summary refs log tree commit diff
path: root/sysdeps/ia64/fpu/s_ilogbl.S
blob: 240da060bfa32277a5eee19ab0c5edcdd3afa7f1 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
.file "ilogbl.s"

// Copyright (C) 2000, 2001, Intel Corporation
// All rights reserved.
// 
// Contributed 2/2/2000 by John Harrison, Ted Kubaska, Bob Norin, Shane Story,
// and Ping Tak Peter Tang of the Computational Software Lab, Intel Corporation.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// * The name of Intel Corporation may not be used to endorse or promote
// products derived from this software without specific prior written
// permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS 
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY 
// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
// 
// Intel Corporation is the author of this code, and requests that all
// problem reports or change requests be submitted to it directly at 
// http://developer.intel.com/opensource.
//
// History
//==============================================================
// 2/03/00  Initial version
// 5/26/00  Fix bug when x a double-extended denormal; 
//          if x=0 call error routine, per C9X
// 8/15/00  Bundle added after call to __libm_error_support to properly
//          set [the previously overwritten] GR_Parameter_RESULT.
// 1/20/01  Fixed result for x=0

.align 32
.global ilogbl#

.section .text
.proc  ilogbl#
.align 32

// API
//==============================================================
// int = ilogbl(double_extended)

// Overview of operation
//==============================================================
// ilogbl computes log2(x) as an int
// and returns it in r8

// ilogbl is similar to logbl but differs in the  following ways:
//         +-inf
//            ilogbl: returns INT_MAX
//             logbl: returns +inf
//         Nan  returns FP_ILOGBNAN (which is either INT_MAX or INT_MIN)
//            ilogbl: returns INT_MAX (7fffffff)
//             logbl: returns QNAN (quieted SNAN)
//         0    returns FP_ILOGB0 (which is either INT_MIN or -INT_MAX)
//            ilogbl: returns INT_MIN (80000000)
//             logbl: returns -inf

// Registers used
//==============================================================

// general local registers: 
// ar.pfs r32
// r33 -> r37
// r38 -> r41 used as parameters to error path

// predicate registers used: 
// p6 - x nan, inf
// p7 - x 0
// p8 - x norm, unorm
// p9 - x unorm

// floating-point registers used: 
// f8 - f10

#include "libm_support.h"

GR_SAVE_PFS         = r32
GR_SAVE_B0          = r34
GR_SAVE_GP          = r35
GR_Parameter_X      = r38
GR_Parameter_Y      = r39
GR_Parameter_RESULT = r40
GR_Parameter_TAG    = r41

FR_X                = f8
FR_Y                = f0
FR_RESULT           = f0


ilogbl: 

// Form signexp of 2^64 in case need to scale denormal
{ .mmf
      alloc          r32=ar.pfs,1,5,4,0
(p0)  mov      r37 = 0x1003f
(p0)  fnorm    f9 = f8 ;;
}

// Form 2^64 in case need to scale denormal
{ .mfi
(p0)  setf.exp f10 = r37
(p0)  fclass.m.unc p7, p8 = f8, 0xe3        
(p0)  mov      r34 = 0xffff ;;
}

// qnan snan inf norm     unorm 0 -+
// 1    1    1   0        0     0 11
// e                      3
// X ZERO, returns INT_MIN
// X INF or NAN, returns INT_MAX

{ .mfi
(p0)  mov      r35 = 0x1ffff
(p8)    fclass.m.unc p6, p8 = f8, 0x07
	nop.i 999 ;;
}
{ .mlx
	nop.m 999
(p7)    movl r8 = 0x000000007fffffff ;;       
}

{ .mib
	nop.m 999
	nop.i 999
(p6)    br.cond.spnt  L(ILOGB_ZERO) ;;
}

// Test for denormal
{ .mfi
	nop.m 999
(p8)    fclass.m.unc p9, p0 = f9, 0x0b        
	nop.i 999 ;;
}

L(ILOGB_COMMON):
// X NORMAL returns true exponent
{ .mmi
	nop.m 999 
(p8)    getf.exp r33 = f9
	nop.i 999 ;;
}

// If denormal add 64 to exponent bias for scaling
{ .mfb
(p9)    add     r34 = 64, r34   
	nop.f 999 
(p9)    br.cond.spnt  L(ILOGB_DENORM) ;;
}

{ .mmi
(p8)    and      r36 = r35, r33
	nop.m 999
	nop.i 999 ;;
}

{ .mib
(p8)    sub r8 = r36, r34                  
	nop.i 999
(p0)    br.ret.sptk    b0 ;;                     
}

L(ILOGB_DENORM):
// Here if x denormal
// Form x * 2^64 which is normal
// Return to common code
{ .mfb
        cmp.eq p8,p9 = r0,r0
        fmpy f9 = f9, f10
        br.cond.sptk  L(ILOGB_COMMON) ;;
}

// X ZERO
// return INT_MIN, call error support
L(ILOGB_ZERO): 
{.mlx
      mov            GR_Parameter_TAG = 156
(p6)  movl r33 = 0x0000000080000000 ;;
};;
.endp ilogbl
ASM_SIZE_DIRECTIVE(ilogbl)

.proc __libm_error_region
__libm_error_region:
.prologue
{ .mfi
        add   GR_Parameter_Y=-32,sp             // Parameter 2 value
        nop.f 0
.save   ar.pfs,GR_SAVE_PFS
        mov  GR_SAVE_PFS=ar.pfs                 // Save ar.pfs
}
{ .mfi
.fframe 64
        add sp=-64,sp                           // Create new stack
        nop.f 0
        mov GR_SAVE_GP=gp                       // Save gp
};;
{ .mmi
        stfe [GR_Parameter_Y] = FR_Y,16         // Save Parameter 2 on stack
        add GR_Parameter_X = 16,sp              // Parameter 1 address
.save   b0, GR_SAVE_B0
        mov GR_SAVE_B0=b0                       // Save b0
};;
.body
{ .mib
        stfe [GR_Parameter_X] = FR_X            // Store Parameter 1 on stack
        add   GR_Parameter_RESULT = 0,GR_Parameter_Y
        nop.b 0                                 // Parameter 3 address
}
{ .mib
        stfe [GR_Parameter_Y] = FR_RESULT      // Store Parameter 3 on stack
        add   GR_Parameter_Y = -16,GR_Parameter_Y
        br.call.sptk b0=__libm_error_support#  // Call error handling function
};;
{ .mmi
        nop.m 0
        nop.m 0
        add   GR_Parameter_RESULT = 48,sp
};;
{ .mmi
        mov   r8 = r33                         // Store result
.restore sp
        add   sp = 64,sp                       // Restore stack pointer
        mov   b0 = GR_SAVE_B0                  // Restore return address
};;
{ .mib
        mov   gp = GR_SAVE_GP                  // Restore gp
        mov   ar.pfs = GR_SAVE_PFS             // Restore ar.pfs
        br.ret.sptk     b0                     // Return
};;

.endp __libm_error_region
ASM_SIZE_DIRECTIVE(__libm_error_region)

.type   __libm_error_support#,@function
.global __libm_error_support#