sl@0
|
1 |
// Copyright (c) 1995-2009 Nokia Corporation and/or its subsidiary(-ies).
|
sl@0
|
2 |
// All rights reserved.
|
sl@0
|
3 |
// This component and the accompanying materials are made available
|
sl@0
|
4 |
// under the terms of the License "Eclipse Public License v1.0"
|
sl@0
|
5 |
// which accompanies this distribution, and is available
|
sl@0
|
6 |
// at the URL "http://www.eclipse.org/legal/epl-v10.html".
|
sl@0
|
7 |
//
|
sl@0
|
8 |
// Initial Contributors:
|
sl@0
|
9 |
// Nokia Corporation - initial contribution.
|
sl@0
|
10 |
//
|
sl@0
|
11 |
// Contributors:
|
sl@0
|
12 |
//
|
sl@0
|
13 |
// Description:
|
sl@0
|
14 |
// e32\common\arm\cgcchelp.cia
|
sl@0
|
15 |
//
|
sl@0
|
16 |
//
|
sl@0
|
17 |
|
sl@0
|
18 |
#include "../common.h"
|
sl@0
|
19 |
#ifdef __KERNEL_MODE__
|
sl@0
|
20 |
#include "nkern.h"
|
sl@0
|
21 |
#endif
|
sl@0
|
22 |
|
sl@0
|
23 |
extern "C" {
|
sl@0
|
24 |
#ifdef __GCC32__
|
sl@0
|
25 |
EXPORT_C __NAKED__ TInt __divsi3(TInt /*dividend*/,TInt /*divisor*/)
|
sl@0
|
26 |
//
|
sl@0
|
27 |
// Signed divide of r0 by r1: returns quotient in r0
|
sl@0
|
28 |
// Quotient is truncated (rounded towards zero).
|
sl@0
|
29 |
// Destroys r2, r3 and ip
|
sl@0
|
30 |
// Negates dividend and divisor, then does an unsigned divide; signs
|
sl@0
|
31 |
// get sorted out again at the end.
|
sl@0
|
32 |
//
|
sl@0
|
33 |
// Have to calculate the sign of the result for the end of the calculation.
|
sl@0
|
34 |
// Store this in the LSB of ip which also saves the old lr.
|
sl@0
|
35 |
//
|
sl@0
|
36 |
{
|
sl@0
|
37 |
|
sl@0
|
38 |
asm("STMFD sp!, {lr} ");
|
sl@0
|
39 |
asm("ANDS r3, r1, #0x80000000 "); // r3 bit 31=sign of divisor, rest of r3=0
|
sl@0
|
40 |
asm("RSBMI r1, r1, #0 "); // r1=ABS(divisor)
|
sl@0
|
41 |
asm("EORS ip, r3, r0, ASR #32 "); // ip bit 31=sign of quotient, all other bits=carry=sign of dividend
|
sl@0
|
42 |
asm("RSBCS r0, r0, #0 "); // r0=ABS(dividend)
|
sl@0
|
43 |
asm(".EXTERN ");
|
sl@0
|
44 |
asm("BL __umodsi3_start ");
|
sl@0
|
45 |
asm("MOV r0, r3 ");
|
sl@0
|
46 |
asm("CMP ip, #0 "); // test sign of quotient
|
sl@0
|
47 |
asm("RSBMI r0, r0, #0 "); // negate if necessary
|
sl@0
|
48 |
__POPRET("");
|
sl@0
|
49 |
}
|
sl@0
|
50 |
|
sl@0
|
51 |
|
sl@0
|
52 |
EXPORT_C __NAKED__ TInt __modsi3(TInt /*dividend*/,TInt /*divisor*/)
|
sl@0
|
53 |
//
|
sl@0
|
54 |
// Signed divide of r0 by r1: returns remainder in r0
|
sl@0
|
55 |
// Sign of remainder = sign of dividend.
|
sl@0
|
56 |
// Destroys r2, r3 and ip
|
sl@0
|
57 |
// Negates dividend and divisor, then does an unsigned divide; signs
|
sl@0
|
58 |
// get sorted out again at the end.
|
sl@0
|
59 |
//
|
sl@0
|
60 |
// Have to save sign of dividend in order to apply sign to remainder
|
sl@0
|
61 |
// at the end of the calculation. Store this in the LSB of ip which also
|
sl@0
|
62 |
// saves the old lr.
|
sl@0
|
63 |
//
|
sl@0
|
64 |
{
|
sl@0
|
65 |
|
sl@0
|
66 |
asm("STMFD sp!, {lr} ");
|
sl@0
|
67 |
asm("MOVS r1, r1 ");
|
sl@0
|
68 |
asm("RSBMI r1, r1, #0 ");
|
sl@0
|
69 |
asm("MOVS ip, r0 ");
|
sl@0
|
70 |
asm("RSBMI r0, r0, #0 ");
|
sl@0
|
71 |
asm(".EXTERN ");
|
sl@0
|
72 |
asm("BL __umodsi3_start ");
|
sl@0
|
73 |
asm("MOVS ip, ip ");
|
sl@0
|
74 |
asm("RSBMI r0, r0, #0 ");
|
sl@0
|
75 |
__POPRET("");
|
sl@0
|
76 |
}
|
sl@0
|
77 |
|
sl@0
|
78 |
EXPORT_C __NAKED__ TUint __udivsi3(TUint /*dividend*/,TUint /*divisor*/)
|
sl@0
|
79 |
//
|
sl@0
|
80 |
// Unsigned divide of r0 by r1: returns quotient in r0
|
sl@0
|
81 |
// Quotient is truncated (rounded towards zero).
|
sl@0
|
82 |
// Destroys r2, r3 and ip
|
sl@0
|
83 |
//
|
sl@0
|
84 |
{
|
sl@0
|
85 |
|
sl@0
|
86 |
asm("MOV ip, lr ");
|
sl@0
|
87 |
asm(".EXTERN ");
|
sl@0
|
88 |
asm("BL __umodsi3_start ");
|
sl@0
|
89 |
asm("MOV r0, r3 ");
|
sl@0
|
90 |
__JUMP(,ip);
|
sl@0
|
91 |
}
|
sl@0
|
92 |
|
sl@0
|
93 |
|
sl@0
|
94 |
EXPORT_C __NAKED__ long long __divdi3(long long /*dividend*/, long long /*divisor*/)
|
sl@0
|
95 |
//
|
sl@0
|
96 |
// Dividend in r1:r0, divisor in r3:r2, Return quotient in r1:r0
|
sl@0
|
97 |
//
|
sl@0
|
98 |
{
|
sl@0
|
99 |
asm("stmfd sp!, {r4-r8,lr} ");
|
sl@0
|
100 |
asm("eor r8, r1, r3 "); // sign of result into r8
|
sl@0
|
101 |
asm("movs r1, r1 ");
|
sl@0
|
102 |
asm("bpl 1f ");
|
sl@0
|
103 |
asm("rsbs r0, r0, #0 "); // ABS(dividend)
|
sl@0
|
104 |
asm("rsc r1, r1, #0 ");
|
sl@0
|
105 |
asm("1: ");
|
sl@0
|
106 |
asm("movs r3, r3 ");
|
sl@0
|
107 |
asm("bpl 2f ");
|
sl@0
|
108 |
asm("rsbs r2, r2, #0 "); // ABS(divisor)
|
sl@0
|
109 |
asm("rsc r3, r3, #0 ");
|
sl@0
|
110 |
asm("2: ");
|
sl@0
|
111 |
asm("bl UDiv01 "); // do the division, result in r4,r5
|
sl@0
|
112 |
asm("eors r0, r4, r8, asr #32 "); // quotient into r1:r0, inverted if quotient -ve
|
sl@0
|
113 |
asm("eors r1, r5, r8, asr #32 ");
|
sl@0
|
114 |
asm("adcs r0, r0, #0 "); // if quotient -ve, add 1
|
sl@0
|
115 |
asm("adcs r1, r1, #0 ");
|
sl@0
|
116 |
__POPRET("r4-r8,");
|
sl@0
|
117 |
}
|
sl@0
|
118 |
|
sl@0
|
119 |
EXPORT_C __NAKED__ long long __moddi3(long long /*dividend*/, long long /*divisor*/) /* signed */
|
sl@0
|
120 |
{
|
sl@0
|
121 |
asm("stmfd sp!, {r4-r8,lr} ");
|
sl@0
|
122 |
asm("movs r8, r1 "); // sign of remainder (=sign of dividend) into r8
|
sl@0
|
123 |
asm("bpl 1f ");
|
sl@0
|
124 |
asm("rsbs r0, r0, #0 "); // ABS(dividend)
|
sl@0
|
125 |
asm("rsc r1, r1, #0 ");
|
sl@0
|
126 |
asm("1: ");
|
sl@0
|
127 |
asm("movs r3, r3 ");
|
sl@0
|
128 |
asm("bpl 2f ");
|
sl@0
|
129 |
asm("rsbs r2, r2, #0 "); // ABS(divisor)
|
sl@0
|
130 |
asm("rsc r3, r3, #0 ");
|
sl@0
|
131 |
asm("2: ");
|
sl@0
|
132 |
asm("bl UDiv01 "); // do the division, remainder in r3,r6
|
sl@0
|
133 |
asm("eors r0, r3, r8, asr #32 "); // remainder into r1:r0, inverted if dividend -ve
|
sl@0
|
134 |
asm("eors r1, r6, r8, asr #32 ");
|
sl@0
|
135 |
asm("adcs r0, r0, #0 "); // if dividend -ve, add 1
|
sl@0
|
136 |
asm("adcs r1, r1, #0 ");
|
sl@0
|
137 |
__POPRET("r4-r8,");
|
sl@0
|
138 |
}
|
sl@0
|
139 |
|
sl@0
|
140 |
EXPORT_C __NAKED__ long long __umoddi3(unsigned long long /*dividend*/, unsigned long long /*divisor*/) /* unsigned */
|
sl@0
|
141 |
{
|
sl@0
|
142 |
asm("stmfd sp!, {r4-r7,lr} ");
|
sl@0
|
143 |
asm("bl UDiv01 "); // do the division, remainder in r6:r3
|
sl@0
|
144 |
asm("mov r0, r3 ");
|
sl@0
|
145 |
asm("mov r1, r6 ");
|
sl@0
|
146 |
__POPRET("r4-r7,");
|
sl@0
|
147 |
}
|
sl@0
|
148 |
|
sl@0
|
149 |
EXPORT_C __NAKED__ long long __ashrdi3(long long /*value*/, unsigned int /*count*/)
|
sl@0
|
150 |
{
|
sl@0
|
151 |
asm("cmp r2, #63 ");
|
sl@0
|
152 |
asm("movhi r2, #63 "); // count>63 same as count=63
|
sl@0
|
153 |
asm("cmp r2, #32 ");
|
sl@0
|
154 |
asm("bcs Asr01 "); // jump if shift count >=32
|
sl@0
|
155 |
asm("rsb r12, r2, #32 "); // r12=32-shift count
|
sl@0
|
156 |
asm("mov r0, r0, lsr r2 "); // shift ls word right
|
sl@0
|
157 |
asm("orr r0, r0, r1, lsl r12 "); // or in bits shifted out of ms word
|
sl@0
|
158 |
asm("mov r1, r1, asr r2 "); // shift ms word right
|
sl@0
|
159 |
__JUMP(,lr);
|
sl@0
|
160 |
asm("Asr01: ");
|
sl@0
|
161 |
asm("sub r2, r2, #32 "); // r2=shift count-32
|
sl@0
|
162 |
asm("mov r0, r1, asr r2 "); // ls word = ms word >> (count-32)
|
sl@0
|
163 |
asm("mov r1, r1, asr #32 "); // ms word of result=sign extension of r1
|
sl@0
|
164 |
__JUMP(,lr);
|
sl@0
|
165 |
}
|
sl@0
|
166 |
|
sl@0
|
167 |
EXPORT_C __NAKED__ long long __ashldi3(long long /*value*/, unsigned int /*count*/)
|
sl@0
|
168 |
{
|
sl@0
|
169 |
asm("cmp r2, #63 ");
|
sl@0
|
170 |
asm("movhi r2, #64 "); // count>63 same as count=64
|
sl@0
|
171 |
asm("cmp r2, #32 ");
|
sl@0
|
172 |
asm("bcs Asl01 "); // jump if shift count >=32
|
sl@0
|
173 |
asm("rsb r12, r2, #32 "); // r12=32-shift count
|
sl@0
|
174 |
asm("mov r1, r1, asl r2 "); // shift ms word left
|
sl@0
|
175 |
asm("orr r1, r1, r0, lsr r12 "); // or in bits shifted out of ls word
|
sl@0
|
176 |
asm("mov r0, r0, asl r2 "); // shift ls word left
|
sl@0
|
177 |
__JUMP(,lr);
|
sl@0
|
178 |
asm("Asl01: ");
|
sl@0
|
179 |
asm("sub r2, r2, #32 "); // r2=shift count-32
|
sl@0
|
180 |
asm("mov r1, r0, asl r2 "); // result ms word = ls word << (count-32)
|
sl@0
|
181 |
asm("mov r0, #0 "); // ls word of result is zero
|
sl@0
|
182 |
__JUMP(,lr);
|
sl@0
|
183 |
}
|
sl@0
|
184 |
|
sl@0
|
185 |
EXPORT_C __NAKED__ unsigned long long __lshrdi3(unsigned long long /*value*/, unsigned int /*count*/)
|
sl@0
|
186 |
{
|
sl@0
|
187 |
asm("cmp r2, #63 ");
|
sl@0
|
188 |
asm("movhi r2, #64 "); // count>63 same as count=64
|
sl@0
|
189 |
asm("cmp r2, #32 ");
|
sl@0
|
190 |
asm("bcs Lsr01 "); // jump if shift count >=32
|
sl@0
|
191 |
asm("rsb r12, r2, #32 "); // r12=32-shift count
|
sl@0
|
192 |
asm("mov r0, r0, lsr r2 "); // shift ls word right
|
sl@0
|
193 |
asm("orr r0, r0, r1, lsl r12 "); // or in bits shifted out of ms word
|
sl@0
|
194 |
asm("mov r1, r1, lsr r2 "); // shift ms word right
|
sl@0
|
195 |
__JUMP(,lr);
|
sl@0
|
196 |
asm("Lsr01: ");
|
sl@0
|
197 |
asm("sub r2, r2, #32 "); // r2=shift count-32
|
sl@0
|
198 |
asm("mov r0, r1, lsr r2 "); // ls word = ms word >> (count-32)
|
sl@0
|
199 |
asm("mov r1, #0 "); // ms word of result = 0
|
sl@0
|
200 |
__JUMP(,lr);
|
sl@0
|
201 |
}
|
sl@0
|
202 |
|
sl@0
|
203 |
EXPORT_C __NAKED__ long long __muldi3(long long /*multiplicand*/, long long /*multiplier*/)
|
sl@0
|
204 |
{
|
sl@0
|
205 |
asm("mul r1, r2, r1 "); // r1=low2*high1
|
sl@0
|
206 |
asm("mov ip, r0 "); // ip=low1
|
sl@0
|
207 |
asm("mla r1, r0, r3, r1 "); // r1+=low1*high2
|
sl@0
|
208 |
asm("mov r0, #0 ");
|
sl@0
|
209 |
asm("umlal r0, r1, r2, ip "); // r1:r0 += high1*low1
|
sl@0
|
210 |
__JUMP(,lr);
|
sl@0
|
211 |
}
|
sl@0
|
212 |
|
sl@0
|
213 |
EXPORT_C __NAKED__ long long __negdi2(long long /*argument*/)
|
sl@0
|
214 |
{
|
sl@0
|
215 |
asm("rsbs r0, r0, #0 "); // r0=0-r0, set carry
|
sl@0
|
216 |
asm("rscs r1, r1, #0 "); // r1=0-r1-(1-C)
|
sl@0
|
217 |
__JUMP(,lr);
|
sl@0
|
218 |
}
|
sl@0
|
219 |
|
sl@0
|
220 |
EXPORT_C __NAKED__ unsigned long long __udivmoddi4 (unsigned long long /*dividend*/,
|
sl@0
|
221 |
unsigned long long /*divisor*/,
|
sl@0
|
222 |
unsigned long long* /*p_remainder*/)
|
sl@0
|
223 |
{
|
sl@0
|
224 |
asm("stmfd sp!, {r4-r7,lr} ");
|
sl@0
|
225 |
asm("bl UDiv01 "); // do the division, quotient in r5:r4 remainder in r6:r3
|
sl@0
|
226 |
asm("ldr r7, [sp, #20] "); // r7=p_remainder
|
sl@0
|
227 |
asm("mov r0, r4 "); // r0=quotient low
|
sl@0
|
228 |
asm("stmia r7, {r3,r6} "); // store remainder
|
sl@0
|
229 |
asm("mov r1, r5 "); // r0=quotient high
|
sl@0
|
230 |
__POPRET("r4-r7,");
|
sl@0
|
231 |
}
|
sl@0
|
232 |
|
sl@0
|
233 |
EXPORT_C __NAKED__ int __cmpdi2(long long /*a*/, long long /*b*/)
|
sl@0
|
234 |
{
|
sl@0
|
235 |
// return 0 if a<b, 1 if a=b, 2 if a>b
|
sl@0
|
236 |
asm("subs r0, r2, r0 ");
|
sl@0
|
237 |
asm("sbcs r1, r3, r1 "); // r1:r0 = b-a, set flags
|
sl@0
|
238 |
asm("movlt r0, #2 "); // if b<a r0=2
|
sl@0
|
239 |
__JUMP(lt,lr); // if b<a return
|
sl@0
|
240 |
asm("cmpeq r0, #0 "); // if top word of difference=0, look at bottom
|
sl@0
|
241 |
asm("moveq r0, #1 "); // if a=b, r0=1
|
sl@0
|
242 |
asm("movne r0, #0 "); // else r=0
|
sl@0
|
243 |
__JUMP(,lr);
|
sl@0
|
244 |
}
|
sl@0
|
245 |
|
sl@0
|
246 |
EXPORT_C __NAKED__ int __ucmpdi2(unsigned long long /*a*/, unsigned long long /*b*/)
|
sl@0
|
247 |
{
|
sl@0
|
248 |
// return 0 if a<b, 1 if a=b, 2 if a>b
|
sl@0
|
249 |
asm("cmp r1, r3 ");
|
sl@0
|
250 |
asm("cmpeq r0, r2 "); // compare r1:r0 - r3:r2
|
sl@0
|
251 |
asm("movhi r0, #2 "); // r0=2 if a>b
|
sl@0
|
252 |
asm("moveq r0, #1 "); // r0=1 if a=b
|
sl@0
|
253 |
asm("movlo r0, #0 "); // r0=0 if a<b
|
sl@0
|
254 |
__JUMP(,lr);
|
sl@0
|
255 |
}
|
sl@0
|
256 |
#endif
|
sl@0
|
257 |
|
sl@0
|
258 |
#if defined(__GCC32__)
|
sl@0
|
259 |
void __division_by_zero();
|
sl@0
|
260 |
#define DIV_BY_ZERO " __division_by_zero "
|
sl@0
|
261 |
#elif defined(__ARMCC__)
|
sl@0
|
262 |
void __rt_div0 (void);
|
sl@0
|
263 |
#define DIV_BY_ZERO " __cpp(__rt_div0) "
|
sl@0
|
264 |
#endif
|
sl@0
|
265 |
|
sl@0
|
266 |
EXPORT_C __NAKED__ TUint __umodsi3(TUint /*dividend*/,TUint /*divisor*/)
|
sl@0
|
267 |
//
|
sl@0
|
268 |
// Unsigned divide of r0 by r1: returns remainder in r0, quotient in r3
|
sl@0
|
269 |
// Sign of remainder = sign of dividend.
|
sl@0
|
270 |
// Destroys r2, r3
|
sl@0
|
271 |
//
|
sl@0
|
272 |
{
|
sl@0
|
273 |
|
sl@0
|
274 |
asm("__umodsi3_start:");
|
sl@0
|
275 |
//
|
sl@0
|
276 |
// Use lookup table for divisors less than 17, and jump to
|
sl@0
|
277 |
// an optimised routine if available
|
sl@0
|
278 |
//
|
sl@0
|
279 |
asm("MOV r3, #0 ");
|
sl@0
|
280 |
asm("CMP r1, #16 ");
|
sl@0
|
281 |
asm("LDRLS r3, [pc, #modtable - . - 8]");
|
sl@0
|
282 |
asm("LDRLS r3, [r3, r1, asl #2] ");
|
sl@0
|
283 |
asm("CMP r3, #0 ");
|
sl@0
|
284 |
__JUMP(NE,r3);
|
sl@0
|
285 |
//
|
sl@0
|
286 |
// r3 must be zero when entering this point
|
sl@0
|
287 |
//
|
sl@0
|
288 |
asm("MOV r2, r1 ");
|
sl@0
|
289 |
|
sl@0
|
290 |
asm("__umodsi3_loop: ");
|
sl@0
|
291 |
asm("CMP r2, r0, LSR #8 ");
|
sl@0
|
292 |
asm("MOVLS r2, r2, LSL #8 ");
|
sl@0
|
293 |
asm("BLO __umodsi3_loop ");
|
sl@0
|
294 |
|
sl@0
|
295 |
asm("CMP r2, r0, LSR #1 ");
|
sl@0
|
296 |
asm("BHI __umodsi3_jump7 ");
|
sl@0
|
297 |
asm("CMP r2, r0, LSR #2 ");
|
sl@0
|
298 |
asm("BHI __umodsi3_jump6 ");
|
sl@0
|
299 |
asm("CMP r2, r0, LSR #3 ");
|
sl@0
|
300 |
asm("BHI __umodsi3_jump5 ");
|
sl@0
|
301 |
asm("CMP r2, r0, LSR #4 ");
|
sl@0
|
302 |
asm("BHI __umodsi3_jump4 ");
|
sl@0
|
303 |
asm("CMP r2, r0, LSR #5 ");
|
sl@0
|
304 |
asm("BHI __umodsi3_jump3 ");
|
sl@0
|
305 |
asm("CMP r2, r0, LSR #6 ");
|
sl@0
|
306 |
asm("BHI __umodsi3_jump2 ");
|
sl@0
|
307 |
asm("CMP r2, r0, LSR #7 ");
|
sl@0
|
308 |
asm("BHI __umodsi3_jump1 ");
|
sl@0
|
309 |
|
sl@0
|
310 |
asm("__umodsi3_loop2: ");
|
sl@0
|
311 |
asm("MOVHI r2, r2, LSR #8 ");
|
sl@0
|
312 |
|
sl@0
|
313 |
asm("CMP r0, r2, LSL #7 ");
|
sl@0
|
314 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
315 |
asm("SUBCS r0, r0, r2, LSL #7 ");
|
sl@0
|
316 |
asm("CMP r0, r2, LSL #6 ");
|
sl@0
|
317 |
|
sl@0
|
318 |
asm("__umodsi3_jump1: ");
|
sl@0
|
319 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
320 |
asm("SUBCS r0, r0, r2, LSL #6 ");
|
sl@0
|
321 |
asm("CMP r0, r2, LSL #5 ");
|
sl@0
|
322 |
asm("__umodsi3_jump2: ");
|
sl@0
|
323 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
324 |
asm("SUBCS r0, r0, r2, LSL #5 ");
|
sl@0
|
325 |
asm("CMP r0, r2, LSL #4 ");
|
sl@0
|
326 |
asm("__umodsi3_jump3: ");
|
sl@0
|
327 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
328 |
asm("SUBCS r0, r0, r2, LSL #4 ");
|
sl@0
|
329 |
asm("CMP r0, r2, LSL #3 ");
|
sl@0
|
330 |
asm("__umodsi3_jump4: ");
|
sl@0
|
331 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
332 |
asm("SUBCS r0, r0, r2, LSL #3 ");
|
sl@0
|
333 |
asm("CMP r0, r2, LSL #2 ");
|
sl@0
|
334 |
asm("__umodsi3_jump5: ");
|
sl@0
|
335 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
336 |
asm("SUBCS r0, r0, r2, LSL #2 ");
|
sl@0
|
337 |
asm("CMP r0, r2, LSL #1 ");
|
sl@0
|
338 |
asm("__umodsi3_jump6: ");
|
sl@0
|
339 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
340 |
asm("SUBCS r0, r0, r2, LSL #1 ");
|
sl@0
|
341 |
asm("__umodsi3_jump7: ");
|
sl@0
|
342 |
asm("CMP r0, r2 ");
|
sl@0
|
343 |
asm("ADC r3, r3, r3 ");
|
sl@0
|
344 |
asm("SUBCS r0, r0, r2 ");
|
sl@0
|
345 |
|
sl@0
|
346 |
asm("CMP r2, r1 ");
|
sl@0
|
347 |
asm("BNE __umodsi3_loop2 ");
|
sl@0
|
348 |
|
sl@0
|
349 |
__JUMP(,lr);
|
sl@0
|
350 |
|
sl@0
|
351 |
asm("modtable: ");
|
sl@0
|
352 |
asm(".word mod_jump_table ");
|
sl@0
|
353 |
//
|
sl@0
|
354 |
// Lookup for optimised divide routines
|
sl@0
|
355 |
//
|
sl@0
|
356 |
asm("mod_jump_table: ");
|
sl@0
|
357 |
asm(".word " DIV_BY_ZERO); // 0
|
sl@0
|
358 |
asm(".word __mod1 "); // 1
|
sl@0
|
359 |
asm(".word __mod2 "); // 2
|
sl@0
|
360 |
asm(".word 0 "); // 3
|
sl@0
|
361 |
asm(".word __mod4 "); // 4
|
sl@0
|
362 |
asm(".word __mod5 "); // 5
|
sl@0
|
363 |
asm(".word 0 "); // 6
|
sl@0
|
364 |
asm(".word __mod7 "); // 7
|
sl@0
|
365 |
asm(".word __mod8 "); // 8
|
sl@0
|
366 |
asm(".word 0 "); // 9
|
sl@0
|
367 |
asm(".word __mod10 "); // 10
|
sl@0
|
368 |
asm(".word 0 "); // 11
|
sl@0
|
369 |
asm(".word 0 "); // 12
|
sl@0
|
370 |
asm(".word 0 "); // 13
|
sl@0
|
371 |
asm(".word 0 "); // 14
|
sl@0
|
372 |
asm(".word 0 "); // 15
|
sl@0
|
373 |
asm(".word __mod16 "); // 16
|
sl@0
|
374 |
|
sl@0
|
375 |
asm("__mod16: ");
|
sl@0
|
376 |
asm("MOV r3,r0,LSR #4 ");
|
sl@0
|
377 |
asm("AND r0,r0,#15 ");
|
sl@0
|
378 |
__JUMP(,lr);
|
sl@0
|
379 |
|
sl@0
|
380 |
asm("__mod1: ");
|
sl@0
|
381 |
asm("MOV r3,r0 ");
|
sl@0
|
382 |
asm("MOV r0,#0 ");
|
sl@0
|
383 |
__JUMP(,lr);
|
sl@0
|
384 |
|
sl@0
|
385 |
asm("__mod2: ");
|
sl@0
|
386 |
asm("MOV r3,r0,LSR #1 ");
|
sl@0
|
387 |
asm("AND r0,r0,#1 ");
|
sl@0
|
388 |
__JUMP(,lr);
|
sl@0
|
389 |
|
sl@0
|
390 |
asm("__mod4: ");
|
sl@0
|
391 |
asm("MOV r3,r0,LSR #2 ");
|
sl@0
|
392 |
asm("AND r0,r0,#3 ");
|
sl@0
|
393 |
__JUMP(,lr);
|
sl@0
|
394 |
|
sl@0
|
395 |
asm("__mod8: ");
|
sl@0
|
396 |
asm("MOV r3,r0,LSR #3 ");
|
sl@0
|
397 |
asm("AND r0,r0,#7 ");
|
sl@0
|
398 |
__JUMP(,lr);
|
sl@0
|
399 |
|
sl@0
|
400 |
asm("__mod10: ");
|
sl@0
|
401 |
asm("MOV r3, r0 ");
|
sl@0
|
402 |
asm("SUB r0, r3, #10 ");
|
sl@0
|
403 |
asm("SUB r3, r3, r3, LSR #2 ");
|
sl@0
|
404 |
asm("ADD r3, r3, r3, LSR #4 ");
|
sl@0
|
405 |
asm("ADD r3, r3, r3, LSR #8 ");
|
sl@0
|
406 |
asm("ADD r3, r3, r3, LSR #16 ");
|
sl@0
|
407 |
asm("MOV r3, r3, LSR #3 ");
|
sl@0
|
408 |
asm("ADD r2, r3, r3, ASL #2 ");
|
sl@0
|
409 |
asm("SUBS r0, r0, r2, ASL #1 ");
|
sl@0
|
410 |
asm("ADDPL r3, r3, #1 ");
|
sl@0
|
411 |
asm("ADDMI r0, r0, #10 ");
|
sl@0
|
412 |
__JUMP(,lr);
|
sl@0
|
413 |
|
sl@0
|
414 |
asm("__mod7: ");
|
sl@0
|
415 |
asm("MOV r3, r0 ");
|
sl@0
|
416 |
asm("SUB r0, r3, #7 ");
|
sl@0
|
417 |
asm("MOV r3, r3, lsr #1 ");
|
sl@0
|
418 |
asm("ADD r3, r3, r3, lsr #3 ");
|
sl@0
|
419 |
asm("ADD r3, r3, r3, lsr #6 ");
|
sl@0
|
420 |
asm("ADD r3, r3, r3, lsr #12 ");
|
sl@0
|
421 |
asm("ADD r3, r3, r3, lsr #24 ");
|
sl@0
|
422 |
asm("MOV r3, r3, lsr #2 ");
|
sl@0
|
423 |
asm("RSB r2, r3, r3, asl #3 ");
|
sl@0
|
424 |
asm("SUBS r0, r0, r2, asl #0 ");
|
sl@0
|
425 |
asm("ADDPL r3, r3, #1 ");
|
sl@0
|
426 |
asm("ADDMI r0, r0, #7 ");
|
sl@0
|
427 |
__JUMP(,lr);
|
sl@0
|
428 |
|
sl@0
|
429 |
asm("__mod5: ");
|
sl@0
|
430 |
asm("MOV r3, r0 ");
|
sl@0
|
431 |
asm("SUB r0, r3, #5 ");
|
sl@0
|
432 |
asm("SUB r3, r3, r3, lsr #2 ");
|
sl@0
|
433 |
asm("ADD r3, r3, r3, lsr #4 ");
|
sl@0
|
434 |
asm("ADD r3, r3, r3, lsr #8 ");
|
sl@0
|
435 |
asm("ADD r3, r3, r3, lsr #16 ");
|
sl@0
|
436 |
asm("MOV r3, r3, lsr #2 ");
|
sl@0
|
437 |
asm("ADD r2, r3, r3, asl #2 ");
|
sl@0
|
438 |
asm("SUBS r0, r0, r2, asl #0 ");
|
sl@0
|
439 |
asm("ADDPL r3, r3, #1 ");
|
sl@0
|
440 |
asm("ADDMI r0, r0, #5 ");
|
sl@0
|
441 |
__JUMP(,lr);
|
sl@0
|
442 |
}
|
sl@0
|
443 |
|
sl@0
|
444 |
|
sl@0
|
445 |
EXPORT_C __NAKED__ unsigned long long __udivdi3(unsigned long long /*dividend*/, unsigned long long /*divisor*/)
|
sl@0
|
446 |
//
|
sl@0
|
447 |
// Dividend in r1:r0, divisor in r3:r2, Return quotient in r1:r0
|
sl@0
|
448 |
//
|
sl@0
|
449 |
{
|
sl@0
|
450 |
asm("stmfd sp!, {r4-r7,lr} ");
|
sl@0
|
451 |
asm("bl UDiv01 "); // do the division, result in r4,r5
|
sl@0
|
452 |
asm("mov r0, r4 ");
|
sl@0
|
453 |
asm("mov r1, r5 ");
|
sl@0
|
454 |
__POPRET("r4-r7,");
|
sl@0
|
455 |
|
sl@0
|
456 |
// Unsigned 64-bit division. Dividend in r0,r1, divisor in r2,r3
|
sl@0
|
457 |
// Quotient returned in r4,r5, Remainder in r3,r6
|
sl@0
|
458 |
// Registers r0-r7,r12 used, r8-r11 unmodified
|
sl@0
|
459 |
asm(".global UDiv01 ");
|
sl@0
|
460 |
asm("UDiv01: ");
|
sl@0
|
461 |
asm("movs r3, r3 "); // check if divisor fits in 32 bits
|
sl@0
|
462 |
asm("bne udiv64a "); // branch if not
|
sl@0
|
463 |
asm("movs r2, r2 "); // check if divisor fits in 31 bits
|
sl@0
|
464 |
asm("bmi udiv64e "); // branch if not
|
sl@0
|
465 |
asm("beq udiv64_divby0 "); // if divisor=0, branch to error routine
|
sl@0
|
466 |
|
sl@0
|
467 |
// Divisor is <0x80000000
|
sl@0
|
468 |
// This means that a 32-bit accumulator is sufficient
|
sl@0
|
469 |
asm("mov r4, #0 "); // use r3 as acc, result in r4, r5
|
sl@0
|
470 |
asm("mov r5, #0 ");
|
sl@0
|
471 |
asm("mov r6, #8 "); // do 2 set of 32 iterations
|
sl@0
|
472 |
asm("udiv64b: ");
|
sl@0
|
473 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
474 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
475 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
476 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
477 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
478 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
479 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
480 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
481 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
482 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
483 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
484 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
485 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
486 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
487 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
488 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
489 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
490 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
491 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
492 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
493 |
asm("subs r6, r6, #1 "); // loop
|
sl@0
|
494 |
asm("bne udiv64b ");
|
sl@0
|
495 |
asm("mov r6, #8 "); // 2nd set of 32 iterations
|
sl@0
|
496 |
asm("udiv64c: ");
|
sl@0
|
497 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
498 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
499 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
500 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
501 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
502 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
503 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
504 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
505 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
506 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
507 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
508 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
509 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
510 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
511 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
512 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
513 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
514 |
asm("subs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
515 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
516 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
517 |
asm("subs r6, r6, #1 "); // loop
|
sl@0
|
518 |
asm("bne udiv64c ");
|
sl@0
|
519 |
__JUMP(,lr);
|
sl@0
|
520 |
|
sl@0
|
521 |
// 2^31 <= Divisor < 2^32
|
sl@0
|
522 |
// Need 33-bit accumulator - use carry flag as 33rd bit
|
sl@0
|
523 |
asm("udiv64e: ");
|
sl@0
|
524 |
asm("mov r4, #0 "); // use r3 as acc, result in r4, r5
|
sl@0
|
525 |
asm("mov r5, #0 ");
|
sl@0
|
526 |
asm("mov r6, #8 "); // do 2 set of 32 iterations
|
sl@0
|
527 |
asm("udiv64f: ");
|
sl@0
|
528 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
529 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
530 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
531 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
532 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
533 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
534 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
535 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
536 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
537 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
538 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
539 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
540 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
541 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
542 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
543 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
544 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
545 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
546 |
asm("adds r1, r1, r1 "); // shift dividend left into acc
|
sl@0
|
547 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
548 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
549 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
550 |
asm("adc r5, r5, r5 "); // shift result bit left into quotient
|
sl@0
|
551 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
552 |
asm("subs r6, r6, #1 "); // loop
|
sl@0
|
553 |
asm("bne udiv64f ");
|
sl@0
|
554 |
asm("mov r6, #8 "); // 2nd set of 32 iterations
|
sl@0
|
555 |
asm("udiv64g: ");
|
sl@0
|
556 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
557 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
558 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
559 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
560 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
561 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
562 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
563 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
564 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
565 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
566 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
567 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
568 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
569 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
570 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
571 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
572 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
573 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
574 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
575 |
asm("adcs r3, r3, r3 ");
|
sl@0
|
576 |
asm("subcs r3, r3, r2 ");
|
sl@0
|
577 |
asm("subccs r3, r3, r2 "); // subtract divisor from acc
|
sl@0
|
578 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
579 |
asm("addcc r3, r3, r2 "); // if borrow, add back
|
sl@0
|
580 |
asm("subs r6, r6, #1 "); // loop
|
sl@0
|
581 |
asm("bne udiv64g ");
|
sl@0
|
582 |
__JUMP(,lr);
|
sl@0
|
583 |
|
sl@0
|
584 |
// Divisor >= 2^32, so quotient < 2^32
|
sl@0
|
585 |
// Use 64 bit accumulator, 32 bit quotient
|
sl@0
|
586 |
asm("udiv64a: ");
|
sl@0
|
587 |
asm("mov r4, #0 "); // quotient in r4, use r1, r6 as accumulator
|
sl@0
|
588 |
asm("mov r6, #0 ");
|
sl@0
|
589 |
asm("mov r5, #8 "); // do 32 iterations
|
sl@0
|
590 |
asm("udiv64d: ");
|
sl@0
|
591 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
592 |
asm("adcs r1, r1, r1 ");
|
sl@0
|
593 |
asm("adcs r6, r6, r6 ");
|
sl@0
|
594 |
asm("subs r7, r1, r2 "); // subtract divisor from acc, result into r7,r12
|
sl@0
|
595 |
asm("sbcs r12, r6, r3 ");
|
sl@0
|
596 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
597 |
asm("movcs r1, r7 "); // if no borrow, update acc
|
sl@0
|
598 |
asm("movcs r6, r12 ");
|
sl@0
|
599 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
600 |
asm("adcs r1, r1, r1 ");
|
sl@0
|
601 |
asm("adcs r6, r6, r6 ");
|
sl@0
|
602 |
asm("subs r7, r1, r2 "); // subtract divisor from acc, result into r7,r12
|
sl@0
|
603 |
asm("sbcs r12, r6, r3 ");
|
sl@0
|
604 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
605 |
asm("movcs r1, r7 "); // if no borrow, update acc
|
sl@0
|
606 |
asm("movcs r6, r12 ");
|
sl@0
|
607 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
608 |
asm("adcs r1, r1, r1 ");
|
sl@0
|
609 |
asm("adcs r6, r6, r6 ");
|
sl@0
|
610 |
asm("subs r7, r1, r2 "); // subtract divisor from acc, result into r7,r12
|
sl@0
|
611 |
asm("sbcs r12, r6, r3 ");
|
sl@0
|
612 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
613 |
asm("movcs r1, r7 "); // if no borrow, update acc
|
sl@0
|
614 |
asm("movcs r6, r12 ");
|
sl@0
|
615 |
asm("adds r0, r0, r0 "); // shift dividend left into acc
|
sl@0
|
616 |
asm("adcs r1, r1, r1 ");
|
sl@0
|
617 |
asm("adcs r6, r6, r6 ");
|
sl@0
|
618 |
asm("subs r7, r1, r2 "); // subtract divisor from acc, result into r7,r12
|
sl@0
|
619 |
asm("sbcs r12, r6, r3 ");
|
sl@0
|
620 |
asm("adc r4, r4, r4 "); // shift result bit left into quotient
|
sl@0
|
621 |
asm("movcs r1, r7 "); // if no borrow, update acc
|
sl@0
|
622 |
asm("movcs r6, r12 ");
|
sl@0
|
623 |
asm("subs r5, r5, #1 "); // loop
|
sl@0
|
624 |
asm("bne udiv64d ");
|
sl@0
|
625 |
asm("mov r3, r1 "); // remainder in r3,r6
|
sl@0
|
626 |
__JUMP(,lr);
|
sl@0
|
627 |
|
sl@0
|
628 |
asm("udiv64_divby0: ");
|
sl@0
|
629 |
asm("stmfd sp!, {r11,lr} ");
|
sl@0
|
630 |
__EH_FRAME_PUSH2(r11,lr)
|
sl@0
|
631 |
asm("mov r11, sp ");
|
sl@0
|
632 |
asm("bic sp, sp, #4 ");
|
sl@0
|
633 |
asm("bl " DIV_BY_ZERO);
|
sl@0
|
634 |
asm("mov sp, r11 ");
|
sl@0
|
635 |
__POPRET("r11,");
|
sl@0
|
636 |
}
|
sl@0
|
637 |
|
sl@0
|
638 |
}
|
sl@0
|
639 |
|