summaryrefslogtreecommitdiff
path: root/sysdeps/sparc/sparc32/sparcv8
diff options
context:
space:
mode:
Diffstat (limited to 'sysdeps/sparc/sparc32/sparcv8')
-rw-r--r--sysdeps/sparc/sparc32/sparcv8/addmul_1.S124
-rw-r--r--sysdeps/sparc/sparc32/sparcv8/mul_1.S99
-rw-r--r--sysdeps/sparc/sparc32/sparcv8/submul_1.S58
-rw-r--r--sysdeps/sparc/sparc32/sparcv8/udiv_qrnnd.S186
4 files changed, 467 insertions, 0 deletions
diff --git a/sysdeps/sparc/sparc32/sparcv8/addmul_1.S b/sysdeps/sparc/sparc32/sparcv8/addmul_1.S
new file mode 100644
index 0000000000..fb9ea7cf0e
--- /dev/null
+++ b/sysdeps/sparc/sparc32/sparcv8/addmul_1.S
@@ -0,0 +1,124 @@
+! SPARC v8 __mpn_addmul_1 -- Multiply a limb vector with a limb and
+! add the result to a second limb vector.
+
+! Copyright (C) 1992, 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.
+
+
+! INPUT PARAMETERS
+! res_ptr o0
+! s1_ptr o1
+! size o2
+! s2_limb o3
+
+#include "sysdep.h"
+
+.text
+ .align 4
+ .global C_SYMBOL_NAME(__mpn_addmul_1)
+C_SYMBOL_NAME(__mpn_addmul_1):
+ orcc %g0,%g0,%g2
+ ld [%o1+0],%o4 ! 1
+
+ sll %o2,4,%g1
+ and %g1,(4-1)<<4,%g1
+#if PIC
+ mov %o7,%g4 ! Save return address register
+ call 1f
+ add %o7,LL-1f,%g3
+1: mov %g4,%o7 ! Restore return address register
+#else
+ sethi %hi(LL),%g3
+ or %g3,%lo(LL),%g3
+#endif
+ jmp %g3+%g1
+ nop
+LL:
+LL00: add %o0,-4,%o0
+ b Loop00 /* 4, 8, 12, ... */
+ add %o1,-4,%o1
+ nop
+LL01: b Loop01 /* 1, 5, 9, ... */
+ nop
+ nop
+ nop
+LL10: add %o0,-12,%o0 /* 2, 6, 10, ... */
+ b Loop10
+ add %o1,4,%o1
+ nop
+LL11: add %o0,-8,%o0 /* 3, 7, 11, ... */
+ b Loop11
+ add %o1,-8,%o1
+ nop
+
+1: addcc %g3,%g2,%g3 ! 1
+ ld [%o1+4],%o4 ! 2
+ rd %y,%g2 ! 1
+ addx %g0,%g2,%g2
+ ld [%o0+0],%g1 ! 2
+ addcc %g1,%g3,%g3
+ st %g3,[%o0+0] ! 1
+Loop00: umul %o4,%o3,%g3 ! 2
+ ld [%o0+4],%g1 ! 2
+ addxcc %g3,%g2,%g3 ! 2
+ ld [%o1+8],%o4 ! 3
+ rd %y,%g2 ! 2
+ addx %g0,%g2,%g2
+ nop
+ addcc %g1,%g3,%g3
+ st %g3,[%o0+4] ! 2
+Loop11: umul %o4,%o3,%g3 ! 3
+ addxcc %g3,%g2,%g3 ! 3
+ ld [%o1+12],%o4 ! 4
+ rd %y,%g2 ! 3
+ add %o1,16,%o1
+ addx %g0,%g2,%g2
+ ld [%o0+8],%g1 ! 2
+ addcc %g1,%g3,%g3
+ st %g3,[%o0+8] ! 3
+Loop10: umul %o4,%o3,%g3 ! 4
+ addxcc %g3,%g2,%g3 ! 4
+ ld [%o1+0],%o4 ! 1
+ rd %y,%g2 ! 4
+ addx %g0,%g2,%g2
+ ld [%o0+12],%g1 ! 2
+ addcc %g1,%g3,%g3
+ st %g3,[%o0+12] ! 4
+ add %o0,16,%o0
+ addx %g0,%g2,%g2
+Loop01: addcc %o2,-4,%o2
+ bg 1b
+ umul %o4,%o3,%g3 ! 1
+
+ addcc %g3,%g2,%g3 ! 4
+ rd %y,%g2 ! 4
+ addx %g0,%g2,%g2
+ ld [%o0+0],%g1 ! 2
+ addcc %g1,%g3,%g3
+ st %g3,[%o0+0] ! 4
+ addx %g0,%g2,%o0
+
+ retl
+ nop
+
+
+! umul, ld, addxcc, rd, st
+
+! umul, ld, addxcc, rd, ld, addcc, st, addx
+
diff --git a/sysdeps/sparc/sparc32/sparcv8/mul_1.S b/sysdeps/sparc/sparc32/sparcv8/mul_1.S
new file mode 100644
index 0000000000..b641feb453
--- /dev/null
+++ b/sysdeps/sparc/sparc32/sparcv8/mul_1.S
@@ -0,0 +1,99 @@
+! SPARC v8 __mpn_mul_1 -- Multiply a limb vector with a single limb and
+! store the product in a second limb vector.
+
+! Copyright (C) 1992, 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.
+
+
+! INPUT PARAMETERS
+! res_ptr o0
+! s1_ptr o1
+! size o2
+! s2_limb o3
+
+#include "sysdep.h"
+
+.text
+ .align 8
+ .global C_SYMBOL_NAME(__mpn_mul_1)
+C_SYMBOL_NAME(__mpn_mul_1):
+ sll %o2,4,%g1
+ and %g1,(4-1)<<4,%g1
+#if PIC
+ mov %o7,%g4 ! Save return address register
+ call 1f
+ add %o7,LL-1f,%g3
+1: mov %g4,%o7 ! Restore return address register
+#else
+ sethi %hi(LL),%g3
+ or %g3,%lo(LL),%g3
+#endif
+ jmp %g3+%g1
+ ld [%o1+0],%o4 ! 1
+LL:
+LL00: add %o0,-4,%o0
+ add %o1,-4,%o1
+ b Loop00 /* 4, 8, 12, ... */
+ orcc %g0,%g0,%g2
+LL01: b Loop01 /* 1, 5, 9, ... */
+ orcc %g0,%g0,%g2
+ nop
+ nop
+LL10: add %o0,-12,%o0 /* 2, 6, 10, ... */
+ add %o1,4,%o1
+ b Loop10
+ orcc %g0,%g0,%g2
+ nop
+LL11: add %o0,-8,%o0 /* 3, 7, 11, ... */
+ add %o1,-8,%o1
+ b Loop11
+ orcc %g0,%g0,%g2
+
+Loop: addcc %g3,%g2,%g3 ! 1
+ ld [%o1+4],%o4 ! 2
+ st %g3,[%o0+0] ! 1
+ rd %y,%g2 ! 1
+Loop00: umul %o4,%o3,%g3 ! 2
+ addxcc %g3,%g2,%g3 ! 2
+ ld [%o1+8],%o4 ! 3
+ st %g3,[%o0+4] ! 2
+ rd %y,%g2 ! 2
+Loop11: umul %o4,%o3,%g3 ! 3
+ addxcc %g3,%g2,%g3 ! 3
+ ld [%o1+12],%o4 ! 4
+ add %o1,16,%o1
+ st %g3,[%o0+8] ! 3
+ rd %y,%g2 ! 3
+Loop10: umul %o4,%o3,%g3 ! 4
+ addxcc %g3,%g2,%g3 ! 4
+ ld [%o1+0],%o4 ! 1
+ st %g3,[%o0+12] ! 4
+ add %o0,16,%o0
+ rd %y,%g2 ! 4
+ addx %g0,%g2,%g2
+Loop01: addcc %o2,-4,%o2
+ bg Loop
+ umul %o4,%o3,%g3 ! 1
+
+ addcc %g3,%g2,%g3 ! 4
+ st %g3,[%o0+0] ! 4
+ rd %y,%g2 ! 4
+
+ retl
+ addx %g0,%g2,%o0
diff --git a/sysdeps/sparc/sparc32/sparcv8/submul_1.S b/sysdeps/sparc/sparc32/sparcv8/submul_1.S
new file mode 100644
index 0000000000..e40119d011
--- /dev/null
+++ b/sysdeps/sparc/sparc32/sparcv8/submul_1.S
@@ -0,0 +1,58 @@
+! SPARC v8 __mpn_submul_1 -- Multiply a limb vector with a limb and
+! subtract the result from a second limb vector.
+
+! Copyright (C) 1992, 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.
+
+
+! INPUT PARAMETERS
+! res_ptr o0
+! s1_ptr o1
+! size o2
+! s2_limb o3
+
+#include "sysdep.h"
+
+.text
+ .align 4
+ .global C_SYMBOL_NAME(__mpn_submul_1)
+C_SYMBOL_NAME(__mpn_submul_1):
+ sub %g0,%o2,%o2 ! negate ...
+ sll %o2,2,%o2 ! ... and scale size
+ sub %o1,%o2,%o1 ! o1 is offset s1_ptr
+ sub %o0,%o2,%g1 ! g1 is offset res_ptr
+
+ mov 0,%o0 ! clear cy_limb
+
+Loop: ld [%o1+%o2],%o4
+ ld [%g1+%o2],%g2
+ umul %o4,%o3,%o5
+ rd %y,%g3
+ addcc %o5,%o0,%o5
+ addx %g3,0,%o0
+ subcc %g2,%o5,%g2
+ addx %o0,0,%o0
+ st %g2,[%g1+%o2]
+
+ addcc %o2,4,%o2
+ bne Loop
+ nop
+
+ retl
+ nop
diff --git a/sysdeps/sparc/sparc32/sparcv8/udiv_qrnnd.S b/sysdeps/sparc/sparc32/sparcv8/udiv_qrnnd.S
new file mode 100644
index 0000000000..49c2398806
--- /dev/null
+++ b/sysdeps/sparc/sparc32/sparcv8/udiv_qrnnd.S
@@ -0,0 +1,186 @@
+! SPARC __udiv_qrnnd division support, used from longlong.h.
+
+! Copyright (C) 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., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+
+! INPUT PARAMETERS
+! rem_ptr o0
+! n1 o1
+! n0 o2
+! d o3
+
+#include "sysdep.h"
+
+ .text
+ .align 4
+ .global C_SYMBOL_NAME(__udiv_qrnnd)
+C_SYMBOL_NAME(__udiv_qrnnd):
+ tst %o3
+ bneg Largedivisor
+ mov 8,%g1
+
+ b Lp1
+ addxcc %o2,%o2,%o2
+
+Lplop: bcc Ln1
+ addxcc %o2,%o2,%o2
+Lp1: addx %o1,%o1,%o1
+ subcc %o1,%o3,%o4
+ bcc Ln2
+ addxcc %o2,%o2,%o2
+Lp2: addx %o1,%o1,%o1
+ subcc %o1,%o3,%o4
+ bcc Ln3
+ addxcc %o2,%o2,%o2
+Lp3: addx %o1,%o1,%o1
+ subcc %o1,%o3,%o4
+ bcc Ln4
+ addxcc %o2,%o2,%o2
+Lp4: addx %o1,%o1,%o1
+ addcc %g1,-1,%g1
+ bne Lplop
+ subcc %o1,%o3,%o4
+ bcc Ln5
+ addxcc %o2,%o2,%o2
+Lp5: st %o1,[%o0]
+ retl
+ xnor %g0,%o2,%o0
+
+Lnlop: bcc Lp1
+ addxcc %o2,%o2,%o2
+Ln1: addx %o4,%o4,%o4
+ subcc %o4,%o3,%o1
+ bcc Lp2
+ addxcc %o2,%o2,%o2
+Ln2: addx %o4,%o4,%o4
+ subcc %o4,%o3,%o1
+ bcc Lp3
+ addxcc %o2,%o2,%o2
+Ln3: addx %o4,%o4,%o4
+ subcc %o4,%o3,%o1
+ bcc Lp4
+ addxcc %o2,%o2,%o2
+Ln4: addx %o4,%o4,%o4
+ addcc %g1,-1,%g1
+ bne Lnlop
+ subcc %o4,%o3,%o1
+ bcc Lp5
+ addxcc %o2,%o2,%o2
+Ln5: st %o4,[%o0]
+ retl
+ xnor %g0,%o2,%o0
+
+Largedivisor:
+ and %o2,1,%o5 ! %o5 = n0 & 1
+
+ srl %o2,1,%o2
+ sll %o1,31,%g2
+ or %g2,%o2,%o2 ! %o2 = lo(n1n0 >> 1)
+ srl %o1,1,%o1 ! %o1 = hi(n1n0 >> 1)
+
+ and %o3,1,%g2
+ srl %o3,1,%g3 ! %g3 = floor(d / 2)
+ add %g3,%g2,%g3 ! %g3 = ceil(d / 2)
+
+ b LLp1
+ addxcc %o2,%o2,%o2
+
+LLplop: bcc LLn1
+ addxcc %o2,%o2,%o2
+LLp1: addx %o1,%o1,%o1
+ subcc %o1,%g3,%o4
+ bcc LLn2
+ addxcc %o2,%o2,%o2
+LLp2: addx %o1,%o1,%o1
+ subcc %o1,%g3,%o4
+ bcc LLn3
+ addxcc %o2,%o2,%o2
+LLp3: addx %o1,%o1,%o1
+ subcc %o1,%g3,%o4
+ bcc LLn4
+ addxcc %o2,%o2,%o2
+LLp4: addx %o1,%o1,%o1
+ addcc %g1,-1,%g1
+ bne LLplop
+ subcc %o1,%g3,%o4
+ bcc LLn5
+ addxcc %o2,%o2,%o2
+LLp5: add %o1,%o1,%o1 ! << 1
+ tst %g2
+ bne Oddp
+ add %o5,%o1,%o1
+ st %o1,[%o0]
+ retl
+ xnor %g0,%o2,%o0
+
+LLnlop: bcc LLp1
+ addxcc %o2,%o2,%o2
+LLn1: addx %o4,%o4,%o4
+ subcc %o4,%g3,%o1
+ bcc LLp2
+ addxcc %o2,%o2,%o2
+LLn2: addx %o4,%o4,%o4
+ subcc %o4,%g3,%o1
+ bcc LLp3
+ addxcc %o2,%o2,%o2
+LLn3: addx %o4,%o4,%o4
+ subcc %o4,%g3,%o1
+ bcc LLp4
+ addxcc %o2,%o2,%o2
+LLn4: addx %o4,%o4,%o4
+ addcc %g1,-1,%g1
+ bne LLnlop
+ subcc %o4,%g3,%o1
+ bcc LLp5
+ addxcc %o2,%o2,%o2
+LLn5: add %o4,%o4,%o4 ! << 1
+ tst %g2
+ bne Oddn
+ add %o5,%o4,%o4
+ st %o4,[%o0]
+ retl
+ xnor %g0,%o2,%o0
+
+Oddp: xnor %g0,%o2,%o2
+ ! q' in %o2. r' in %o1
+ addcc %o1,%o2,%o1
+ bcc LLp6
+ addx %o2,0,%o2
+ sub %o1,%o3,%o1
+LLp6: subcc %o1,%o3,%g0
+ bcs LLp7
+ subx %o2,-1,%o2
+ sub %o1,%o3,%o1
+LLp7: st %o1,[%o0]
+ retl
+ mov %o2,%o0
+
+Oddn: xnor %g0,%o2,%o2
+ ! q' in %o2. r' in %o4
+ addcc %o4,%o2,%o4
+ bcc LLn6
+ addx %o2,0,%o2
+ sub %o4,%o3,%o4
+LLn6: subcc %o4,%o3,%g0
+ bcs LLn7
+ subx %o2,-1,%o2
+ sub %o4,%o3,%o4
+LLn7: st %o4,[%o0]
+ retl
+ mov %o2,%o0