import some math funcs from glibc

6 files changed, 534 insertions, 1 deletions

diff --git a/libc/sysdeps/linux/hppa/Makefile.arch b/libc/sysdeps/linux/hppa/Makefile.arch
index 26aeb7f8c..3509f6900 100644
--- a/libc/sysdeps/linux/hppa/Makefile.arch
+++ b/libc/sysdeps/linux/hppa/Makefile.arch
@@ -7,6 +7,7 @@
 
 CSRC := __syscall_error.c brk.c mmap.c syscall.c
 
-SSRC := __longjmp.S bsd-_setjmp.S bsd-setjmp.S clone.S setjmp.S
+SSRC := __longjmp.S bsd-_setjmp.S bsd-setjmp.S clone.S setjmp.S \
+	add_n.S lshift.S rshift.S sub_n.S udiv_qrnnd.S
 
 include $(top_srcdir)libc/sysdeps/linux/Makefile.arch
diff --git a/libc/sysdeps/linux/hppa/add_n.S b/libc/sysdeps/linux/hppa/add_n.S
new file mode 100644
index 000000000..a396b3471
--- /dev/null
+++ b/libc/sysdeps/linux/hppa/add_n.S
@@ -0,0 +1,58 @@
+;! HP-PA  __mpn_add_n -- Add two limb vectors of the same length > 0 and store
+;! sum in a third limb vector.
+
+;! Copyright (C) 1992, 1994 Free Software Foundation, Inc.
+
+;! This file is part of the GNU MP Library.
+
+;! The GNU MP Library is free software; you can redistribute it and/or modify
+;! it under the terms of the GNU Lesser General Public License as published by
+;! the Free Software Foundation; either version 2.1 of the License, or (at your
+;! option) any later version.
+
+;! The GNU MP Library is distributed in the hope that it will be useful, but
+;! WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+;! or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
+;! License for more details.
+
+;! You should have received a copy of the GNU Lesser General Public License
+;! along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
+;! the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+;! MA 02111-1307, USA.
+
+
+;! INPUT PARAMETERS
+;! res_ptr	gr26
+;! s1_ptr	gr25
+;! s2_ptr	gr24
+;! size		gr23
+
+;! One might want to unroll this as for other processors, but it turns
+;! out that the data cache contention after a store makes such
+;! unrolling useless.  We can't come under 5 cycles/limb anyway.
+
+	.text
+	.export		__mpn_add_n
+__mpn_add_n:
+	.proc
+	.callinfo	frame=0,no_calls
+	.entry
+
+	ldws,ma		4(%r25),%r21
+	ldws,ma		4(%r24),%r20
+
+	addib,=		-1,%r23,L$end	;! check for (SIZE == 1)
+	 add		%r21,%r20,%r28	;! add first limbs ignoring cy
+
+L$loop:	ldws,ma		4(%r25),%r21
+	ldws,ma		4(%r24),%r20
+	stws,ma		%r28,4(%r26)
+	addib,<>	-1,%r23,L$loop
+	 addc		%r21,%r20,%r28
+
+L$end:	stws		%r28,0(%r26)
+	bv		0(%r2)
+	 addc		%r0,%r0,%r28
+
+	.exit
+	.procend
diff --git a/libc/sysdeps/linux/hppa/lshift.S b/libc/sysdeps/linux/hppa/lshift.S
new file mode 100644
index 000000000..151b283e5
--- /dev/null
+++ b/libc/sysdeps/linux/hppa/lshift.S
@@ -0,0 +1,66 @@
+;! HP-PA  __mpn_lshift --
+
+;! Copyright (C) 1992, 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 Lesser General Public License as published by
+;! the Free Software Foundation; either version 2.1 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 Lesser General Public
+;! License for more details.
+
+;! You should have received a copy of the GNU Lesser 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	gr26
+;! s_ptr	gr25
+;! size		gr24
+;! cnt		gr23
+
+	.text
+	.export		__mpn_lshift
+__mpn_lshift:
+	.proc
+	.callinfo	frame=64,no_calls
+	.entry
+
+	sh2add		%r24,%r25,%r25
+	sh2add		%r24,%r26,%r26
+	ldws,mb		-4(%r25),%r22
+	subi		32,%r23,%r1
+	mtsar		%r1
+	addib,=		-1,%r24,L$0004
+	vshd		%r0,%r22,%r28		;! compute carry out limb
+	ldws,mb		-4(%r25),%r29
+	addib,=		-1,%r24,L$0002
+	vshd		%r22,%r29,%r20
+
+L$loop:	ldws,mb		-4(%r25),%r22
+	stws,mb		%r20,-4(%r26)
+	addib,=		-1,%r24,L$0003
+	vshd		%r29,%r22,%r20
+	ldws,mb		-4(%r25),%r29
+	stws,mb		%r20,-4(%r26)
+	addib,<>	-1,%r24,L$loop
+	vshd		%r22,%r29,%r20
+
+L$0002:	stws,mb		%r20,-4(%r26)
+	vshd		%r29,%r0,%r20
+	bv		0(%r2)
+	stw		%r20,-4(%r26)
+L$0003:	stws,mb		%r20,-4(%r26)
+L$0004:	vshd		%r22,%r0,%r20
+	bv		0(%r2)
+	stw		%r20,-4(%r26)
+
+	.exit
+	.procend
diff --git a/libc/sysdeps/linux/hppa/rshift.S b/libc/sysdeps/linux/hppa/rshift.S
new file mode 100644
index 000000000..dff189dc4
--- /dev/null
+++ b/libc/sysdeps/linux/hppa/rshift.S
@@ -0,0 +1,63 @@
+;! HP-PA  __mpn_rshift -- 
+
+;! Copyright (C) 1992, 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 Lesser General Public License as published by
+;! the Free Software Foundation; either version 2.1 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 Lesser General Public
+;! License for more details.
+
+;! You should have received a copy of the GNU Lesser 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	gr26
+;! s_ptr	gr25
+;! size		gr24
+;! cnt		gr23
+
+	.text
+	.export		__mpn_rshift
+__mpn_rshift:
+	.proc
+	.callinfo	frame=64,no_calls
+	.entry
+
+	ldws,ma		4(%r25),%r22
+	mtsar		%r23
+	addib,=		-1,%r24,L$0004
+	vshd		%r22,%r0,%r28		;! compute carry out limb
+	ldws,ma		4(%r25),%r29
+	addib,=		-1,%r24,L$0002
+	vshd		%r29,%r22,%r20
+
+L$loop:	ldws,ma		4(%r25),%r22
+	stws,ma		%r20,4(%r26)
+	addib,=		-1,%r24,L$0003
+	vshd		%r22,%r29,%r20
+	ldws,ma		4(%r25),%r29
+	stws,ma		%r20,4(%r26)
+	addib,<>	-1,%r24,L$loop
+	vshd		%r29,%r22,%r20
+
+L$0002:	stws,ma		%r20,4(%r26)
+	vshd		%r0,%r29,%r20
+	bv		0(%r2)
+	stw		%r20,0(%r26)
+L$0003:	stws,ma		%r20,4(%r26)
+L$0004:	vshd		%r0,%r22,%r20
+	bv		0(%r2)
+	stw		%r20,0(%r26)
+
+	.exit
+	.procend
diff --git a/libc/sysdeps/linux/hppa/sub_n.S b/libc/sysdeps/linux/hppa/sub_n.S
new file mode 100644
index 000000000..7764961a2
--- /dev/null
+++ b/libc/sysdeps/linux/hppa/sub_n.S
@@ -0,0 +1,59 @@
+;! HP-PA  __mpn_sub_n -- Subtract two limb vectors of the same length > 0 and
+;! store difference in a third limb vector.
+
+;! Copyright (C) 1992, 1994 Free Software Foundation, Inc.
+
+;! This file is part of the GNU MP Library.
+
+;! The GNU MP Library is free software; you can redistribute it and/or modify
+;! it under the terms of the GNU Lesser General Public License as published by
+;! the Free Software Foundation; either version 2.1 of the License, or (at your
+;! option) any later version.
+
+;! The GNU MP Library is distributed in the hope that it will be useful, but
+;! WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+;! or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
+;! License for more details.
+
+;! You should have received a copy of the GNU Lesser General Public License
+;! along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
+;! the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+;! MA 02111-1307, USA.
+
+
+;! INPUT PARAMETERS
+;! res_ptr	gr26
+;! s1_ptr	gr25
+;! s2_ptr	gr24
+;! size		gr23
+
+;! One might want to unroll this as for other processors, but it turns
+;! out that the data cache contention after a store makes such
+;! unrolling useless.  We can't come under 5 cycles/limb anyway.
+
+	.text
+	.export		__mpn_sub_n
+__mpn_sub_n:
+	.proc
+	.callinfo	frame=0,no_calls
+	.entry
+
+	ldws,ma		4(%r25),%r21
+	ldws,ma		4(%r24),%r20
+
+	addib,=		-1,%r23,L$end	;! check for (SIZE == 1)
+	 sub		%r21,%r20,%r28	;! subtract first limbs ignoring cy
+
+L$loop:	ldws,ma		4(%r25),%r21
+	ldws,ma		4(%r24),%r20
+	stws,ma		%r28,4(%r26)
+	addib,<>	-1,%r23,L$loop
+	 subb		%r21,%r20,%r28
+
+L$end:	stws		%r28,0(%r26)
+	addc		%r0,%r0,%r28
+	bv		0(%r2)
+	 subi		1,%r28,%r28
+
+	.exit
+	.procend
diff --git a/libc/sysdeps/linux/hppa/udiv_qrnnd.S b/libc/sysdeps/linux/hppa/udiv_qrnnd.S
new file mode 100644
index 000000000..8e9c07a20
--- /dev/null
+++ b/libc/sysdeps/linux/hppa/udiv_qrnnd.S
@@ -0,0 +1,286 @@
+;! HP-PA  __udiv_qrnnd division support, used from longlong.h.
+;! This version runs fast on pre-PA7000 CPUs.
+
+;! Copyright (C) 1993, 1994 Free Software Foundation, Inc.
+
+;! This file is part of the GNU MP Library.
+
+;! The GNU MP Library is free software; you can redistribute it and/or modify
+;! it under the terms of the GNU Lesser General Public License as published by
+;! the Free Software Foundation; either version 2.1 of the License, or (at your
+;! option) any later version.
+
+;! The GNU MP Library is distributed in the hope that it will be useful, but
+;! WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+;! or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
+;! License for more details.
+
+;! You should have received a copy of the GNU Lesser General Public License
+;! along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
+;! the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+;! MA 02111-1307, USA.
+
+
+;! INPUT PARAMETERS
+;! rem_ptr	gr26
+;! n1		gr25
+;! n0		gr24
+;! d		gr23
+
+;! The code size is a bit excessive.  We could merge the last two ds;addc
+;! sequences by simply moving the "bb,< Odd" instruction down.  The only
+;! trouble is the FFFFFFFF code that would need some hacking.
+
+	.text
+	.export		__udiv_qrnnd
+__udiv_qrnnd:
+	.proc
+	.callinfo	frame=0,no_calls
+	.entry
+
+	comb,<		%r23,%r0,L$largedivisor
+	 sub		%r0,%r23,%r1		;! clear cy as side-effect
+	ds		%r0,%r1,%r0
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r23,%r25
+	addc		%r24,%r24,%r28
+	ds		%r25,%r23,%r25
+	comclr,>=	%r25,%r0,%r0
+	addl		%r25,%r23,%r25
+	stws		%r25,0(%r26)
+	bv		0(%r2)
+	 addc		%r28,%r28,%r28
+
+L$largedivisor:
+	extru		%r24,31,1,%r20		;! r20 = n0 & 1
+	bb,<		%r23,31,L$odd
+	 extru		%r23,30,31,%r22		;! r22 = d >> 1
+	shd		%r25,%r24,1,%r24	;! r24 = new n0
+	extru		%r25,30,31,%r25		;! r25 = new n1
+	sub		%r0,%r22,%r21
+	ds		%r0,%r21,%r0
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	comclr,>=	%r25,%r0,%r0
+	addl		%r25,%r22,%r25
+	sh1addl		%r25,%r20,%r25
+	stws		%r25,0(%r26)
+	bv		0(%r2)
+	 addc		%r24,%r24,%r28
+
+L$odd:	addib,sv,n	1,%r22,L$FF..		;! r22 = (d / 2 + 1)
+	shd		%r25,%r24,1,%r24	;! r24 = new n0
+	extru		%r25,30,31,%r25		;! r25 = new n1
+	sub		%r0,%r22,%r21
+	ds		%r0,%r21,%r0
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r24
+	ds		%r25,%r22,%r25
+	addc		%r24,%r24,%r28
+	comclr,>=	%r25,%r0,%r0
+	addl		%r25,%r22,%r25
+	sh1addl		%r25,%r20,%r25
+;! We have computed (n1,,n0) / (d + 1), q' = r28, r' = r25
+	add,nuv		%r28,%r25,%r25
+	addl		%r25,%r1,%r25
+	addc		%r0,%r28,%r28
+	sub,<<		%r25,%r23,%r0
+	addl		%r25,%r1,%r25
+	stws		%r25,0(%r26)
+	bv		0(%r2)
+	 addc		%r0,%r28,%r28
+
+;! This is just a special case of the code above.
+;! We come here when d == 0xFFFFFFFF
+L$FF..:	add,uv		%r25,%r24,%r24
+	sub,<<		%r24,%r23,%r0
+	ldo		1(%r24),%r24
+	stws		%r24,0(%r26)
+	bv		0(%r2)
+	 addc		%r0,%r25,%r28
+
+	.exit
+	.procend