|
DataMuseum.dkPresents historical artifacts from the history of: DKUUG/EUUG Conference tapes |
This is an automatic "excavation" of a thematic subset of
See our Wiki for more about DKUUG/EUUG Conference tapes Excavated with: AutoArchaeologist - Free & Open Source Software. |
top - metrics - downloadIndex: T a
Length: 7412 (0x1cf4) Types: TextFile Names: »alloca.s«
└─⟦a0efdde77⟧ Bits:30001252 EUUGD11 Tape, 1987 Spring Conference Helsinki └─⟦this⟧ »EUUGD11/gnu-31mar87/emacs/src/alloca.s«
/* `alloca' standard 4.2 subroutine for 68000's and 16000's and pyramids. Also has _setjmp and _longjmp for pyramids. Copyright (C) 1985, 1986 Free Software Foundation, Inc. This file is part of GNU Emacs. GNU Emacs is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY. No author or distributor accepts responsibility to anyone for the consequences of using it or for whether it serves any particular purpose or works at all, unless he says so in writing. Refer to the GNU Emacs General Public License for full details. Everyone is granted permission to copy, modify and redistribute GNU Emacs, but only under the conditions described in the GNU Emacs General Public License. A copy of this license is supposed to have been given to you along with GNU Emacs so you can know your rights and responsibilities. It should be in a file named COPYING. Among other things, the copyright notice and this notice must be preserved on all copies. */ /* Both 68000 systems I have run this on have had broken versions of alloca. Also, I am told that non-berkeley systems do not have it at all. So replace whatever system-provided alloca there may be on all 68000 systems. */ #include "config.h" #ifndef HAVE_ALLOCA /* define this to use system's alloca */ #ifndef hp9000 #ifndef m68k #ifndef m68000 #ifndef WICAT #ifndef ns16000 #ifndef sequent #ifndef pyramid #ifndef ATT3B5 you lose!! #endif /* ATT3B5 */ #endif /* pyramid */ #endif /* sequent */ #endif /* ns16000 */ #endif /* WICAT */ #endif /* m68000 */ #endif /* m68k */ #endif /* hp9000 */ #ifdef hp9000 #ifdef OLD_HP_ASSEMBLER data text globl _alloca _alloca move.l (sp)+,a0 ; pop return addr from top of stack move.l (sp)+,d0 ; pop size in bytes from top of stack add.l #ROUND,d0 ; round size up to long word and.l #MASK,d0 ; mask out lower two bits of size sub.l d0,sp ; allocate by moving stack pointer tst.b PROBE(sp) ; stack probe to allocate pages move.l sp,d0 ; return pointer add.l #-4,sp ; new top of stack jmp (a0) ; not a normal return MASK equ -4 ; Longword alignment ROUND equ 3 ; ditto PROBE equ -128 ; safety buffer for C compiler scratch data #else /* new hp assembler syntax */ /* The new compiler does "move.m <registers> (%sp)" to save registers, so we must copy the saved registers when we mung the sp. The old compiler did "move.m <register> <offset>(%a6)", which gave us no trouble */ text set PROBE,-128 # safety for C frame temporaries set MAXREG,10 # d2-d7, a2-a5 may have been saved global _alloca _alloca: mov.l (%sp)+,%a0 # return addess mov.l (%sp)+,%d0 # number of bytes to allocate mov.l %sp,%a1 # save old sp for register copy mov.l %sp,%d1 # compute new sp sub.l %d0,%d1 # space requested and.l &-4,%d1 # round down to longword sub.l &MAXREG*4,%d1 # space for saving registers mov.l %d1,%sp # save new value of sp tst.b PROBE(%sp) # create pages (sigh) move.w &MAXREG-1,%d0 copy_regs_loop: # save caller's saved registers mov.l (%a1)+,(%sp)+ dbra %d0,copy_regs_loop mov.l %sp,%d0 # return value mov.l %d1,%sp add.l &-4,%sp # adjust tos jmp (%a0) # rts #endif /* new hp assembler */ #else #ifdef m68k /* SGS assembler totally different */ file "alloca.s" global alloca alloca: mov.l (%sp)+,%a1 # pop return addr from top of stack mov.l (%sp)+,%d0 # pop size in bytes from top of stack add.l &R%1,%d0 # round size up to long word and.l &-4,%d0 # mask out lower two bits of size sub.l %d0,%sp # allocate by moving stack pointer tst.b P%1(%sp) # stack probe to allocate pages mov.l %sp,%a0 # return pointer as pointer mov.l %sp,%d0 # return pointer as int to avoid disaster add.l &-4,%sp # new top of stack jmp (%a1) # not a normal return set S%1,64 # safety factor for C compiler scratch set R%1,3+S%1 # add to size for rounding set P%1,-132 # probe this far below current top of stack #else /* not m68k */ #ifdef m68000 #ifdef WICAT /* * Registers are saved after the corresponding link so we have to explicitly * move them to the top of the stack where they are expected to be. * Since we do not know how many registers were saved in the calling function * we must assume the maximum possible (d2-d7,a2-a5). Hence, we end up * wasting some space on the stack. * * The large probe (tst.b) attempts to make up for the fact that we have * potentially used up the space that the caller probed for its own needs. */ .procss m0 .config "68000 1" .module _alloca MAXREG: .const 10 .sect text .global _alloca _alloca: move.l (sp)+,a0 ; pop return address move.l (sp)+,d0 ; pop allocation size move.l sp,d1 ; get current SP value sub.l d0,d1 ; adjust to reflect required size... sub.l #MAXREG*4,d1 ; ...and space needed for registers and.l #-4,d1 ; backup to longword boundry move.l sp,a1 ; save old SP value for register copy move.l d1,sp ; set the new SP value tst.b -4096(sp) ; grab an extra page (to cover caller) move.w #MAXREG-1,d0 ; # of longwords to copy loop: move.l (a1)+,(sp)+ ; copy registers... dbra d0,loop ; ...til there are no more move.l sp,d0 ; end of register area is addr for new space move.l d1,sp ; reset the SP subq.l #4,sp ; adjust to account for caller argument pop jmp (a0) ; return .end _alloca #else /* Some systems want the _, some do not. Win with both kinds. */ .globl _alloca _alloca: .globl alloca alloca: movl sp@+,a0 movl a7,d0 subl sp@,d0 andl #~3,d0 movl d0,sp tstb sp@(0) /* Make stack pages exist */ /* Needed on certain systems that lack true demand paging */ addql #4,d0 jmp a0@ #endif /* not WICAT */ #endif /* m68000 */ #endif /* not m68k */ #endif /* not hp9000 */ #ifdef ns16000 .text .align 2 /* Some systems want the _, some do not. Win with both kinds. */ .globl _alloca _alloca: .globl alloca alloca: /* Two different assembler syntaxes are used for the same code on different systems. */ #ifdef sequent #define IM #define REGISTER(x) x #else #define IM $ #define REGISTER(x) 0(x) #endif /* * The ns16000 is a little more difficult, need to copy regs. * Also the code assumes direct linkage call sequence (no mod table crap). * We have to copy registers, and therefore waste 32 bytes. * * Stack layout: * new sp -> junk * registers (copy) * r0 -> new data * | (orig retval) * | (orig arg) * old sp -> regs (orig) * local data * fp -> old fp */ movd tos,r1 /* pop return addr */ negd tos,r0 /* pop amount to allocate */ sprd sp,r2 addd r2,r0 bicb IM/**/3,r0 /* 4-byte align */ lprd sp,r0 adjspb IM/**/36 /* space for regs, +4 for caller to pop */ movmd 0(r2),4(sp),IM/**/4 /* copy regs */ movmd 0x10(r2),0x14(sp),IM/**/4 jump REGISTER(r1) /* funky return */ #endif /* ns16000 */ #ifdef pyramid .globl _alloca _alloca: addw $3,pr0 # add 3 (dec) to first argument bicw $3,pr0 # then clear its last 2 bits subw pr0,sp # subtract from SP the val in PR0 movw sp,pr0 # ret. current SP ret .globl __longjmp .globl _longjmp .globl __setjmp .globl _setjmp __longjmp: jump _longjmp __setjmp: jump _setjmp #endif /* pyramid */ #ifdef ATT3B5 .align 4 .globl alloca alloca: movw %ap, %r8 subw2 $9*4, %r8 movw 0(%r8), %r1 /* pc */ movw 4(%r8), %fp movw 8(%r8), %sp addw2 %r0, %sp /* make room */ movw %sp, %r0 /* return value */ jmp (%r1) /* continue... */ #endif /* ATT3B5 */ #endif /* not HAVE_ALLOCA */