diff options
author | Michael Shalayeff <mickey@cvs.openbsd.org> | 1998-06-23 18:56:55 +0000 |
---|---|---|
committer | Michael Shalayeff <mickey@cvs.openbsd.org> | 1998-06-23 18:56:55 +0000 |
commit | 3eb962f33405f5d242202b0c16e31ce87c4b6537 (patch) | |
tree | 182734bab9c92f9dbb32d613e13728d11066c66d /sys | |
parent | 1d6e6684a878474c571fc7313a9648765932986e (diff) |
initial import
Diffstat (limited to 'sys')
-rw-r--r-- | sys/lib/libkern/arch/hppa/Makefile.inc | 10 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/bcopy.S | 252 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/bzero.S | 115 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/divI.S | 211 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/divU.S | 199 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/ffs.S | 52 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/mulI.S | 480 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/mulU.S | 131 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/prefix.h | 87 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/remI.S | 164 | ||||
-rw-r--r-- | sys/lib/libkern/arch/hppa/remU.S | 152 |
11 files changed, 1853 insertions, 0 deletions
diff --git a/sys/lib/libkern/arch/hppa/Makefile.inc b/sys/lib/libkern/arch/hppa/Makefile.inc new file mode 100644 index 00000000000..5364c28199b --- /dev/null +++ b/sys/lib/libkern/arch/hppa/Makefile.inc @@ -0,0 +1,10 @@ +# $OpenBSD: Makefile.inc,v 1.1 1998/06/23 18:56:53 mickey Exp $ + +SRCS+= __main.c imax.c imin.c lmax.c lmin.c max.c min.c ulmax.c ulmin.c \ + random.c ashrdi3.c + +# `source' files built from m4 source +SRCS+= divU.S divI.S +SRCS+= remU.S remI.S +SRCS+= mulU.S mulI.S + diff --git a/sys/lib/libkern/arch/hppa/bcopy.S b/sys/lib/libkern/arch/hppa/bcopy.S new file mode 100644 index 00000000000..18a632e1665 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/bcopy.S @@ -0,0 +1,252 @@ +/* $OpenBSD: bcopy.S,v 1.1 1998/06/23 18:56:53 mickey Exp $ */ + +/* + * (c) Copyright 1988 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +/* + * Copyright (c) 1990,1994 The University of Utah and + * the Computer Systems Laboratory (CSL). All rights reserved. + * + * THE UNIVERSITY OF UTAH AND CSL PROVIDE THIS SOFTWARE IN ITS "AS IS" + * CONDITION, AND DISCLAIM ANY LIABILITY OF ANY KIND FOR ANY DAMAGES + * WHATSOEVER RESULTING FROM ITS USE. + * + * CSL requests users of this software to return to csl-dist@cs.utah.edu any + * improvements that they make and grant CSL redistribution rights. + * + * Utah $Hdr: bcopy.s 1.10 94/12/14$ + * Author: Bob Wheeler, University of Utah CSL + */ + +#include <machine/asm.h> + +/* + * void + * memcpy(dst, src, count) + * vm_offset_t dst; + * vm_offset_t src; + * int count; + */ +ENTRY(memcpy) + copy arg0,arg3 + copy arg1,arg0 + copy arg3,arg1 + /* And fall into.... */ + +/* + * void + * bcopy(src, dst, count) + * vm_offset_t src; + * vm_offset_t dst; + * int count; + */ +ALTENTRY(bcopy) + comb,>=,n r0,arg2,$bcopy_exit + + /* + * See if the source and destination are word aligned and if the count + * is an integer number of words. If so then we can use an optimized + * routine. If not then branch to bcopy_checkalign and see what we can + * do there. + */ + + or arg0,arg1,t1 + or t1,arg2,t2 + extru,= t2,31,2,r0 + b,n $bcopy_checkalign + + addib,<,n -16,arg2,$bcopy_movewords + + /* + * We can move the data in 4 word moves. We'll use 4 registers to + * avoid interlock and pipeline stalls. + */ + +$bcopy_loop16 + + ldwm 16(arg0),t1 + ldw -12(arg0),t2 + ldw -8(arg0),t3 + ldw -4(arg0),t4 + stwm t1,16(arg1) + stw t2,-12(arg1) + stw t3,-8(arg1) + addib,>= -16,arg2,$bcopy_loop16 + stw t4,-4(arg1) + + + /* + * We have already decremented the count by 16, add 12 to it and then + * we can test if there is at least 1 word left to move. + */ + +$bcopy_movewords + addib,<,n 12,arg2,$bcopy_exit + + /* + * Clean up any remaining words that were not moved in the 16 byte + * moves + */ + +$bcopy_loop4 + ldwm 4(arg0),t1 + addib,>= -4,arg2,$bcopy_loop4 + stwm t1,4(arg1) + + b,n $bcopy_exit + + +$bcopy_checkalign + + /* + * The source or destination is not word aligned or the count is not + * an integral number of words. If we are dealing with less than 16 + * bytes then just do it byte by byte. Otherwise, see if the data has + * the same basic alignment. We will add in the byte offset to size to + * keep track of what we have to move even though the stbys instruction + * won't physically move it. + */ + + comib,>= 15,arg2,$bcopy_byte + extru arg0,31,2,t1 + extru arg1,31,2,t2 + add arg2,t2,arg2 + comb,<> t2,t1,$bcopy_unaligned + dep 0,31,2,arg0 + + /* + * the source and destination have the same basic alignment. We will + * move the data in blocks of 16 bytes as long as we can and then + * we'll go to the 4 byte moves. + */ + + addib,<,n -16,arg2,$bcopy_aligned2 + +$bcopy_loop_aligned4 + ldwm 16(arg0),t1 + ldw -12(arg0),t2 + ldw -8(arg0),t3 + ldw -4(arg0),t4 + stbys,b,m t1,4(arg1) + stwm t2,4(arg1) + stwm t3,4(arg1) + addib,>= -16,arg2,$bcopy_loop_aligned4 + stwm t4,4(arg1) + + /* + * see if there is anything left that needs to be moved in a word move. + * Since the count was decremented by 16, add 12 to test if there are + * any full word moves left to do. + */ + +$bcopy_aligned2 + addib,<,n 12,arg2,$bcopy_cleanup + +$bcopy_loop_aligned2 + ldws,ma 4(arg0),t1 + addib,>= -4,arg2,$bcopy_loop_aligned2 + stbys,b,m t1,4(arg1) + + /* + * move the last bytes that may be unaligned on a word boundary + */ + +$bcopy_cleanup + addib,=,n 4,arg2,$bcopy_exit + ldws 0(arg0),t1 + add arg1,arg2,arg1 + b $bcopy_exit + stbys,e t1,0(arg1) + + /* + * The source and destination are not alligned on the same boundary + * types. We will have to shift the data around. Figure out the shift + * amount and load it into cr11. + */ + +$bcopy_unaligned + sub,>= t2,t1,t3 + ldwm 4(arg0),t1 + zdep t3,28,29,t4 + mtctl t4,11 + + /* + * see if we can do some of this work in blocks of 16 bytes + */ + + addib,<,n -16,arg2,$bcopy_unaligned_words + +$bcopy_unaligned4 + ldwm 16(arg0),t2 + ldw -12(arg0),t3 + ldw -8(arg0),t4 + ldw -4(arg0),r1 + vshd t1,t2,r28 + stbys,b,m r28,4(arg1) + vshd t2,t3,r28 + stwm r28,4(arg1) + vshd t3,t4,r28 + stwm r28,4(arg1) + vshd t4,r1,r28 + stwm r28,4(arg1) + addib,>= -16,arg2,$bcopy_unaligned4 + copy r1,t1 + + /* + * see if there is a full word that we can transfer + */ + +$bcopy_unaligned_words + addib,<,n 12,arg2,$bcopy_unaligned_cleanup1 + +$bcopy_unaligned_loop + ldwm 4(arg0),t2 + vshd t1,t2,t3 + addib,< -4,arg2,$bcopy_unaligned_cleanup2 + stbys,b,m t3,4(arg1) + + ldwm 4(arg0),t1 + vshd t2,t1,t3 + addib,>= -4,arg2,$bcopy_unaligned_loop + stbys,b,m t3,4(arg1) + +$bcopy_unaligned_cleanup1 + copy t1,t2 + +$bcopy_unaligned_cleanup2 + addib,<=,n 4,arg2,$bcopy_exit + add arg1,arg2,arg1 + mfctl sar,t3 + extru t3,28,2,t3 + sub,<= arg2,t3,r0 + ldwm 4(arg0),t1 + vshd t2,t1,t3 + b $bcopy_exit + stbys,e t3,0(arg1) + + /* + * move data one byte at a time + */ + +$bcopy_byte + comb,>=,n r0,arg2,$bcopy_exit + +$bcopy_loop_byte + ldbs,ma 1(arg0),t1 + addib,> -1,arg2,$bcopy_loop_byte + stbs,ma t1,1(arg1) + +$bcopy_exit +EXIT(bcopy) diff --git a/sys/lib/libkern/arch/hppa/bzero.S b/sys/lib/libkern/arch/hppa/bzero.S new file mode 100644 index 00000000000..34a22fcf450 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/bzero.S @@ -0,0 +1,115 @@ +/* $OpenBSD: bzero.S,v 1.1 1998/06/23 18:56:53 mickey Exp $ */ + +/* + * (c) Copyright 1988 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +/* + * Copyright (c) 1990,1994 The University of Utah and + * the Computer Systems Laboratory (CSL). All rights reserved. + * + * THE UNIVERSITY OF UTAH AND CSL PROVIDE THIS SOFTWARE IN ITS "AS IS" + * CONDITION, AND DISCLAIM ANY LIABILITY OF ANY KIND FOR ANY DAMAGES + * WHATSOEVER RESULTING FROM ITS USE. + * + * CSL requests users of this software to return to csl-dist@cs.utah.edu any + * improvements that they make and grant CSL redistribution rights. + * + * Utah $Hdr: bzero.s 1.9 94/12/14$ + * Author: Bob Wheeler, University of Utah CSL + */ + +#include <machine/asm.h> + +/* + * void + * blkclr(dst, count) + * void *dst; + * u_int count; + */ +ENTRY(blkclr) + +/* + * void + * bzero(dst, count) + * vm_offset_t dst; + * int count; + */ +ALTENTRY(bzero) + comb,>=,n r0,arg1,$bzero_exit + + /* + * If we need to clear less than a word do it a byte at a time + */ + + comib,>,n 4,arg1,$bzero_bytes + + /* + * Since we are only clearing memory the alignment restrictions + * are simplified. Figure out how many "extra" bytes we need to + * store with stbys. + */ + + extru arg0,31,2,t2 + add arg1,t2,arg1 + + /* + * We will zero the destination in blocks of 16 bytes as long as we + * can and then we'll go to the 4 byte moves. + */ + + addib,<,n -16,arg1,$bzero_word + +$bzero_loop_16 + stbys,b,m r0,4(arg0) + stwm r0,4(arg0) + stwm r0,4(arg0) + addib,>= -16,arg1,$bzero_loop_16 + stwm r0,4(arg0) + + /* + * see if there is anything left that needs to be zeroed in a word + * move. Since the count was decremented by 16, add 12 to test if + * there are any full word moves left to do. + */ + +$bzero_word + addib,<,n 12,arg1,$bzero_cleanup + +$bzero_loop_4 + addib,>= -4,arg1,$bzero_loop_4 + stbys,b,m r0,4(arg0) + + /* + * zero the last bytes that may be unaligned on a word boundary + */ + +$bzero_cleanup + addib,=,n 4,arg1,$bzero_exit + add arg0,arg1,arg0 + b $bzero_exit + stbys,e r0,0(arg0) + b,n $bzero_exit + + + /* + * zero by bytes + */ + +$bzero_bytes + addib,> -1,arg1,$bzero_bytes + stbs,ma r0,1(arg0) + +$bzero_exit +EXIT(bzero) diff --git a/sys/lib/libkern/arch/hppa/divI.S b/sys/lib/libkern/arch/hppa/divI.S new file mode 100644 index 00000000000..62ff31a547a --- /dev/null +++ b/sys/lib/libkern/arch/hppa/divI.S @@ -0,0 +1,211 @@ +/* $OpenBSD: divI.S,v 1.1 1998/06/23 18:56:53 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * + * ROUTINES: $$divI, $$divoI + * + * Single precision divide for singed binary integers. + * + * The quotient is truncated towards zero. + * The sign of the quotient is the XOR of the signs of the dividend and + * divisor. + * Divide by zero is trapped. + * Divide of -2**31 by -1 is trapped for $$divoI but not for $$divI. + * + * INPUT REGISTERS: + * arg0 == dividend + * arg1 == divisor + * r31 == return pc + * sr0 == return space when called externally + * + * OUTPUT REGISTERS: + * arg0 = undefined + * arg1 = undefined + * ret1 = quotient + * + * OTHER REGISTERS AFFECTED: + * r1 = undefined + * + * SIDE EFFECTS: + * Causes a trap under the following conditions: + * divisor is zero (traps with ADDIT,= 0,25,0) + * dividend==-2**31 and divisor==-1 and routine is $$divoI + * (traps with ADDO 26,25,0) + * Changes memory at the following places: + * NONE + * + * PERMISSIBLE CONTEXT: + * Unwindable. + * Suitable for internal or external millicode. + * Assumes the special millicode register conventions. + * + * DISCUSSION: + * Branchs to other millicode routines using BE + * $$div_# for # being 2,3,4,5,6,7,8,9,10,12,14,15 + * + * For selected divisors, calls a divide by constant routine written by + * Karl Pettis. Eligible divisors are 1..15 excluding 11 and 13. + * + * The only overflow case is -2**31 divided by -1. + * Both routines return -2**31 but only $$divoI traps. + */ + +DEFINE(temp,r1) +DEFINE(retreg,ret1) ; r29 + .import $$divI_2,millicode + .import $$divI_3,millicode + .import $$divI_4,millicode + .import $$divI_5,millicode + .import $$divI_6,millicode + .import $$divI_7,millicode + .import $$divI_8,millicode + .import $$divI_9,millicode + .import $$divI_10,millicode + .import $$divI_12,millicode + .import $$divI_14,millicode + .import $$divI_15,millicode + .export $$divI,millicode + .export $$divoI,millicode + .proc + .callinfo millicode +$$divoI + comib,=,n -1,arg1,negative1 ; when divisor == -1 +$$divI + comib,>>=,n 15,arg1,small_divisor + add,>= 0,arg0,retreg ; move dividend, if retreg < 0, +normal + sub 0,retreg,retreg ; make it positive + sub 0,arg1,temp ; clear carry, + ; negate the divisor + ds 0,temp,0 ; set V-bit to the comple- + ; ment of the divisor sign + add retreg,retreg,retreg ; shift msb bit into carry + ds r0,arg1,temp ; 1st divide step, if no carry + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 2nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 3rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 4th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 5th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 6th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 7th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 8th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 9th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 10th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 11th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 12th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 13th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 14th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 15th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 16th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 17th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 18th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 19th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 20th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 21st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 22nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 23rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 24th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 25th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 26th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 27th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 28th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 29th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 30th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 31st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 32nd divide step, + addc retreg,retreg,retreg ; shift last retreg bit into retreg + xor,>= arg0,arg1,0 ; get correct sign of quotient + sub 0,retreg,retreg ; based on operand signs + MILLIRETN + nop +;______________________________________________________________________ +small_divisor + blr,n arg1,r0 + nop +; table for divisor == 0,1, ... ,15 + addit,= 0,arg1,r0 ; trap if divisor == 0 + nop + MILLIRET ; divisor == 1 + copy arg0,retreg + MILLI_BEN($$divI_2) ; divisor == 2 + nop + MILLI_BEN($$divI_3) ; divisor == 3 + nop + MILLI_BEN($$divI_4) ; divisor == 4 + nop + MILLI_BEN($$divI_5) ; divisor == 5 + nop + MILLI_BEN($$divI_6) ; divisor == 6 + nop + MILLI_BEN($$divI_7) ; divisor == 7 + nop + MILLI_BEN($$divI_8) ; divisor == 8 + nop + MILLI_BEN($$divI_9) ; divisor == 9 + nop + MILLI_BEN($$divI_10) ; divisor == 10 + nop + b normal ; divisor == 11 + add,>= 0,arg0,retreg + MILLI_BEN($$divI_12) ; divisor == 12 + nop + b normal ; divisor == 13 + add,>= 0,arg0,retreg + MILLI_BEN($$divI_14) ; divisor == 14 + nop + MILLI_BEN($$divI_15) ; divisor == 15 + nop +;______________________________________________________________________ +negative1 + sub 0,arg0,retreg ; result is negation of dividend + MILLIRET + addo arg0,arg1,r0 ; trap iff dividend==0x80000000 && divisor==-1 + .procend + .end diff --git a/sys/lib/libkern/arch/hppa/divU.S b/sys/lib/libkern/arch/hppa/divU.S new file mode 100644 index 00000000000..025ce1bdf11 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/divU.S @@ -0,0 +1,199 @@ +/* $OpenBSD: divU.S,v 1.1 1998/06/23 18:56:53 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * ROUTINE: $$divU + * + * Single precision divide for unsigned integers. + * + * Quotient is truncated towards zero. + * Traps on divide by zero. + * + * INPUT REGISTERS: + * arg0 == dividend + * arg1 == divisor + * r31 == return pc + * sr0 == return space when called externally + * + * OUTPUT REGISTERS: + * arg0 = undefined + * arg1 = undefined + * ret1 = quotient + * + * OTHER REGISTERS AFFECTED: + * r1 = undefined + * + * SIDE EFFECTS: + * Causes a trap under the following conditions: + * divisor is zero + * Changes memory at the following places: + * NONE + * + * PERMISSIBLE CONTEXT: + * Unwindable. + * Does not create a stack frame. + * Suitable for internal or external millicode. + * Assumes the special millicode register conventions. + * + * DISCUSSION: + * Branchs to other millicode routines using BE: + * $$divU_# for 3,5,6,7,9,10,12,14,15 + * + * For selected small divisors calls the special divide by constant + * routines written by Karl Pettis. These are: 3,5,6,7,9,10,12,14,15. + * + */ + +DEFINE(temp,r1) +DEFINE(retreg,ret1) ; r29 + .subspa $MILLICODE$ + .export $$divU,millicode + .import $$divU_3,millicode + .import $$divU_5,millicode + .import $$divU_6,millicode + .import $$divU_7,millicode + .import $$divU_9,millicode + .import $$divU_10,millicode + .import $$divU_12,millicode + .import $$divU_14,millicode + .import $$divU_15,millicode + .proc + .callinfo millicode +$$divU +; The subtract is not nullified since it does no harm and can be used +; by the two cases that branch back to "normal". + comib,>= 15,arg1,special_divisor + sub r0,arg1,temp ; clear carry, negate the divisor + ds r0,temp,r0 ; set V-bit to 1 +normal + add arg0,arg0,retreg ; shift msb bit into carry + ds r0,arg1,temp ; 1st divide step, if no carry + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 2nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 3rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 4th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 5th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 6th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 7th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 8th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 9th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 10th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 11th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 12th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 13th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 14th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 15th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 16th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 17th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 18th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 19th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 20th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 21st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 22nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 23rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 24th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 25th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 26th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 27th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 28th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 29th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 30th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 31st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds temp,arg1,temp ; 32nd divide step, + MILLIRET + addc retreg,retreg,retreg ; shift last retreg bit into retreg +;_____________________________________________________________________________ +; handle the cases where divisor is a small constant or has high bit on +special_divisor + blr arg1,r0 + comib,>,n 0,arg1,big_divisor ; nullify previous instruction +zero_divisor ; this label is here to provide external visibility + addit,= 0,arg1,0 ; trap for zero dvr + nop + MILLIRET ; divisor == 1 + copy arg0,retreg + MILLIRET ; divisor == 2 + extru arg0,30,31,retreg + MILLI_BEN($$divU_3) ; divisor == 3 + nop + MILLIRET ; divisor == 4 + extru arg0,29,30,retreg + MILLI_BEN($$divU_5) ; divisor == 5 + nop + MILLI_BEN($$divU_6) ; divisor == 6 + nop + MILLI_BEN($$divU_7) ; divisor == 7 + nop + MILLIRET ; divisor == 8 + extru arg0,28,29,retreg + MILLI_BEN($$divU_9) ; divisor == 9 + nop + MILLI_BEN($$divU_10) ; divisor == 10 + nop + b normal ; divisor == 11 + ds r0,temp,r0 ; set V-bit to 1 + MILLI_BEN($$divU_12) ; divisor == 12 + nop + b normal ; divisor == 13 + ds r0,temp,r0 ; set V-bit to 1 + MILLI_BEN($$divU_14) ; divisor == 14 + nop + MILLI_BEN($$divU_15) ; divisor == 15 + nop +;_____________________________________________________________________________ +; Handle the case where the high bit is on in the divisor. +; Compute: if( dividend>=divisor) quotient=1; else quotient=0; +; Note: dividend>==divisor iff dividend-divisor does not borrow +; and not borrow iff carry +big_divisor + sub arg0,arg1,r0 + MILLIRET + addc r0,r0,retreg + .procend + .end diff --git a/sys/lib/libkern/arch/hppa/ffs.S b/sys/lib/libkern/arch/hppa/ffs.S new file mode 100644 index 00000000000..df21ca06dc5 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/ffs.S @@ -0,0 +1,52 @@ +/* $OpenBSD: ffs.S,v 1.1 1998/06/23 18:56:53 mickey Exp $ */ + +/* + * Copyright (c) 1990, 1991, 1992, 1994, The University of Utah and + * the Computer Systems Laboratory at the University of Utah (CSL). + * All rights reserved. + * + * Permission to use, copy, modify and distribute this software and its + * documentation is hereby granted, provided that both the copyright + * notice and this permission notice appear in all copies of the + * software, derivative works or modified versions, and any portions + * thereof, and that both notices appear in supporting documentation. + * + * THE UNIVERSITY OF UTAH AND CSL ALLOW FREE USE OF THIS SOFTWARE IN ITS "AS + * IS" CONDITION. THE UNIVERSITY OF UTAH AND CSL DISCLAIM ANY LIABILITY OF + * ANY KIND FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. + * + * CSL requests users of this software to return to csl-dist@cs.utah.edu any + * improvements that they make and grant CSL redistribution rights. + * + */ + +#include <machine/asm.h> + +/* + * ffs(bitmask) + * + * Return the position of the "most significant" bit in `bitmask'. + * Since this is similar to the VAX ffs instruction, bits in a word + * are numbered as "32, 31, ... 1", 0 is returned if no bits are set. + */ + +ENTRY(ffs) + comb,= arg0,r0,ffsdone ; If arg0 is 0 + or r0,r0,ret0 ; return 0 + ldi 32,ret0 ; Set return to high bit + extru,= arg0,31,16,r0 ; If low 16 bits are non-zero + addi,tr -16,ret0,ret0 ; subtract 16 from bitpos + shd r0,arg0,16,arg0 ; else shift right 16 bits + extru,= arg0,31,8,r0 ; If low 8 bits are non-zero + addi,tr -8,ret0,ret0 ; subtract 8 from bitpos + shd r0,arg0,8,arg0 ; else shift right 8 bits + extru,= arg0,31,4,r0 ; If low 4 bits are non-zero + addi,tr -4,ret0,ret0 ; subtract 4 from bitpos + shd r0,arg0,4,arg0 ; else shift right 4 bits + extru,= arg0,31,2,r0 ; If low 2 bits are non-zero + addi,tr -2,ret0,ret0 ; subtract 2 from bitpos + shd r0,arg0,2,arg0 ; else shift right 2 bits + extru,= arg0,31,1,r0 ; If low bit is non-zero + addi -1,ret0,ret0 ; subtract 1 from bitpos +ffsdone: +EXIT(ffs) diff --git a/sys/lib/libkern/arch/hppa/mulI.S b/sys/lib/libkern/arch/hppa/mulI.S new file mode 100644 index 00000000000..3bf9c1d215b --- /dev/null +++ b/sys/lib/libkern/arch/hppa/mulI.S @@ -0,0 +1,480 @@ +/* $OpenBSD: mulI.S,v 1.1 1998/06/23 18:56:54 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * Author: Dan Zuras + * Editor: Pratap Subrahmanyam + * + * ROUTINE: $$mulI + * + * + * DESCRIPTION: + * + * $$mulI multiplies two single word integers, giving a single + * word result. + * + * + * INPUT REGISTERS: + * + * arg0 = Operand 1 + * arg1 = Operand 2 + * r31 == return pc + * sr0 == return space when called externally + * + * + * OUTPUT REGISTERS: + * + * arg0 = undefined + * arg1 = undefined + * ret1 = result + * + * OTHER REGISTERS AFFECTED: + * + * r1 = undefined + * + * SIDE EFFECTS: + * + * Causes a trap under the following conditions: NONE + * Changes memory at the following places: NONE + * + * PERMISSIBLE CONTEXT: + * + * Unwindable + * Does not create a stack frame + * Is usable for internal or external microcode + * + * DISCUSSION: + * + * Calls other millicode routines via r31: NONE + * Calls other millicode routines: NONE + * + */ + + +#define a0 %arg0 +#define a1 %arg1 +#define rtn %mrp +#define t0 %r1 +#define r %ret1 + +#define a0__128a0 zdep a0,24,25,a0 +#define a0__256a0 zdep a0,23,24,a0 +#define a1_ne_0_b_l0 comb,<> a1,0,l0 +#define a1_ne_0_b_l1 comb,<> a1,0,l1 +#define a1_ne_0_b_l2 comb,<> a1,0,l2 +#define b_n_ret_t0 b,n ret_t0 +#define b_e_shift b e_shift +#define b_e_t0ma0 b e_t0ma0 +#define b_e_t0 b e_t0 +#define b_e_t0a0 b e_t0a0 +#define b_e_t02a0 b e_t02a0 +#define b_e_t04a0 b e_t04a0 +#define b_e_2t0 b e_2t0 +#define b_e_2t0a0 b e_2t0a0 +#define b_e_2t04a0 b e2t04a0 +#define b_e_3t0 b e_3t0 +#define b_e_4t0 b e_4t0 +#define b_e_4t0a0 b e_4t0a0 +#define b_e_4t08a0 b e4t08a0 +#define b_e_5t0 b e_5t0 +#define b_e_8t0 b e_8t0 +#define b_e_8t0a0 b e_8t0a0 +#define r__r_a0 add r,a0,r +#define r__r_2a0 sh1add a0,r,r +#define r__r_4a0 sh2add a0,r,r +#define r__r_8a0 sh3add a0,r,r +#define r__r_t0 add r,t0,r +#define r__r_2t0 sh1add t0,r,r +#define r__r_4t0 sh2add t0,r,r +#define r__r_8t0 sh3add t0,r,r +; Use MILLIRET and MILLIRETN instead. +; #define return bv 0(rtn) +; #define return_n bv,n 0(rtn) +#define t0__3a0 sh1add a0,a0,t0 +#define t0__4a0 sh2add a0,0,t0 +#define t0__5a0 sh2add a0,a0,t0 +#define t0__8a0 sh3add a0,0,t0 +#define t0__9a0 sh3add a0,a0,t0 +#define t0__16a0 zdep a0,27,28,t0 +#define t0__32a0 zdep a0,26,27,t0 +#define t0__64a0 zdep a0,25,26,t0 +#define t0__128a0 zdep a0,24,25,t0 +#define t0__t0ma0 sub t0,a0,t0 +#define t0__t0_a0 add t0,a0,t0 +#define t0__t0_2a0 sh1add a0,t0,t0 +#define t0__t0_4a0 sh2add a0,t0,t0 +#define t0__t0_8a0 sh3add a0,t0,t0 +#define t0__2t0_a0 sh1add t0,a0,t0 +#define t0__3t0 sh1add t0,t0,t0 +#define t0__4t0 sh2add t0,0,t0 +#define t0__4t0_a0 sh2add t0,a0,t0 +#define t0__5t0 sh2add t0,t0,t0 +#define t0__8t0 sh3add t0,0,t0 +#define t0__8t0_a0 sh3add t0,a0,t0 +#define t0__9t0 sh3add t0,t0,t0 +#define t0__16t0 zdep t0,27,28,t0 +#define t0__32t0 zdep t0,26,27,t0 +#define t0__256a0 zdep a0,23,24,t0 + + + .subspa $MILLICODE$ + .align 16 + .proc + .callinfo millicode + .export $$mulI, millicode +$$mulI + combt,<<= a1,a0,l4 ; swap args if unsigned a1>a0 + copy 0,r ; zero out the result + xor a0,a1,a0 ; swap a0 & a1 using the + xor a0,a1,a1 ; old xor trick + xor a0,a1,a0 +l4 combt,<= 0,a0,l3 ; if a0>=0 then proceed like unsigned + zdep a1,30,8,t0 ; t0 = (a1&0xff)<<1 ********* + sub,> 0,a1,t0 ; otherwise negate both and + combt,<=,n a0,t0,l2 ; swap back if |a0|<|a1| + sub 0,a0,a1 + movb,tr,n t0,a0,l2 ; 10th inst. + +l0 r__r_t0 ; add in this partial product +l1 a0__256a0 ; a0 <<= 8 ****************** +l2 zdep a1,30,8,t0 ; t0 = (a1&0xff)<<1 ********* +l3 blr t0,0 ; case on these 8 bits ****** + extru a1,23,24,a1 ; a1 >>= 8 ****************** + +;16 insts before this. +; a0 <<= 8 ************************** +x0 a1_ne_0_b_l2 ! a0__256a0 ! MILLIRETN ! nop +x1 a1_ne_0_b_l1 ! r__r_a0 ! MILLIRETN ! nop +x2 a1_ne_0_b_l1 ! r__r_2a0 ! MILLIRETN ! nop +x3 a1_ne_0_b_l0 ! t0__3a0 ! MILLIRET ! r__r_t0 +x4 a1_ne_0_b_l1 ! r__r_4a0 ! MILLIRETN ! nop +x5 a1_ne_0_b_l0 ! t0__5a0 ! MILLIRET ! r__r_t0 +x6 t0__3a0 ! a1_ne_0_b_l1 ! r__r_2t0 ! MILLIRETN +x7 t0__3a0 ! a1_ne_0_b_l0 ! r__r_4a0 ! b_n_ret_t0 +x8 a1_ne_0_b_l1 ! r__r_8a0 ! MILLIRETN ! nop +x9 a1_ne_0_b_l0 ! t0__9a0 ! MILLIRET ! r__r_t0 +x10 t0__5a0 ! a1_ne_0_b_l1 ! r__r_2t0 ! MILLIRETN +x11 t0__3a0 ! a1_ne_0_b_l0 ! r__r_8a0 ! b_n_ret_t0 +x12 t0__3a0 ! a1_ne_0_b_l1 ! r__r_4t0 ! MILLIRETN +x13 t0__5a0 ! a1_ne_0_b_l0 ! r__r_8a0 ! b_n_ret_t0 +x14 t0__3a0 ! t0__2t0_a0 ! b_e_shift ! r__r_2t0 +x15 t0__5a0 ! a1_ne_0_b_l0 ! t0__3t0 ! b_n_ret_t0 +x16 t0__16a0 ! a1_ne_0_b_l1 ! r__r_t0 ! MILLIRETN +x17 t0__9a0 ! a1_ne_0_b_l0 ! t0__t0_8a0 ! b_n_ret_t0 +x18 t0__9a0 ! a1_ne_0_b_l1 ! r__r_2t0 ! MILLIRETN +x19 t0__9a0 ! a1_ne_0_b_l0 ! t0__2t0_a0 ! b_n_ret_t0 +x20 t0__5a0 ! a1_ne_0_b_l1 ! r__r_4t0 ! MILLIRETN +x21 t0__5a0 ! a1_ne_0_b_l0 ! t0__4t0_a0 ! b_n_ret_t0 +x22 t0__5a0 ! t0__2t0_a0 ! b_e_shift ! r__r_2t0 +x23 t0__5a0 ! t0__2t0_a0 ! b_e_t0 ! t0__2t0_a0 +x24 t0__3a0 ! a1_ne_0_b_l1 ! r__r_8t0 ! MILLIRETN +x25 t0__5a0 ! a1_ne_0_b_l0 ! t0__5t0 ! b_n_ret_t0 +x26 t0__3a0 ! t0__4t0_a0 ! b_e_shift ! r__r_2t0 +x27 t0__3a0 ! a1_ne_0_b_l0 ! t0__9t0 ! b_n_ret_t0 +x28 t0__3a0 ! t0__2t0_a0 ! b_e_shift ! r__r_4t0 +x29 t0__3a0 ! t0__2t0_a0 ! b_e_t0 ! t0__4t0_a0 +x30 t0__5a0 ! t0__3t0 ! b_e_shift ! r__r_2t0 +x31 t0__32a0 ! a1_ne_0_b_l0 ! t0__t0ma0 ! b_n_ret_t0 +x32 t0__32a0 ! a1_ne_0_b_l1 ! r__r_t0 ! MILLIRETN +x33 t0__8a0 ! a1_ne_0_b_l0 ! t0__4t0_a0 ! b_n_ret_t0 +x34 t0__16a0 ! t0__t0_a0 ! b_e_shift ! r__r_2t0 +x35 t0__9a0 ! t0__3t0 ! b_e_t0 ! t0__t0_8a0 +x36 t0__9a0 ! a1_ne_0_b_l1 ! r__r_4t0 ! MILLIRETN +x37 t0__9a0 ! a1_ne_0_b_l0 ! t0__4t0_a0 ! b_n_ret_t0 +x38 t0__9a0 ! t0__2t0_a0 ! b_e_shift ! r__r_2t0 +x39 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__2t0_a0 +x40 t0__5a0 ! a1_ne_0_b_l1 ! r__r_8t0 ! MILLIRETN +x41 t0__5a0 ! a1_ne_0_b_l0 ! t0__8t0_a0 ! b_n_ret_t0 +x42 t0__5a0 ! t0__4t0_a0 ! b_e_shift ! r__r_2t0 +x43 t0__5a0 ! t0__4t0_a0 ! b_e_t0 ! t0__2t0_a0 +x44 t0__5a0 ! t0__2t0_a0 ! b_e_shift ! r__r_4t0 +x45 t0__9a0 ! a1_ne_0_b_l0 ! t0__5t0 ! b_n_ret_t0 +x46 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__t0_a0 +x47 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__t0_2a0 +x48 t0__3a0 ! a1_ne_0_b_l0 ! t0__16t0 ! b_n_ret_t0 +x49 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__t0_4a0 +x50 t0__5a0 ! t0__5t0 ! b_e_shift ! r__r_2t0 +x51 t0__9a0 ! t0__t0_8a0 ! b_e_t0 ! t0__3t0 +x52 t0__3a0 ! t0__4t0_a0 ! b_e_shift ! r__r_4t0 +x53 t0__3a0 ! t0__4t0_a0 ! b_e_t0 ! t0__4t0_a0 +x54 t0__9a0 ! t0__3t0 ! b_e_shift ! r__r_2t0 +x55 t0__9a0 ! t0__3t0 ! b_e_t0 ! t0__2t0_a0 +x56 t0__3a0 ! t0__2t0_a0 ! b_e_shift ! r__r_8t0 +x57 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__3t0 +x58 t0__3a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__4t0_a0 +x59 t0__9a0 ! t0__2t0_a0 ! b_e_t02a0 ! t0__3t0 +x60 t0__5a0 ! t0__3t0 ! b_e_shift ! r__r_4t0 +x61 t0__5a0 ! t0__3t0 ! b_e_t0 ! t0__4t0_a0 +x62 t0__32a0 ! t0__t0ma0 ! b_e_shift ! r__r_2t0 +x63 t0__64a0 ! a1_ne_0_b_l0 ! t0__t0ma0 ! b_n_ret_t0 +x64 t0__64a0 ! a1_ne_0_b_l1 ! r__r_t0 ! MILLIRETN +x65 t0__8a0 ! a1_ne_0_b_l0 ! t0__8t0_a0 ! b_n_ret_t0 +x66 t0__32a0 ! t0__t0_a0 ! b_e_shift ! r__r_2t0 +x67 t0__8a0 ! t0__4t0_a0 ! b_e_t0 ! t0__2t0_a0 +x68 t0__8a0 ! t0__2t0_a0 ! b_e_shift ! r__r_4t0 +x69 t0__8a0 ! t0__2t0_a0 ! b_e_t0 ! t0__4t0_a0 +x70 t0__64a0 ! t0__t0_4a0 ! b_e_t0 ! t0__t0_2a0 +x71 t0__9a0 ! t0__8t0 ! b_e_t0 ! t0__t0ma0 +x72 t0__9a0 ! a1_ne_0_b_l1 ! r__r_8t0 ! MILLIRETN +x73 t0__9a0 ! t0__8t0_a0 ! b_e_shift ! r__r_t0 +x74 t0__9a0 ! t0__4t0_a0 ! b_e_shift ! r__r_2t0 +x75 t0__9a0 ! t0__4t0_a0 ! b_e_t0 ! t0__2t0_a0 +x76 t0__9a0 ! t0__2t0_a0 ! b_e_shift ! r__r_4t0 +x77 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__4t0_a0 +x78 t0__9a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__2t0_a0 +x79 t0__16a0 ! t0__5t0 ! b_e_t0 ! t0__t0ma0 +x80 t0__16a0 ! t0__5t0 ! b_e_shift ! r__r_t0 +x81 t0__9a0 ! t0__9t0 ! b_e_shift ! r__r_t0 +x82 t0__5a0 ! t0__8t0_a0 ! b_e_shift ! r__r_2t0 +x83 t0__5a0 ! t0__8t0_a0 ! b_e_t0 ! t0__2t0_a0 +x84 t0__5a0 ! t0__4t0_a0 ! b_e_shift ! r__r_4t0 +x85 t0__8a0 ! t0__2t0_a0 ! b_e_t0 ! t0__5t0 +x86 t0__5a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__2t0_a0 +x87 t0__9a0 ! t0__9t0 ! b_e_t02a0 ! t0__t0_4a0 +x88 t0__5a0 ! t0__2t0_a0 ! b_e_shift ! r__r_8t0 +x89 t0__5a0 ! t0__2t0_a0 ! b_e_t0 ! t0__8t0_a0 +x90 t0__9a0 ! t0__5t0 ! b_e_shift ! r__r_2t0 +x91 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__2t0_a0 +x92 t0__5a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__2t0_a0 +x93 t0__32a0 ! t0__t0ma0 ! b_e_t0 ! t0__3t0 +x94 t0__9a0 ! t0__5t0 ! b_e_2t0 ! t0__t0_2a0 +x95 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__5t0 +x96 t0__8a0 ! t0__3t0 ! b_e_shift ! r__r_4t0 +x97 t0__8a0 ! t0__3t0 ! b_e_t0 ! t0__4t0_a0 +x98 t0__32a0 ! t0__3t0 ! b_e_t0 ! t0__t0_2a0 +x99 t0__8a0 ! t0__4t0_a0 ! b_e_t0 ! t0__3t0 +x100 t0__5a0 ! t0__5t0 ! b_e_shift ! r__r_4t0 +x101 t0__5a0 ! t0__5t0 ! b_e_t0 ! t0__4t0_a0 +x102 t0__32a0 ! t0__t0_2a0 ! b_e_t0 ! t0__3t0 +x103 t0__5a0 ! t0__5t0 ! b_e_t02a0 ! t0__4t0_a0 +x104 t0__3a0 ! t0__4t0_a0 ! b_e_shift ! r__r_8t0 +x105 t0__5a0 ! t0__4t0_a0 ! b_e_t0 ! t0__5t0 +x106 t0__3a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__4t0_a0 +x107 t0__9a0 ! t0__t0_4a0 ! b_e_t02a0 ! t0__8t0_a0 +x108 t0__9a0 ! t0__3t0 ! b_e_shift ! r__r_4t0 +x109 t0__9a0 ! t0__3t0 ! b_e_t0 ! t0__4t0_a0 +x110 t0__9a0 ! t0__3t0 ! b_e_2t0 ! t0__2t0_a0 +x111 t0__9a0 ! t0__4t0_a0 ! b_e_t0 ! t0__3t0 +x112 t0__3a0 ! t0__2t0_a0 ! b_e_t0 ! t0__16t0 +x113 t0__9a0 ! t0__4t0_a0 ! b_e_t02a0 ! t0__3t0 +x114 t0__9a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__3t0 +x115 t0__9a0 ! t0__2t0_a0 ! b_e_2t0a0 ! t0__3t0 +x116 t0__3a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__4t0_a0 +x117 t0__3a0 ! t0__4t0_a0 ! b_e_t0 ! t0__9t0 +x118 t0__3a0 ! t0__4t0_a0 ! b_e_t0a0 ! t0__9t0 +x119 t0__3a0 ! t0__4t0_a0 ! b_e_t02a0 ! t0__9t0 +x120 t0__5a0 ! t0__3t0 ! b_e_shift ! r__r_8t0 +x121 t0__5a0 ! t0__3t0 ! b_e_t0 ! t0__8t0_a0 +x122 t0__5a0 ! t0__3t0 ! b_e_2t0 ! t0__4t0_a0 +x123 t0__5a0 ! t0__8t0_a0 ! b_e_t0 ! t0__3t0 +x124 t0__32a0 ! t0__t0ma0 ! b_e_shift ! r__r_4t0 +x125 t0__5a0 ! t0__5t0 ! b_e_t0 ! t0__5t0 +x126 t0__64a0 ! t0__t0ma0 ! b_e_shift ! r__r_2t0 +x127 t0__128a0 ! a1_ne_0_b_l0 ! t0__t0ma0 ! b_n_ret_t0 +x128 t0__128a0 ! a1_ne_0_b_l1 ! r__r_t0 ! MILLIRETN +x129 t0__128a0 ! a1_ne_0_b_l0 ! t0__t0_a0 ! b_n_ret_t0 +x130 t0__64a0 ! t0__t0_a0 ! b_e_shift ! r__r_2t0 +x131 t0__8a0 ! t0__8t0_a0 ! b_e_t0 ! t0__2t0_a0 +x132 t0__8a0 ! t0__4t0_a0 ! b_e_shift ! r__r_4t0 +x133 t0__8a0 ! t0__4t0_a0 ! b_e_t0 ! t0__4t0_a0 +x134 t0__8a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__2t0_a0 +x135 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__3t0 +x136 t0__8a0 ! t0__2t0_a0 ! b_e_shift ! r__r_8t0 +x137 t0__8a0 ! t0__2t0_a0 ! b_e_t0 ! t0__8t0_a0 +x138 t0__8a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__4t0_a0 +x139 t0__8a0 ! t0__2t0_a0 ! b_e_2t0a0 ! t0__4t0_a0 +x140 t0__3a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__5t0 +x141 t0__8a0 ! t0__2t0_a0 ! b_e_4t0a0 ! t0__2t0_a0 +x142 t0__9a0 ! t0__8t0 ! b_e_2t0 ! t0__t0ma0 +x143 t0__16a0 ! t0__9t0 ! b_e_t0 ! t0__t0ma0 +x144 t0__9a0 ! t0__8t0 ! b_e_shift ! r__r_2t0 +x145 t0__9a0 ! t0__8t0 ! b_e_t0 ! t0__2t0_a0 +x146 t0__9a0 ! t0__8t0_a0 ! b_e_shift ! r__r_2t0 +x147 t0__9a0 ! t0__8t0_a0 ! b_e_t0 ! t0__2t0_a0 +x148 t0__9a0 ! t0__4t0_a0 ! b_e_shift ! r__r_4t0 +x149 t0__9a0 ! t0__4t0_a0 ! b_e_t0 ! t0__4t0_a0 +x150 t0__9a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__2t0_a0 +x151 t0__9a0 ! t0__4t0_a0 ! b_e_2t0a0 ! t0__2t0_a0 +x152 t0__9a0 ! t0__2t0_a0 ! b_e_shift ! r__r_8t0 +x153 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__8t0_a0 +x154 t0__9a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__4t0_a0 +x155 t0__32a0 ! t0__t0ma0 ! b_e_t0 ! t0__5t0 +x156 t0__9a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__2t0_a0 +x157 t0__32a0 ! t0__t0ma0 ! b_e_t02a0 ! t0__5t0 +x158 t0__16a0 ! t0__5t0 ! b_e_2t0 ! t0__t0ma0 +x159 t0__32a0 ! t0__5t0 ! b_e_t0 ! t0__t0ma0 +x160 t0__5a0 ! t0__4t0 ! b_e_shift ! r__r_8t0 +x161 t0__8a0 ! t0__5t0 ! b_e_t0 ! t0__4t0_a0 +x162 t0__9a0 ! t0__9t0 ! b_e_shift ! r__r_2t0 +x163 t0__9a0 ! t0__9t0 ! b_e_t0 ! t0__2t0_a0 +x164 t0__5a0 ! t0__8t0_a0 ! b_e_shift ! r__r_4t0 +x165 t0__8a0 ! t0__4t0_a0 ! b_e_t0 ! t0__5t0 +x166 t0__5a0 ! t0__8t0_a0 ! b_e_2t0 ! t0__2t0_a0 +x167 t0__5a0 ! t0__8t0_a0 ! b_e_2t0a0 ! t0__2t0_a0 +x168 t0__5a0 ! t0__4t0_a0 ! b_e_shift ! r__r_8t0 +x169 t0__5a0 ! t0__4t0_a0 ! b_e_t0 ! t0__8t0_a0 +x170 t0__32a0 ! t0__t0_2a0 ! b_e_t0 ! t0__5t0 +x171 t0__9a0 ! t0__2t0_a0 ! b_e_t0 ! t0__9t0 +x172 t0__5a0 ! t0__4t0_a0 ! b_e_4t0 ! t0__2t0_a0 +x173 t0__9a0 ! t0__2t0_a0 ! b_e_t02a0 ! t0__9t0 +x174 t0__32a0 ! t0__t0_2a0 ! b_e_t04a0 ! t0__5t0 +x175 t0__8a0 ! t0__2t0_a0 ! b_e_5t0 ! t0__2t0_a0 +x176 t0__5a0 ! t0__4t0_a0 ! b_e_8t0 ! t0__t0_a0 +x177 t0__5a0 ! t0__4t0_a0 ! b_e_8t0a0 ! t0__t0_a0 +x178 t0__5a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__8t0_a0 +x179 t0__5a0 ! t0__2t0_a0 ! b_e_2t0a0 ! t0__8t0_a0 +x180 t0__9a0 ! t0__5t0 ! b_e_shift ! r__r_4t0 +x181 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__4t0_a0 +x182 t0__9a0 ! t0__5t0 ! b_e_2t0 ! t0__2t0_a0 +x183 t0__9a0 ! t0__5t0 ! b_e_2t0a0 ! t0__2t0_a0 +x184 t0__5a0 ! t0__9t0 ! b_e_4t0 ! t0__t0_a0 +x185 t0__9a0 ! t0__4t0_a0 ! b_e_t0 ! t0__5t0 +x186 t0__32a0 ! t0__t0ma0 ! b_e_2t0 ! t0__3t0 +x187 t0__9a0 ! t0__4t0_a0 ! b_e_t02a0 ! t0__5t0 +x188 t0__9a0 ! t0__5t0 ! b_e_4t0 ! t0__t0_2a0 +x189 t0__5a0 ! t0__4t0_a0 ! b_e_t0 ! t0__9t0 +x190 t0__9a0 ! t0__2t0_a0 ! b_e_2t0 ! t0__5t0 +x191 t0__64a0 ! t0__3t0 ! b_e_t0 ! t0__t0ma0 +x192 t0__8a0 ! t0__3t0 ! b_e_shift ! r__r_8t0 +x193 t0__8a0 ! t0__3t0 ! b_e_t0 ! t0__8t0_a0 +x194 t0__8a0 ! t0__3t0 ! b_e_2t0 ! t0__4t0_a0 +x195 t0__8a0 ! t0__8t0_a0 ! b_e_t0 ! t0__3t0 +x196 t0__8a0 ! t0__3t0 ! b_e_4t0 ! t0__2t0_a0 +x197 t0__8a0 ! t0__3t0 ! b_e_4t0a0 ! t0__2t0_a0 +x198 t0__64a0 ! t0__t0_2a0 ! b_e_t0 ! t0__3t0 +x199 t0__8a0 ! t0__4t0_a0 ! b_e_2t0a0 ! t0__3t0 +x200 t0__5a0 ! t0__5t0 ! b_e_shift ! r__r_8t0 +x201 t0__5a0 ! t0__5t0 ! b_e_t0 ! t0__8t0_a0 +x202 t0__5a0 ! t0__5t0 ! b_e_2t0 ! t0__4t0_a0 +x203 t0__5a0 ! t0__5t0 ! b_e_2t0a0 ! t0__4t0_a0 +x204 t0__8a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__3t0 +x205 t0__5a0 ! t0__8t0_a0 ! b_e_t0 ! t0__5t0 +x206 t0__64a0 ! t0__t0_4a0 ! b_e_t02a0 ! t0__3t0 +x207 t0__8a0 ! t0__2t0_a0 ! b_e_3t0 ! t0__4t0_a0 +x208 t0__5a0 ! t0__5t0 ! b_e_8t0 ! t0__t0_a0 +x209 t0__5a0 ! t0__5t0 ! b_e_8t0a0 ! t0__t0_a0 +x210 t0__5a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__5t0 +x211 t0__5a0 ! t0__4t0_a0 ! b_e_2t0a0 ! t0__5t0 +x212 t0__3a0 ! t0__4t0_a0 ! b_e_4t0 ! t0__4t0_a0 +x213 t0__3a0 ! t0__4t0_a0 ! b_e_4t0a0 ! t0__4t0_a0 +x214 t0__9a0 ! t0__t0_4a0 ! b_e_2t04a0 ! t0__8t0_a0 +x215 t0__5a0 ! t0__4t0_a0 ! b_e_5t0 ! t0__2t0_a0 +x216 t0__9a0 ! t0__3t0 ! b_e_shift ! r__r_8t0 +x217 t0__9a0 ! t0__3t0 ! b_e_t0 ! t0__8t0_a0 +x218 t0__9a0 ! t0__3t0 ! b_e_2t0 ! t0__4t0_a0 +x219 t0__9a0 ! t0__8t0_a0 ! b_e_t0 ! t0__3t0 +x220 t0__3a0 ! t0__9t0 ! b_e_4t0 ! t0__2t0_a0 +x221 t0__3a0 ! t0__9t0 ! b_e_4t0a0 ! t0__2t0_a0 +x222 t0__9a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__3t0 +x223 t0__9a0 ! t0__4t0_a0 ! b_e_2t0a0 ! t0__3t0 +x224 t0__9a0 ! t0__3t0 ! b_e_8t0 ! t0__t0_a0 +x225 t0__9a0 ! t0__5t0 ! b_e_t0 ! t0__5t0 +x226 t0__3a0 ! t0__2t0_a0 ! b_e_t02a0 ! t0__32t0 +x227 t0__9a0 ! t0__5t0 ! b_e_t02a0 ! t0__5t0 +x228 t0__9a0 ! t0__2t0_a0 ! b_e_4t0 ! t0__3t0 +x229 t0__9a0 ! t0__2t0_a0 ! b_e_4t0a0 ! t0__3t0 +x230 t0__9a0 ! t0__5t0 ! b_e_5t0 ! t0__t0_a0 +x231 t0__9a0 ! t0__2t0_a0 ! b_e_3t0 ! t0__4t0_a0 +x232 t0__3a0 ! t0__2t0_a0 ! b_e_8t0 ! t0__4t0_a0 +x233 t0__3a0 ! t0__2t0_a0 ! b_e_8t0a0 ! t0__4t0_a0 +x234 t0__3a0 ! t0__4t0_a0 ! b_e_2t0 ! t0__9t0 +x235 t0__3a0 ! t0__4t0_a0 ! b_e_2t0a0 ! t0__9t0 +x236 t0__9a0 ! t0__2t0_a0 ! b_e_4t08a0 ! t0__3t0 +x237 t0__16a0 ! t0__5t0 ! b_e_3t0 ! t0__t0ma0 +x238 t0__3a0 ! t0__4t0_a0 ! b_e_2t04a0 ! t0__9t0 +x239 t0__16a0 ! t0__5t0 ! b_e_t0ma0 ! t0__3t0 +x240 t0__9a0 ! t0__t0_a0 ! b_e_8t0 ! t0__3t0 +x241 t0__9a0 ! t0__t0_a0 ! b_e_8t0a0 ! t0__3t0 +x242 t0__5a0 ! t0__3t0 ! b_e_2t0 ! t0__8t0_a0 +x243 t0__9a0 ! t0__9t0 ! b_e_t0 ! t0__3t0 +x244 t0__5a0 ! t0__3t0 ! b_e_4t0 ! t0__4t0_a0 +x245 t0__8a0 ! t0__3t0 ! b_e_5t0 ! t0__2t0_a0 +x246 t0__5a0 ! t0__8t0_a0 ! b_e_2t0 ! t0__3t0 +x247 t0__5a0 ! t0__8t0_a0 ! b_e_2t0a0 ! t0__3t0 +x248 t0__32a0 ! t0__t0ma0 ! b_e_shift ! r__r_8t0 +x249 t0__32a0 ! t0__t0ma0 ! b_e_t0 ! t0__8t0_a0 +x250 t0__5a0 ! t0__5t0 ! b_e_2t0 ! t0__5t0 +x251 t0__5a0 ! t0__5t0 ! b_e_2t0a0 ! t0__5t0 +x252 t0__64a0 ! t0__t0ma0 ! b_e_shift ! r__r_4t0 +x253 t0__64a0 ! t0__t0ma0 ! b_e_t0 ! t0__4t0_a0 +x254 t0__128a0 ! t0__t0ma0 ! b_e_shift ! r__r_2t0 +x255 t0__256a0 ! a1_ne_0_b_l0 ! t0__t0ma0 ! b_n_ret_t0 +;1040 insts before this. +ret_t0 MILLIRET +e_t0 r__r_t0 +e_shift a1_ne_0_b_l2 + a0__256a0 ; a0 <<= 8 *********** + MILLIRETN +e_t0ma0 a1_ne_0_b_l0 + t0__t0ma0 + MILLIRET + r__r_t0 +e_t0a0 a1_ne_0_b_l0 + t0__t0_a0 + MILLIRET + r__r_t0 +e_t02a0 a1_ne_0_b_l0 + t0__t0_2a0 + MILLIRET + r__r_t0 +e_t04a0 a1_ne_0_b_l0 + t0__t0_4a0 + MILLIRET + r__r_t0 +e_2t0 a1_ne_0_b_l1 + r__r_2t0 + MILLIRETN +e_2t0a0 a1_ne_0_b_l0 + t0__2t0_a0 + MILLIRET + r__r_t0 +e2t04a0 t0__t0_2a0 + a1_ne_0_b_l1 + r__r_2t0 + MILLIRETN +e_3t0 a1_ne_0_b_l0 + t0__3t0 + MILLIRET + r__r_t0 +e_4t0 a1_ne_0_b_l1 + r__r_4t0 + MILLIRETN +e_4t0a0 a1_ne_0_b_l0 + t0__4t0_a0 + MILLIRET + r__r_t0 +e4t08a0 t0__t0_2a0 + a1_ne_0_b_l1 + r__r_4t0 + MILLIRETN +e_5t0 a1_ne_0_b_l0 + t0__5t0 + MILLIRET + r__r_t0 +e_8t0 a1_ne_0_b_l1 + r__r_8t0 + MILLIRETN +e_8t0a0 a1_ne_0_b_l0 + t0__8t0_a0 + MILLIRET + r__r_t0 + + .procend + .end + diff --git a/sys/lib/libkern/arch/hppa/mulU.S b/sys/lib/libkern/arch/hppa/mulU.S new file mode 100644 index 00000000000..10147a8bb42 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/mulU.S @@ -0,0 +1,131 @@ +/* $OpenBSD: mulU.S,v 1.1 1998/06/23 18:56:54 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * + * Editor: Michael J. Eager + * + * + * ROUTINE: $$mulU + * + * + * DESCRIPTION: + * + * $$mulU multiplies two single word unsigned integers, giving a single + * word result. + * + * + * INPUT REGISTERS: + * + * arg0 = Operand 1 + * arg1 = Operand 2 + * r31 == return pc + * sr0 == return space when called externally + * + * + * OUTPUT REGISTERS: + * + * arg0 = undefined + * arg1 = undefined + * ret1 = result + * + * OTHER REGISTERS AFFECTED: + * + * r1 = undefined + * + * SIDE EFFECTS: + * + * Causes a trap under the following conditions: NONE + * Changes memory at the following places: NONE + * + * PERMISSIBLE CONTEXT: + * + * Unwindable + * Does not create a stack frame + * Is usable for internal or external microcode + * + * DISCUSSION: + * + * Calls other millicode routines via r31: NONE + * Calls other millicode routines: NONE + */ + +DEFINE(tmp,r1) +DEFINE(retreg,ret1) + + .subspa $MILLICODE$ + .proc + .callinfo millicode + + .entry +$$mulU + .export $$mulU,MILLICODE + + combt,<<= arg1,arg0,$loop + add 0,0,retreg + add 0,arg1,tmp + add 0,arg0,arg1 + add 0,tmp,arg0 +$loop extru arg1,31,4,tmp + blr tmp,0 + extru arg1,27,28,arg1 + comb,=,n arg1,0,$return + b,n $shmcand + movb,tr arg0,tmp,$addtmp + comb,= arg1,0,$return +$two b $endloop + sh1add arg0,0,tmp + b $endloop + sh1add arg0,arg0,tmp + b $endloop + sh2add arg0,0,tmp + b $endloop + sh2add arg0,arg0,tmp + b $fiftnx + sh1add arg0,0,tmp + movb,tr arg0,tmp,$fiftnx + sh2add arg0,retreg,retreg + b $endloop + sh3add arg0,0,tmp + b $endloop + sh3add arg0,arg0,tmp + b $two + sh3add arg0,retreg,retreg + movb,tr arg0,tmp,$fiftnx + sh3add arg0,retreg,retreg + b $fiftnx + sh2add arg0,0,tmp + sh2add arg0,arg0,tmp + b $endloop + sh3add arg0,retreg,retreg + sh1add,tr arg0,0,tmp + sh2add arg0,arg0,tmp +$fiftnx sh1add tmp,tmp,tmp +$endloop comb,= arg1,0,$return +$addtmp add tmp,retreg,retreg +$shmcand addb,tr arg0,arg0,$loop + sh3add arg0,0,arg0 + .exit +$return MILLIRET + nop + .procend +#ifdef milliext + .origin 0x00000100 +#endif + .end diff --git a/sys/lib/libkern/arch/hppa/prefix.h b/sys/lib/libkern/arch/hppa/prefix.h new file mode 100644 index 00000000000..442f1fd2ce0 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/prefix.h @@ -0,0 +1,87 @@ +/* $OpenBSD: prefix.h,v 1.1 1998/06/23 18:56:54 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +/* + * STANDARD INCLUDE FILE FOR MILLICODE + * Every source file must include this file. + * + * Hardware General Registers + * + * Frame Offsets (millicode convention!) + * Used when calling other millicode routines. + * Stack unwinding is dependent upon these definitions. + * r31_slot .equ -20 + * sr0_slot .equ -16 + */ + +#include <machine/asm.h> + +#define DEFINE(name, value)name: .EQU value +#ifdef milliext +#ifdef PIC +#define MILLI_BE(lbl) \ + BL .+8,r1\ + ! ADDIL L%lbl-labl/**/lbl,r1\ + ! .LABEL labl/**/lbl\ + ! BE R%lbl-labl/**/lbl(sr7,r1) + +#define MILLI_BEN(lbl) \ + BL .+8,r1\ + ! ADDIL L%lbl-labl/**/lbl,r1\ + ! .LABEL labl/**/lbl\ + ! BE,N R%lbl-labl/**/lbl(sr7,r1) + +#define MILLI_BLE(lbl) \ + BL .+8,r1\ + ! ADDIL L%lbl-labl/**/lbl,r1\ + ! .LABEL labl/**/lbl \ + ! BLE R%lbl-labl/**/lbl(sr7,r1) + +#define MILLI_BLEN(lbl) \ + BL .+8,r1\ + ! ADDIL L%lbl-labl/**/lbl,r1\ + ! .LABEL labl/**/lbl\ + ! BLE,N R%lbl-labl/**/lbl(sr7,r1) +#else +#define MILLI_BE(lbl) BE lbl(sr7,r0) +#define MILLI_BEN(lbl) BE,n lbl(sr7,r0) +#define MILLI_BLE(lbl) BLE lbl(sr7,r0) +#define MILLI_BLEN(lbl) BLE,n lbl(sr7,r0) +#endif + +#define MILLIRETN BE,n 0(sr0,r31) +#define MILLIRET BE 0(sr0,r31) +#define MILLI_RETN BE,n 0(sr0,r31) +#define MILLI_RET BE 0(sr0,r31) + +#else +#define MILLI_BE(lbl) B lbl +#define MILLI_BEN(lbl) B,n lbl +#define MILLI_BLE(lbl) BL lbl,r31 +#define MILLI_BLEN(lbl) BL,n lbl,r31 +#define MILLIRETN BV,n 0(r31) +#define MILLIRET BV 0(r31) +#define MILLI_RETN BV,n 0(r31) +#define MILLI_RET BV 0(r31) +#endif +;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + .SPACE $TEXT$ + .subspa $MILLICODE$ +;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +; VERSION is used wherever ".version" can appear in a routine +;#define VERSION .version +#define VERSION ; diff --git a/sys/lib/libkern/arch/hppa/remI.S b/sys/lib/libkern/arch/hppa/remI.S new file mode 100644 index 00000000000..cc7b92f779f --- /dev/null +++ b/sys/lib/libkern/arch/hppa/remI.S @@ -0,0 +1,164 @@ +/* $OpenBSD: remI.S,v 1.1 1998/06/23 18:56:54 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * + * Editor: Michael J. Eager + * + * + * ROUTINE: $$remI + * + * DESCRIPTION: + * + * $$remI returns the remainder of the division of two signed 32-bit + * integers. The sign of the remainder is the same as the sign of + * the dividend. + * + * + * INPUT REGISTERS: + * + * arg0 == dividend + * arg1 == divisor + * r31 == return pc + * sr0 == return space when called externally + * + * OUTPUT REGISTERS: + * + * arg0 = destroyed + * arg1 = destroyed + * ret1 = remainder + * + * OTHER REGISTERS AFFECTED: + * + * r1 = undefined + * + * SIDE EFFECTS: + * + * Causes a trap under the following conditions: DIVIDE BY ZERO + * Changes memory at the following places: NONE + * + * PERMISSIBLE CONTEXT: + * + * Unwindable + * Does not create a stack frame + * Is usable for internal or external microcode + * + * DISCUSSION: + * + * Calls other millicode routines via r31: NONE + * Calls other millicode routines: NONE + */ + +DEFINE(tmp,r1) +DEFINE(retreg,ret1) + + .subspa $MILLICODE$ + .proc + .callinfo millicode + .entry +$$remI + .export $$remI,MILLICODE + addit,= 0,arg1,0 ; trap if div by zero + add,>= 0,arg0,retreg ; move dividend, if retreg < 0, + sub 0,retreg,retreg ; make it positive + sub 0,arg1, tmp ; clear carry, + ; negate the divisor + ds 0, tmp,0 ; set V-bit to the comple- + ; ment of the divisor sign + or 0,0, tmp ; clear tmp + add retreg,retreg,retreg ; shift msb bit into carry + ds tmp,arg1, tmp ; 1st divide step, if no carry + ; out, msb of quotient = 0 + addc retreg,retreg,retreg ; shift retreg with/into carry +tt1 ds tmp,arg1, tmp ; 2nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 3rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 4th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 5th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 6th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 7th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 8th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 9th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 10th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 11th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 12th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 13th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 14th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 15th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 16th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 17th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 18th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 19th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 20th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 21st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 22nd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 23rd divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 24th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 25th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 26th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 27th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 28th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 29th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 30th divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 31st divide step + addc retreg,retreg,retreg ; shift retreg with/into carry + ds tmp,arg1, tmp ; 32nd divide step, + addc retreg,retreg,retreg ; shift last bit into retreg + movb,>=,n tmp,retreg,finish ; branch if pos. tmp + add,< arg1,0,0 ; if arg1 > 0, add arg1 + add,tr tmp,arg1,retreg ; for correcting remainder tmp + sub tmp,arg1,retreg ; else add absolute value arg1 +finish add,>= arg0,0,0 ; set sign of remainder + sub 0,retreg,retreg ; to sign of dividend + .exit + MILLIRET + nop + .procend +#ifdef milliext + .origin 0x00000200 +#endif + .end diff --git a/sys/lib/libkern/arch/hppa/remU.S b/sys/lib/libkern/arch/hppa/remU.S new file mode 100644 index 00000000000..b347a6d49f0 --- /dev/null +++ b/sys/lib/libkern/arch/hppa/remU.S @@ -0,0 +1,152 @@ +/* $OpenBSD: remU.S,v 1.1 1998/06/23 18:56:54 mickey Exp $ */ + +/* + * (c) Copyright 1985 HEWLETT-PACKARD COMPANY + * + * To anyone who acknowledges that this file is provided "AS IS" + * without any express or implied warranty: + * permission to use, copy, modify, and distribute this file + * for any purpose is hereby granted without fee, provided that + * the above copyright notice and this notice appears in all + * copies, and that the name of Hewlett-Packard Company not be + * used in advertising or publicity pertaining to distribution + * of the software without specific, written prior permission. + * Hewlett-Packard Company makes no representations about the + * suitability of this software for any purpose. + */ + +#include "prefix.h" + +/* + * Author: Committee + * + * + * ROUTINE: $$remU + * + * Single precision divide for remainder with unsigned binary integers. + * + * The remainder must be dividend-(dividend/divisor)*divisor. + * Divide by zero is trapped. + * + * INPUT REGISTERS: + * arg0 == dividend + * arg1 == divisor + * r31 == return pc + * sr0 == return space when called externally + * + * OUTPUT REGISTERS: + * arg0 = undefined + * arg1 = undefined + * ret1 = remainder + * + * OTHER REGISTERS AFFECTED: + * r1 = undefined + * + * SIDE EFFECTS: + * Causes a trap under the following conditions: DIVIDE BY ZERO + * Changes memory at the following places: NONE + * + * PERMISSIBLE CONTEXT: + * Unwindable. + * Does not create a stack frame. + * Suitable for internal or external millicode. + * Assumes the special millicode register conventions. + * + * DISCUSSION: + * Calls other millicode routines using r31: NONE + * Calls other millicode routines: NONE + */ + +DEFINE(temp,r1) +DEFINE(rmndr,ret1) ; r29 + .subspa $MILLICODE$ + .export $$remU,millicode + .proc + .callinfo millicode + + .entry +$$remU + + comib,>=,n 0,arg1,special_case + sub r0,arg1,rmndr ; clear carry, negate the divisor + ds r0,rmndr,r0 ; set V-bit to 1 + add arg0,arg0,temp ; shift msb bit into carry + ds r0,arg1,rmndr ; 1st divide step, if no carry + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 2nd divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 3rd divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 4th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 5th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 6th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 7th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 8th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 9th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 10th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 11th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 12th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 13th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 14th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 15th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 16th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 17th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 18th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 19th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 20th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 21st divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 22nd divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 23rd divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 24th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 25th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 26th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 27th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 28th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 29th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 30th divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 31st divide step + addc temp,temp,temp ; shift temp with/into carry + ds rmndr,arg1,rmndr ; 32nd divide step, + comiclr,<= 0,rmndr,r0 + add rmndr,arg1,rmndr ; correction + MILLIRETN + nop +; Putting >= on the last DS and deleting COMICLR does not work! + + +special_case + addit,= 0,arg1,r0 ; trap on div by zero + sub,>>= arg0,arg1,rmndr + copy arg0,rmndr + .exit + MILLIRETN + nop + .procend + .end |