1dnl  HP-PA  mpn_lshift -- Shift a number left.
2dnl  Optimized for the PA7100, where is runs at 3.25 cycles/limb.
3
4dnl  Copyright 1992, 1994, 2000-2003 Free Software Foundation, Inc.
5
6dnl  This file is part of the GNU MP Library.
7dnl
8dnl  The GNU MP Library is free software; you can redistribute it and/or modify
9dnl  it under the terms of either:
10dnl
11dnl    * the GNU Lesser General Public License as published by the Free
12dnl      Software Foundation; either version 3 of the License, or (at your
13dnl      option) any later version.
14dnl
15dnl  or
16dnl
17dnl    * the GNU General Public License as published by the Free Software
18dnl      Foundation; either version 2 of the License, or (at your option) any
19dnl      later version.
20dnl
21dnl  or both in parallel, as here.
22dnl
23dnl  The GNU MP Library is distributed in the hope that it will be useful, but
24dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
25dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
26dnl  for more details.
27dnl
28dnl  You should have received copies of the GNU General Public License and the
29dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
30dnl  see https://www.gnu.org/licenses/.
31
32include(`../config.m4')
33
34C INPUT PARAMETERS
35C res_ptr r26
36C s_ptr             r25
37C size              r24
38C cnt               r23
39
40ASM_START()
41PROLOGUE(mpn_lshift)
42          sh2add              %r24,%r25,%r25
43          sh2add              %r24,%r26,%r26
44          ldws,mb             -4(0,%r25),%r22
45          subi                32,%r23,%r1
46          mtsar               %r1
47          addib,=             -1,%r24,L(0004)
48          vshd                %r0,%r22,%r28                 C compute carry out limb
49          ldws,mb             -4(0,%r25),%r29
50          addib,<=  -5,%r24,L(rest)
51          vshd                %r22,%r29,%r20
52
53LDEF(loop)
54          ldws,mb             -4(0,%r25),%r22
55          stws,mb             %r20,-4(0,%r26)
56          vshd                %r29,%r22,%r20
57          ldws,mb             -4(0,%r25),%r29
58          stws,mb             %r20,-4(0,%r26)
59          vshd                %r22,%r29,%r20
60          ldws,mb             -4(0,%r25),%r22
61          stws,mb             %r20,-4(0,%r26)
62          vshd                %r29,%r22,%r20
63          ldws,mb             -4(0,%r25),%r29
64          stws,mb             %r20,-4(0,%r26)
65          addib,>             -4,%r24,L(loop)
66          vshd                %r22,%r29,%r20
67
68LDEF(rest)
69          addib,=             4,%r24,L(end1)
70          nop
71
72LDEF(eloop)
73          ldws,mb             -4(0,%r25),%r22
74          stws,mb             %r20,-4(0,%r26)
75          addib,<=  -1,%r24,L(end2)
76          vshd                %r29,%r22,%r20
77          ldws,mb             -4(0,%r25),%r29
78          stws,mb             %r20,-4(0,%r26)
79          addib,>             -1,%r24,L(eloop)
80          vshd                %r22,%r29,%r20
81
82LDEF(end1)
83          stws,mb             %r20,-4(0,%r26)
84          vshd                %r29,%r0,%r20
85          bv                  0(%r2)
86          stw                 %r20,-4(0,%r26)
87
88LDEF(end2)
89          stws,mb             %r20,-4(0,%r26)
90
91LDEF(0004)
92          vshd                %r22,%r0,%r20
93          bv                  0(%r2)
94          stw                 %r20,-4(0,%r26)
95EPILOGUE()
96