singrdk/base/Kernel/Native/ix64/_memcpy.asm

614 lines
22 KiB
NASM
Raw Permalink Normal View History

2008-11-17 18:29:00 -05:00
;*******************************************************************************
;memcpy.asm - contains memcpy and memmove routines
;
; Copyright (c) Microsoft Corporation. All rights reserved.
;
;Purpose:
; memcpy() copies a source memory buffer to a destination buffer.
; Overlapping buffers are not treated specially, so propogation may occur.
; memmove() copies a source memory buffer to a destination buffer.
; Overlapping buffers are treated specially, to avoid propogation.
;
;*******************************************************************************
.code
include hal.inc
;***
;memcpy - Copy source buffer to destination buffer
;
;Purpose:
; memcpy() copies a source memory buffer to a destination memory buffer.
; This routine does NOT recognize overlapping buffers, and thus can lead
; to propogation.
; For cases where propogation must be avoided, memmove() must be used.
;
; Algorithm:
;
; void * memcpy(void * dst, void * src, size_t count)
; {
; void * ret = dst;
;
; /*
; * copy from lower addresses to higher addresses
; */
; while (count--)
; *dst++ = *src++;
;
; return(ret);
; }
;
;memmove - Copy source buffer to destination buffer
;
;Purpose:
; memmove() copies a source memory buffer to a destination memory buffer.
; This routine recognize overlapping buffers to avoid propogation.
; For cases where propogation is not a problem, memcpy() can be used.
;
; Algorithm:
;
; void * memmove(void * dst, void * src, size_t count)
; {
; void * ret = dst;
;
; if (dst <= src || dst >= (src + count)) {
; /*
; * Non-Overlapping Buffers
; * copy from lower addresses to higher addresses
; */
; while (count--)
; *dst++ = *src++;
; }
; else {
; /*
; * Overlapping Buffers
; * copy from higher addresses to lower addresses
; */
; dst += count - 1;
; src += count - 1;
;
; while (count--)
; *dst-- = *src--;
; }
;
; return(ret);
; }
;
;
;Entry:
; void *dst = pointer to destination buffer
; const void *src = pointer to source buffer
; size_t count = number of bytes to copy
;
;Exit:
; Returns a pointer to the destination buffer in AX/DX:AX
;
;Uses:
; CX, DX
;
;Exceptions:
;*******************************************************************************
% public memcpy
;; memcopy (dst,src,count)
memcpy proc ;;frame
;;PrologPush rbp ; create ebp chain entry
;;SetFramePointer rbp ; set new ebp
;;.endprolog
;;spin: jmp spin
;;mov [esp+8], rcx ; spill dest
push rdi ;U - save rdi
push rsi ;V - save rsi
mov r9,rcx ; save off dst in scratch reg
mov rsi,rdx ;U - rsi = source
mov rdi,rcx ;U - rdi = dest
mov rcx,r8 ;V - rdx = number of bytes to move
cmp r8,2h
jnz hack
;spin: jmp spin
hack:
;
; Check for overlapping buffers:
; If (dst <= src) Or (dst >= src + Count) Then
; Do normal (Upwards) Copy
; Else
; Do Downwards Copy to avoid propagation
;
mov rax,r8 ;V - rax = byte count...
mov rdx,r8 ;U - rdx = byte count...
add rax,rsi ;V - rax = point past source end
cmp rdi,rsi ;U - dst <= src ?
jbe CopyUp ;V - yes, copy toward higher addresses
cmp rdi,rax ;U - dst < (src + count) ?
jb CopyDown ;V - yes, copy toward lower addresses
;
; Copy toward higher addresses.
;
;
; The algorithm for forward moves is to align the destination to a dword
; boundary and so we can move dwords with an aligned destination. This
; occurs in 3 steps.
;
; - move x = ((4 - Dest & 3) & 3) bytes
; - move y = ((L-x) >> 2) dwords
; - move (L - x - y*4) bytes
;
CopyUp:
test rdi,11b ;U - destination dword aligned?
jnz short CopyLeadUp ;V - if we are not dword aligned already, align
shr rcx,2 ;U - shift down to dword count
and rdx,11b ;V - trailing byte count
cmp rcx,8 ;U - test if small enough for unwind copy
jb short CopyUnwindUp ;V - if so, then jump
rep movsd ;N - move all of our dwords
lea r10, TrailUpVec
jmp [rdx*8 + r10] ;N - process trailing bytes
;
; Code to do optimal memory copies for non-dword-aligned destinations.
;
; The following length check is done for two reasons:
;
; 1. to ensure that the actual move length is greater than any possiale
; alignment move, and
;
; 2. to skip the multiple move logic for small moves where it would
; be faster to move the bytes with one instruction.
;
align 8
CopyLeadUp:
mov rax,rdi ;U - get destination offset
mov rdx,11b ;V - prepare for mask
sub rcx,4 ;U - check for really short string - sub for adjust
jb ByteCopyUp ;V - branch to just copy bytes
and rax,11b ;U - get offset within first dword
add rcx,rax ;V - update size after leading bytes copied
lea r10,LeadUpVec
jmp [rax*8-8 + r10] ;N - process leading bytes
align 8
ByteCopyUp:
lea r10, TrailUpVec
jmp [rcx*8+32 + r10] ;N - process just bytes
align 8
CopyUnwindUp:
lea r10,UnwindUpVec
jmp [rcx*8 + r10] ;N - unwind dword copy
align 8
LeadUpVec dq LeadUp1, LeadUp2, LeadUp3
align 8
LeadUp1:
and rdx,rcx ;U - trailing byte count
mov al,[rsi] ;V - get first byte from source
mov [rdi],al ;U - write second byte to destination
mov al,[rsi+1] ;V - get second byte from source
mov [rdi+1],al ;U - write second byte to destination
mov al,[rsi+2] ;V - get third byte from source
shr rcx,2 ;U - shift down to dword count
mov [rdi+2],al ;V - write third byte to destination
add rsi,3 ;U - advance source pointer
add rdi,3 ;V - advance destination pointer
cmp rcx,8 ;U - test if small enough for unwind copy
jb short CopyUnwindUp ;V - if so, then jump
rep movsd ;N - move all of our dwords
;;jmp TrailUpVec[rdx*4] ;N - process trailing bytes
lea r10, TrailUpVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
LeadUp2:
and rdx,rcx ;U - trailing byte count
mov al,[rsi] ;V - get first byte from source
mov [rdi],al ;U - write second byte to destination
mov al,[rsi+1] ;V - get second byte from source
shr rcx,2 ;U - shift down to dword count
mov [rdi+1],al ;V - write second byte to destination
add rsi,2 ;U - advance source pointer
add rdi,2 ;V - advance destination pointer
cmp rcx,8 ;U - test if small enough for unwind copy
jb CopyUnwindUp ;V - if so, then jump
rep movsd ;N - move all of our dwords
;jmp qword ptr TrailUpVec[rdx*4] ;N - process trailing bytes
lea r10, TrailUpVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
LeadUp3:
and rdx,rcx ;U - trailing byte count
mov al,[rsi] ;V - get first byte from source
mov [rdi],al ;U - write second byte to destination
add rsi,1 ;V - advance source pointer
shr rcx,2 ;U - shift down to dword count
add rdi,1 ;V - advance destination pointer
cmp rcx,8 ;U - test if small enough for unwind copy
jb CopyUnwindUp ;V - if so, then jump
rep movsd ;N - move all of our dwords
;jmp qword ptr TrailUpVec[rdx*4] ;N - process trailing bytes
lea r10, TrailUpVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
UnwindUpVec dq UnwindUp0, UnwindUp1, UnwindUp2, UnwindUp3
dq UnwindUp4, UnwindUp5, UnwindUp6, UnwindUp7
UnwindUp7:
mov eax,[rsi+rcx*4-28] ;U - get dword from source
;V - spare
mov [rdi+rcx*4-28],eax ;U - put dword into destination
UnwindUp6:
mov eax,[rsi+rcx*4-24] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-24],eax ;U - put dword into destination
UnwindUp5:
mov eax,[rsi+rcx*4-20] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-20],eax ;U - put dword into destination
UnwindUp4:
mov eax,[rsi+rcx*4-16] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-16],eax ;U - put dword into destination
UnwindUp3:
mov eax,[rsi+rcx*4-12] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-12],eax ;U - put dword into destination
UnwindUp2:
mov eax,[rsi+rcx*4-8] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-8],eax ;U - put dword into destination
UnwindUp1:
mov eax,[rsi+rcx*4-4] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4-4],eax ;U - put dword into destination
lea rax,[rcx*4] ;V - compute update for pointer
add rsi,rax ;U - update source pointer
add rdi,rax ;V - update destination pointer
UnwindUp0:
;jmp qword ptr TrailUpVec[rdx*4] ;N - process trailing bytes
lea r10, TrailUpVec
jmp [rdx*8 + r10] ;N - process trailing bytes
;-----------------------------------------------------------------------------
align 8
TrailUpVec dq TrailUp0, TrailUp1, TrailUp2, TrailUp3
align 8
TrailUp0:
mov rax, r9 ;U - return pointer to destination
pop rsi ;V - restore rsi
pop rdi ;U - restore rdi
;V - spare
ret
align 8
TrailUp1:
mov al,[rsi] ;U - get byte from source
;V - spare
mov [rdi],al ;U - put byte in destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
align 8
TrailUp2:
mov al,[rsi] ;U - get first byte from source
;V - spare
mov [rdi],al ;U - put first byte into destination
mov al,[rsi+1] ;V - get second byte from source
mov [rdi+1],al ;U - put second byte into destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
align 8
TrailUp3:
mov al,[rsi] ;U - get first byte from source
;V - spare
mov [rdi],al ;U - put first byte into destination
mov al,[rsi+1] ;V - get second byte from source
mov [rdi+1],al ;U - put second byte into destination
mov al,[rsi+2] ;V - get third byte from source
mov [rdi+2],al ;U - put third byte into destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
;-----------------------------------------------------------------------------
;-----------------------------------------------------------------------------
;-----------------------------------------------------------------------------
;
; Copy down to avoid propogation in overlapping buffers.
;
align 8
CopyDown:
lea rsi,[rsi+rcx-4] ;U - point to 4 bytes before src buffer end
lea rdi,[rdi+rcx-4] ;V - point to 4 bytes before dest buffer end
;
; See if the destination start is dword aligned
;
test rdi,11b ;U - test if dword aligned
jnz short CopyLeadDown ;V - if not, jump
shr rcx,2 ;U - shift down to dword count
and rdx,11b ;V - trailing byte count
cmp rcx,8 ;U - test if small enough for unwind copy
jb short CopyUnwindDown ;V - if so, then jump
std ;N - set direction flag
rep movsd ;N - move all of our dwords
cld ;N - clear direction flag back
;jmp qword ptr TrailDownVec[rdx*4] ;N - process trailing bytes
lea r10, TrailDownVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
CopyUnwindDown:
neg rcx ;U - negate dword count for table merging
;V - spare
;jmp qword ptr UnwindDownVec[rcx*4+28] ;N - unwind copy
lea r10,UnwindDownVec
jmp [rcx*8+56 + r10] ;N - unwind copy
align 8
CopyLeadDown:
mov rax,rdi ;U - get destination offset
mov rdx,11b ;V - prepare for mask
cmp rcx,4 ;U - check for really short string
jb short ByteCopyDown ;V - branch to just copy bytes
and rax,11b ;U - get offset within first dword
sub rcx,rax ;U - to update size after lead copied
;jmp qword ptr LeadDownVec[rax*8-8] ;N - process leading bytes
lea r10,LeadDownVec
jmp [rax*8-8 + r10] ;N - process leading bytes
align 8
ByteCopyDown:
;jmp qword ptr TrailDownVec[rcx*8] ;N - process just bytes
lea r10, TrailDownVec
jmp [rcx*8 + r10] ;N - process just bytes
align 8
LeadDownVec dq LeadDown1, LeadDown2, LeadDown3
align 8
LeadDown1:
mov al,[rsi+3] ;U - load first byte
and rdx,rcx ;V - trailing byte count
mov [rdi+3],al ;U - write out first byte
sub rsi,1 ;V - point to last src dword
shr rcx,2 ;U - shift down to dword count
sub rdi,1 ;V - point to last dest dword
cmp rcx,8 ;U - test if small enough for unwind copy
jb CopyUnwindDown ;V - if so, then jump
std ;N - set direction flag
rep movsd ;N - move all of our dwords
cld ;N - clear direction flag
;jmp qword ptr TrailDownVec[rdx*4] ;N - process trailing bytes
lea r10, TrailDownVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
LeadDown2:
mov al,[rsi+3] ;U - load first byte
and rdx,rcx ;V - trailing byte count
mov [rdi+3],al ;U - write out first byte
mov al,[rsi+2] ;V - get second byte from source
shr rcx,2 ;U - shift down to dword count
mov [rdi+2],al ;V - write second byte to destination
sub rsi,2 ;U - point to last src dword
sub rdi,2 ;V - point to last dest dword
cmp rcx,8 ;U - test if small enough for unwind copy
jb CopyUnwindDown ;V - if so, then jump
std ;N - set direction flag
rep movsd ;N - move all of our dwords
cld ;N - clear direction flag
;jmp qword ptr TrailDownVec[rdx*4] ;N - process trailing bytes
lea r10, TrailDownVec
jmp [rdx*8 + r10] ;N - process trailing bytes
align 8
LeadDown3:
mov al,[rsi+3] ;U - load first byte
and rdx,rcx ;V - trailing byte count
mov [rdi+3],al ;U - write out first byte
mov al,[rsi+2] ;V - get second byte from source
mov [rdi+2],al ;U - write second byte to destination
mov al,[rsi+1] ;V - get third byte from source
shr rcx,2 ;U - shift down to dword count
mov [rdi+1],al ;V - write third byte to destination
sub rsi,3 ;U - point to last src dword
sub rdi,3 ;V - point to last dest dword
cmp rcx,8 ;U - test if small enough for unwind copy
jb CopyUnwindDown ;V - if so, then jump
std ;N - set direction flag
rep movsd ;N - move all of our dwords
cld ;N - clear direction flag
;jmp qword ptr TrailDownVec[rdx*4] ;N - process trailing bytes
lea r10, TrailDownVec
jmp [rdx*8 + r10] ;N - process trailing bytes
;------------------------------------------------------------------
align 8
UnwindDownVec dq UnwindDown7, UnwindDown6, UnwindDown5, UnwindDown4
dq UnwindDown3, UnwindDown2, UnwindDown1, UnwindDown0
UnwindDown7:
mov rax,[rsi+rcx*4+28] ;U - get dword from source
;V - spare
mov [rdi+rcx*4+28],rax ;U - put dword into destination
UnwindDown6:
mov rax,[rsi+rcx*4+24] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+24],rax ;U - put dword into destination
UnwindDown5:
mov rax,[rsi+rcx*4+20] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+20],rax ;U - put dword into destination
UnwindDown4:
mov rax,[rsi+rcx*4+16] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+16],rax ;U - put dword into destination
UnwindDown3:
mov rax,[rsi+rcx*4+12] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+12],rax ;U - put dword into destination
UnwindDown2:
mov rax,[rsi+rcx*4+8] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+8],rax ;U - put dword into destination
UnwindDown1:
mov rax,[rsi+rcx*4+4] ;U(entry)/V(not) - get dword from source
;V(entry) - spare
mov [rdi+rcx*4+4],rax ;U - put dword into destination
lea rax,[rcx*4] ;V - compute update for pointer
add rsi,rax ;U - update source pointer
add rdi,rax ;V - update destination pointer
UnwindDown0:
;jmp qword ptr TrailDownVec[rdx*4] ;N - process trailing bytes
lea r10, TrailDownVec
jmp [rdx*8 + r10] ;N - process trailing bytes
;-----------------------------------------------------------------------------
align 8
TrailDownVec dq TrailDown0, TrailDown1, TrailDown2, TrailDown3
align 8
TrailDown0:
mov rax,r9 ;U - return pointer to destination
;V - spare
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
align 8
TrailDown1:
mov al,[rsi+3] ;U - get byte from source
;V - spare
mov [rdi+3],al ;U - put byte in destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
align 8
TrailDown2:
mov al,[rsi+3] ;U - get first byte from source
;V - spare
mov [rdi+3],al ;U - put first byte into destination
mov al,[rsi+2] ;V - get second byte from source
mov [rdi+2],al ;U - put second byte into destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
align 8
TrailDown3:
mov al,[rsi+3] ;U - get first byte from source
;V - spare
mov [rdi+3],al ;U - put first byte into destination
mov al,[rsi+2] ;V - get second byte from source
mov [rdi+2],al ;U - put second byte into destination
mov al,[rsi+1] ;V - get third byte from source
mov [rdi+1],al ;U - put third byte into destination
mov rax,r9 ;V - return pointer to destination
pop rsi ;U - restore rsi
pop rdi ;V - restore rdi
ret
memcpy endp
memmove proc \
dst:ptr byte, \
src:ptr byte, \
count:DWORD
jmp memcpy
memmove endp
end