aboutsummaryrefslogtreecommitdiff
path: root/sysdeps/sparc/sparc32/sub_n.S
diff options
context:
space:
mode:
Diffstat (limited to 'sysdeps/sparc/sparc32/sub_n.S')
-rw-r--r--sysdeps/sparc/sparc32/sub_n.S406
1 files changed, 212 insertions, 194 deletions
diff --git a/sysdeps/sparc/sparc32/sub_n.S b/sysdeps/sparc/sparc32/sub_n.S
index b7a11958e2..74400600bd 100644
--- a/sysdeps/sparc/sparc32/sub_n.S
+++ b/sysdeps/sparc/sparc32/sub_n.S
@@ -1,20 +1,20 @@
! SPARC __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,
@@ -22,290 +22,308 @@
! INPUT PARAMETERS
-#define res_ptr %o0
-#define s1_ptr %o1
-#define s2_ptr %o2
-#define size %o3
+#define RES_PTR %o0
+#define S1_PTR %o1
+#define S2_PTR %o2
+#define SIZE %o3
-#include "sysdep.h"
+#include <sysdep.h>
- .text
- .align 4
- .global C_SYMBOL_NAME(__mpn_sub_n)
-C_SYMBOL_NAME(__mpn_sub_n):
- xor s2_ptr,res_ptr,%g1
+ENTRY(__mpn_sub_n)
+ xor S2_PTR,RES_PTR,%g1
andcc %g1,4,%g0
- bne L1 ! branch if alignment differs
+ bne LOC(1) ! branch if alignment differs
nop
! ** V1a **
- andcc res_ptr,4,%g0 ! res_ptr unaligned? Side effect: cy=0
- be L_v1 ! if no, branch
+ andcc RES_PTR,4,%g0 ! RES_PTR unaligned? Side effect: cy=0
+ be LOC(v1) ! if no, branch
nop
-/* Add least significant limb separately to align res_ptr and s2_ptr */
- ld [s1_ptr],%g4
- add s1_ptr,4,s1_ptr
- ld [s2_ptr],%g2
- add s2_ptr,4,s2_ptr
- add size,-1,size
+/* Add least significant limb separately to align RES_PTR and S2_PTR */
+ ld [S1_PTR],%g4
+ add S1_PTR,4,S1_PTR
+ ld [S2_PTR],%g2
+ add S2_PTR,4,S2_PTR
+ add SIZE,-1,SIZE
subcc %g4,%g2,%o4
- st %o4,[res_ptr]
- add res_ptr,4,res_ptr
-L_v1: addx %g0,%g0,%o4 ! save cy in register
- cmp size,2 ! if size < 2 ...
- bl Lend2 ! ... branch to tail code
+ st %o4,[RES_PTR]
+ add RES_PTR,4,RES_PTR
+LOC(v1):
+ addx %g0,%g0,%o4 ! save cy in register
+ cmp SIZE,2 ! if SIZE < 2 ...
+ bl LOC(end2) ! ... branch to tail code
subcc %g0,%o4,%g0 ! restore cy
- ld [s1_ptr+0],%g4
- addcc size,-10,size
- ld [s1_ptr+4],%g1
- ldd [s2_ptr+0],%g2
- blt Lfin1
+ ld [S1_PTR+0],%g4
+ addcc SIZE,-10,SIZE
+ ld [S1_PTR+4],%g1
+ ldd [S2_PTR+0],%g2
+ blt LOC(fin1)
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 8 limbs until less than 8 limbs remain */
-Loop1: subxcc %g4,%g2,%o4
- ld [s1_ptr+8],%g4
+LOC(loop1):
+ subxcc %g4,%g2,%o4
+ ld [S1_PTR+8],%g4
subxcc %g1,%g3,%o5
- ld [s1_ptr+12],%g1
- ldd [s2_ptr+8],%g2
- std %o4,[res_ptr+0]
+ ld [S1_PTR+12],%g1
+ ldd [S2_PTR+8],%g2
+ std %o4,[RES_PTR+0]
subxcc %g4,%g2,%o4
- ld [s1_ptr+16],%g4
+ ld [S1_PTR+16],%g4
subxcc %g1,%g3,%o5
- ld [s1_ptr+20],%g1
- ldd [s2_ptr+16],%g2
- std %o4,[res_ptr+8]
+ ld [S1_PTR+20],%g1
+ ldd [S2_PTR+16],%g2
+ std %o4,[RES_PTR+8]
subxcc %g4,%g2,%o4
- ld [s1_ptr+24],%g4
+ ld [S1_PTR+24],%g4
subxcc %g1,%g3,%o5
- ld [s1_ptr+28],%g1
- ldd [s2_ptr+24],%g2
- std %o4,[res_ptr+16]
+ ld [S1_PTR+28],%g1
+ ldd [S2_PTR+24],%g2
+ std %o4,[RES_PTR+16]
subxcc %g4,%g2,%o4
- ld [s1_ptr+32],%g4
+ ld [S1_PTR+32],%g4
subxcc %g1,%g3,%o5
- ld [s1_ptr+36],%g1
- ldd [s2_ptr+32],%g2
- std %o4,[res_ptr+24]
+ ld [S1_PTR+36],%g1
+ ldd [S2_PTR+32],%g2
+ std %o4,[RES_PTR+24]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-8,size
- add s1_ptr,32,s1_ptr
- add s2_ptr,32,s2_ptr
- add res_ptr,32,res_ptr
- bge Loop1
+ addcc SIZE,-8,SIZE
+ add S1_PTR,32,S1_PTR
+ add S2_PTR,32,S2_PTR
+ add RES_PTR,32,RES_PTR
+ bge LOC(loop1)
subcc %g0,%o4,%g0 ! restore cy
-Lfin1: addcc size,8-2,size
- blt Lend1
+LOC(fin1):
+ addcc SIZE,8-2,SIZE
+ blt LOC(end1)
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 2 limbs until less than 2 limbs remain */
-Loope1: subxcc %g4,%g2,%o4
- ld [s1_ptr+8],%g4
+LOC(loope1):
+ subxcc %g4,%g2,%o4
+ ld [S1_PTR+8],%g4
subxcc %g1,%g3,%o5
- ld [s1_ptr+12],%g1
- ldd [s2_ptr+8],%g2
- std %o4,[res_ptr+0]
+ ld [S1_PTR+12],%g1
+ ldd [S2_PTR+8],%g2
+ std %o4,[RES_PTR+0]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-2,size
- add s1_ptr,8,s1_ptr
- add s2_ptr,8,s2_ptr
- add res_ptr,8,res_ptr
- bge Loope1
+ addcc SIZE,-2,SIZE
+ add S1_PTR,8,S1_PTR
+ add S2_PTR,8,S2_PTR
+ add RES_PTR,8,RES_PTR
+ bge LOC(loope1)
subcc %g0,%o4,%g0 ! restore cy
-Lend1: subxcc %g4,%g2,%o4
+LOC(end1):
+ subxcc %g4,%g2,%o4
subxcc %g1,%g3,%o5
- std %o4,[res_ptr+0]
+ std %o4,[RES_PTR+0]
addx %g0,%g0,%o4 ! save cy in register
- andcc size,1,%g0
- be Lret1
+ andcc SIZE,1,%g0
+ be LOC(ret1)
subcc %g0,%o4,%g0 ! restore cy
/* Add last limb */
- ld [s1_ptr+8],%g4
- ld [s2_ptr+8],%g2
+ ld [S1_PTR+8],%g4
+ ld [S2_PTR+8],%g2
subxcc %g4,%g2,%o4
- st %o4,[res_ptr+8]
+ st %o4,[RES_PTR+8]
-Lret1: retl
+LOC(ret1):
+ retl
addx %g0,%g0,%o0 ! return carry-out from most sign. limb
-L1: xor s1_ptr,res_ptr,%g1
+LOC(1): xor S1_PTR,RES_PTR,%g1
andcc %g1,4,%g0
- bne L2
+ bne LOC(2)
nop
! ** V1b **
- andcc res_ptr,4,%g0 ! res_ptr unaligned? Side effect: cy=0
- be L_v1b ! if no, branch
+ andcc RES_PTR,4,%g0 ! RES_PTR unaligned? Side effect: cy=0
+ be LOC(v1b) ! if no, branch
nop
-/* Add least significant limb separately to align res_ptr and s1_ptr */
- ld [s2_ptr],%g4
- add s2_ptr,4,s2_ptr
- ld [s1_ptr],%g2
- add s1_ptr,4,s1_ptr
- add size,-1,size
+/* Add least significant limb separately to align RES_PTR and S1_PTR */
+ ld [S2_PTR],%g4
+ add S2_PTR,4,S2_PTR
+ ld [S1_PTR],%g2
+ add S1_PTR,4,S1_PTR
+ add SIZE,-1,SIZE
subcc %g2,%g4,%o4
- st %o4,[res_ptr]
- add res_ptr,4,res_ptr
-L_v1b: addx %g0,%g0,%o4 ! save cy in register
- cmp size,2 ! if size < 2 ...
- bl Lend2 ! ... branch to tail code
+ st %o4,[RES_PTR]
+ add RES_PTR,4,RES_PTR
+LOC(v1b):
+ addx %g0,%g0,%o4 ! save cy in register
+ cmp SIZE,2 ! if SIZE < 2 ...
+ bl LOC(end2) ! ... branch to tail code
subcc %g0,%o4,%g0 ! restore cy
- ld [s2_ptr+0],%g4
- addcc size,-10,size
- ld [s2_ptr+4],%g1
- ldd [s1_ptr+0],%g2
- blt Lfin1b
+ ld [S2_PTR+0],%g4
+ addcc SIZE,-10,SIZE
+ ld [S2_PTR+4],%g1
+ ldd [S1_PTR+0],%g2
+ blt LOC(fin1b)
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 8 limbs until less than 8 limbs remain */
-Loop1b: subxcc %g2,%g4,%o4
- ld [s2_ptr+8],%g4
+LOC(loop1b):
+ subxcc %g2,%g4,%o4
+ ld [S2_PTR+8],%g4
subxcc %g3,%g1,%o5
- ld [s2_ptr+12],%g1
- ldd [s1_ptr+8],%g2
- std %o4,[res_ptr+0]
+ ld [S2_PTR+12],%g1
+ ldd [S1_PTR+8],%g2
+ std %o4,[RES_PTR+0]
subxcc %g2,%g4,%o4
- ld [s2_ptr+16],%g4
+ ld [S2_PTR+16],%g4
subxcc %g3,%g1,%o5
- ld [s2_ptr+20],%g1
- ldd [s1_ptr+16],%g2
- std %o4,[res_ptr+8]
+ ld [S2_PTR+20],%g1
+ ldd [S1_PTR+16],%g2
+ std %o4,[RES_PTR+8]
subxcc %g2,%g4,%o4
- ld [s2_ptr+24],%g4
+ ld [S2_PTR+24],%g4
subxcc %g3,%g1,%o5
- ld [s2_ptr+28],%g1
- ldd [s1_ptr+24],%g2
- std %o4,[res_ptr+16]
+ ld [S2_PTR+28],%g1
+ ldd [S1_PTR+24],%g2
+ std %o4,[RES_PTR+16]
subxcc %g2,%g4,%o4
- ld [s2_ptr+32],%g4
+ ld [S2_PTR+32],%g4
subxcc %g3,%g1,%o5
- ld [s2_ptr+36],%g1
- ldd [s1_ptr+32],%g2
- std %o4,[res_ptr+24]
+ ld [S2_PTR+36],%g1
+ ldd [S1_PTR+32],%g2
+ std %o4,[RES_PTR+24]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-8,size
- add s1_ptr,32,s1_ptr
- add s2_ptr,32,s2_ptr
- add res_ptr,32,res_ptr
- bge Loop1b
+ addcc SIZE,-8,SIZE
+ add S1_PTR,32,S1_PTR
+ add S2_PTR,32,S2_PTR
+ add RES_PTR,32,RES_PTR
+ bge LOC(loop1b)
subcc %g0,%o4,%g0 ! restore cy
-Lfin1b: addcc size,8-2,size
- blt Lend1b
+LOC(fin1b):
+ addcc SIZE,8-2,SIZE
+ blt LOC(end1b)
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 2 limbs until less than 2 limbs remain */
-Loope1b:subxcc %g2,%g4,%o4
- ld [s2_ptr+8],%g4
+LOC(loope1b):
+ subxcc %g2,%g4,%o4
+ ld [S2_PTR+8],%g4
subxcc %g3,%g1,%o5
- ld [s2_ptr+12],%g1
- ldd [s1_ptr+8],%g2
- std %o4,[res_ptr+0]
+ ld [S2_PTR+12],%g1
+ ldd [S1_PTR+8],%g2
+ std %o4,[RES_PTR+0]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-2,size
- add s1_ptr,8,s1_ptr
- add s2_ptr,8,s2_ptr
- add res_ptr,8,res_ptr
- bge Loope1b
+ addcc SIZE,-2,SIZE
+ add S1_PTR,8,S1_PTR
+ add S2_PTR,8,S2_PTR
+ add RES_PTR,8,RES_PTR
+ bge LOC(loope1b)
subcc %g0,%o4,%g0 ! restore cy
-Lend1b: subxcc %g2,%g4,%o4
+LOC(end1b):
+ subxcc %g2,%g4,%o4
subxcc %g3,%g1,%o5
- std %o4,[res_ptr+0]
+ std %o4,[RES_PTR+0]
addx %g0,%g0,%o4 ! save cy in register
- andcc size,1,%g0
- be Lret1b
+ andcc SIZE,1,%g0
+ be LOC(ret1b)
subcc %g0,%o4,%g0 ! restore cy
/* Add last limb */
- ld [s2_ptr+8],%g4
- ld [s1_ptr+8],%g2
+ ld [S2_PTR+8],%g4
+ ld [S1_PTR+8],%g2
subxcc %g2,%g4,%o4
- st %o4,[res_ptr+8]
+ st %o4,[RES_PTR+8]
-Lret1b: retl
+LOC(ret1b):
+ retl
addx %g0,%g0,%o0 ! return carry-out from most sign. limb
! ** 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
+/* 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. */
+ of S1_PTR and S2_PTR are the same. */
-L2: cmp size,1
- be Ljone
+LOC(2): cmp SIZE,1
+ be LOC(jone)
nop
- andcc s1_ptr,4,%g0 ! s1_ptr unaligned? Side effect: cy=0
- be L_v2 ! if no, branch
+ andcc S1_PTR,4,%g0 ! S1_PTR unaligned? Side effect: cy=0
+ be LOC(v2) ! if no, branch
nop
-/* Add least significant limb separately to align s1_ptr and s2_ptr */
- ld [s1_ptr],%g4
- add s1_ptr,4,s1_ptr
- ld [s2_ptr],%g2
- add s2_ptr,4,s2_ptr
- add size,-1,size
+/* Add least significant limb separately to align S1_PTR and S2_PTR */
+ ld [S1_PTR],%g4
+ add S1_PTR,4,S1_PTR
+ ld [S2_PTR],%g2
+ add S2_PTR,4,S2_PTR
+ add SIZE,-1,SIZE
subcc %g4,%g2,%o4
- st %o4,[res_ptr]
- add res_ptr,4,res_ptr
+ st %o4,[RES_PTR]
+ add RES_PTR,4,RES_PTR
-L_v2: addx %g0,%g0,%o4 ! save cy in register
- addcc size,-8,size
- blt Lfin2
+LOC(v2):
+ addx %g0,%g0,%o4 ! save cy in register
+ addcc SIZE,-8,SIZE
+ blt LOC(fin2)
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 8 limbs until less than 8 limbs remain */
-Loop2: ldd [s1_ptr+0],%g2
- ldd [s2_ptr+0],%o4
+LOC(loop2):
+ ldd [S1_PTR+0],%g2
+ ldd [S2_PTR+0],%o4
subxcc %g2,%o4,%g2
- st %g2,[res_ptr+0]
+ st %g2,[RES_PTR+0]
subxcc %g3,%o5,%g3
- st %g3,[res_ptr+4]
- ldd [s1_ptr+8],%g2
- ldd [s2_ptr+8],%o4
+ st %g3,[RES_PTR+4]
+ ldd [S1_PTR+8],%g2
+ ldd [S2_PTR+8],%o4
subxcc %g2,%o4,%g2
- st %g2,[res_ptr+8]
+ st %g2,[RES_PTR+8]
subxcc %g3,%o5,%g3
- st %g3,[res_ptr+12]
- ldd [s1_ptr+16],%g2
- ldd [s2_ptr+16],%o4
+ st %g3,[RES_PTR+12]
+ ldd [S1_PTR+16],%g2
+ ldd [S2_PTR+16],%o4
subxcc %g2,%o4,%g2
- st %g2,[res_ptr+16]
+ st %g2,[RES_PTR+16]
subxcc %g3,%o5,%g3
- st %g3,[res_ptr+20]
- ldd [s1_ptr+24],%g2
- ldd [s2_ptr+24],%o4
+ st %g3,[RES_PTR+20]
+ ldd [S1_PTR+24],%g2
+ ldd [S2_PTR+24],%o4
subxcc %g2,%o4,%g2
- st %g2,[res_ptr+24]
+ st %g2,[RES_PTR+24]
subxcc %g3,%o5,%g3
- st %g3,[res_ptr+28]
+ st %g3,[RES_PTR+28]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-8,size
- add s1_ptr,32,s1_ptr
- add s2_ptr,32,s2_ptr
- add res_ptr,32,res_ptr
- bge Loop2
+ addcc SIZE,-8,SIZE
+ add S1_PTR,32,S1_PTR
+ add S2_PTR,32,S2_PTR
+ add RES_PTR,32,RES_PTR
+ bge LOC(loop2)
subcc %g0,%o4,%g0 ! restore cy
-Lfin2: addcc size,8-2,size
- blt Lend2
+LOC(fin2):
+ addcc SIZE,8-2,SIZE
+ blt LOC(end2)
subcc %g0,%o4,%g0 ! restore cy
-Loope2: ldd [s1_ptr+0],%g2
- ldd [s2_ptr+0],%o4
+LOC(loope2):
+ ldd [S1_PTR+0],%g2
+ ldd [S2_PTR+0],%o4
subxcc %g2,%o4,%g2
- st %g2,[res_ptr+0]
+ st %g2,[RES_PTR+0]
subxcc %g3,%o5,%g3
- st %g3,[res_ptr+4]
+ st %g3,[RES_PTR+4]
addx %g0,%g0,%o4 ! save cy in register
- addcc size,-2,size
- add s1_ptr,8,s1_ptr
- add s2_ptr,8,s2_ptr
- add res_ptr,8,res_ptr
- bge Loope2
+ addcc SIZE,-2,SIZE
+ add S1_PTR,8,S1_PTR
+ add S2_PTR,8,S2_PTR
+ add RES_PTR,8,RES_PTR
+ bge LOC(loope2)
subcc %g0,%o4,%g0 ! restore cy
-Lend2: andcc size,1,%g0
- be Lret2
+LOC(end2):
+ andcc SIZE,1,%g0
+ be LOC(ret2)
subcc %g0,%o4,%g0 ! restore cy
/* Add last limb */
-Ljone: ld [s1_ptr],%g4
- ld [s2_ptr],%g2
+LOC(jone):
+ ld [S1_PTR],%g4
+ ld [S2_PTR],%g2
subxcc %g4,%g2,%o4
- st %o4,[res_ptr]
+ st %o4,[RES_PTR]
-Lret2: retl
+LOC(ret2):
+ retl
addx %g0,%g0,%o0 ! return carry-out from most sign. limb
+
+END(__mpn_sub_n)
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-10 04:11:59 +0000