![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to objc-msg-hppa.s CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Apple Darwin 0.x] / objc|
Return to objc-msg-hppa.s CVS log | Up to [Apple Darwin 0.x] / objc |
Darwin 0.1 In-kernel Objective-C runtime
/*
* Copyright (c) 1999 Apple Computer, Inc. All rights reserved.
*
* @APPLE_LICENSE_HEADER_START@
*
* "Portions Copyright (c) 1999 Apple Computer, Inc. All Rights
* Reserved. This file contains Original Code and/or Modifications of
* Original Code as defined in and that are subject to the Apple Public
* Source License Version 1.0 (the 'License'). You may not use this file
* except in compliance with the License. Please obtain a copy of the
* License at http://www.apple.com/publicsource and read it before using
* this file.
*
* The Original Code and all software distributed under the License are
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
* License for the specific language governing rights and limitations
* under the License."
*
* @APPLE_LICENSE_HEADER_END@
*/
#ifdef KERNEL
#define OBJC_LOCK_ROUTINE _simple_lock
#else
#ifdef DYLIB
#define OBJC_LOCK_ROUTINE _spin_lock$non_lazy_ptr
#else
#define OBJC_LOCK_ROUTINE _spin_lock
#endif
#endif /* KERNEL */
#define isa 0
#define cache 32
#define mask 0
#define buckets 8
#define method_name 0
#define method_imp 4
;; Optimized specifically for HP7100 (Gecko architecture)
;; It is assumed, that the cache line containing __objc_multithread_mask
;; is not missing. This version has only two (2) back-to-back insn
;; output dependencies, of which one is in the inner loop at 2:.
;; This version has also no read insns with output dependency on the
;; following insn. It is optimized with regard to branch prediction for
;; hit in the first probe (average probe length assumed to be .45), that
;; is why the final branch is moved before _objc_msgSend.
;; non-pic: 14 clocks best case, 4 clock / probe
;; (4miss + 1hit) + (2hit) / probe
;; pic: 16 clocks best case, 4 clocks / probe
;; (4miss + 2hit) + (2hit) / probe
.text
.align 2
.globl _objc_msgSend
LX0: bv,n 0(%r1) ; goto *imp; (nullify delay)
_objc_msgSend:
comib,=,n 0,%r26, Lnull ; <?not taken?>
#ifdef DYLIB
bl LX1,%r21
ldw isa(0,%r26),%r19 ; class = self->isa;
LX1:
depi 0,31,2,%r21
addil L`__objc_multithread_mask$non_lazy_ptr-LX1,%r21
ldw cache(0,%r19),%r20 ; cache = class->cache
ldw R`__objc_multithread_mask$non_lazy_ptr-LX1(%r1),%r22
ldw mask(0,%r20),%r21 ; mask = cache->mask
ldw 0(%r22),%r22 ; <indirect non lazy pointer>
ldo buckets(%r20),%r20 ; buckets = cache->buckets
comib,=,n 0,%r22, SendLocking ; <?not taken?>
#else
ldil L`__objc_multithread_mask,%r1
ldw isa(0,%r26),%r19 ; class = self->isa;
ldw R`__objc_multithread_mask(%r1),%r22
ldw cache(0,%r19),%r20 ; cache = class->cache
comib,=,n 0,%r22, SendLocking ; <?not taken?>
ldw mask(0,%r20),%r21 ; mask = cache->mask
ldo buckets(%r20),%r20 ; buckets = cache->buckets
#endif
and %r21,%r25,%r22 ; index = selector & mask;
2: sh3add %r22,%r20,%r1 ; bucket = (index * 4) + buckets
ldw method_name(0,%r1),%r19 ; op = bucket->selector_name
ldo 1(%r22),%r22 ; index += 1
comb,= %r25,%r19,LX0 ; if (selector == op) goto Lexit1 <?taken?>
ldw method_imp(0,%r1),%r1 ; <delay slot> imp = bucket->selector_imp
comib,<> 0,%r19,2b ; if (op != 0) goto 2 <?taken?>
and %r22,%r21,%r22 ; <delay slot> index &= mask
b,n L2 ; else goto cacheMiss
L2:
; We have to save all the register based arguments (including floating
; point) before calling _class_lookupMethodAndLoadCache. This is because
; we do not know what arguments were passed to us, and the arguments are
; not guaranteed to be saved across procedure calls (they are all caller-saved)
; We also have to save the return address (since we did not save it on entry).
copy %r30,%r19
ldo 128(%r30),%r30 ; Allocate space on stack
stwm %r2,4(0,%r19) ; Save return pointer
stwm %r23,4(0,%r19) ; Save old args
stwm %r24,4(0,%r19) ;
stwm %r25,4(0,%r19) ;
stwm %r26,4(0,%r19) ;
#ifndef KERNEL
fstds,mb %fr4,4(0,%r19) ; Save floating point args
fstds,mb %fr5,8(0,%r19) ; mb (modify before) is used instead
fstds,mb %fr6,8(0,%r19) ; of ma (as is implicit in above
fstds,mb %fr7,8(0,%r19) ; stores) with an initial value of 4
; so that doubles are aligned
; to 8 byte boundaries.
; Arg 1 (selector) is the same
#endif /* KERNEL */
stw %r28,8(0,%r19) ; save return struct ptr
jbsr __class_lookupMethodAndLoadCache,%r2,0f
ldw isa(0,%r26),%r26 ; <delay slot> arg 0 = self->isa
ldo -128(%r30),%r30 ; deallocate
copy %r30,%r19 ;
ldwm 4(0,%r19),%r2 ; restore everything
ldwm 4(0,%r19),%r23 ;
ldwm 4(0,%r19),%r24 ;
ldwm 4(0,%r19),%r25 ;
ldwm 4(0,%r19),%r26 ;
#ifndef KERNEL
fldds,mb 4(0,%r19),%fr4 ; see comment above about alignment
fldds,mb 8(0,%r19),%fr5 ;
fldds,mb 8(0,%r19),%fr6 ;
fldds,mb 8(0,%r19),%fr7 ;
#endif /* KERNEL */
ldw 8(0,%r19),%r20 ; get ret structure ptr
copy %r28,%r19
copy %r20,%r28 ; restore ret structure ptr
bv,n 0(%r19) ; goto *imp (nullify delay)
; stub for far call
.align 2
#ifdef DYLIB
0: bl LX5,%r1
nop
LX5: depi 0,31,2,%r1
addil L`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX5,%r1
ldw R`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX5(%r1),%r1
be,n 0(4,%r1)
#else
0: ldil L`__class_lookupMethodAndLoadCache,%r1
be,n R`__class_lookupMethodAndLoadCache(4,%r1)
#endif
.align 2
Lnull:
bv 0(%r2) ; return null
copy 0,%r28 ; <delay slot> return val = 0
; Locking version of objc_msgSend
; uses spin_lock() to lock the mutex.
.align 2
SendLocking:
copy %r30,%r19
ldo 128(%r30),%r30 ; Allocate space on stack
stwm %r2,4(0,%r19) ; Save return pointer
stwm %r23,4(0,%r19) ; Save old args
stwm %r24,4(0,%r19) ;
stwm %r25,4(0,%r19) ;
stwm %r26,4(0,%r19) ;
stwm %r28,4(0,%r19) ; save return struct ptr
#ifndef KERNEL
fstds,ma %fr4,8(0,%r19) ; Save floating point args
fstds,ma %fr5,8(0,%r19) ;
fstds,ma %fr6,8(0,%r19) ;
fstds,ma %fr7,8(0,%r19) ;
#endif /* KERNEL */
#ifdef DYLIB
0: bl LX6,%r24
nop
LX6: depi 0,31,2,%r24
addil L`_messageLock$non_lazy_ptr-LX6,%r24
ldw R`_messageLock$non_lazy_ptr-LX6(%r1),%r26
addil L`OBJC_LOCK_ROUTINE -LX6,%r24 ; call spin_lock() with _messageLock
ldw R`OBJC_LOCK_ROUTINE -LX6(%r1),%r24
ble 0(%sr4,%r24)
#else
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r26
ldil L`OBJC_LOCK_ROUTINE,%r1 ; call spin_lock() with _messageLock
ble R`OBJC_LOCK_ROUTINE(%sr4,%r1)
#endif
copy %r31,%r2
ldw -112(%r30),%r26 ; restore arg0
ldw -108(%r30),%r28 ; and ret0
ldw isa(0,%r26),%r19 ; class = self->isa;
ldw cache(0,%r19),%r20 ; cache = class->cache
ldw mask(0,%r20),%r21 ; mask = cache->mask
ldo buckets(%r20),%r20 ; buckets = cache->buckets
and %r21,%r25,%r22 ; index = selector & mask;
LL1: sh3add %r22,%r20,%r1 ; bucket = (index * 4) + buckets
ldw method_name(0,%r1),%r19 ; op = bucket->selector_name
ldo 1(%r22),%r22 ; index += 1
comb,=,n %r25,%r19,LLhit1 ; if (selector == op) goto LLhit1
comib,<> 0,%r19,LL1 ; if (op != 0) goto LL1
and %r22,%r21,%r22 ; <delay slot> index &= mask
b,n LL2 ; else goto cacheMiss
LLhit1:
ldw method_imp(0,%r1),%r19 ; imp = bucket->selector_imp
#if KERNEL
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r20
addi 0xc,%r20,%r20
depi 0,31,4,%r20
zdepi 1,31,1,%r1
stw %r1,0(0,%r20)
#else
#ifdef DYLIB
bl LX7,%r21
nop
LX7: depi 0,31,2,%r21
addil L`_messageLock$non_lazy_ptr-LX7,%r21
ldw R`_messageLock$non_lazy_ptr-LX7(%r1),%r21
stw %r0,0(%r21) ; unlock the lock
#else
ldil L`_messageLock,%r1
stw %r0,R`_messageLock(%r1) ; unlock the lock
#endif
#endif
ldwm -128(%r30),%r2 ; restore original rp and deallocate
bv,n 0(%r19) ; goto *imp; (nullify delay)
LL2:
jbsr __class_lookupMethodAndLoadCache,%r2,0f
ldw isa(0,%r26),%r26 ; <delay slot> arg 0 = self->isa
ldo -128(%r30),%r30 ; deallocate
copy %r30,%r19 ;
ldwm 4(0,%r19),%r2 ; restore everything
ldwm 4(0,%r19),%r23 ;
ldwm 4(0,%r19),%r24 ;
ldwm 4(0,%r19),%r25 ;
ldwm 4(0,%r19),%r26 ;
ldwm 4(0,%r19),%r20 ; get ret structure ptr
#ifndef KERNEL
fldds,ma 8(0,%r19),%fr4 ;
fldds,ma 8(0,%r19),%fr5 ;
fldds,ma 8(0,%r19),%fr6 ;
fldds,ma 8(0,%r19),%fr7 ;
#endif /* KERNEL */
copy %r28,%r19
copy %r20,%r28 ; restore ret structure ptr
#if KERNEL
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r20
addi 0xc,%r20,%r20
depi 0,31,4,%r20
zdepi 1,31,1,%r1
stw %r1,0(0,%r20)
#else
#ifdef DYLIB
bl LX8,%r19
nop
LX8: depi 0,31,2,%r19
addil L`_messageLock$non_lazy_ptr-LX8,%r19
ldw _messageLock$non_lazy_ptr-LX8(%r1),%r19
stw %r0,0(%r19) ; unlock the lock
#else
ldil L`_messageLock,%r1
stw %r0,R`_messageLock(%r1) ; unlock the lock
#endif
#endif
bv,n 0(%r19) ; goto *imp (nullify delay)
; stub for far call
.align 2
#ifdef DYLIB
0: bl LX2,%r1
nop
LX2: depi 0,31,2,%r1
addil L`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX2,%r1
ldw R`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX2(%r1),%r1
be,n 0(4,%r1)
#else
0: ldil L`__class_lookupMethodAndLoadCache,%r1
be,n R`__class_lookupMethodAndLoadCache(4,%r1)
#endif
#define receiver 0
#define class 4
.globl _objc_msgSendSuper
.align 2
_objc_msgSendSuper:
ldil L`__objc_multithread_mask,%r1
ldw R`__objc_multithread_mask(%r1),%r19
combt,= %r0,%r19,SuperLocking ;
ldw class(0,%r26),%r19 ; class = caller->class;
ldw cache(0,%r19),%r20 ; cache = class->cache
ldw mask(0,%r20),%r21 ; mask = cache->mask
ldo buckets(%r20),%r20 ; buckets = cache->buckets
and %r21,%r25,%r22 ; index = selector & mask;
LS1: sh3add %r22,%r20,%r1 ; bucket = (index * 4) + buckets
ldw method_name(0,%r1),%r19 ; op = bucket->selector_name
ldo 1(%r22),%r22 ; index += 1
comb,= %r25,%r19,LShit1 ; if (selector == op) goto LShit1
ldw method_imp(0,%r1),%r1 ; <delay slot> imp = bucket->selector_imp
comib,<> 0,%r19,LS1 ; if (op != 0) goto LS1
and %r22,%r21,%r22 ; <delay slot> index &= mask
b,n LS2 ; else goto cacheMiss
LShit1:
bv 0(%r1) ; goto *imp; (nullify delay)
ldw receiver(0,%r26),%r26 ; <delay slot> self = caller->receiver;
;
LS2: ;
copy %r30,%r19
ldo 128(%r30),%r30 ; Allocate space on stack
stwm %r2,4(0,%r19) ; Save return pointer
stwm %r23,4(0,%r19) ; Save old args
stwm %r24,4(0,%r19) ;
stwm %r25,4(0,%r19) ;
stwm %r26,4(0,%r19) ;
#ifndef KERNEL
fstds,mb %fr4,4(0,%r19) ; Save floating point args
fstds,mb %fr5,8(0,%r19) ; mb (modify before) is used instead
fstds,mb %fr6,8(0,%r19) ; of ma (as is implicit in above
fstds,mb %fr7,8(0,%r19) ; stores) with an initial value of 4
; so that doubles are aligned
; to 8 byte boundaries.
; Arg 1 (selector) is the same
#endif /* KERNEL */
stw %r28,8(0,%r19) ; save return struct ptr
jbsr __class_lookupMethodAndLoadCache,%r2,0f
ldw class(0,%r26),%r26 ; <delay slot> arg 0 = caller->class;
ldo -128(%r30),%r30 ; deallocate
copy %r30,%r19 ;
ldwm 4(0,%r19),%r2 ; restore everything
ldwm 4(0,%r19),%r23 ;
ldwm 4(0,%r19),%r24 ;
ldwm 4(0,%r19),%r25 ;
ldwm 4(0,%r19),%r26 ;
#ifndef KERNEL
fldds,mb 4(0,%r19),%fr4 ; see comment above about alignment
fldds,mb 8(0,%r19),%fr5 ;
fldds,mb 8(0,%r19),%fr6 ;
fldds,mb 8(0,%r19),%fr7 ;
#endif /* KERNEL */
ldw 8(0,%r19),%r20 ; get ret structure ptr
ldw receiver(0,%r26),%r26 ; self = caller->receiver;
copy %r28,%r19
copy %r20,%r28
bv,n 0(%r19) ; goto *imp (nullify delay)
; stub for far call
.align 2
#ifdef DYLIB
0: bl LX3,%r1
nop
LX3: depi 0,31,2,%r1
addil L`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX3,%r1
ldw R`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX3(%r1),%r1
be,n 0(4,%r1)
#else
0: ldil L`__class_lookupMethodAndLoadCache,%r1
be,n R`__class_lookupMethodAndLoadCache(4,%r1)
#endif
; locking version of objc_msgSendSuper
; uses spin_lock() to lock the lock.
.align 2
SuperLocking:
copy %r30,%r19
ldo 128(%r30),%r30 ; Allocate space on stack
stwm %r2,4(0,%r19) ; Save return pointer
stwm %r23,4(0,%r19) ; Save old args
stwm %r24,4(0,%r19) ;
stwm %r25,4(0,%r19) ;
stwm %r26,4(0,%r19) ;
stwm %r28,4(0,%r19) ; save return struct ptr
#ifndef KERNEL
fstds,ma %fr4,8(0,%r19) ; Save floating point args
fstds,ma %fr5,8(0,%r19) ;
fstds,ma %fr6,8(0,%r19) ;
fstds,ma %fr7,8(0,%r19) ;
#endif /* KERNEL */
#ifdef DYLIB
0: bl LX9,%r24
nop
LX9: depi 0,31,2,%r24
addil L`_messageLock$non_lazy_ptr-LX9,%r24
ldw R`_messageLock$non_lazy_ptr-LX9(%r1),%r26
addil L`OBJC_LOCK_ROUTINE-LX9,%r24 ; call spin_lock() with _messageLock
ldw R`OBJC_LOCK_ROUTINE-LX9(%r1),%r24
ble 0(%sr4,%r24)
#else
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r26
ldil L`OBJC_LOCK_ROUTINE,%r1 ; call spin_lock() with _messageLock
ble R`OBJC_LOCK_ROUTINE(%sr4,%r1)
#endif
copy %r31,%r2
ldw -112(%r30),%r26 ; restore arg0
ldw -108(%r30),%r28 ; and ret0 (spin_lock doesnt
; touch anything else)
ldw class(0,%r26),%r19 ; class = caller->class;
ldw cache(0,%r19),%r20 ; cache = class->cache
ldw mask(0,%r20),%r21 ; mask = cache->mask
ldo buckets(%r20),%r20 ; buckets = cache->buckets
and %r21,%r25,%r22 ; index = selector & mask;
LLS1: sh3add %r22,%r20,%r1 ; bucket = (index * 4) + buckets
ldw method_name(0,%r1),%r19 ; op = bucket->selector_name
ldo 1(%r22),%r22 ; index += 1
comb,=,n %r25,%r19,LLShit1 ; if (selector == op) goto LShit1
comib,<> 0,%r19,LLS1 ; if (op != 0) goto LS1
and %r22,%r21,%r22 ; <delay slot> index &= mask
b,n LLS2 ; else goto cacheMiss
LLShit1:
ldw method_imp(0,%r1),%r19 ; imp = bucket->selector_imp
ldw receiver(0,%r26),%r26 ; self = caller->receiver;
#if KERNEL
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r20
addi 0xc,%r20,%r20
depi 0,31,4,%r20
zdepi 1,31,1,%r1
stw %r1,0(0,%r20)
#else
#ifdef DYLIB
bl LX10,%r19
nop
LX10: depi 0,31,2,%r19
addil L`_messageLock$non_lazy_ptr-LX10,%r19
ldw R`_messageLock$non_lazy_ptr-LX10(%r1),%r19
stw %r0,0(%r19) ; unlock the lock
#else
ldil L`_messageLock,%r1
stw %r0,R`_messageLock(%r1) ; unlock the lock
#endif
#endif
ldwm -128(%r30),%r2 ; restore original rp and deallocate
bv,n 0(%r19) ; goto *imp; (nullify delay)
#ifdef MONINIT
.space 128 ; /* area for moninitobjc to write */
#endif
;
LLS2: ;
jbsr __class_lookupMethodAndLoadCache,%r2,0f
ldw class(0,%r26),%r26 ; <delay slot> arg 0 = caller->class;
ldo -128(%r30),%r30 ; deallocate
copy %r30,%r19 ;
ldwm 4(0,%r19),%r2 ; restore everything
ldwm 4(0,%r19),%r23 ;
ldwm 4(0,%r19),%r24 ;
ldwm 4(0,%r19),%r25 ;
ldwm 4(0,%r19),%r26 ;
ldwm 4(0,%r19),%r20 ; get ret structure ptr
#ifndef KERNEL
fldds,ma 8(0,%r19),%fr4 ;
fldds,ma 8(0,%r19),%fr5 ;
fldds,ma 8(0,%r19),%fr6 ;
fldds,ma 8(0,%r19),%fr7 ;
#endif /* KERNEL */
ldw receiver(0,%r26),%r26 ; self = caller->receiver;
copy %r28,%r19
copy %r20,%r28
#if KERNEL
ldil L`_messageLock,%r1
ldo R`_messageLock(%r1),%r20
addi 0xc,%r20,%r20
depi 0,31,4,%r20
zdepi 1,31,1,%r1
stw %r1,0(0,%r20)
#else
ldil L`_messageLock,%r1
stw %r0,R`_messageLock(%r1) ; unlock the lock
#endif
bv,n 0(%r19) ; goto *imp (nullify delay)
; stub for far call
.align 2
#ifdef DYLIB
0: bl LX4,%r1
nop
LX4: depi 0,31,2,%r1
addil L`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX4,%r1
ldw R`__class_lookupMethodAndLoadCache$non_lazy_ptr-LX4(%r1),%r1
be,n 0(4,%r1)
#else
0: ldil L`__class_lookupMethodAndLoadCache,%r1
be,n R`__class_lookupMethodAndLoadCache(4,%r1)
#endif
.objc_meth_var_names
L30: .ascii "forward::\0"
.objc_message_refs
.align 2
L31: .long L30
.cstring
L32: .ascii "Does not recognize selector %s\0"
.text
.align 2
;
; NOTE: Because the stack grows from low mem to high mem on this machine
; and the args go the other way, the marg_list pointer is to the first argument
; and subsequent arguments are at NEGATIVE offsets from the marg_list.
; This means that marg_getValue() and related macros will have to be adjusted
; appropriately.
;
.globl __objc_msgForward
__objc_msgForward:
stw %r2,-20(0,%r30) ; save rp
ldo 64(%r30),%r30 ; create frame area (no locals needed)
ldil L`L31,%r1
ldo R`L31(%r1),%r19
ldw 0(0,%r19),%r19
combt,=,n %r19, %r25,L34 ; if (sel==@selector(forward::))
ldo -112(%r30),%r20 ; ptr to arg3 homing area
stwm %r23,4(0,%r20) ; Mirror registers onto stack
stwm %r24,4(0,%r20) ;
stwm %r25,4(0,%r20) ;
stwm %r26,4(0,%r20) ;
copy %r25,%r24
copy %r19,%r25 ; [self forward:sel :marg_list]
bl _objc_msgSend,%r2
copy %r20,%r23 ; <delay slot> copy original sel
ldo -64(%r30),%r30 ; deallocate
ldw -20(0,%r30),%r2 ; restore rp
bv,n 0(%r2) ; return
L34:
ldil L`L32,%r1
ldo R`L32(%r1),%r25
ldil L`__objc_error,%r1
be R`__objc_error(%sr4,%r1) ; __objc_error never returns, so no
copy %r19,%r24 ; need to clean up.
; Algorithm is as follows:
; . Calculate how much stack size is needed for any arguments not in the
; general registers and allocate space on stack.
; . Restore general argument regs from the bottom of the marg_list.
; . Restore fp argument regs from the same area.
; (The first two args in the marg list are always old obj and old SEL.)
; . Copy any remaining args from the marg_list to the new frame
; . Call the new method.
.align 2
.globl _objc_msgSendv
_objc_msgSendv:
; objc_msgSendv(self, sel, size, margs)
stw %r2,-20(0,%r30) ;
copy %r4, %r1 ; stanard prologue
copy %r30,%r4 ;
stw %r1,0(0,%r4) ;
ldo 99(%r24),%r19 ; Calculate frame size, rounded
depi 0,31,6,%r19 ; up to 64 byte boundary...
add %r19,%r30,%r30 ; Allocate frame area (no locals)
copy %r24,%r20 ; r20 now holds arg size
ldo -16(%r23),%r21 ; r21 now holds marg_list+16
ldws 0(0,%r21),%r23 ; Get old general register args (dont
ldws 4(0,%r21),%r24 ; need first two: always self & SEL)
#ifndef KERNEL
fldds 0(0,%r21),%fr7 ; Mirror to fp regs
fldws 4(0,%r21),%fr6 ;
#endif /* KERNEL */
ldo -52(%r30),%r22 ; newly allocated stack area.
ldo -16(%r20),%r20 ; Size -= 16
comibf,<,n 0,%r20,L36
L35: ldws,mb -4(0,%r21),%r19 ; while(size>0)
addibf,<= -4,%r20,L35 ; { *(dest--) = *(src--); size-=4; }
stws,ma %r19,-4(0,%r22) ; <delay slot>
L36: bl _objc_msgSend,%r2
nop
copy %r4,%r30 ; deallocate
ldw -20(0,%r30), %r2
bv 0(%r2)
ldw 0(0,%r30), %r4
#ifdef DYLIB
.non_lazy_symbol_pointer
_messageLock$non_lazy_ptr:
.indirect_symbol _messageLock
.long 0
_spin_lock$non_lazy_ptr:
.indirect_symbol _spin_lock
.long 0
__objc_multithread_mask$non_lazy_ptr:
.indirect_symbol __objc_multithread_mask
.long 0
__class_lookupMethodAndLoadCache$non_lazy_ptr:
.indirect_symbol __class_lookupMethodAndLoadCache
.long 0
#endif
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.