupt-handler:epilogue/disp8
  # - key pressed
  # if (al == 0x2a) shift = true, return  # left shift pressed
  {
    3c/compare-al-and 0x2a/imm8
    75/jump-if-!= break/disp8
    # *shift = 1
    c7 0/subop/copy *Keyboard-shift-pressed? 1/imm32
    # return
    eb/jump $keyboard-interrupt-handler:epilogue/disp8
  }
  # if (al == 0x36) shift = true, return  # right shift pressed
  {
    3c/compare-al-and 0x36/imm8
    75/jump-if-!= break/disp8
    # *shift = 1
    c7 0/subop/copy *Keyboard-shift-pressed? 1/imm32
    # return
    eb/jump $keyboard-interrupt-handler:epilogue/disp8
  }
  # if (al == 0x1d) ctrl = true, return
  {
    3c/compare-al-and 0x1d/imm8
    75/jump-if-!= break/disp8
    # *ctrl = 1
    c7 0/subop/copy *Keyboard-ctrl-pressed? 1/imm32
    # return
    eb/jump $keyboard-interrupt-handler:epilogue/disp8
  }
  # - convert key to character
  # if (shift) use keyboard shift map
  {
    81 7/subop/compare *Keyboard-shift-pressed? 0/imm32
    74/jump-if-= break/disp8
    # sigils don't currently support labels inside *(eax+label)
    05/add-to-eax Keyboard-shift-map/imm32
    8a/byte-> *eax 0/r32/al
    eb/jump $keyboard-interrupt-handler:select-map-done/disp8
  }
  # if (ctrl) al = *(ctrl map + al)
  {
    81 7/subop/compare *Keyboard-ctrl-pressed? 0/imm32
    74/jump-if-= break/disp8
    05/add-to-eax Keyboard-ctrl-map/imm32
    8a/byte-> *eax 0/r32/al
    eb/jump $keyboard-interrupt-handler:select-map-done/disp8
  }
  # otherwise al = *(normal map + al)
  05/add-to-eax Keyboard-normal-map/imm32
  8a/byte-> *eax 0/r32/al
$keyboard-interrupt-handler:select-map-done:
  # - if there's no character mapping, return
  {
    3c/compare-al-and 0/imm8
    74/jump-if-= break/disp8
    # - store al in keyboard buffer
    88/<- *ecx 0/r32/al
    # increment index
    fe/increment-byte *Keyboard-buffer:write
    # clear top nibble of index (keyboard buffer is circular)
    80 4/subop/and-byte *Keyboard-buffer:write 0x0f/imm8
  }
$keyboard-interrupt-handler:epilogue:
  # epilogue
  9d/pop-flags
  61/pop-all-registers
  fb/enable-interrupts
  cf/return-from-interrupt

== data
Keyboard-shift-pressed?:  # boolean
  0/imm32

Keyboard-ctrl-pressed?:  # boolean
  0/imm32

# var keyboard circular buffer
Keyboard-buffer:write:  # nibble
  0/imm32
Keyboard-buffer:read:  # nibble
  0/imm32
Keyboard-buffer:data:  # byte[16]
  00 00 00 00
  00 00 00 00
  00 00 00 00
  00 00 00 00

# Keyboard maps for translating keys to ASCII {{{
Keyboard-normal-map:
00
#  es
   1b
#     |<--- digits -------------->| -  =  backspace
      31 32 33 34 35 36 37 38 39 30 2d 3d 08
# 0f
# tab q  w  e  r  t  y  u  i  o  p  [  ]
   09 71 77 65 72 74 79 75 69 6f 70 5b 5d
# 1c
#                                         enter (newline)
                                          0a 00
# 1e
#     a  s  d  f  g  h  j  k  l  ;  '  `     \
      61 73 64 66 67 68 6a 6b 6c 3b 27 60 00 5c
                                        # ^ left shift
# 2c
#     z  x  c  v  b  n  m  ,  .  /     *
      7a 78 63 76 62 6e 6d 2c 2e 2f 00 2a
                                  # ^ right shift
# 38
#                          space
                        00 20
# 3a
                              00 00 00 00 00 00
00 00 00 00 00 00 00 00
# 48
#                       ↑*       ←*    →*       ↓*
                        82 00 00 80 00 83 00 00 81
# 51
   00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

# * - Not a valid ASCII/Unicode value.

Keyboard-shift-map:
00
#  es
   1b
#     !  @  #  $  %  ^  &  *  (  )  _  +  backspace
      21 40 23 24 25 53 26 2a 28 29 5f 2b 08
# 0f
# tab Q  W  E  R  T  Y  U  I  O  P  {  }
   09 51 57 45 52 54 59 55 49 5f 50 7b 7d
# 1c
#                                         enter (newline)
                                          0a 00
# 1e
#     A  S  D  F  G  H  J  K  L  :  "  ~     |
      41 53 44 46 47 48 4a 4b 4c 3a 22 7e 00 7c
# 2c
#     Z  X  C  V  B  N  M  <  >  ?     *
      5a 58 43 56 42 4e 4d 3c 3e 3f 00 2a
# 38
#                          space
                        00 20
# 3a
                              00 00 00 00 00 00
00 00 00 00 00 00 00 00
# 48
#                       ↑*       ←*    →*       ↓*
                        82 00 00 80 00 83 00 00 81
# 51
   00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

# * - Not a valid ASCII/Unicode value.

Keyboard-ctrl-map:
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
# 10
#     ^q ^w ^e ^r ^t ^y ^u tb ^o ^p
      11 17 05 12 14 19 15 09 1f 10 00 00
# 1c
#                                         carriage-return
                                          0d 00
# 1e
#     ^a ^s ^d ^f ^g ^h ^j ^k ^l             ^\
      01 13 04 06 07 08 0a 0b 0c 00 00 00 00 1c
# 2c
#     ^z ^x ^c ^v ^b ^n ^m       ^/
      1a 18 03 16 02 0e 0d 00 00 1f 00 00
# 38
#                          space
                        00 20
# 3a
                              00 00 00 00 00 00
00 00 00 00 00 00 00 00
# 48
#                       ↑*       ←*    →*       ↓*
                        82 00 00 80 00 83 00 00 81
# 51
   00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

# * - Not a valid ASCII/Unicode value.
# }}}

Video-mode-info:
# video mode info {{{
  0/imm16  # attributes
  00  # winA
  00  # winB
# 04
  0/imm16  # granularity
  0/imm16  # winsize
# 08
  0/imm16  # segmentA
  0/imm16  # segmentB
# 0c
  0/imm32  # realFctPtr (who knows)
# 10
  0/imm16  # pitch
  0/imm16  # Xres
  0/imm16  # Yres
  0/imm16  # Wchar Ychar
# 18
  00  # planes
  00  # bpp
  00  # banks
  00  # memory_model
# 1c
  00  # bank_size
  00  # image_pages
  00  # reserved
# 1f
  0/imm16  # red_mask red_position
  0/imm16  # green_mask green_position
  0/imm16  # blue_mask blue_position
  0/imm16  # rsv_mask rsv_position
  00  # directcolor_attributes
# 28
Video-memory-addr:
  0/imm32  # physbase

# 2c
# reserved for video mode info
                                    00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
# }}}

Font:
# Bitmaps for some ASCII characters (soon Unicode) {{{
# Part of GNU Unifont
# 8px wide, 16px tall
# Based on http://unifoundry.com/pub/unifont/unifont-13.0.05/font-builds/unifont-13.0.05.hex.gz
# See https://en.wikipedia.org/wiki/GNU_Unifont#The_.hex_font_format
# Website: http://unifoundry.com/unifont/index.html
# License: http://unifoundry.com/LICENSE.txt (GPL v2)
# Each line below is a bitmap for a single character.
#   Each byte is a bitmap for a single row of 8 pixels.

# some unprintable ASCII chars
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
# 0x20 = space
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
# !
  00 00 00 00 08 08 08 08 08 08 08 00 08 08 00 00
# "
  00 00 22 22 22 22 00 00 00 00 00 00 00 00 00 00
# 0x23 = '#'
  00 00 00 00 12 12 12 7e 24 24 7e 48 48 48 00 00
# $
  00 00 00 00 08 3e 49 48 38 0e 09 49 3e 08 00 00
# %
  00 00 00 00 31 4a 4a 34 08 08 16 29 29 46 00 00
# &
  00 00 00 00 1c 22 22 14 18 29 45 42 46 39 00 00
# '
  00 00 08 08 08 08 00 00 00 00 00 00 00 00 00 00
# (
  00 00 00 04 08 08 10 10 10 10 10 10 08 08 04 00
# )
  00 00 00 20 10 10 08 08 08 08 08 08 10 10 20 00
# *
  00 00 00 00 00 00 08 49 2a 1c 2a 49 08 00 00 00
# +
  00 00 00 00 00 00 08 08 08 7f 08 08 08 00 00 00
# ,
  00 00 00 00 00 00 00 00 00 00 00 00 18 08 08 10
# -
  00 00 00 00 00 00 00 00 00 3c 00 00 00 00 00 00
# .
  00 00 00 00 00 00 00 00 00 00 00 00 18 18 00 00
# /
  00 00 00 00 02 02 04 08 08 10 10 20 40 40 00 00
# 0x30 = '0'
  00 00 00 00 18 24 42 46 4a 52 62 42 24 18 00 00
# 1
  00 00 00 00 08 18 28 08 08 08 08 08 08 3e 00 00
# 2
  00 00 00 00 3c 42 42 02 0c 10 20 40 40 7e 00 00
# 3
  00 00 00 00 3c 42 42 02 1c 02 02 42 42 3c 00 00
# 4
  00 00 00 00 04 0c 14 24 44 44 7e 04 04 04 00 00
# 5
  00 00 00 00 7e 40 40 40 7c 02 02 02 42 3c 00 00
# 6
  00 00 00 00 1c 20 40 40 7c 42 42 42 42 3c 00 00
# 7
  00 00 00 00 7e 02 02 04 04 04 08 08 08 08 00 00
# 8
  00 00 00 00 3c 42 42 42 3c 42 42 42 42 3c 00 00
# 9
  00 00 00 00 3c 42 42 42 3e 02 02 02 04 38 00 00
# :
  00 00 00 00 00 00 18 18 00 00 00 18 18 00 00 00
# ;
  00 00 00 00 00 00 18 18 00 00 00 18 08 08 10 00
# <
  00 00 00 00 00 02 04 08 10 20 10 08 04 02 00 00
# =
  00 00 00 00 00 00 00 7e 00 00 00 7e 00 00 00 00
# >
  00 00 00 00 00 40 20 10 08 04 08 10 20 40 00 00
# ?
  00 00 00 00 3c 42 42 02 04 08 08 00 08 08 00 00
# 0x40 = @
  00 00 00 00 1c 22 4a 56 52 52 52 4e 20 1e 00 00
# A
  00 00 00 00 18 24 24 42 42 7e 42 42 42 42 00 00
# B
  00 00 00 00 7c 42 42 42 7c 42 42 42 42 7c 00 00
# C
  00 00 00 00 3c 42 42 40 40 40 40 42 42 3c 00 00
# D
  00 00 00 00 78 44 42 42 42 42 42 42 44 78 00 00
# E
  00 00 00 00 7e 40 40 40 7c 40 40 40 40 7e 00 00
# F
  00 00 00 00 7e 40 40 40 7c 40 40 40 40 40 00 00
# G
  00 00 00 00 3c 42 42 40 40 4e 42 42 46 3a 00 00
# H
  00 00 00 00 42 42 42 42 7e 42 42 42 42 42 00 00
# I
  00 00 00 00 3e 08 08 08 08 08 08 08 08 3e 00 00
# J
  00 00 00 00 1f 04 04 04 04 04 04 44 44 38 00 00
# K
  00 00 00 00 42 44 48 50 60 60 50 48 44 42 00 00
# L
  00 00 00 00 40 40 40 40 40 40 40 40 40 7e 00 00
# M
  00 00 00 00 42 42 66 66 5a 5a 42 42 42 42 00 00
# N
  00 00 00 00 42 62 62 52 52 4a 4a 46 46 42 00 00
# O
  00 00 00 00 3c 42 42 42 42 42 42 42 42 3c 00 00
# 0x50 = P
  00 00 00 00 7c 42 42 42 7c 40 40 40 40 40 00 00
# Q
  00 00 00 00 3c 42 42 42 42 42 42 5a 66 3c 03 00
# R
  00 00 00 00 7c 42 42 42 7c 48 44 44 42 42 00 00
# S
  00 00 00 00 3c 42 42 40 30 0c 02 42 42 3c 00 00
# T
  00 00 00 00 7f 08 08 08 08 08 08 08 08 08 00 00
# U
  00 00 00 00 42 42 42 42 42 42 42 42 42 3c 00 00
# V
  00 00 00 00 41 41 41 22 22 22 14 14 08 08 00 00
# W
  00 00 00 00 42 42 42 42 5a 5a 66 66 42 42 00 00
# X
  00 00 00 00 42 42 24 24 18 18 24 24 42 42 00 00
# Y
  00 00 00 00 41 41 22 22 14 08 08 08 08 08 00 00
# Z
  00 00 00 00 7e 02 02 04 08 10 20 40 40 7e 00 00
# [
  00 00 00 0e 08 08 08 08 08 08 08 08 08 08 0e 00
# \
  00 00 00 00 40 40 20 10 10 08 08 04 02 02 00 00
# ]
  00 00 00 70 10 10 10 10 10 10 10 10 10 10 70 00
# ^
  00 00 18 24 42 00 00 00 00 00 00 00 00 00 00 00
# _
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 7f 00
# 0x60 = backtick
  00 20 10 08 00 00 00 00 00 00 00 00 00 00 00 00
# a
  00 00 00 00 00 00 3c 42 02 3e 42 42 46 3a 00 00
# b
  00 00 00 40 40 40 5c 62 42 42 42 42 62 5c 00 00
# c
  00 00 00 00 00 00 3c 42 40 40 40 40 42 3c 00 00
# d
  00 00 00 02 02 02 3a 46 42 42 42 42 46 3a 00 00
# e
  00 00 00 00 00 00 3c 42 42 7e 40 40 42 3c 00 00
# f
  00 00 00 0c 10 10 10 7c 10 10 10 10 10 10 00 00
# g
  00 00 00 00 00 02 3a 44 44 44 38 20 3c 42 42 3c
# h
  00 00 00 40 40 40 5c 62 42 42 42 42 42 42 00 00
# i
  00 00 00 08 08 00 18 08 08 08 08 08 08 3e 00 00
# j
  00 00 00 04 04 00 0c 04 04 04 04 04 04 04 48 30
# k
  00 00 00 40 40 40 44 48 50 60 50 48 44 42 00 00
# l
  00 00 00 18 08 08 08 08 08 08 08 08 08 3e 00 00
# m
  00 00 00 00 00 00 76 49 49 49 49 49 49 49 00 00
# n
  00 00 00 00 00 00 5c 62 42 42 42 42 42 42 00 00
# o
  00 00 00 00 00 00 3c 42 42 42 42 42 42 3c 00 00
# 0x70 = p
  00 00 00 00 00 00 5c 62 42 42 42 42 62 5c 40 40
# q
  00 00 00 00 00 00 3a 46 42 42 42 42 46 3a 02 02
# r
  00 00 00 00 00 00 5c 62 42 40 40 40 40 40 00 00
# s
  00 00 00 00 00 00 3c 42 40 30 0c 02 42 3c 00 00
# t
  00 00 00 00 10 10 10 7c 10 10 10 10 10 0c 00 00
# u
  00 00 00 00 00 00 42 42 42 42 42 42 46 3a 00 00
# v
  00 00 00 00 00 00 42 42 42 24 24 24 18 18 00 00
# w
  00 00 00 00 00 00 41 49 49 49 49 49 49 36 00 00
# x
  00 00 00 00 00 00 42 42 24 18 18 24 42 42 00 00
# y
  00 00 00 00 00 00 42 42 42 42 42 26 1a 02 02 3c
# z
  00 00 00 00 00 00 7e 02 04 08 10 20 40 7e 00 00
# {
  00 00 00 0c 10 10 08 08 10 20 10 08 08 10 10 0c
# |
  00 00 08 08 08 08 08 08 08 08 08 08 08 08 08 08
# }
  00 00 00 30 08 08 10 10 08 04 08 10 10 08 08 30
# ~
  00 00 00 31 49 46 00 00 00 00 00 00 00 00 00 00
# 0x7f = del (unused)
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
# }}}

## Controlling IDE (ATA) hard disks
# Uses 28-bit PIO mode.
# Inspired by https://colorforth.github.io/ide.html
#
# Resources:
#   https://wiki.osdev.org/ATA_PIO_Mode
#   https://forum.osdev.org/viewtopic.php?f=1&p=167798
#   read-sector, according to https://www.scs.stanford.edu/11wi-cs140/pintos/specs/ata-3-std.pdf

== data

# code disk
# All ports are 8-bit except data-port, which is 16-bit.
Primary-bus-primary-drive:
  # command-port: int (write)
  0x1f7/imm32
  # status-port: int (read)
  0x1f7/imm32
  # alternative-status-port: int (read)
  0x3f6/imm32
  # error-port: int (read)
  0x1f1/imm32
  # drive-and-head-port: int
  0x1f6/imm32
  # sector-count-port: int
  0x1f2/imm32
  # lba-low-port: int
  0x1f3/imm32
  # lba-mid-port: int
  0x1f4/imm32
  # lba-high-port: int
  0x1f5/imm32
  # data-port: int
  0x1f0/imm32
  # drive-code: byte                # only drive-specific field
  0xe0/imm32  # LBA mode also enabled

# data disk
# All ports are 8-bit except data-port, which is 16-bit.
Primary-bus-secondary-drive:
  # command-port: int (write)
  0x1f7/imm32
  # status-port: int (read)
  0x1f7/imm32
  # alternative-status-port: int (read)
  0x3f6/imm32
  # error-port: int (read)
  0x1f1/imm32
  # drive-and-head-port: int
  0x1f6/imm32
  # sector-count-port: int
  0x1f2/imm32
  # lba-low-port: int
  0x1f3/imm32
  # lba-mid-port: int
  0x1f4/imm32
  # lba-high-port: int
  0x1f5/imm32
  # data-port: int
  0x1f0/imm32
  # drive-code: byte                # only drive-specific field
  0xf0/imm32  # LBA mode also enabled

== code

load-sectors:  # disk: (addr disk), lba: int, n: int, out: (addr stream byte)
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  51/push-ecx
  52/push-edx
  # check for drive
  (drive-exists? *(ebp+8))  # => eax
  3d/compare-eax-and 0/imm32/false
  0f 84/jump-if-= $load-sectors:end/disp32
  # kick off read
  (ata-drive-select *(ebp+8) *(ebp+0xc))
  (clear-ata-error *(ebp+8))
  (ata-sector-count *(ebp+8) *(ebp+0x10))
  (ata-lba *(ebp+8) *(ebp+0xc))
  (ata-command *(ebp+8) 0x20)  # read sectors with retries
  # for each sector
  {
    # poll for results
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "waiting for sector.." 7 0)
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "." 7 0)
    (while-ata-busy *(ebp+8))
    (until-ata-data-available *(ebp+8))
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "reading\n" 7 0)
    # var data-port/edx = disk->data-port
    8b/-> *(ebp+8) 0/r32/eax
    8b/-> *(eax+0x24) 2/r32/edx
    # emit results
    31/xor %eax 0/r32/eax
    b9/copy-to-ecx 0x200/imm32  # 512 bytes per sector
    {
      81 7/subop/compare %ecx 0/imm32
      74/jump-if-= break/disp8
      66 ed/read-port-dx-into-ax
      # write 2 bytes to stream one at a time
      (append-byte *(ebp+0x14) %eax)
      49/decrement-ecx
      c1/shift 5/subop/right-padding-zeroes %eax 8/imm8
      (append-byte *(ebp+0x14) %eax)
      49/decrement-ecx
      eb/jump loop/disp8
    }
    # next sector
    ff 1/subop/decrement *(ebp+0x10)
#?     (draw-int32-decimal-wrapping-right-then-down-from-cursor-over-full-screen 0 *(ebp+0x10) 0xc 0)
    81 7/subop/compare *(ebp+0x10) 0/imm32
    7e/jump-if-<= break/disp8
    (wait-400ns *(ebp+8))
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "next sector\n" 7 0)
    e9/jump loop/disp32
  }
$load-sectors:end:
  # . restore registers
  5a/pop-to-edx
  59/pop-to-ecx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

store-sectors:  # disk: (addr disk), lba: int, n: int, in: (addr stream byte)
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  51/push-ecx
  52/push-edx
  53/push-ebx
  # check for drive
  (drive-exists? *(ebp+8))  # => eax
  3d/compare-eax-and 0/imm32/false
  0f 84/jump-if-= $store-sectors:end/disp32
  # kick off write
  (ata-drive-select *(ebp+8) *(ebp+0xc))
  (clear-ata-error *(ebp+8))
  (ata-sector-count *(ebp+8) *(ebp+0x10))
  (ata-lba *(ebp+8) *(ebp+0xc))
  (ata-command *(ebp+8) 0x30)  # write sectors with retries
  # for each sector
#?   (set-cursor-position 0 0 0)
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "0" 7 0)
  {
    # wait
    (while-ata-busy *(ebp+8))
    (until-ata-ready-for-data *(ebp+8))
    # var data-port/edx = disk->data-port
    8b/-> *(ebp+8) 0/r32/eax
    8b/-> *(eax+0x24) 2/r32/edx
    # send data
    b9/copy-to-ecx 0x200/imm32  # 512 bytes per sector
    # . var first-byte/ebx: byte
    # . when it's more than 0xff, we're at an even-numbered byte
    bb/copy-to-ebx 0xffff/imm32
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "D" 7 0)
$store-sectors:store-sector:
    {
      81 7/subop/compare %ecx 0/imm32
      74/jump-if-= break/disp8
      # this loop is slow, but the ATA spec also requires a small delay
      (stream-empty? *(ebp+0x14))  # => eax
      3d/compare-eax-and 0/imm32/false
      75/jump-if-!= break/disp8
      # read byte from stream
      (read-byte *(ebp+0x14))  # => eax
      # if we're at an odd-numbered byte, save it to first-byte
      81 7/subop/compare %ebx 0xff/imm32
      {
        7e/jump-if-<= break/disp8
        89/<- %ebx 0/r32/eax
        eb/jump $store-sectors:store-sector/disp8
      }
      # otherwise OR it with first-byte and write it out
      c1/shift 4/subop/left %eax 8/imm8
      09/or %eax 3/r32/ebx
      66 ef/write-ax-into-port-dx
      49/decrement-ecx
      49/decrement-ecx
      # reset first-byte
      bb/copy-to-ebx 0xffff/imm32
      eb/jump loop/disp8
    }
    # write out final first-byte if necessary
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "I" 7 0)
    81 7/subop/compare %ebx 0xff/imm32
    {
      7f/jump-if-> break/disp8
      89/<- %eax 3/r32/ebx
      66 ef/write-ax-into-port-dx
      49/decrement-ecx
      49/decrement-ecx
    }
    # pad zeroes
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "P" 7 0)
    31/xor %eax 0/r32/eax
    {
      81 7/subop/compare %ecx 0/imm32
      74/jump-if-= break/disp8
      66 ef/write-ax-into-port-dx
      49/decrement-ecx
      49/decrement-ecx
      eb/jump loop/disp8
    }
    # next sector
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "N" 7 0)
    ff 1/subop/decrement *(ebp+0x10)
    81 7/subop/compare *(ebp+0x10) 0/imm32
    7e/jump-if-<= break/disp8
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "W" 7 0)
    (wait-400ns *(ebp+8))
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "L" 7 0)
    e9/jump loop/disp32
  }
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "F" 7 0)
  (flush-ata-cache *(ebp+8))
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "Y" 7 0)
$store-sectors:end:
  # . restore registers
  5b/pop-to-ebx
  5a/pop-to-edx
  59/pop-to-ecx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

# disk helpers {{{

drive-exists?:  # disk: (addr disk) -> _/eax: boolean
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  52/push-edx
  # check for floating bus
  {
    31/xor %eax 0/r32/eax
    ba/copy-to-edx 0x1f7/imm32
    ec/read-port-dx-into-al
    3d/compare-eax-and 0xff/imm32
    # if eax is 0xff, primary bus has no drives
    b8/copy-to-eax 0/imm32/false
    0f 84/jump-if-= $drive-exists?:end/disp32
  }
  # identify
  (ata-drive-select *(ebp+8) 0)
  (ata-sector-count *(ebp+8) 0)
  (ata-lba *(ebp+8) 0)
  (ata-command *(ebp+8) 0xec)  # identify
  # var status-port/edx = disk->status-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+4) 2/r32/edx  # 4 = status-port offset
  # read status port
  # TODO: might need to spin here for 400ns: https://wiki.osdev.org/index.php?title=ATA_PIO_Mode&oldid=25664#400ns_delays
  31/xor %eax 0/r32/eax
  ec/read-port-dx-into-al
  # if eax is 0, drive does not exist
  3d/compare-eax-and 0/imm32
  {
    74/jump-if-= break/disp8
    b8/copy-to-eax 1/imm32/true
    eb/jump $drive-exists?:complete-identify/disp8
  }
  # TODO: might need to perform remaining steps at https://wiki.osdev.org/index.php?title=ATA_PIO_Mode&oldid=25664#IDENTIFY_command
  b8/copy-to-eax 0/imm32/false
$drive-exists?:complete-identify:
  50/push-eax
  # var data-port/edx = disk->data-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+0x24) 2/r32/edx  # 0x24 = data-port offset
  # clear FIFO from the drive
  b9/copy-to-ecx 0x200/imm32
  {
    81 7/subop/compare %ecx 0/imm32
    74/jump-if-= break/disp8
    # read 4 bytes
    ed/read-port-dx-into-eax
    49/decrement-ecx
    49/decrement-ecx
    49/decrement-ecx
    49/decrement-ecx
    eb/jump loop/disp8
  }
  58/pop-to-eax
$drive-exists?:end:
  # . restore registers
  5a/pop-to-edx
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

ata-drive-select:  # disk: (addr disk), lba: int
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  52/push-edx
  56/push-esi
  # esi = disk
  8b/-> *(ebp+8) 6/r32/esi
  # var drive-head/edx: byte = lba >> 24
  8b/-> *(ebp+0xc) 2/r32/edx
  c1/shift 5/subop/right-padding-zeroes %edx 0x18/imm8
  # var drive-code/eax: byte = disk->drive-code | drive-head
  8b/-> *(esi+0x28) 0/r32/eax  # 0x28 = drive-code offset
  09/or= %eax 2/r32/edx
  # var drive-and-head-port/edx: int
  8b/-> *(esi+0x10) 2/r32/edx  # 0x10 = drive-and-head-port offset
  ee/write-al-into-port-dx
$ata-drive-select:end:
  # . restore registers
  5e/pop-to-esi
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

clear-ata-error:  # disk: (addr disk)
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  52/push-edx
  # var error-port/edx = disk->error-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+0xc) 2/r32/edx  # 0xc = error-port offset
  #
  b8/copy-to-eax 0/imm32
  ee/write-al-into-port-dx
$ata-error:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

ata-sector-count:  # disk: (addr disk), n: byte
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  52/push-edx
  # var sector-count-port/edx = disk->sector-count-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+0x14) 2/r32/edx  # 0x14 = sector-count-port offset
  #
  8b/-> *(ebp+0xc) 0/r32/eax
  ee/write-al-into-port-dx
$ata-sector-count:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

ata-lba:  # disk: (addr disk), lba: int
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  52/push-edx
  # var port/edx = disk->port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+0x18) 2/r32/edx  # 0x18 = lba-low-port offset
  # eax = lba
  8b/-> *(ebp+0xc) 0/r32/eax
  # lo
  ee/write-al-into-port-dx
  # mid
  42/increment-dx  # lba-mid-port
  c1/shift 5/subop/right-padding-zeroes %eax 8/imm8
  ee/write-al-into-port-dx
  # hi
  42/increment-dx  # lba-high-port
  c1/shift 5/subop/right-padding-zeroes %eax 8/imm8
  ee/write-al-into-port-dx
$ata-lba:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

ata-command:  # disk: (addr disk), cmd: byte
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  52/push-edx
  # var command-port/edx = disk->command-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+0) 2/r32/edx  # 0 = command-port offset
  #
  8b/-> *(ebp+0xc) 0/r32/eax
  ee/write-al-into-port-dx
$ata-command:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

while-ata-busy:  # disk: (addr disk)
  # . save registers
  50/push-eax
  52/push-edx
  # var status-port/edx = disk->status-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+4) 2/r32/edx  # 4 = status-port offset
  {
    ec/read-port-dx-into-al
    a8/test-bits-in-al 0x80/imm8/bsy  # set zf if bit 7 (most significant) is not set
    75/jump-if-zf-not-set-and-bit-7-set loop/disp8
  }
$while-ata-busy:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  c3/return

until-ata-data-available:  # disk: (addr disk)
  # . save registers
  50/push-eax
  52/push-edx
  # var status-port/edx = disk->status-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+4) 2/r32/edx  # 4 = status-port offset
  {
    ec/read-port-dx-into-al
    a8/test-bits-in-al 8/imm8/drq  # set zf if bit 3 is not set
    74/jump-if-zf-set-and-bit-3-not-set loop/disp8
  }
$until-ata-data-available:end:
  # . restore registers
  5a/pop-to-edx
  58/pop-to-eax
  # . epilogue
  c3/return

until-ata-ready-for-data:
  (until-ata-data-available)
  c3/return

# https://wiki.osdev.org/index.php?title=ATA_PIO_Mode&oldid=25664#400ns_delays
wait-400ns:  # disk: (addr disk)
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  51/push-ecx
  52/push-edx
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "waiting 400ns\n" 7 0)
  # var status-port/edx = disk->status-port
  8b/-> *(ebp+8) 0/r32/eax
  8b/-> *(eax+4) 2/r32/edx  # 4 = status-port offset
  #
  b9/copy-to-ecx 0x10/imm32
  {
    81 7/subop/compare %ecx 0/imm32
    74/jump-if-= break/disp8
#?     (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "." 7 0)
    ec/read-port-dx-into-al
    49/decrement-ecx
    eb/jump loop/disp8
  }
$wait-400ns:end:
  # . restore registers
  5a/pop-to-edx
  59/pop-to-ecx
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

# Flush cache isn't in ATA 3, but it shows up by the ATA 5 spec:
#   http://hddguru.com/download/documentation/ATA-ATAPI-standard-5/ATA-ATAPI-5.pdf
flush-ata-cache:  # disk: (addr disk)
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  #
  (ata-drive-select *(ebp+8) 0)
  (ata-command *(ebp+8) 0xe7)  # flush cache
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "W" 7 0)
  (while-ata-busy *(ebp+8))
#?   (draw-text-wrapping-right-then-down-from-cursor-over-full-screen 0 "X" 7 0)
  # TODO: seems unneeded? works for a single sector but Qemu hangs with multiple
  # sectors. Data is still written.
#?   (until-ata-ready-for-data *(ebp+8))
$flush-ata-cache:end:
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

# }}}

## Controlling a PS/2 mouse
# Uses no IRQs, just polling.
# Thanks Dave Long: https://github.com/jtauber/cleese/blob/master/necco/kernel/bochs/py8042.py
#
# Resources:
#   https://wiki.osdev.org/Mouse_Input

# results x/eax, y/ecx range from -256 to +255
# See https://wiki.osdev.org/index.php?title=Mouse_Input&oldid=25663#Format_of_First_3_Packet_Bytes
read-mouse-event:  # -> _/eax: int, _/ecx: int
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  52/push-edx
  53/push-ebx
  # if no event, return 0, 0
  b8/copy-to-eax 0/imm32
  b9/copy-to-ecx 0/imm32
  (any-mouse-event?)  # => eax
  3d/compare-eax-and 0/imm32/false
  74/jump-if-= $read-mouse-event:end/disp8
  # var f1/edx: byte = inb(0x60)
  31/xor %eax 0/r32/eax
  e4/read-port-into-al 0x60/imm8
  89/<- %edx 0/r32/eax
  (wait-for-mouse-event)
  # var dx/ebx: byte = inb(0x60)
  31/xor %eax 0/r32/eax
  e4/read-port-into-al 0x60/imm8
  89/<- %ebx 0/r32/eax
  (wait-for-mouse-event)
  # var dy/ecx: byte = inb(0x60)
  31/xor %eax 0/r32/eax
  e4/read-port-into-al 0x60/imm8
  89/<- %ecx 0/r32/eax
  # eax = dx
  89/<- %eax 3/r32/ebx
  # if (f1 & 0x10) dx = -dx
  {
    f6 0/subop/test-bits %dl 0x10/imm8
    74/jump-if-zero break/disp8
    0d/or-eax-with 0xffffff00/imm32
  }
  # if (f1 & 0x20) dy = -dy
  {
    f6 0/subop/test-bits %dl 0x20/imm8
    74/jump-if-zero break/disp8
    81 1/subop/or %ecx 0xffffff00/imm32
  }
$read-mouse-event:end:
  # . restore registers
  5b/pop-to-ebx
  5a/pop-to-edx
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

# mouse helpers {{{

wait-for-mouse-event:
  # . save registers
  50/push-eax
  #
  {
    (any-mouse-event?)  # => eax
    3d/compare-eax-and 0/imm32/false
    74/jump-if-= loop/disp8
  }
$wait-for-mouse-event:end:
  # . restore registers
  58/pop-to-eax
  # .
  c3/return

any-mouse-event?:  # -> _/eax: boolean
  31/xor %eax 0/r32/eax
  # 0x1 bit: there's data from the keyboard controller
  # 0x20 bit: it's data from the aux port (the mouse)
  e4/read-port-into-al 0x60/imm8
  24/and-al-with 0x21/imm8
  3c/compare-al-with 0x21/imm8
  0f 94/set-byte-if-= %al
  c3/return

initialize-mouse:
  (enable-keyboard-controller-aux-device)
  # tell mouse to use default settings
  (send-mouse-command 0xf6)
  # enable mouse
  (send-mouse-command 0xf4)
  c3/return

enable-keyboard-controller-aux-device:
  (command-keyboard-controller 0xa8)
  c3/return

send-mouse-command:  # command: byte
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  #
  (command-keyboard-controller 0xd4)
  (send-keyboard-controller-data *(ebp+8))
  (wait-for-ack-from-mouse)
$send-mouse-command:end:
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

wait-for-ack-from-mouse:
  # . save registers
  50/push-eax
  {
    (read-keyboard-controller-data)  # => eax
    3d/compare-eax-with 0xfa/imm32
    75/jump-if-!= loop/disp8
  }
$wait-for-ack-from-mouse:end:
  # . restore registers
  58/pop-eax
  c3/return

command-keyboard-controller:  # command: byte
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  #
  (poll-keyboard-controller-to-write)
  8b/-> *(ebp+8) 0/r32/eax
  e6/write-al-into-port 0x64/imm8
$command-keyboard-controller:end:
  # . restore registers
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

send-keyboard-controller-data:  # data: byte
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  #
  (poll-keyboard-controller-to-write)
  8b/-> *(ebp+8) 0/r32/eax
  e6/write-al-into-port 0x60/imm8
$send-keyboard-controller-data:end:
  # . restore registers
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

read-keyboard-controller-data:  # -> _/eax: byte
  (poll-keyboard-controller-to-read-data-port)
  31/xor %eax 0/r32/eax
  e4/read-port-into-al 0x60/imm8
  c3/return

poll-keyboard-controller-to-write:
  # . save registers
  50/push-eax
  # "All output to port 0x60 or 0x64 must be preceded by waiting for bit 1
  # (value=2) of port 0x64 to become clear."
  #   https://wiki.osdev.org/index.php?title=Mouse_Input&oldid=25663#Waiting_to_Send_Bytes_to_Port_0x60_and_0x64
  {
    e4/read-port-into-al 0x64/imm8
    a8/test-bits-in-al 2/imm8  # set zf if bit 1 (second-least significant) is not set
    75/jump-if-zf-not-set-and-bit-1-set loop/disp8
  }
$poll-keyboard-controller-to-write:end:
  # . restore registers
  58/pop-to-eax
  # . epilogue
  c3/return

poll-keyboard-controller-to-read-data-port:
  # . prologue
  55/push-ebp
  89/<- %ebp 4/r32/esp
  # . save registers
  50/push-eax
  # "Bytes cannot be read from port 0x60 until bit 0 (value=1) of port 0x64 is set."
  #   https://wiki.osdev.org/index.php?title=Mouse_Input&oldid=25663#Waiting_to_Send_Bytes_to_Port_0x60_and_0x64
  {
    e4/read-port-into-al 0x64/imm8
    a8/test-bits-in-al 1/imm8  # set zf if bit 0 (least significant) is not set
    74/jump-if-zf-set-and-bit-0-not-set loop/disp8
  }
$poll-keyboard-controller-to-read-data-port:end:
  # . restore registers
  58/pop-to-eax
  # . epilogue
  89/<- %esp 5/r32/ebp
  5d/pop-to-ebp
  c3/return

# }}}

# vim:ft=subx