1dnl  HP-PA 2.0 mpn_rshift -- Right shift.
2
3dnl  Copyright 1997, 2000, 2002, 2003 Free Software Foundation, Inc.
4
5dnl  This file is part of the GNU MP Library.
6dnl
7dnl  The GNU MP Library is free software; you can redistribute it and/or modify
8dnl  it under the terms of either:
9dnl
10dnl    * the GNU Lesser General Public License as published by the Free
11dnl      Software Foundation; either version 3 of the License, or (at your
12dnl      option) any later version.
13dnl
14dnl  or
15dnl
16dnl    * the GNU General Public License as published by the Free Software
17dnl      Foundation; either version 2 of the License, or (at your option) any
18dnl      later version.
19dnl
20dnl  or both in parallel, as here.
21dnl
22dnl  The GNU MP Library is distributed in the hope that it will be useful, but
23dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
24dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
25dnl  for more details.
26dnl
27dnl  You should have received copies of the GNU General Public License and the
28dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
29dnl  see https://www.gnu.org/licenses/.
30
31
32dnl  This runs at 1.5 cycles/limb on PA8000 and 1.0 cycles/limb on PA8500.
33
34include(`../config.m4')
35
36dnl  INPUT PARAMETERS
37define(`rp',`%r26')
38define(`up',`%r25')
39define(`n',`%r24')
40define(`cnt',`%r23')
41
42ifdef(`HAVE_ABI_2_0w',
43`       .level  2.0w
44',`     .level  2.0
45')
46PROLOGUE(mpn_rshift)
47	mtsar		cnt
48	ldd		0(up), %r21
49	addib,=		-1, n, L(end)
50	shrpd		%r21, %r0, %sar, %r29	C compute carry out limb
51	depw,z		n, 31, 3, %r28		C r28 = (size & 7)
52	sub		%r0, n, %r22
53	depw,z		%r22, 28, 3, %r22	C r22 = 8 * (-size & 7)
54	sub		up, %r22, up		C offset up
55	blr		%r28, %r0		C branch into jump table
56	sub		rp, %r22, rp		C offset rp
57	b		L(0)
58	nop
59	b		L(1)
60	copy		%r21, %r20
61	b		L(2)
62	nop
63	b		L(3)
64	copy		%r21, %r20
65	b		L(4)
66	nop
67	b		L(5)
68	copy		%r21, %r20
69	b		L(6)
70	nop
71	b		L(7)
72	copy		%r21, %r20
73
74LDEF(loop)
75LDEF(0)	ldd		8(up), %r20
76	shrpd		%r20, %r21, %sar, %r21
77	std		%r21, 0(rp)
78LDEF(7)	ldd		16(up), %r21
79	shrpd		%r21, %r20, %sar, %r20
80	std		%r20, 8(rp)
81LDEF(6)	ldd		24(up), %r20
82	shrpd		%r20, %r21, %sar, %r21
83	std		%r21, 16(rp)
84LDEF(5)	ldd		32(up), %r21
85	shrpd		%r21, %r20, %sar, %r20
86	std		%r20, 24(rp)
87LDEF(4)	ldd		40(up), %r20
88	shrpd		%r20, %r21, %sar, %r21
89	std		%r21, 32(rp)
90LDEF(3)	ldd		48(up), %r21
91	shrpd		%r21, %r20, %sar, %r20
92	std		%r20, 40(rp)
93LDEF(2)	ldd		56(up), %r20
94	shrpd		%r20, %r21, %sar, %r21
95	std		%r21, 48(rp)
96LDEF(1)	ldd		64(up), %r21
97	ldo		64(up), up
98	shrpd		%r21, %r20, %sar, %r20
99	std		%r20, 56(rp)
100	addib,>		-8, n, L(loop)
101	ldo		64(rp), rp
102
103LDEF(end)
104	shrpd		%r0, %r21, %sar, %r21
105	std		%r21, 0(rp)
106	bve		(%r2)
107ifdef(`HAVE_ABI_2_0w',
108`	copy		%r29,%r28
109',`	extrd,u		%r29, 31, 32, %r28
110')
111EPILOGUE(mpn_rshift)
112