--- /dev/null
+; HP-PA 7100/7200 __mpn_addmul_1 -- Multiply a limb vector with a limb and
+; add the result to a second limb vector.
+
+; Copyright (C) 1995 Free Software Foundation, Inc.
+
+; This file is part of the GNU MP Library.
+
+; The GNU MP Library is free software; you can redistribute it and/or modify
+; it under the terms of the GNU Library General Public License as published by
+; the Free Software Foundation; either version 2 of the License, or (at your
+; option) any later version.
+
+; The GNU MP Library is distributed in the hope that it will be useful, but
+; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+; License for more details.
+
+; You should have received a copy of the GNU Library General Public License
+; along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+; the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+; MA 02111-1307, USA.
+
+; INPUT PARAMETERS
+#define res_ptr %r26
+#define s1_ptr %r25
+#define size %r24
+#define s2_limb %r23
+
+#define cylimb %r28
+#define s0 %r19
+#define s1 %r20
+#define s2 %r3
+#define s3 %r4
+#define lo0 %r21
+#define lo1 %r5
+#define lo2 %r6
+#define lo3 %r7
+#define hi0 %r22
+#define hi1 %r23 /* safe to reuse */
+#define hi2 %r29
+#define hi3 %r1
+
+ .code
+ .export __mpn_addmul_1
+__mpn_addmul_1
+ .proc
+ .callinfo frame=128,no_calls
+ .entry
+
+ ldo 128(%r30),%r30
+ stws s2_limb,-16(%r30)
+ add %r0,%r0,cylimb ; clear cy and cylimb
+ addib,< -4,size,L$few_limbs
+ fldws -16(%r30),%fr31R
+
+ ldo -112(%r30),%r31
+ stw %r3,-96(%r30)
+ stw %r4,-92(%r30)
+ stw %r5,-88(%r30)
+ stw %r6,-84(%r30)
+ stw %r7,-80(%r30)
+
+ bb,>=,n s1_ptr,29,L$0
+
+ fldws,ma 4(s1_ptr),%fr4
+ ldws 0(res_ptr),s0
+ xmpyu %fr4,%fr31R,%fr5
+ fstds %fr5,-16(%r31)
+ ldws -16(%r31),cylimb
+ ldws -12(%r31),lo0
+ add s0,lo0,s0
+ addib,< -1,size,L$few_limbs
+ stws,ma s0,4(res_ptr)
+
+; start software pipeline ----------------------------------------------------
+L$0 fldds,ma 8(s1_ptr),%fr4
+ fldds,ma 8(s1_ptr),%fr8
+
+ xmpyu %fr4L,%fr31R,%fr5
+ xmpyu %fr4R,%fr31R,%fr6
+ xmpyu %fr8L,%fr31R,%fr9
+ xmpyu %fr8R,%fr31R,%fr10
+
+ fstds %fr5,-16(%r31)
+ fstds %fr6,-8(%r31)
+ fstds %fr9,0(%r31)
+ fstds %fr10,8(%r31)
+
+ ldws -16(%r31),hi0
+ ldws -12(%r31),lo0
+ ldws -8(%r31),hi1
+ ldws -4(%r31),lo1
+ ldws 0(%r31),hi2
+ ldws 4(%r31),lo2
+ ldws 8(%r31),hi3
+ ldws 12(%r31),lo3
+
+ addc lo0,cylimb,lo0
+ addc lo1,hi0,lo1
+ addc lo2,hi1,lo2
+ addc lo3,hi2,lo3
+
+ addib,< -4,size,L$end
+ addc %r0,hi3,cylimb ; propagate carry into cylimb
+; main loop ------------------------------------------------------------------
+L$loop fldds,ma 8(s1_ptr),%fr4
+ fldds,ma 8(s1_ptr),%fr8
+
+ ldws 0(res_ptr),s0
+ xmpyu %fr4L,%fr31R,%fr5
+ ldws 4(res_ptr),s1
+ xmpyu %fr4R,%fr31R,%fr6
+ ldws 8(res_ptr),s2
+ xmpyu %fr8L,%fr31R,%fr9
+ ldws 12(res_ptr),s3
+ xmpyu %fr8R,%fr31R,%fr10
+
+ fstds %fr5,-16(%r31)
+ add s0,lo0,s0
+ fstds %fr6,-8(%r31)
+ addc s1,lo1,s1
+ fstds %fr9,0(%r31)
+ addc s2,lo2,s2
+ fstds %fr10,8(%r31)
+ addc s3,lo3,s3
+
+ ldws -16(%r31),hi0
+ ldws -12(%r31),lo0
+ ldws -8(%r31),hi1
+ ldws -4(%r31),lo1
+ ldws 0(%r31),hi2
+ ldws 4(%r31),lo2
+ ldws 8(%r31),hi3
+ ldws 12(%r31),lo3
+
+ addc lo0,cylimb,lo0
+ stws,ma s0,4(res_ptr)
+ addc lo1,hi0,lo1
+ stws,ma s1,4(res_ptr)
+ addc lo2,hi1,lo2
+ stws,ma s2,4(res_ptr)
+ addc lo3,hi2,lo3
+ stws,ma s3,4(res_ptr)
+
+ addib,>= -4,size,L$loop
+ addc %r0,hi3,cylimb ; propagate carry into cylimb
+; finish software pipeline ---------------------------------------------------
+L$end ldws 0(res_ptr),s0
+ ldws 4(res_ptr),s1
+ ldws 8(res_ptr),s2
+ ldws 12(res_ptr),s3
+
+ add s0,lo0,s0
+ stws,ma s0,4(res_ptr)
+ addc s1,lo1,s1
+ stws,ma s1,4(res_ptr)
+ addc s2,lo2,s2
+ stws,ma s2,4(res_ptr)
+ addc s3,lo3,s3
+ stws,ma s3,4(res_ptr)
+
+; restore callee-saves registers ---------------------------------------------
+ ldw -96(%r30),%r3
+ ldw -92(%r30),%r4
+ ldw -88(%r30),%r5
+ ldw -84(%r30),%r6
+ ldw -80(%r30),%r7
+
+L$few_limbs
+ addib,=,n 4,size,L$ret
+L$loop2 fldws,ma 4(s1_ptr),%fr4
+ ldws 0(res_ptr),s0
+ xmpyu %fr4,%fr31R,%fr5
+ fstds %fr5,-16(%r30)
+ ldws -16(%r30),hi0
+ ldws -12(%r30),lo0
+ addc lo0,cylimb,lo0
+ addc %r0,hi0,cylimb
+ add s0,lo0,s0
+ stws,ma s0,4(res_ptr)
+ addib,<> -1,size,L$loop2
+ nop
+
+L$ret addc %r0,cylimb,cylimb
+ bv 0(%r2)
+ ldo -128(%r30),%r30
+
+ .exit
+ .procend
--- /dev/null
+; HP-PA 7100/7200 __mpn_submul_1 -- Multiply a limb vector with a limb and
+; subtract the result from a second limb vector.
+
+; Copyright (C) 1995 Free Software Foundation, Inc.
+
+; This file is part of the GNU MP Library.
+
+; The GNU MP Library is free software; you can redistribute it and/or modify
+; it under the terms of the GNU Library General Public License as published by
+; the Free Software Foundation; either version 2 of the License, or (at your
+; option) any later version.
+
+; The GNU MP Library is distributed in the hope that it will be useful, but
+; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+; License for more details.
+
+; You should have received a copy of the GNU Library General Public License
+; along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+; the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+; MA 02111-1307, USA.
+
+; INPUT PARAMETERS
+#define res_ptr %r26
+#define s1_ptr %r25
+#define size %r24
+#define s2_limb %r23
+
+#define cylimb %r28
+#define s0 %r19
+#define s1 %r20
+#define s2 %r3
+#define s3 %r4
+#define lo0 %r21
+#define lo1 %r5
+#define lo2 %r6
+#define lo3 %r7
+#define hi0 %r22
+#define hi1 %r23 /* safe to reuse */
+#define hi2 %r29
+#define hi3 %r1
+
+ .code
+ .export __mpn_submul_1
+__mpn_submul_1
+ .proc
+ .callinfo frame=128,no_calls
+ .entry
+
+ ldo 128(%r30),%r30
+ stws s2_limb,-16(%r30)
+ add %r0,%r0,cylimb ; clear cy and cylimb
+ addib,< -4,size,L$few_limbs
+ fldws -16(%r30),%fr31R
+
+ ldo -112(%r30),%r31
+ stw %r3,-96(%r30)
+ stw %r4,-92(%r30)
+ stw %r5,-88(%r30)
+ stw %r6,-84(%r30)
+ stw %r7,-80(%r30)
+
+ bb,>=,n s1_ptr,29,L$0
+
+ fldws,ma 4(s1_ptr),%fr4
+ ldws 0(res_ptr),s0
+ xmpyu %fr4,%fr31R,%fr5
+ fstds %fr5,-16(%r31)
+ ldws -16(%r31),cylimb
+ ldws -12(%r31),lo0
+ sub s0,lo0,s0
+ add s0,lo0,%r0 ; invert cy
+ addib,< -1,size,L$few_limbs
+ stws,ma s0,4(res_ptr)
+
+; start software pipeline ----------------------------------------------------
+L$0 fldds,ma 8(s1_ptr),%fr4
+ fldds,ma 8(s1_ptr),%fr8
+
+ xmpyu %fr4L,%fr31R,%fr5
+ xmpyu %fr4R,%fr31R,%fr6
+ xmpyu %fr8L,%fr31R,%fr9
+ xmpyu %fr8R,%fr31R,%fr10
+
+ fstds %fr5,-16(%r31)
+ fstds %fr6,-8(%r31)
+ fstds %fr9,0(%r31)
+ fstds %fr10,8(%r31)
+
+ ldws -16(%r31),hi0
+ ldws -12(%r31),lo0
+ ldws -8(%r31),hi1
+ ldws -4(%r31),lo1
+ ldws 0(%r31),hi2
+ ldws 4(%r31),lo2
+ ldws 8(%r31),hi3
+ ldws 12(%r31),lo3
+
+ addc lo0,cylimb,lo0
+ addc lo1,hi0,lo1
+ addc lo2,hi1,lo2
+ addc lo3,hi2,lo3
+
+ addib,< -4,size,L$end
+ addc %r0,hi3,cylimb ; propagate carry into cylimb
+; main loop ------------------------------------------------------------------
+L$loop fldds,ma 8(s1_ptr),%fr4
+ fldds,ma 8(s1_ptr),%fr8
+
+ ldws 0(res_ptr),s0
+ xmpyu %fr4L,%fr31R,%fr5
+ ldws 4(res_ptr),s1
+ xmpyu %fr4R,%fr31R,%fr6
+ ldws 8(res_ptr),s2
+ xmpyu %fr8L,%fr31R,%fr9
+ ldws 12(res_ptr),s3
+ xmpyu %fr8R,%fr31R,%fr10
+
+ fstds %fr5,-16(%r31)
+ sub s0,lo0,s0
+ fstds %fr6,-8(%r31)
+ subb s1,lo1,s1
+ fstds %fr9,0(%r31)
+ subb s2,lo2,s2
+ fstds %fr10,8(%r31)
+ subb s3,lo3,s3
+ subb %r0,%r0,lo0 ; these two insns ...
+ add lo0,lo0,%r0 ; ... just invert cy
+
+ ldws -16(%r31),hi0
+ ldws -12(%r31),lo0
+ ldws -8(%r31),hi1
+ ldws -4(%r31),lo1
+ ldws 0(%r31),hi2
+ ldws 4(%r31),lo2
+ ldws 8(%r31),hi3
+ ldws 12(%r31),lo3
+
+ addc lo0,cylimb,lo0
+ stws,ma s0,4(res_ptr)
+ addc lo1,hi0,lo1
+ stws,ma s1,4(res_ptr)
+ addc lo2,hi1,lo2
+ stws,ma s2,4(res_ptr)
+ addc lo3,hi2,lo3
+ stws,ma s3,4(res_ptr)
+
+ addib,>= -4,size,L$loop
+ addc %r0,hi3,cylimb ; propagate carry into cylimb
+; finish software pipeline ---------------------------------------------------
+L$end ldws 0(res_ptr),s0
+ ldws 4(res_ptr),s1
+ ldws 8(res_ptr),s2
+ ldws 12(res_ptr),s3
+
+ sub s0,lo0,s0
+ stws,ma s0,4(res_ptr)
+ subb s1,lo1,s1
+ stws,ma s1,4(res_ptr)
+ subb s2,lo2,s2
+ stws,ma s2,4(res_ptr)
+ subb s3,lo3,s3
+ stws,ma s3,4(res_ptr)
+ subb %r0,%r0,lo0 ; these two insns ...
+ add lo0,lo0,%r0 ; ... invert cy
+
+; restore callee-saves registers ---------------------------------------------
+ ldw -96(%r30),%r3
+ ldw -92(%r30),%r4
+ ldw -88(%r30),%r5
+ ldw -84(%r30),%r6
+ ldw -80(%r30),%r7
+
+L$few_limbs
+ addib,=,n 4,size,L$ret
+L$loop2 fldws,ma 4(s1_ptr),%fr4
+ ldws 0(res_ptr),s0
+ xmpyu %fr4,%fr31R,%fr5
+ fstds %fr5,-16(%r30)
+ ldws -16(%r30),hi0
+ ldws -12(%r30),lo0
+ addc lo0,cylimb,lo0
+ addc %r0,hi0,cylimb
+ sub s0,lo0,s0
+ add s0,lo0,%r0 ; invert cy
+ stws,ma s0,4(res_ptr)
+ addib,<> -1,size,L$loop2
+ nop
+
+L$ret addc %r0,cylimb,cylimb
+ bv 0(%r2)
+ ldo -128(%r30),%r30
+
+ .exit
+ .procend
--- /dev/null
+This directory contains mpn functions for Intel i960 processors.
+
+RELEVANT OPTIMIZATION ISSUES
+
+The code in this directory is not well optimized.
+
+STATUS
+
+The code in this directory has not been tested.
--- /dev/null
+/* mc68020 __mpn_add_n -- Add two limb vectors of the same length > 0 and store
+ sum in a third limb vector.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s1_ptr (sp + 8)
+ s2_ptr (sp + 16)
+ size (sp + 12)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_add_n)
+
+C_SYMBOL_NAME(__mpn_add_n:)
+PROLOG(__mpn_add_n)
+/* Save used registers on the stack. */
+ movel R(d2),MEM_PREDEC(sp)
+ movel R(a2),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. Better use movem? */
+ movel MEM_DISP(sp,12),R(a2)
+ movel MEM_DISP(sp,16),R(a0)
+ movel MEM_DISP(sp,20),R(a1)
+ movel MEM_DISP(sp,24),R(d2)
+
+ eorw #1,R(d2)
+ lsrl #1,R(d2)
+ bcc L(L1)
+ subql #1,R(d2) /* clears cy as side effect */
+
+L(Loop:)
+ movel MEM_POSTINC(a0),R(d0)
+ movel MEM_POSTINC(a1),R(d1)
+ addxl R(d1),R(d0)
+ movel R(d0),MEM_POSTINC(a2)
+L(L1:) movel MEM_POSTINC(a0),R(d0)
+ movel MEM_POSTINC(a1),R(d1)
+ addxl R(d1),R(d0)
+ movel R(d0),MEM_POSTINC(a2)
+
+ dbf R(d2),L(Loop) /* loop until 16 lsb of %4 == -1 */
+ subxl R(d0),R(d0) /* d0 <= -cy; save cy as 0 or -1 in d0 */
+ subl #0x10000,R(d2)
+ bcs L(L2)
+ addl R(d0),R(d0) /* restore cy */
+ bra L(Loop)
+
+L(L2:)
+ negl R(d0)
+
+/* Restore used registers from stack frame. */
+ movel MEM_POSTINC(sp),R(a2)
+ movel MEM_POSTINC(sp),R(d2)
+
+ rts
+EPILOG(__mpn_add_n)
--- /dev/null
+/* mc68020 __mpn_lshift -- Shift left a low-level natural-number integer.
+
+Copyright (C) 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s_ptr (sp + 8)
+ s_size (sp + 16)
+ cnt (sp + 12)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+#define res_ptr a1
+#define s_ptr a0
+#define s_size d6
+#define cnt d4
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_lshift)
+
+C_SYMBOL_NAME(__mpn_lshift:)
+PROLOG(__mpn_lshift)
+
+/* Save used registers on the stack. */
+ moveml R(d2)-R(d6)/R(a2),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. */
+ movel MEM_DISP(sp,28),R(res_ptr)
+ movel MEM_DISP(sp,32),R(s_ptr)
+ movel MEM_DISP(sp,36),R(s_size)
+ movel MEM_DISP(sp,40),R(cnt)
+
+ moveql #1,R(d5)
+ cmpl R(d5),R(cnt)
+ bne L(Lnormal)
+ cmpl R(s_ptr),R(res_ptr)
+ bls L(Lspecial) /* jump if s_ptr >= res_ptr */
+#if (defined (__mc68020__) || defined (__NeXT__) || defined(mc68020))
+ lea MEM_INDX1(s_ptr,s_size,l,4),R(a2)
+#else /* not mc68020 */
+ movel R(s_size),R(d0)
+ asll #2,R(d0)
+ lea MEM_INDX(s_ptr,d0,l),R(a2)
+#endif
+ cmpl R(res_ptr),R(a2)
+ bls L(Lspecial) /* jump if res_ptr >= s_ptr + s_size */
+
+L(Lnormal:)
+ moveql #32,R(d5)
+ subl R(cnt),R(d5)
+
+#if (defined (__mc68020__) || defined (__NeXT__) || defined(mc68020))
+ lea MEM_INDX1(s_ptr,s_size,l,4),R(s_ptr)
+ lea MEM_INDX1(res_ptr,s_size,l,4),R(res_ptr)
+#else /* not mc68000 */
+ movel R(s_size),R(d0)
+ asll #2,R(d0)
+ addl R(s_size),R(s_ptr)
+ addl R(s_size),R(res_ptr)
+#endif
+ movel MEM_PREDEC(s_ptr),R(d2)
+ movel R(d2),R(d0)
+ lsrl R(d5),R(d0) /* compute carry limb */
+
+ lsll R(cnt),R(d2)
+ movel R(d2),R(d1)
+ subql #1,R(s_size)
+ beq L(Lend)
+ lsrl #1,R(s_size)
+ bcs L(L1)
+ subql #1,R(s_size)
+
+L(Loop:)
+ movel MEM_PREDEC(s_ptr),R(d2)
+ movel R(d2),R(d3)
+ lsrl R(d5),R(d3)
+ orl R(d3),R(d1)
+ movel R(d1),MEM_PREDEC(res_ptr)
+ lsll R(cnt),R(d2)
+L(L1:)
+ movel MEM_PREDEC(s_ptr),R(d1)
+ movel R(d1),R(d3)
+ lsrl R(d5),R(d3)
+ orl R(d3),R(d2)
+ movel R(d2),MEM_PREDEC(res_ptr)
+ lsll R(cnt),R(d1)
+
+ dbf R(s_size),L(Loop)
+ subl #0x10000,R(s_size)
+ bcc L(Loop)
+
+L(Lend:)
+ movel R(d1),MEM_PREDEC(res_ptr) /* store least significant limb */
+
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d6)/R(a2)
+ rts
+
+/* We loop from least significant end of the arrays, which is only
+ permissable if the source and destination don't overlap, since the
+ function is documented to work for overlapping source and destination. */
+
+L(Lspecial:)
+ clrl R(d0) /* initialize carry */
+ eorw #1,R(s_size)
+ lsrl #1,R(s_size)
+ bcc L(LL1)
+ subql #1,R(s_size)
+
+L(LLoop:)
+ movel MEM_POSTINC(s_ptr),R(d2)
+ addxl R(d2),R(d2)
+ movel R(d2),MEM_POSTINC(res_ptr)
+L(LL1:)
+ movel MEM_POSTINC(s_ptr),R(d2)
+ addxl R(d2),R(d2)
+ movel R(d2),MEM_POSTINC(res_ptr)
+
+ dbf R(s_size),L(LLoop)
+ addxl R(d0),R(d0) /* save cy in lsb */
+ subl #0x10000,R(s_size)
+ bcs L(LLend)
+ lsrl #1,R(d0) /* restore cy */
+ bra L(LLoop)
+
+L(LLend:)
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d6)/R(a2)
+ rts
+EPILOG(__mpn_lshift)
--- /dev/null
+/* mc68020 __mpn_addmul_1 -- Multiply a limb vector with a limb and add
+ the result to a second limb vector.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s1_ptr (sp + 8)
+ s1_size (sp + 12)
+ s2_limb (sp + 16)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_addmul_1)
+
+C_SYMBOL_NAME(__mpn_addmul_1:)
+PROLOG(__mpn_addmul_1)
+
+#define res_ptr a0
+#define s1_ptr a1
+#define s1_size d2
+#define s2_limb d4
+
+/* Save used registers on the stack. */
+ moveml R(d2)-R(d5),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. Better use movem? */
+ movel MEM_DISP(sp,20),R(res_ptr)
+ movel MEM_DISP(sp,24),R(s1_ptr)
+ movel MEM_DISP(sp,28),R(s1_size)
+ movel MEM_DISP(sp,32),R(s2_limb)
+
+ eorw #1,R(s1_size)
+ clrl R(d1)
+ clrl R(d5)
+ lsrl #1,R(s1_size)
+ bcc L(L1)
+ subql #1,R(s1_size)
+ subl R(d0),R(d0) /* (d0,cy) <= (0,0) */
+
+L(Loop:)
+ movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d1):R(d3)
+ addxl R(d0),R(d3)
+ addxl R(d5),R(d1)
+ addl R(d3),MEM_POSTINC(res_ptr)
+L(L1:) movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d0):R(d3)
+ addxl R(d1),R(d3)
+ addxl R(d5),R(d0)
+ addl R(d3),MEM_POSTINC(res_ptr)
+
+ dbf R(s1_size),L(Loop)
+ addxl R(d5),R(d0)
+ subl #0x10000,R(s1_size)
+ bcc L(Loop)
+
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d5)
+
+ rts
+EPILOG(__mpn_addmul_1)
--- /dev/null
+/* mc68020 __mpn_mul_1 -- Multiply a limb vector with a limb and store
+ the result in a second limb vector.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s1_ptr (sp + 8)
+ s1_size (sp + 12)
+ s2_limb (sp + 16)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_mul_1)
+
+C_SYMBOL_NAME(__mpn_mul_1:)
+PROLOG(__mpn_mul_1)
+
+#define res_ptr a0
+#define s1_ptr a1
+#define s1_size d2
+#define s2_limb d4
+
+/* Save used registers on the stack. */
+ moveml R(d2)-R(d4),MEM_PREDEC(sp)
+#if 0
+ movel R(d2),MEM_PREDEC(sp)
+ movel R(d3),MEM_PREDEC(sp)
+ movel R(d4),MEM_PREDEC(sp)
+#endif
+
+/* Copy the arguments to registers. Better use movem? */
+ movel MEM_DISP(sp,16),R(res_ptr)
+ movel MEM_DISP(sp,20),R(s1_ptr)
+ movel MEM_DISP(sp,24),R(s1_size)
+ movel MEM_DISP(sp,28),R(s2_limb)
+
+ eorw #1,R(s1_size)
+ clrl R(d1)
+ lsrl #1,R(s1_size)
+ bcc L(L1)
+ subql #1,R(s1_size)
+ subl R(d0),R(d0) /* (d0,cy) <= (0,0) */
+
+L(Loop:)
+ movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d1):R(d3)
+ addxl R(d0),R(d3)
+ movel R(d3),MEM_POSTINC(res_ptr)
+L(L1:) movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d0):R(d3)
+ addxl R(d1),R(d3)
+ movel R(d3),MEM_POSTINC(res_ptr)
+
+ dbf R(s1_size),L(Loop)
+ clrl R(d3)
+ addxl R(d3),R(d0)
+ subl #0x10000,R(s1_size)
+ bcc L(Loop)
+
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d4)
+#if 0
+ movel MEM_POSTINC(sp),R(d4)
+ movel MEM_POSTINC(sp),R(d3)
+ movel MEM_POSTINC(sp),R(d2)
+#endif
+ rts
+EPILOG(__mpn_mul_1)
--- /dev/null
+/* mc68020 __mpn_submul_1 -- Multiply a limb vector with a limb and subtract
+ the result from a second limb vector.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s1_ptr (sp + 8)
+ s1_size (sp + 12)
+ s2_limb (sp + 16)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_submul_1)
+
+C_SYMBOL_NAME(__mpn_submul_1:)
+PROLOG(__mpn_submul_1)
+
+#define res_ptr a0
+#define s1_ptr a1
+#define s1_size d2
+#define s2_limb d4
+
+/* Save used registers on the stack. */
+ moveml R(d2)-R(d5),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. Better use movem? */
+ movel MEM_DISP(sp,20),R(res_ptr)
+ movel MEM_DISP(sp,24),R(s1_ptr)
+ movel MEM_DISP(sp,28),R(s1_size)
+ movel MEM_DISP(sp,32),R(s2_limb)
+
+ eorw #1,R(s1_size)
+ clrl R(d1)
+ clrl R(d5)
+ lsrl #1,R(s1_size)
+ bcc L(L1)
+ subql #1,R(s1_size)
+ subl R(d0),R(d0) /* (d0,cy) <= (0,0) */
+
+L(Loop:)
+ movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d1):R(d3)
+ addxl R(d0),R(d3)
+ addxl R(d5),R(d1)
+ subl R(d3),MEM_POSTINC(res_ptr)
+L(L1:) movel MEM_POSTINC(s1_ptr),R(d3)
+ mulul R(s2_limb),R(d0):R(d3)
+ addxl R(d1),R(d3)
+ addxl R(d5),R(d0)
+ subl R(d3),MEM_POSTINC(res_ptr)
+
+ dbf R(s1_size),L(Loop)
+ addxl R(d5),R(d0)
+ subl #0x10000,R(s1_size)
+ bcc L(Loop)
+
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d5)
+
+ rts
+EPILOG(__mpn_submul_1)
--- /dev/null
+/* mc68020 __mpn_rshift -- Shift right a low-level natural-number integer.
+
+Copyright (C) 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s_ptr (sp + 8)
+ s_size (sp + 16)
+ cnt (sp + 12)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+#define res_ptr a1
+#define s_ptr a0
+#define s_size d6
+#define cnt d4
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_rshift)
+
+C_SYMBOL_NAME(__mpn_rshift:)
+PROLOG(__mpn_rshift)
+/* Save used registers on the stack. */
+ moveml R(d2)-R(d6)/R(a2),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. */
+ movel MEM_DISP(sp,28),R(res_ptr)
+ movel MEM_DISP(sp,32),R(s_ptr)
+ movel MEM_DISP(sp,36),R(s_size)
+ movel MEM_DISP(sp,40),R(cnt)
+
+ moveql #1,R(d5)
+ cmpl R(d5),R(cnt)
+ bne L(Lnormal)
+ cmpl R(res_ptr),R(s_ptr)
+ bls L(Lspecial) /* jump if res_ptr >= s_ptr */
+#if (defined (__mc68020__) || defined (__NeXT__) || defined(mc68020))
+ lea MEM_INDX1(res_ptr,s_size,l,4),R(a2)
+#else /* not mc68020 */
+ movel R(s_size),R(d0)
+ asll #2,R(d0)
+ lea MEM_INDX(res_ptr,d0,l),R(a2)
+#endif
+ cmpl R(s_ptr),R(a2)
+ bls L(Lspecial) /* jump if s_ptr >= res_ptr + s_size */
+
+L(Lnormal:)
+ moveql #32,R(d5)
+ subl R(cnt),R(d5)
+ movel MEM_POSTINC(s_ptr),R(d2)
+ movel R(d2),R(d0)
+ lsll R(d5),R(d0) /* compute carry limb */
+
+ lsrl R(cnt),R(d2)
+ movel R(d2),R(d1)
+ subql #1,R(s_size)
+ beq L(Lend)
+ lsrl #1,R(s_size)
+ bcs L(L1)
+ subql #1,R(s_size)
+
+L(Loop:)
+ movel MEM_POSTINC(s_ptr),R(d2)
+ movel R(d2),R(d3)
+ lsll R(d5),R(d3)
+ orl R(d3),R(d1)
+ movel R(d1),MEM_POSTINC(res_ptr)
+ lsrl R(cnt),R(d2)
+L(L1:)
+ movel MEM_POSTINC(s_ptr),R(d1)
+ movel R(d1),R(d3)
+ lsll R(d5),R(d3)
+ orl R(d3),R(d2)
+ movel R(d2),MEM_POSTINC(res_ptr)
+ lsrl R(cnt),R(d1)
+
+ dbf R(s_size),L(Loop)
+ subl #0x10000,R(s_size)
+ bcc L(Loop)
+
+L(Lend:)
+ movel R(d1),MEM(res_ptr) /* store most significant limb */
+
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d6)/R(a2)
+ rts
+
+/* We loop from most significant end of the arrays, which is only
+ permissable if the source and destination don't overlap, since the
+ function is documented to work for overlapping source and destination. */
+
+L(Lspecial:)
+#if (defined (__mc68020__) || defined (__NeXT__) || defined(mc68020))
+ lea MEM_INDX1(s_ptr,s_size,l,4),R(s_ptr)
+ lea MEM_INDX1(res_ptr,s_size,l,4),R(res_ptr)
+#else /* not mc68000 */
+ movel R(s_size),R(d0)
+ asll #2,R(d0)
+ addl R(s_size),R(s_ptr)
+ addl R(s_size),R(res_ptr)
+#endif
+
+ clrl R(d0) /* initialize carry */
+ eorw #1,R(s_size)
+ lsrl #1,R(s_size)
+ bcc L(LL1)
+ subql #1,R(s_size)
+
+L(LLoop:)
+ movel MEM_PREDEC(s_ptr),R(d2)
+ roxrl #1,R(d2)
+ movel R(d2),MEM_PREDEC(res_ptr)
+L(LL1:)
+ movel MEM_PREDEC(s_ptr),R(d2)
+ roxrl #1,R(d2)
+ movel R(d2),MEM_PREDEC(res_ptr)
+
+ dbf R(s_size),L(LLoop)
+ roxrl #1,R(d0) /* save cy in msb */
+ subl #0x10000,R(s_size)
+ bcs L(LLend)
+ addl R(d0),R(d0) /* restore cy */
+ bra L(LLoop)
+
+L(LLend:)
+/* Restore used registers from stack frame. */
+ moveml MEM_POSTINC(sp),R(d2)-R(d6)/R(a2)
+ rts
+EPILOG(__mpn_rshift)
--- /dev/null
+/* mc68020 __mpn_sub_n -- Subtract two limb vectors of the same length > 0 and
+ store difference in a third limb vector.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+/*
+ INPUT PARAMETERS
+ res_ptr (sp + 4)
+ s1_ptr (sp + 8)
+ s2_ptr (sp + 16)
+ size (sp + 12)
+*/
+
+#include "sysdep.h"
+#include "asm-syntax.h"
+
+ TEXT
+ ALIGN
+ GLOBL C_SYMBOL_NAME(__mpn_sub_n)
+
+C_SYMBOL_NAME(__mpn_sub_n:)
+PROLOG(__mpn_sub_n)
+/* Save used registers on the stack. */
+ movel R(d2),MEM_PREDEC(sp)
+ movel R(a2),MEM_PREDEC(sp)
+
+/* Copy the arguments to registers. Better use movem? */
+ movel MEM_DISP(sp,12),R(a2)
+ movel MEM_DISP(sp,16),R(a0)
+ movel MEM_DISP(sp,20),R(a1)
+ movel MEM_DISP(sp,24),R(d2)
+
+ eorw #1,R(d2)
+ lsrl #1,R(d2)
+ bcc L(L1)
+ subql #1,R(d2) /* clears cy as side effect */
+
+L(Loop:)
+ movel MEM_POSTINC(a0),R(d0)
+ movel MEM_POSTINC(a1),R(d1)
+ subxl R(d1),R(d0)
+ movel R(d0),MEM_POSTINC(a2)
+L(L1:) movel MEM_POSTINC(a0),R(d0)
+ movel MEM_POSTINC(a1),R(d1)
+ subxl R(d1),R(d0)
+ movel R(d0),MEM_POSTINC(a2)
+
+ dbf R(d2),L(Loop) /* loop until 16 lsb of %4 == -1 */
+ subxl R(d0),R(d0) /* d0 <= -cy; save cy as 0 or -1 in d0 */
+ subl #0x10000,R(d2)
+ bcs L(L2)
+ addl R(d0),R(d0) /* restore cy */
+ bra L(Loop)
+
+L(L2:)
+ negl R(d0)
+
+/* Restore used registers from stack frame. */
+ movel MEM_POSTINC(sp),R(a2)
+ movel MEM_POSTINC(sp),R(d2)
+
+ rts
+EPILOG(__mpn_sub_n)
--- /dev/null
+/* asm.h -- Definitions for 68k syntax variations.
+
+Copyright (C) 1992, 1994, 1996 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+#undef ALIGN
+
+#ifdef MIT_SYNTAX
+#define PROLOG(name)
+#define EPILOG(name)
+#define R(r)r
+#define MEM(base)base@
+#define MEM_DISP(base,displacement)base@(displacement)
+#define MEM_INDX(base,idx,size_suffix)base@(idx:size_suffix)
+#define MEM_INDX1(base,idx,size_suffix,scale)base@(idx:size_suffix:scale)
+#define MEM_PREDEC(memory_base)memory_base@-
+#define MEM_POSTINC(memory_base)memory_base@+
+#define L(label) label
+#define TEXT .text
+#define ALIGN .even
+#define GLOBL .globl
+#define moveql moveq
+/* Use variable sized opcodes. */
+#define bcc jcc
+#define bcs jcs
+#define bls jls
+#define beq jeq
+#define bne jne
+#define bra jra
+#endif
+
+#ifdef SONY_SYNTAX
+#define PROLOG(name)
+#define EPILOG(name)
+#define R(r)r
+#define MEM(base)(base)
+#define MEM_DISP(base,displacement)(displacement,base)
+#define MEM_INDX(base,idx,size_suffix)(base,idx.size_suffix)
+#define MEM_INDX1(base,idx,size_suffix,scale)(base,idx.size_suffix*scale)
+#define MEM_PREDEC(memory_base)-(memory_base)
+#define MEM_POSTINC(memory_base)(memory_base)+
+#define L(label) label
+#define TEXT .text
+#define ALIGN .even
+#define GLOBL .globl
+#endif
+
+#ifdef MOTOROLA_SYNTAX
+#define PROLOG(name)
+#define EPILOG(name)
+#define R(r)r
+#define MEM(base)(base)
+#define MEM_DISP(base,displacement)(displacement,base)
+#define MEM_INDX(base,idx,size_suffix)(base,idx.size_suffix)
+#define MEM_INDX1(base,idx,size_suffix,scale)(base,idx.size_suffix*scale)
+#define MEM_PREDEC(memory_base)-(memory_base)
+#define MEM_POSTINC(memory_base)(memory_base)+
+#define L(label) label
+#define TEXT
+#define ALIGN
+#define GLOBL XDEF
+#define lea LEA
+#define movel MOVE.L
+#define moveml MOVEM.L
+#define moveql MOVEQ.L
+#define cmpl CMP.L
+#define orl OR.L
+#define clrl CLR.L
+#define eorw EOR.W
+#define lsrl LSR.L
+#define lsll LSL.L
+#define roxrl ROXR.L
+#define roxll ROXL.L
+#define addl ADD.L
+#define addxl ADDX.L
+#define addql ADDQ.L
+#define subl SUB.L
+#define subxl SUBX.L
+#define subql SUBQ.L
+#define negl NEG.L
+#define mulul MULU.L
+#define bcc BCC
+#define bcs BCS
+#define bls BLS
+#define beq BEQ
+#define bne BNE
+#define bra BRA
+#define dbf DBF
+#define rts RTS
+#define d0 D0
+#define d1 D1
+#define d2 D2
+#define d3 D3
+#define d4 D4
+#define d5 D5
+#define d6 D6
+#define d7 D7
+#define a0 A0
+#define a1 A1
+#define a2 A2
+#define a3 A3
+#define a4 A4
+#define a5 A5
+#define a6 A6
+#define a7 A7
+#define sp SP
+#endif
+
+#ifdef ELF_SYNTAX
+#define PROLOG(name) .type name,@function
+#define EPILOG(name) .size name,.-name
+#define MEM(base)(R(base))
+#define MEM_DISP(base,displacement)(displacement,R(base))
+#define MEM_PREDEC(memory_base)-(R(memory_base))
+#define MEM_POSTINC(memory_base)(R(memory_base))+
+#ifdef __STDC__
+#define R_(r)%##r
+#define R(r)R_(r)
+#define MEM_INDX_(base,idx,size_suffix)(R(base),R(idx##.##size_suffix))
+#define MEM_INDX(base,idx,size_suffix)MEM_INDX_(base,idx,size_suffix)
+#define MEM_INDX1_(base,idx,size_suffix,scale)(R(base),R(idx##.##size_suffix*scale))
+#define MEM_INDX1(base,idx,size_suffix,scale)MEM_INDX1_(base,idx,size_suffix,scale)
+#define L(label) .##label
+#else
+#define R(r)%/**/r
+#define MEM_INDX(base,idx,size_suffix)(R(base),R(idx).size_suffix)
+#define MEM_INDX1(base,idx,size_suffix,scale)(R(base),R(idx).size_suffix*scale)
+#define L(label) ./**/label
+#endif
+#define TEXT .text
+#define ALIGN .align 2
+#define GLOBL .globl
+#define bcc jbcc
+#define bcs jbcs
+#define bls jbls
+#define beq jbeq
+#define bne jbne
+#define bra jbra
+#endif
+
+#if defined (SONY_SYNTAX) || defined (ELF_SYNTAX)
+#define movel move.l
+#define moveml movem.l
+#define moveql moveq.l
+#define cmpl cmp.l
+#define orl or.l
+#define clrl clr.l
+#define eorw eor.w
+#define lsrl lsr.l
+#define lsll lsl.l
+#define roxrl roxr.l
+#define roxll roxl.l
+#define addl add.l
+#define addxl addx.l
+#define addql addq.l
+#define subl sub.l
+#define subxl subx.l
+#define subql subq.l
+#define negl neg.l
+#define mulul mulu.l
+#endif
--- /dev/null
+; mc88110 __mpn_add_n -- Add two limb vectors of the same length > 0 and store
+; sum in a third limb vector.
+
+; Copyright (C) 1995, 1996 Free Software Foundation, Inc.
+
+; This file is part of the GNU MP Library.
+
+; The GNU MP Library is free software; you can redistribute it and/or modify
+; it under the terms of the GNU Library General Public License as published by
+; the Free Software Foundation; either version 2 of the License, or (at your
+; option) any later version.
+
+; The GNU MP Library is distributed in the hope that it will be useful, but
+; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+; License for more details.
+
+; You should have received a copy of the GNU Library General Public License
+; along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+; the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+; MA 02111-1307, USA.
+
+
+; INPUT PARAMETERS
+#define res_ptr r2
+#define s1_ptr r3
+#define s2_ptr r4
+#define size r5
+
+#include "sysdep.h"
+
+ text
+ align 16
+ global C_SYMBOL_NAME(__mpn_add_n)
+C_SYMBOL_NAME(__mpn_add_n):
+ addu.co r0,r0,r0 ; clear cy flag
+ xor r12,s2_ptr,res_ptr
+ bb1 2,r12,L1
+; ** V1a **
+L0: bb0 2,res_ptr,L_v1 ; branch if res_ptr is aligned?
+/* Add least significant limb separately to align res_ptr and s2_ptr */
+ ld r10,s1_ptr,0
+ addu s1_ptr,s1_ptr,4
+ ld r8,s2_ptr,0
+ addu s2_ptr,s2_ptr,4
+ subu size,size,1
+ addu.co r6,r10,r8
+ st r6,res_ptr,0
+ addu res_ptr,res_ptr,4
+L_v1: cmp r12,size,2
+ bb1 lt,r12,Lend2
+
+ ld r10,s1_ptr,0
+ ld r12,s1_ptr,4
+ ld.d r8,s2_ptr,0
+ subu size,size,10
+ bcnd lt0,size,Lfin1
+/* Add blocks of 8 limbs until less than 8 limbs remain */
+ align 8
+Loop1: subu size,size,8
+ addu.cio r6,r10,r8
+ ld r10,s1_ptr,8
+ addu.cio r7,r12,r9
+ ld r12,s1_ptr,12
+ ld.d r8,s2_ptr,8
+ st.d r6,res_ptr,0
+ addu.cio r6,r10,r8
+ ld r10,s1_ptr,16
+ addu.cio r7,r12,r9
+ ld r12,s1_ptr,20
+ ld.d r8,s2_ptr,16
+ st.d r6,res_ptr,8
+ addu.cio r6,r10,r8
+ ld r10,s1_ptr,24
+ addu.cio r7,r12,r9
+ ld r12,s1_ptr,28
+ ld.d r8,s2_ptr,24
+ st.d r6,res_ptr,16
+ addu.cio r6,r10,r8
+ ld r10,s1_ptr,32
+ addu.cio r7,r12,r9
+ ld r12,s1_ptr,36
+ addu s1_ptr,s1_ptr,32
+ ld.d r8,s2_ptr,32
+ addu s2_ptr,s2_ptr,32
+ st.d r6,res_ptr,24
+ addu res_ptr,res_ptr,32
+ bcnd ge0,size,Loop1
+
+Lfin1: addu size,size,8-2
+ bcnd lt0,size,Lend1
+/* Add blocks of 2 limbs until less than 2 limbs remain */
+Loope1: addu.cio r6,r10,r8
+ ld r10,s1_ptr,8
+ addu.cio r7,r12,r9
+ ld r12,s1_ptr,12
+ ld.d r8,s2_ptr,8
+ st.d r6,res_ptr,0
+ subu size,size,2
+ addu s1_ptr,s1_ptr,8
+ addu s2_ptr,s2_ptr,8
+ addu res_ptr,res_ptr,8
+ bcnd ge0,size,Loope1
+Lend1: addu.cio r6,r10,r8
+ addu.cio r7,r12,r9
+ st.d r6,res_ptr,0
+
+ bb0 0,size,Lret1
+/* Add last limb */
+ ld r10,s1_ptr,8
+ ld r8,s2_ptr,8
+ addu.cio r6,r10,r8
+ st r6,res_ptr,8
+
+Lret1: jmp.n r1
+ addu.ci r2,r0,r0 ; return carry-out from most sign. limb
+
+L1: xor r12,s1_ptr,res_ptr
+ bb1 2,r12,L2
+; ** V1b **
+ or r12,r0,s2_ptr
+ or s2_ptr,r0,s1_ptr
+ or s1_ptr,r0,r12
+ br L0
+
+; ** V2 **
+/* If we come here, the alignment of s1_ptr and res_ptr as well as the
+ alignment of s2_ptr and res_ptr differ. Since there are only two ways
+ things can be aligned (that we care about) we now know that the alignment
+ of s1_ptr and s2_ptr are the same. */
+
+L2: cmp r12,size,1
+ bb1 eq,r12,Ljone
+ bb0 2,s1_ptr,L_v2 ; branch if s1_ptr is aligned
+/* Add least significant limb separately to align res_ptr and s2_ptr */
+ ld r10,s1_ptr,0
+ addu s1_ptr,s1_ptr,4
+ ld r8,s2_ptr,0
+ addu s2_ptr,s2_ptr,4
+ subu size,size,1
+ addu.co r6,r10,r8
+ st r6,res_ptr,0
+ addu res_ptr,res_ptr,4
+
+L_v2: subu size,size,8
+ bcnd lt0,size,Lfin2
+/* Add blocks of 8 limbs until less than 8 limbs remain */
+ align 8
+Loop2: subu size,size,8
+ ld.d r8,s1_ptr,0
+ ld.d r6,s2_ptr,0
+ addu.cio r8,r8,r6
+ st r8,res_ptr,0
+ addu.cio r9,r9,r7
+ st r9,res_ptr,4
+ ld.d r8,s1_ptr,8
+ ld.d r6,s2_ptr,8
+ addu.cio r8,r8,r6
+ st r8,res_ptr,8
+ addu.cio r9,r9,r7
+ st r9,res_ptr,12
+ ld.d r8,s1_ptr,16
+ ld.d r6,s2_ptr,16
+ addu.cio r8,r8,r6
+ st r8,res_ptr,16
+ addu.cio r9,r9,r7
+ st r9,res_ptr,20
+ ld.d r8,s1_ptr,24
+ ld.d r6,s2_ptr,24
+ addu.cio r8,r8,r6
+ st r8,res_ptr,24
+ addu.cio r9,r9,r7
+ st r9,res_ptr,28
+ addu s1_ptr,s1_ptr,32
+ addu s2_ptr,s2_ptr,32
+ addu res_ptr,res_ptr,32
+ bcnd ge0,size,Loop2
+
+Lfin2: addu size,size,8-2
+ bcnd lt0,size,Lend2
+Loope2: ld.d r8,s1_ptr,0
+ ld.d r6,s2_ptr,0
+ addu.cio r8,r8,r6
+ st r8,res_ptr,0
+ addu.cio r9,r9,r7
+ st r9,res_ptr,4
+ subu size,size,2
+ addu s1_ptr,s1_ptr,8
+ addu s2_ptr,s2_ptr,8
+ addu res_ptr,res_ptr,8
+ bcnd ge0,size,Loope2
+Lend2: bb0 0,size,Lret2
+/* Add last limb */
+Ljone: ld r10,s1_ptr,0
+ ld r8,s2_ptr,0
+ addu.cio r6,r10,r8
+ st r6,res_ptr,0
+
+Lret2: jmp.n r1
+ addu.ci r2,r0,r0 ; return carry-out from most sign. limb
--- /dev/null
+; mc88110 __mpn_sub_n -- Subtract two limb vectors of the same length > 0 and
+; store difference in a third limb vector.
+
+; Copyright (C) 1995, 1996 Free Software Foundation, Inc.
+
+; This file is part of the GNU MP Library.
+
+; The GNU MP Library is free software; you can redistribute it and/or modify
+; it under the terms of the GNU Library General Public License as published by
+; the Free Software Foundation; either version 2 of the License, or (at your
+; option) any later version.
+
+; The GNU MP Library is distributed in the hope that it will be useful, but
+; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+; License for more details.
+
+; You should have received a copy of the GNU Library General Public License
+; along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+; the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+; MA 02111-1307, USA.
+
+
+; INPUT PARAMETERS
+#define res_ptr r2
+#define s1_ptr r3
+#define s2_ptr r4
+#define size r5
+
+#include "sysdep.h"
+
+ text
+ align 16
+ global C_SYMBOL_NAME(__mpn_sub_n)
+C_SYMBOL_NAME(__mpn_sub_n):
+ subu.co r0,r0,r0 ; set cy flag
+ xor r12,s2_ptr,res_ptr
+ bb1 2,r12,L1
+; ** V1a **
+L0: bb0 2,res_ptr,L_v1 ; branch if res_ptr is aligned
+/* Add least significant limb separately to align res_ptr and s2_ptr */
+ ld r10,s1_ptr,0
+ addu s1_ptr,s1_ptr,4
+ ld r8,s2_ptr,0
+ addu s2_ptr,s2_ptr,4
+ subu size,size,1
+ subu.co r6,r10,r8
+ st r6,res_ptr,0
+ addu res_ptr,res_ptr,4
+L_v1: cmp r12,size,2
+ bb1 lt,r12,Lend2
+
+ ld r10,s1_ptr,0
+ ld r12,s1_ptr,4
+ ld.d r8,s2_ptr,0
+ subu size,size,10
+ bcnd lt0,size,Lfin1
+/* Add blocks of 8 limbs until less than 8 limbs remain */
+ align 8
+Loop1: subu size,size,8
+ subu.cio r6,r10,r8
+ ld r10,s1_ptr,8
+ subu.cio r7,r12,r9
+ ld r12,s1_ptr,12
+ ld.d r8,s2_ptr,8
+ st.d r6,res_ptr,0
+ subu.cio r6,r10,r8
+ ld r10,s1_ptr,16
+ subu.cio r7,r12,r9
+ ld r12,s1_ptr,20
+ ld.d r8,s2_ptr,16
+ st.d r6,res_ptr,8
+ subu.cio r6,r10,r8
+ ld r10,s1_ptr,24
+ subu.cio r7,r12,r9
+ ld r12,s1_ptr,28
+ ld.d r8,s2_ptr,24
+ st.d r6,res_ptr,16
+ subu.cio r6,r10,r8
+ ld r10,s1_ptr,32
+ subu.cio r7,r12,r9
+ ld r12,s1_ptr,36
+ addu s1_ptr,s1_ptr,32
+ ld.d r8,s2_ptr,32
+ addu s2_ptr,s2_ptr,32
+ st.d r6,res_ptr,24
+ addu res_ptr,res_ptr,32
+ bcnd ge0,size,Loop1
+
+Lfin1: addu size,size,8-2
+ bcnd lt0,size,Lend1
+/* Add blocks of 2 limbs until less than 2 limbs remain */
+Loope1: subu.cio r6,r10,r8
+ ld r10,s1_ptr,8
+ subu.cio r7,r12,r9
+ ld r12,s1_ptr,12
+ ld.d r8,s2_ptr,8
+ st.d r6,res_ptr,0
+ subu size,size,2
+ addu s1_ptr,s1_ptr,8
+ addu s2_ptr,s2_ptr,8
+ addu res_ptr,res_ptr,8
+ bcnd ge0,size,Loope1
+Lend1: subu.cio r6,r10,r8
+ subu.cio r7,r12,r9
+ st.d r6,res_ptr,0
+
+ bb0 0,size,Lret1
+/* Add last limb */
+ ld r10,s1_ptr,8
+ ld r8,s2_ptr,8
+ subu.cio r6,r10,r8
+ st r6,res_ptr,8
+
+Lret1: addu.ci r2,r0,r0 ; return carry-out from most sign. limb
+ jmp.n r1
+ xor r2,r2,1
+
+L1: xor r12,s1_ptr,res_ptr
+ bb1 2,r12,L2
+; ** V1b **
+ bb0 2,res_ptr,L_v1b ; branch if res_ptr is aligned
+/* Add least significant limb separately to align res_ptr and s1_ptr */
+ ld r10,s2_ptr,0
+ addu s2_ptr,s2_ptr,4
+ ld r8,s1_ptr,0
+ addu s1_ptr,s1_ptr,4
+ subu size,size,1
+ subu.co r6,r8,r10
+ st r6,res_ptr,0
+ addu res_ptr,res_ptr,4
+L_v1b: cmp r12,size,2
+ bb1 lt,r12,Lend2
+
+ ld r10,s2_ptr,0
+ ld r12,s2_ptr,4
+ ld.d r8,s1_ptr,0
+ subu size,size,10
+ bcnd lt0,size,Lfin1b
+/* Add blocks of 8 limbs until less than 8 limbs remain */
+ align 8
+Loop1b: subu size,size,8
+ subu.cio r6,r8,r10
+ ld r10,s2_ptr,8
+ subu.cio r7,r9,r12
+ ld r12,s2_ptr,12
+ ld.d r8,s1_ptr,8
+ st.d r6,res_ptr,0
+ subu.cio r6,r8,r10
+ ld r10,s2_ptr,16
+ subu.cio r7,r9,r12
+ ld r12,s2_ptr,20
+ ld.d r8,s1_ptr,16
+ st.d r6,res_ptr,8
+ subu.cio r6,r8,r10
+ ld r10,s2_ptr,24
+ subu.cio r7,r9,r12
+ ld r12,s2_ptr,28
+ ld.d r8,s1_ptr,24
+ st.d r6,res_ptr,16
+ subu.cio r6,r8,r10
+ ld r10,s2_ptr,32
+ subu.cio r7,r9,r12
+ ld r12,s2_ptr,36
+ addu s2_ptr,s2_ptr,32
+ ld.d r8,s1_ptr,32
+ addu s1_ptr,s1_ptr,32
+ st.d r6,res_ptr,24
+ addu res_ptr,res_ptr,32
+ bcnd ge0,size,Loop1b
+
+Lfin1b: addu size,size,8-2
+ bcnd lt0,size,Lend1b
+/* Add blocks of 2 limbs until less than 2 limbs remain */
+Loope1b:subu.cio r6,r8,r10
+ ld r10,s2_ptr,8
+ subu.cio r7,r9,r12
+ ld r12,s2_ptr,12
+ ld.d r8,s1_ptr,8
+ st.d r6,res_ptr,0
+ subu size,size,2
+ addu s1_ptr,s1_ptr,8
+ addu s2_ptr,s2_ptr,8
+ addu res_ptr,res_ptr,8
+ bcnd ge0,size,Loope1b
+Lend1b: subu.cio r6,r8,r10
+ subu.cio r7,r9,r12
+ st.d r6,res_ptr,0
+
+ bb0 0,size,Lret1b
+/* Add last limb */
+ ld r10,s2_ptr,8
+ ld r8,s1_ptr,8
+ subu.cio r6,r8,r10
+ st r6,res_ptr,8
+
+Lret1b: addu.ci r2,r0,r0 ; return carry-out from most sign. limb
+ jmp.n r1
+ xor r2,r2,1
+
+; ** V2 **
+/* If we come here, the alignment of s1_ptr and res_ptr as well as the
+ alignment of s2_ptr and res_ptr differ. Since there are only two ways
+ things can be aligned (that we care about) we now know that the alignment
+ of s1_ptr and s2_ptr are the same. */
+
+L2: cmp r12,size,1
+ bb1 eq,r12,Ljone
+ bb0 2,s1_ptr,L_v2 ; branch if s1_ptr is aligned
+/* Add least significant limb separately to align res_ptr and s2_ptr */
+ ld r10,s1_ptr,0
+ addu s1_ptr,s1_ptr,4
+ ld r8,s2_ptr,0
+ addu s2_ptr,s2_ptr,4
+ subu size,size,1
+ subu.co r6,r10,r8
+ st r6,res_ptr,0
+ addu res_ptr,res_ptr,4
+
+L_v2: subu size,size,8
+ bcnd lt0,size,Lfin2
+/* Add blocks of 8 limbs until less than 8 limbs remain */
+ align 8
+Loop2: subu size,size,8
+ ld.d r8,s1_ptr,0
+ ld.d r6,s2_ptr,0
+ subu.cio r8,r8,r6
+ st r8,res_ptr,0
+ subu.cio r9,r9,r7
+ st r9,res_ptr,4
+ ld.d r8,s1_ptr,8
+ ld.d r6,s2_ptr,8
+ subu.cio r8,r8,r6
+ st r8,res_ptr,8
+ subu.cio r9,r9,r7
+ st r9,res_ptr,12
+ ld.d r8,s1_ptr,16
+ ld.d r6,s2_ptr,16
+ subu.cio r8,r8,r6
+ st r8,res_ptr,16
+ subu.cio r9,r9,r7
+ st r9,res_ptr,20
+ ld.d r8,s1_ptr,24
+ ld.d r6,s2_ptr,24
+ subu.cio r8,r8,r6
+ st r8,res_ptr,24
+ subu.cio r9,r9,r7
+ st r9,res_ptr,28
+ addu s1_ptr,s1_ptr,32
+ addu s2_ptr,s2_ptr,32
+ addu res_ptr,res_ptr,32
+ bcnd ge0,size,Loop2
+
+Lfin2: addu size,size,8-2
+ bcnd lt0,size,Lend2
+Loope2: ld.d r8,s1_ptr,0
+ ld.d r6,s2_ptr,0
+ subu.cio r8,r8,r6
+ st r8,res_ptr,0
+ subu.cio r9,r9,r7
+ st r9,res_ptr,4
+ subu size,size,2
+ addu s1_ptr,s1_ptr,8
+ addu s2_ptr,s2_ptr,8
+ addu res_ptr,res_ptr,8
+ bcnd ge0,size,Loope2
+Lend2: bb0 0,size,Lret2
+/* Add last limb */
+Ljone: ld r10,s1_ptr,0
+ ld r8,s2_ptr,0
+ subu.cio r6,r10,r8
+ st r6,res_ptr,0
+
+Lret2: addu.ci r2,r0,r0 ; return carry-out from most sign. limb
+ jmp.n r1
+ xor r2,r2,1
--- /dev/null
+This directory contains mpn functions optimized for MIPS3. Example of
+processors that implement MIPS3 are R4000, R4400, R4600, R4700, and R8000.
+
+RELEVANT OPTIMIZATION ISSUES
+
+1. On the R4000 and R4400, branches, both the plain and the "likely" ones,
+ take 3 cycles to execute. (The fastest possible loop will take 4 cycles,
+ because of the delay insn.)
+
+ On the R4600, branches takes a single cycle
+
+ On the R8000, branches often take no noticable cycles, as they are
+ executed in a separate function unit..
+
+2. The R4000 and R4400 have a load latency of 4 cycles.
+
+3. On the R4000 and R4400, multiplies take a data-dependent number of
+ cycles, contrary to the SGI documentation. There seem to be 3 or 4
+ possible latencies.
+
+STATUS
+
+Good...
--- /dev/null
+/* gmp-mparam.h -- Compiler/machine parameter header file.
+
+Copyright (C) 1991, 1993, 1994 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+#define BITS_PER_MP_LIMB 64
+#define BYTES_PER_MP_LIMB 8
+#define BITS_PER_LONGINT 32
+#define BITS_PER_INT 32
+#define BITS_PER_SHORTINT 16
+#define BITS_PER_CHAR 8
--- /dev/null
+/* gmp-mparam.h -- Compiler/machine parameter header file.
+
+Copyright (C) 1991, 1993, 1994, 1995 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of the GNU Library General Public License as published by
+the Free Software Foundation; either version 2 of the License, or (at your
+option) any later version.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public
+License for more details.
+
+You should have received a copy of the GNU Library General Public License
+along with the GNU MP Library; see the file COPYING.LIB. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+MA 02111-1307, USA. */
+
+#define BITS_PER_MP_LIMB 64
+#define BYTES_PER_MP_LIMB 8
+#define BITS_PER_LONGINT 64
+#define BITS_PER_INT 32
+#define BITS_PER_SHORTINT 16
+#define BITS_PER_CHAR 8
projects / ghc-hetmet.git / commitdiff