andrius
/
stage0-uefi
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

hex0: optimize stack alignment code.

This commit is contained in:
Andrius Štikonas 2023-12-17 17:34:49 +00:00
parent 78a70d0af8
commit ddcc53ab44
4 changed files with 52 additions and 75 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
amd64/Development/hex0.M1
View File

@ -48,7 +48,6 @@ DEFINE mov_al,[rbx] 8A03
DEFINE mov_[rbx], C603 DEFINE mov_[rbx], C603
DEFINE mov_rbx,[rdi+BYTE] 488B5F DEFINE mov_rbx,[rdi+BYTE] 488B5F
DEFINE mov_rcx,[rdi+BYTE] 488B4F DEFINE mov_rcx,[rdi+BYTE] 488B4F
DEFINE mov_rsp,[rsp+BYTE] 488B6424
DEFINE mov_r14,[rdx+BYTE] 4C8B72 DEFINE mov_r14,[rdx+BYTE] 4C8B72
DEFINE pop_rax 58 DEFINE pop_rax 58
DEFINE pop_rbx 5B DEFINE pop_rbx 5B
@ -69,12 +68,10 @@ DEFINE push_rbx 53
DEFINE push_rdi 57 DEFINE push_rdi 57
DEFINE push_rdx 52 DEFINE push_rdx 52
DEFINE push_rsi 56 DEFINE push_rsi 56
DEFINE push_rsp 54
DEFINE push_r12 4154 DEFINE push_r12 4154
DEFINE push_r13 4155 DEFINE push_r13 4155
DEFINE push_r14 4156 DEFINE push_r14 4156
DEFINE push_r15 4157 DEFINE push_r15 4157
DEFINE push_[rsp] FF3424
DEFINE ret C3 DEFINE ret C3
DEFINE ror_r9 49D1C9 DEFINE ror_r9 49D1C9
DEFINE shl_edi, C1E7 DEFINE shl_edi, C1E7
@ -266,8 +263,9 @@ DEFINE xor_r9,r9 4D31C9
jmp !ascii_other jmp !ascii_other
:purge_comment :purge_comment
# Read a byte push_rbx # align stack to 16 bytes
call %read_byte call %read_byte # Read a byte
pop_rbx # deallocate stack
# Loop if not LF # Loop if not LF
cmp_al, !10 cmp_al, !10
@ -293,17 +291,14 @@ DEFINE xor_r9,r9 4D31C9
# Writes byte stored in al # Writes byte stored in al
:write_byte :write_byte
push_rsp # align stack to 16 bytes
push_[rsp] # align stack to 16 bytes
and_rsp, !-16 # align stack to 16 bytes
mov_rcx,r13 # arg1 = fout mov_rcx,r13 # arg1 = fout
push !1 # size = 1 push !1 # size = 1
mov_rdx,rsp # arg2 = &size mov_rdx,rsp # arg2 = &size
push_rax # allocate stack push_rax # allocate stack
mov_r8,rsp # arg3 = &output mov_r8,rsp # arg3 = &output
sub_rsp, !32 # allocate shadow stack space for UEFI function sub_rsp, !40 # allocate shadow stack space for UEFI function
call_[rcx+BYTE] !40 # fout->write() call_[rcx+BYTE] !40 # fout->write()
mov_rsp,[rsp+BYTE] !56 # deallocate stack add_rsp, !56 # deallocate stack
ret # return ret # return
@ -313,12 +308,9 @@ DEFINE xor_r9,r9 4D31C9
mov_rdx,rsp # arg2 = &size mov_rdx,rsp # arg2 = &size
push !0 # allocate stack push !0 # allocate stack
mov_r8,rsp # arg3 = &input mov_r8,rsp # arg3 = &input
push_rsp # align stack to 16 bytes sub_rsp, !40 # allocate shadow stack space for UEFI function
push_[rsp] # align stack to 16 bytes
and_rsp, !-16 # align stack to 16 bytes
sub_rsp, !32 # allocate shadow stack space for UEFI function
call_[rcx+BYTE] !32 # fin->read() call_[rcx+BYTE] !32 # fin->read()
mov_rsp,[rsp+BYTE] !40 # deallocate stack add_rsp, !40 # deallocate stack
pop_rax # save input to rax pop_rax # save input to rax
pop_rsi # save size to rsi pop_rsi # save size to rsi

19
amd64/Development/hex0.S
View File

@ -193,8 +193,9 @@ hex:
jmp ascii_other jmp ascii_other
purge_comment: purge_comment:
# Read a byte push rbx # align stack to 16 bytes
call read_byte call read_byte # Read a byte
pop rbx # deallocate stack
# Loop if not LF (works for CR/LF and LF/CR endings too) # Loop if not LF (works for CR/LF and LF/CR endings too)
cmp al, 10 cmp al, 10
@ -220,17 +221,14 @@ ascii_high:
# Writes byte stored in al # Writes byte stored in al
write_byte: write_byte:
push rsp # align stack to 16 bytes
push [rsp] # align stack to 16 bytes
and rsp, -16 # align stack to 16 bytes
mov rcx, r13 # arg1 = fout mov rcx, r13 # arg1 = fout
push 1 # size = 1 push 1 # size = 1
mov rdx, rsp # arg2 = &size mov rdx, rsp # arg2 = &size
push rax # allocate stack push rax # allocate stack
mov r8, rsp # arg3 = &output mov r8, rsp # arg3 = &output
sub rsp, 32 # allocate shadow stack space for UEFI function sub rsp, 40 # allocate shadow stack space for UEFI function
call [rcx+40] # fout->write() call [rcx+40] # fout->write()
mov rsp, [rsp+56] # deallocate stack add rsp, 56 # deallocate stack
ret # return ret # return
@ -240,12 +238,9 @@ read_byte:
mov rdx, rsp # arg2 = &size mov rdx, rsp # arg2 = &size
push 0 # allocate stack push 0 # allocate stack
mov r8, rsp # arg3 = &input mov r8, rsp # arg3 = &input
push rsp # align stack to 16 bytes sub rsp, 40 # allocate shadow stack space for UEFI function
push [rsp] # align stack to 16 bytes
and rsp, -16 # align stack to 16 bytes
sub rsp, 32 # allocate shadow stack space for UEFI function
call [rcx+32] # fin->read() call [rcx+32] # fin->read()
mov rsp, [rsp+40] # deallocate stack add rsp, 40 # deallocate stack
pop rax # save input to rax pop rax # save input to rax
pop rsi # save size to rsi pop rsi # save size to rsi

19
amd64/Development/hex0.hex2
View File

@ -189,8 +189,9 @@
EB !ascii_other ; jmp !ascii_other EB !ascii_other ; jmp !ascii_other
:purge_comment :purge_comment
# Read a byte 53 ; push_rbx # align stack to 16 bytes
E8 %read_byte ; call %read_byte E8 %read_byte ; call %read_byte # Read a byte
5B ; pop_rbx # deallocate stack
# Loop if not LF # Loop if not LF
3C 0A ; cmp_al, !10 3C 0A ; cmp_al, !10
@ -216,17 +217,14 @@
# Writes byte stored in al # Writes byte stored in al
:write_byte :write_byte
54 ; push_rsp # align stack to 16 bytes
FF3424 ; push_[rsp] # align stack to 16 bytes
4883E4 F0 ; and_rsp, !-16 # align stack to 16 bytes
4C89E9 ; mov_rcx,r13 # arg1 = fout 4C89E9 ; mov_rcx,r13 # arg1 = fout
6A 01 ; push !1 # size = 1 6A 01 ; push !1 # size = 1
4889E2 ; mov_rdx,rsp # arg2 = &size 4889E2 ; mov_rdx,rsp # arg2 = &size
50 ; push_rax # allocate stack 50 ; push_rax # allocate stack
4989E0 ; mov_r8,rsp # arg3 = &output 4989E0 ; mov_r8,rsp # arg3 = &output
4883EC 20 ; sub_rsp, !32 # allocate shadow stack space for UEFI function 4883EC 28 ; sub_rsp, !40 # allocate shadow stack space for UEFI function
FF51 28 ; call_[rcx+BYTE] !40 # fout->write() FF51 28 ; call_[rcx+BYTE] !40 # fout->write()
488B6424 38 ; mov_rsp,[rsp+BYTE] !56 # deallocate stack 4883C4 38 ; add_rsp, !56 # deallocate stack
C3 ; ret # return C3 ; ret # return
@ -236,12 +234,9 @@
4889E2 ; mov_rdx,rsp # arg2 = &size 4889E2 ; mov_rdx,rsp # arg2 = &size
6A 00 ; push !0 # allocate stack 6A 00 ; push !0 # allocate stack
4989E0 ; mov_r8,rsp # arg3 = &input 4989E0 ; mov_r8,rsp # arg3 = &input
54 ; push_rsp # align stack to 16 bytes 4883EC 28 ; sub_rsp, !40 # allocate shadow stack space for UEFI function
FF3424 ; push_[rsp] # align stack to 16 bytes
4883E4 F0 ; and_rsp, !-16 # align stack to 16 bytes
4883EC 20 ; sub_rsp, !32 # allocate shadow stack space for UEFI function
FF51 20 ; call_[rcx+BYTE] !32 # fin->read() FF51 20 ; call_[rcx+BYTE] !32 # fin->read()
488B6424 28 ; mov_rsp,[rsp+BYTE] !40 # deallocate stack 4883C4 28 ; add_rsp, !40 # deallocate stack
58 ; pop_rax # save input to rax 58 ; pop_rax # save input to rax
5E ; pop_rsi # save size to rsi 5E ; pop_rsi # save size to rsi

67
amd64/hex0.hex0
View File

@ -129,9 +129,9 @@ F0 00 # SizeOfOptionalHeader
# [0x148] # [0x148]
# Start of section headers # Start of section headers
00 00 00 00 00 00 00 00 ; Name of the section (empty) but could set to ".text" 00 00 00 00 00 00 00 00 ; Name of the section (empty) but could set to ".text"
FF 01 00 00 ; VirtualSize EF 01 00 00 ; VirtualSize
00 10 00 00 ; VirtualAddress 00 10 00 00 ; VirtualAddress
FF 01 00 00 ; SizeOfRawData EF 01 00 00 ; SizeOfRawData
70 01 00 00 ; PointerToRawData 70 01 00 00 ; PointerToRawData
00 00 00 00 ; PointerToRelocations 00 00 00 00 ; PointerToRelocations
00 00 00 00 ; PointerToLinenumbers 00 00 00 00 ; PointerToLinenumbers
@ -158,7 +158,7 @@ FF 01 00 00 ; SizeOfRawData
# Open Loaded Image protocol # Open Loaded Image protocol
4D89F9 ; mov_r9,r15 # arg4 = image_handle 4D89F9 ; mov_r9,r15 # arg4 = image_handle
488D15 C2010000 ; lea_rdx,[rip+DWORD] %LOADED_IMAGE_PROTOCOL # guid = &LOADED_IMAGE_PROTOCOL 488D15 B2010000 ; lea_rdx,[rip+DWORD] %LOADED_IMAGE_PROTOCOL # guid = &LOADED_IMAGE_PROTOCOL
4C89C9 ; mov_rcx,r9 # arg1 = image_handle 4C89C9 ; mov_rcx,r9 # arg1 = image_handle
50 ; push_rax # allocate stack for image 50 ; push_rax # allocate stack for image
4989E0 ; mov_r8,rsp # arg3 = &image 4989E0 ; mov_r8,rsp # arg3 = &image
@ -171,7 +171,7 @@ FF 01 00 00 ; SizeOfRawData
# Get root file system # Get root file system
4D89F9 ; mov_r9,r15 # arg4 = image_handle 4D89F9 ; mov_r9,r15 # arg4 = image_handle
488D15 AD010000 ; lea_rdx,[rip+DWORD] %SIMPLE_FS_PROTOCOL # guid = &SIMPLE_FS_PROTOCOL 488D15 9D010000 ; lea_rdx,[rip+DWORD] %SIMPLE_FS_PROTOCOL # guid = &SIMPLE_FS_PROTOCOL
488B4F 18 ; mov_rcx,[rdi+BYTE] !24 # arg1 = root_device = image->device 488B4F 18 ; mov_rcx,[rdi+BYTE] !24 # arg1 = root_device = image->device
4989CD ; mov_r13,rcx # save root_device 4989CD ; mov_r13,rcx # save root_device
50 ; push_rax # allocate stack for rootfs 50 ; push_rax # allocate stack for rootfs
@ -250,7 +250,7 @@ FF 01 00 00 ; SizeOfRawData
# :loop [_start+0xDE] # :loop [_start+0xDE]
# Read a byte # Read a byte
E8 79000000 ; call %read_byte E8 72000000 ; call %read_byte
# process byte # process byte
E8 1C000000 ; call %hex E8 1C000000 ; call %hex
@ -277,7 +277,7 @@ FF 01 00 00 ; SizeOfRawData
# flip the toggle # flip the toggle
FFCB ; dec_ebx # rbx = -1 FFCB ; dec_ebx # rbx = -1
E8 39000000 ; call %write_byte E8 3B000000 ; call %write_byte
EB DA ; jmp !loop EB DA ; jmp !loop
@ -292,85 +292,80 @@ FF 01 00 00 ; SizeOfRawData
# deal all ascii less than '0' # deal all ascii less than '0'
3C 30 ; cmp_al, !48 3C 30 ; cmp_al, !48
7C 1F ; jl !ascii_other 7C 21 ; jl !ascii_other
# deal with 0-9 # deal with 0-9
3C 3A ; cmp_al, !58 3C 3A ; cmp_al, !58
7C 1F ; jl !ascii_num 7C 21 ; jl !ascii_num
# deal with all ascii less than 'A' # deal with all ascii less than 'A'
3C 41 ; cmp_al, !65 3C 41 ; cmp_al, !65
7C 17 ; jl !ascii_other 7C 19 ; jl !ascii_other
# deal with 'A'-'F' # deal with 'A'-'F'
3C 47 ; cmp_al, !71 3C 47 ; cmp_al, !71
7C 1C ; jl !ascii_high 7C 1E ; jl !ascii_high
# deal with all ascii less than 'a' # deal with all ascii less than 'a'
3C 61 ; cmp_al, !97 3C 61 ; cmp_al, !97
7C 0F ; jl !ascii_other 7C 11 ; jl !ascii_other
#deal with 'a'-'f' #deal with 'a'-'f'
3C 67 ; cmp_al, !103 3C 67 ; cmp_al, !103
7C 12 ; jl !ascii_low 7C 14 ; jl !ascii_low
# The rest that remains needs to be ignored # The rest that remains needs to be ignored
EB 09 ; jmp !ascii_other EB 0B ; jmp !ascii_other
# :purge_comment [_start+0x126] # :purge_comment [_start+0x126]
# Read a byte 53 ; push_rbx # align stack to 16 bytes
E8 31000000 ; call %read_byte E8 29000000 ; call %read_byte # Read a byte
5B ; pop_rbx # deallocate stack
# Loop if not LF # Loop if not LF
3C 0A ; cmp_al, !10 3C 0A ; cmp_al, !10
75 F7 ; jne !purge_comment 75 F5 ; jne !purge_comment
# Otherwise return -1 # Otherwise return -1
# :ascii_other [_start+0x12F] # :ascii_other [_start+0x131]
6A FF ; push !-1 6A FF ; push !-1
58 ; pop_rax # return = -1 58 ; pop_rax # return = -1
C3 ; ret C3 ; ret
# :ascii_num [_start+0x133] # :ascii_num [_start+0x135]
2C 30 ; sub_al, !48 2C 30 ; sub_al, !48
C3 ; ret C3 ; ret
# :ascii_low [_start+0x136] # :ascii_low [_start+0x138]
2C 20 ; sub_al, !32 # convert to uppercase 2C 20 ; sub_al, !32 # convert to uppercase
# :ascii_high [_start+0x138] # :ascii_high [_start+0x13A]
2C 37 ; sub_al, !55 2C 37 ; sub_al, !55
C3 ; ret C3 ; ret
# Writes byte stored in al # Writes byte stored in al
# :write_byte [_start+0x13B] # :write_byte [_start+0x13D]
54 ; push_rsp # align stack to 16 bytes
FF3424 ; push_[rsp] # align stack to 16 bytes
4883E4 F0 ; and_rsp, !-16 # align stack to 16 bytes
4C89E9 ; mov_rcx,r13 # arg1 = fout 4C89E9 ; mov_rcx,r13 # arg1 = fout
6A 01 ; push !1 # size = 1 6A 01 ; push !1 # size = 1
4889E2 ; mov_rdx,rsp # arg2 = &size 4889E2 ; mov_rdx,rsp # arg2 = &size
50 ; push_rax # allocate stack 50 ; push_rax # allocate stack
4989E0 ; mov_r8,rsp # arg3 = &output 4989E0 ; mov_r8,rsp # arg3 = &output
4883EC 20 ; sub_rsp, !32 # allocate shadow stack space for UEFI function 4883EC 28 ; sub_rsp, !40 # allocate shadow stack space for UEFI function
FF51 28 ; call_[rcx+BYTE] !40 # fout->write() FF51 28 ; call_[rcx+BYTE] !40 # fout->write()
488B6424 38 ; mov_rsp,[rsp+BYTE] !56 # deallocate stack 4883C4 38 ; add_rsp, !56 # deallocate stack
C3 ; ret # return C3 ; ret # return
# :read_byte [_start+0x15C] # :read_byte [_start+0x155]
4C89E1 ; mov_rcx,r12 # arg1 = fin 4C89E1 ; mov_rcx,r12 # arg1 = fin
6A 01 ; push !1 # size = 1 6A 01 ; push !1 # size = 1
4889E2 ; mov_rdx,rsp # arg2 = &size 4889E2 ; mov_rdx,rsp # arg2 = &size
6A 00 ; push_rsi # allocate stack 6A 00 ; push_rsi # allocate stack
4989E0 ; mov_r8,rsp # arg3 = &input 4989E0 ; mov_r8,rsp # arg3 = &input
54 ; push_rsp # align stack to 16 bytes 4883EC 28 ; sub_rsp, !40 # allocate shadow stack space for UEFI function
FF3424 ; push_[rsp] # align stack to 16 bytes
4883E4 F0 ; and_rsp, !-16 # align stack to 16 bytes
4883EC 20 ; sub_rsp, !32 # allocate shadow stack space for UEFI function
FF51 20 ; call_[rcx+BYTE] !32 # fin->read() FF51 20 ; call_[rcx+BYTE] !32 # fin->read()
488B6424 28 ; mov_rsp,[rsp+BYTE] !40 # deallocate stack 4883C4 28 ; add_rsp, !40 # deallocate stack
58 ; pop_rax # save input to rax 58 ; pop_rax # save input to rax
5E ; pop_rsi # save size to rsi 5E ; pop_rsi # save size to rsi
@ -380,7 +375,7 @@ FF 01 00 00 ; SizeOfRawData
C3 ; ret # return C3 ; ret # return
# :terminate [_start+0x184] # :terminate [_start+0x174]
58 ; pop_rax # deallocate stack (return code of read_byte) 58 ; pop_rax # deallocate stack (return code of read_byte)
5E ; pop_rsi # restore rootdir 5E ; pop_rsi # restore rootdir
415E ; pop_r14 # restore system->boot 415E ; pop_r14 # restore system->boot
@ -422,16 +417,16 @@ FF 01 00 00 ; SizeOfRawData
# Protocol GUIDs # Protocol GUIDs
# :LOADED_IMAGE_PROTOCOL [_start+0x1DF] # :LOADED_IMAGE_PROTOCOL [_start+0x1CF]
A1 31 1B 5B ; %0x5b1b31a1 A1 31 1B 5B ; %0x5b1b31a1
62 95 ; $0x9562 62 95 ; $0x9562
D2 11 ; $0x11d2 D2 11 ; $0x11d2
8E 3F 00 A0 C9 69 72 3B ; !0x8e !0x3f !0 !0xa0 !0xc9 !0x69 !0x72 !0x3b 8E 3F 00 A0 C9 69 72 3B ; !0x8e !0x3f !0 !0xa0 !0xc9 !0x69 !0x72 !0x3b
# :SIMPLE_FS_PROTOCOL [_start+0x1EF] # :SIMPLE_FS_PROTOCOL [_start+0x1DF]
22 5B 4E 96 ; %0x964e5b22 22 5B 4E 96 ; %0x964e5b22
59 64 ; $0x6459 59 64 ; $0x6459
D2 11 ; $0x11d2 D2 11 ; $0x11d2
8E 39 00 A0 C9 69 72 3B ; !0x8e !0x39 !0 !0xa0 !0xc9 !0x69 !0x72 !0x3b 8E 39 00 A0 C9 69 72 3B ; !0x8e !0x39 !0 !0xa0 !0xc9 !0x69 !0x72 !0x3b
# :ELF_end [_start+0x1FF] # :ELF_end [_start+0x1EF]