summaryrefslogtreecommitdiff
path: root/gcc/config/sh/lib1funcs-Os-4-200.asm
blob: aae57ccd36ce8d3f06643e6df395973060aa3300 (plain)
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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
/* Copyright (C) 2006, 2009 Free Software Foundation, Inc.

This file is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 3, or (at your option) any
later version.

This file 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
General Public License for more details.

Under Section 7 of GPL version 3, you are granted additional
permissions described in the GCC Runtime Library Exception, version
3.1, as published by the Free Software Foundation.

You should have received a copy of the GNU General Public License and
a copy of the GCC Runtime Library Exception along with this program;
see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
<http://www.gnu.org/licenses/>.  */

/* Moderately Space-optimized libgcc routines for the Renesas SH /
   STMicroelectronics ST40 CPUs.
   Contributed by J"orn Rennecke joern.rennecke@st.com.  */

#include "lib1funcs.h"

#if !__SHMEDIA__
#ifdef L_udivsi3_i4i

/* 88 bytes; sh4-200 cycle counts:
   divisor  >= 2G: 11 cycles
   dividend <  2G: 48 cycles
   dividend >= 2G: divisor != 1: 54 cycles
   dividend >= 2G, divisor == 1: 22 cycles */
#if defined (__SH_FPU_DOUBLE__) || defined (__SH4_SINGLE_ONLY__)
!! args in r4 and r5, result in r0, clobber r1

	.global GLOBAL(udivsi3_i4i)
	FUNC(GLOBAL(udivsi3_i4i))
GLOBAL(udivsi3_i4i):
	mova L1,r0
	cmp/pz r5
	sts fpscr,r1
	lds.l @r0+,fpscr
	sts.l fpul,@-r15
	bf LOCAL(huge_divisor)
	mov.l r1,@-r15
	lds r4,fpul
	cmp/pz r4
#ifdef FMOVD_WORKS
	fmov.d dr0,@-r15
	float fpul,dr0
	fmov.d dr2,@-r15
	bt LOCAL(dividend_adjusted)
	mov #1,r1
	fmov.d @r0,dr2
	cmp/eq r1,r5
	bt LOCAL(div_by_1)
	fadd dr2,dr0
LOCAL(dividend_adjusted):
	lds r5,fpul
	float fpul,dr2
	fdiv dr2,dr0
LOCAL(div_by_1):
	fmov.d @r15+,dr2
	ftrc dr0,fpul
	fmov.d @r15+,dr0
#else /* !FMOVD_WORKS */
	fmov.s DR01,@-r15
	mov #1,r1
	fmov.s DR00,@-r15
	float fpul,dr0
	fmov.s DR21,@-r15
	bt/s LOCAL(dividend_adjusted)
	fmov.s DR20,@-r15
	cmp/eq r1,r5
	bt LOCAL(div_by_1)
	fmov.s @r0+,DR20
	fmov.s @r0,DR21
	fadd dr2,dr0
LOCAL(dividend_adjusted):
	lds r5,fpul
	float fpul,dr2
	fdiv dr2,dr0
LOCAL(div_by_1):
	fmov.s @r15+,DR20
	fmov.s @r15+,DR21
	ftrc dr0,fpul
	fmov.s @r15+,DR00
	fmov.s @r15+,DR01
#endif /* !FMOVD_WORKS */
	lds.l @r15+,fpscr
	sts fpul,r0
	rts
	lds.l @r15+,fpul

#ifdef FMOVD_WORKS
	.p2align 3        ! make double below 8 byte aligned.
#endif
LOCAL(huge_divisor):
	lds r1,fpscr
	add #4,r15
	cmp/hs r5,r4
	rts
	movt r0

	.p2align 2
L1:
#ifndef FMOVD_WORKS
	.long 0x80000
#else
	.long 0x180000
#endif
	.double 4294967296

	ENDFUNC(GLOBAL(udivsi3_i4i))
#elif !defined (__sh1__)  /* !__SH_FPU_DOUBLE__ */

#if 0
/* With 36 bytes, the following would probably be the most compact
   implementation, but with 139 cycles on an sh4-200, it is extremely slow.  */
GLOBAL(udivsi3_i4i):
	mov.l r2,@-r15
	mov #0,r1
	div0u
	mov r1,r2
	mov.l r3,@-r15
	mov r1,r3
	sett
	mov r4,r0
LOCAL(loop):
	rotcr r2
	;
	bt/s LOCAL(end)
	cmp/gt r2,r3
	rotcl r0
	bra LOCAL(loop)
	div1 r5,r1
LOCAL(end):
	rotcl r0
	mov.l @r15+,r3
	rts
	mov.l @r15+,r2
#endif /* 0 */

/* Size: 186 bytes jointly for udivsi3_i4i and sdivsi3_i4i
   sh4-200 run times:
   udiv small divisor: 55 cycles
   udiv large divisor: 52 cycles
   sdiv small divisor, positive result: 59 cycles
   sdiv large divisor, positive result: 56 cycles
   sdiv small divisor, negative result: 65 cycles (*)
   sdiv large divisor, negative result: 62 cycles (*)
   (*): r2 is restored in the rts delay slot and has a lingering latency
        of two more cycles.  */
	.balign 4
	.global	GLOBAL(udivsi3_i4i)
	FUNC(GLOBAL(udivsi3_i4i))
	FUNC(GLOBAL(sdivsi3_i4i))
GLOBAL(udivsi3_i4i):
	sts pr,r1
	mov.l r4,@-r15
	extu.w r5,r0
	cmp/eq r5,r0
	swap.w r4,r0
	shlr16 r4
	bf/s LOCAL(large_divisor)
	div0u
	mov.l r5,@-r15
	shll16 r5
LOCAL(sdiv_small_divisor):
	div1 r5,r4
	bsr LOCAL(div6)
	div1 r5,r4
	div1 r5,r4
	bsr LOCAL(div6)
	div1 r5,r4
	xtrct r4,r0
	xtrct r0,r4
	bsr LOCAL(div7)
	swap.w r4,r4
	div1 r5,r4
	bsr LOCAL(div7)
	div1 r5,r4
	xtrct r4,r0
	mov.l @r15+,r5
	swap.w r0,r0
	mov.l @r15+,r4
	jmp @r1
	rotcl r0
LOCAL(div7):
	div1 r5,r4
LOCAL(div6):
	            div1 r5,r4; div1 r5,r4; div1 r5,r4
	div1 r5,r4; div1 r5,r4; rts;        div1 r5,r4

LOCAL(divx3):
	rotcl r0
	div1 r5,r4
	rotcl r0
	div1 r5,r4
	rotcl r0
	rts
	div1 r5,r4

LOCAL(large_divisor):
	mov.l r5,@-r15
LOCAL(sdiv_large_divisor):
	xor r4,r0
	.rept 4
	rotcl r0
	bsr LOCAL(divx3)
	div1 r5,r4
	.endr
	mov.l @r15+,r5
	mov.l @r15+,r4
	jmp @r1
	rotcl r0
	ENDFUNC(GLOBAL(udivsi3_i4i))

	.global	GLOBAL(sdivsi3_i4i)
GLOBAL(sdivsi3_i4i):
	mov.l r4,@-r15
	cmp/pz r5
	mov.l r5,@-r15
	bt/s LOCAL(pos_divisor)
	cmp/pz r4
	neg r5,r5
	extu.w r5,r0
	bt/s LOCAL(neg_result)
	cmp/eq r5,r0
	neg r4,r4
LOCAL(pos_result):
	swap.w r4,r0
	bra LOCAL(sdiv_check_divisor)
	sts pr,r1
LOCAL(pos_divisor):
	extu.w r5,r0
	bt/s LOCAL(pos_result)
	cmp/eq r5,r0
	neg r4,r4
LOCAL(neg_result):
	mova LOCAL(negate_result),r0
	;
	mov r0,r1
	swap.w r4,r0
	lds r2,macl
	sts pr,r2
LOCAL(sdiv_check_divisor):
	shlr16 r4
	bf/s LOCAL(sdiv_large_divisor)
	div0u
	bra LOCAL(sdiv_small_divisor)
	shll16 r5
	.balign 4
LOCAL(negate_result):
	neg r0,r0
	jmp @r2
	sts macl,r2
	ENDFUNC(GLOBAL(sdivsi3_i4i))
#endif /* !__SH_FPU_DOUBLE__ */
#endif /* L_udivsi3_i4i */

#ifdef L_sdivsi3_i4i
#if defined (__SH_FPU_DOUBLE__) || defined (__SH4_SINGLE_ONLY__)
/* 48 bytes, 45 cycles on sh4-200  */
!! args in r4 and r5, result in r0, clobber r1

	.global GLOBAL(sdivsi3_i4i)
	FUNC(GLOBAL(sdivsi3_i4i))
GLOBAL(sdivsi3_i4i):
	sts.l fpscr,@-r15
	sts fpul,r1
	mova L1,r0
	lds.l @r0+,fpscr
	lds r4,fpul
#ifdef FMOVD_WORKS
	fmov.d dr0,@-r15
	float fpul,dr0
	lds r5,fpul
	fmov.d dr2,@-r15
#else
	fmov.s DR01,@-r15
	fmov.s DR00,@-r15
	float fpul,dr0
	lds r5,fpul
	fmov.s DR21,@-r15
	fmov.s DR20,@-r15
#endif
	float fpul,dr2
	fdiv dr2,dr0
#ifdef FMOVD_WORKS
	fmov.d @r15+,dr2
#else
	fmov.s @r15+,DR20
	fmov.s @r15+,DR21
#endif
	ftrc dr0,fpul
#ifdef FMOVD_WORKS
	fmov.d @r15+,dr0
#else
	fmov.s @r15+,DR00
	fmov.s @r15+,DR01
#endif
	lds.l @r15+,fpscr
	sts fpul,r0
	rts
	lds r1,fpul

	.p2align 2
L1:
#ifndef FMOVD_WORKS
	.long 0x80000
#else
	.long 0x180000
#endif

	ENDFUNC(GLOBAL(sdivsi3_i4i))
#endif /* __SH_FPU_DOUBLE__ */
#endif /* L_sdivsi3_i4i */
#endif /* !__SHMEDIA__ */