os/security/crypto/weakcryptospi/source/bigint/gcchelp.cpp
changeset 0 bde4ae8d615e
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/os/security/crypto/weakcryptospi/source/bigint/gcchelp.cpp	Fri Jun 15 03:10:57 2012 +0200
     1.3 @@ -0,0 +1,395 @@
     1.4 +/*
     1.5 +* Copyright (c) 2005-2009 Nokia Corporation and/or its subsidiary(-ies).
     1.6 +* All rights reserved.
     1.7 +* This component and the accompanying materials are made available
     1.8 +* under the terms of the License "Eclipse Public License v1.0"
     1.9 +* which accompanies this distribution, and is available
    1.10 +* at the URL "http://www.eclipse.org/legal/epl-v10.html".
    1.11 +*
    1.12 +* Initial Contributors:
    1.13 +* Nokia Corporation - initial contribution.
    1.14 +*
    1.15 +* Contributors:
    1.16 +*
    1.17 +* Description: 
    1.18 +*
    1.19 +*/
    1.20 +
    1.21 +
    1.22 +/**
    1.23 + @file
    1.24 +*/
    1.25 +
    1.26 +
    1.27 +#include <kernel\u32std.h>	// need this for __NAKED__, __JUMP and __POPRET
    1.28 +
    1.29 +extern "C" {
    1.30 +#ifdef __GCC32__
    1.31 +__NAKED__ long long __divdi3(long long /*dividend*/, long long /*divisor*/)
    1.32 +//
    1.33 +// Dividend in r1:r0, divisor in r3:r2, Return quotient in r1:r0
    1.34 +//
    1.35 +	{
    1.36 +	asm("stmfd sp!, {r4-r8,lr} ");
    1.37 +	asm("eor r8, r1, r3 ");				// sign of result into r8
    1.38 +	asm("movs r1, r1 ");
    1.39 +	asm("bpl 1f ");
    1.40 +	asm("rsbs r0, r0, #0 ");			// ABS(dividend)
    1.41 +	asm("rsc r1, r1, #0 ");
    1.42 +	asm("1: ");
    1.43 +	asm("movs r3, r3 ");
    1.44 +	asm("bpl 2f ");
    1.45 +	asm("rsbs r2, r2, #0 ");			// ABS(divisor)
    1.46 +	asm("rsc r3, r3, #0 ");
    1.47 +	asm("2: ");
    1.48 +	asm("bl UDiv01 ");					// do the division, result in r4,r5
    1.49 +	asm("eors r0, r4, r8, asr #32 ");	// quotient into r1:r0, inverted if quotient -ve
    1.50 +	asm("eors r1, r5, r8, asr #32 ");
    1.51 +	asm("adcs r0, r0, #0 ");		// if quotient -ve, add 1
    1.52 +	asm("adcs r1, r1, #0 ");
    1.53 +	__POPRET("r4-r8,");
    1.54 +	}
    1.55 +
    1.56 +__NAKED__ long long __moddi3(long long /*dividend*/, long long /*divisor*/)	/* signed */
    1.57 +	{
    1.58 +	asm("stmfd sp!, {r4-r8,lr} ");
    1.59 +	asm("movs r8, r1 ");				// sign of remainder (=sign of dividend) into r8
    1.60 +	asm("bpl 1f ");
    1.61 +	asm("rsbs r0, r0, #0 ");			// ABS(dividend)
    1.62 +	asm("rsc r1, r1, #0 ");
    1.63 +	asm("1: ");
    1.64 +	asm("movs r3, r3 ");
    1.65 +	asm("bpl 2f ");
    1.66 +	asm("rsbs r2, r2, #0 ");			// ABS(divisor)
    1.67 +	asm("rsc r3, r3, #0 ");
    1.68 +	asm("2: ");
    1.69 +	asm("bl UDiv01 ");					// do the division, remainder in r3,r6
    1.70 +	asm("eors r0, r3, r8, asr #32 ");	// remainder into r1:r0, inverted if dividend -ve
    1.71 +	asm("eors r1, r6, r8, asr #32 ");
    1.72 +	asm("adcs r0, r0, #0 ");			// if dividend -ve, add 1
    1.73 +	asm("adcs r1, r1, #0 ");
    1.74 +	__POPRET("r4-r8,");
    1.75 +	}
    1.76 +
    1.77 +__NAKED__ long long __umoddi3(unsigned long long /*dividend*/, unsigned long long /*divisor*/)	/* unsigned */
    1.78 +	{
    1.79 +	asm("stmfd sp!, {r4-r7,lr} ");
    1.80 +	asm("bl UDiv01 ");					// do the division, remainder in r6:r3
    1.81 +	asm("mov r0, r3 ");
    1.82 +	asm("mov r1, r6 ");
    1.83 +	__POPRET("r4-r7,");
    1.84 +	}
    1.85 +
    1.86 +__NAKED__ long long __ashrdi3(long long /*value*/, unsigned int /*count*/)
    1.87 +	{
    1.88 +	asm("cmp r2, #63 ");
    1.89 +	asm("movhi r2, #63 ");			// count>63 same as count=63
    1.90 +	asm("cmp r2, #32 ");
    1.91 +	asm("bcs Asr01 ");				// jump if shift count >=32
    1.92 +	asm("rsb r12, r2, #32 ");		// r12=32-shift count
    1.93 +	asm("mov r0, r0, lsr r2 ");		// shift ls word right
    1.94 +	asm("orr r0, r0, r1, lsl r12 ");	// or in bits shifted out of ms word
    1.95 +	asm("mov r1, r1, asr r2 ");		// shift ms word right
    1.96 +	__JUMP(,lr);
    1.97 +	asm("Asr01: ");
    1.98 +	asm("sub r2, r2, #32 ");		// r2=shift count-32
    1.99 +	asm("mov r0, r1, asr r2 ");		// ls word = ms word >> (count-32)
   1.100 +	asm("mov r1, r1, asr #32 ");	// ms word of result=sign extension of r1
   1.101 +	__JUMP(,lr);
   1.102 +	}
   1.103 +
   1.104 +__NAKED__ long long __ashldi3(long long /*value*/, unsigned int /*count*/)
   1.105 +	{
   1.106 +	asm("cmp r2, #63 ");
   1.107 +	asm("movhi r2, #64 ");			// count>63 same as count=64
   1.108 +	asm("cmp r2, #32 ");
   1.109 +	asm("bcs Asl01 ");				// jump if shift count >=32
   1.110 +	asm("rsb r12, r2, #32 ");		// r12=32-shift count
   1.111 +	asm("mov r1, r1, asl r2 ");		// shift ms word left
   1.112 +	asm("orr r1, r1, r0, lsr r12 ");	// or in bits shifted out of ls word
   1.113 +	asm("mov r0, r0, asl r2 ");		// shift ls word left
   1.114 +	__JUMP(,lr);
   1.115 +	asm("Asl01: ");
   1.116 +	asm("sub r2, r2, #32 ");		// r2=shift count-32
   1.117 +	asm("mov r1, r0, asl r2 ");		// result ms word = ls word << (count-32)
   1.118 +	asm("mov r0, #0 ");				// ls word of result is zero
   1.119 +	__JUMP(,lr);
   1.120 +	}
   1.121 +
   1.122 +__NAKED__ unsigned long long __lshrdi3(unsigned long long /*value*/, unsigned int /*count*/)
   1.123 +	{
   1.124 +	asm("cmp r2, #63 ");
   1.125 +	asm("movhi r2, #64 ");			// count>63 same as count=64
   1.126 +	asm("cmp r2, #32 ");
   1.127 +	asm("bcs Lsr01 ");				// jump if shift count >=32
   1.128 +	asm("rsb r12, r2, #32 ");		// r12=32-shift count
   1.129 +	asm("mov r0, r0, lsr r2 ");		// shift ls word right
   1.130 +	asm("orr r0, r0, r1, lsl r12 ");	// or in bits shifted out of ms word
   1.131 +	asm("mov r1, r1, lsr r2 ");		// shift ms word right
   1.132 +	__JUMP(,lr);
   1.133 +	asm("Lsr01: ");
   1.134 +	asm("sub r2, r2, #32 ");		// r2=shift count-32
   1.135 +	asm("mov r0, r1, lsr r2 ");		// ls word = ms word >> (count-32)
   1.136 +	asm("mov r1, #0 ");				// ms word of result = 0
   1.137 +	__JUMP(,lr);
   1.138 +	}
   1.139 +
   1.140 +__NAKED__ long long __muldi3(long long /*multiplicand*/, long long /*multiplier*/)
   1.141 +	{
   1.142 +	asm("mul r1, r2, r1 ");				// r1=low2*high1
   1.143 +	asm("mov ip, r0 ");					// ip=low1
   1.144 +	asm("mla r1, r0, r3, r1 ");			// r1+=low1*high2
   1.145 +	asm("mov r0, #0 ");
   1.146 +	asm("umlal r0, r1, r2, ip ");		// r1:r0 += high1*low1
   1.147 +	__JUMP(,lr);
   1.148 +	}
   1.149 +
   1.150 +__NAKED__ long long __negdi2(long long /*argument*/)
   1.151 +	{
   1.152 +	asm("rsbs r0, r0, #0 ");		// r0=0-r0, set carry
   1.153 +	asm("rscs r1, r1, #0 ");		// r1=0-r1-(1-C)
   1.154 +	__JUMP(,lr);
   1.155 +	}
   1.156 +
   1.157 +__NAKED__ unsigned long long __udivmoddi4 (unsigned long long /*dividend*/,
   1.158 +													unsigned long long /*divisor*/,
   1.159 +													unsigned long long* /*p_remainder*/)
   1.160 +	{
   1.161 +	asm("stmfd sp!, {r4-r7,lr} ");
   1.162 +	asm("bl UDiv01 ");					// do the division, quotient in r5:r4 remainder in r6:r3
   1.163 +	asm("ldr r7, [sp, #20] ");			// r7=p_remainder
   1.164 +	asm("mov r0, r4 ");					// r0=quotient low
   1.165 +	asm("stmia r7, {r3,r6} ");			// store remainder
   1.166 +	asm("mov r1, r5 ");					// r0=quotient high
   1.167 +	__POPRET("r4-r7,");
   1.168 +	}
   1.169 +
   1.170 +__NAKED__ int __cmpdi2(long long /*a*/, long long /*b*/)
   1.171 +	{
   1.172 +	// return 0 if a<b, 1 if a=b, 2 if a>b
   1.173 +	asm("subs r0, r2, r0 ");
   1.174 +	asm("sbcs r1, r3, r1 ");			// r1:r0 = b-a, set flags
   1.175 +	asm("movlt r0, #2 ");				// if b<a r0=2
   1.176 +	__JUMP(lt,lr);						// if b<a return
   1.177 +	asm("cmpeq r0, #0 ");				// if top word of difference=0, look at bottom
   1.178 +	asm("moveq r0, #1 ");				// if a=b, r0=1
   1.179 +	asm("movne r0, #0 ");				// else r=0
   1.180 +	__JUMP(,lr);
   1.181 +	}
   1.182 +
   1.183 +__NAKED__ int __ucmpdi2(unsigned long long /*a*/, unsigned long long /*b*/)
   1.184 +	{
   1.185 +	// return 0 if a<b, 1 if a=b, 2 if a>b
   1.186 +	asm("cmp r1, r3 ");
   1.187 +	asm("cmpeq r0, r2 ");				// compare r1:r0 - r3:r2
   1.188 +	asm("movhi r0, #2 ");				// r0=2 if a>b
   1.189 +	asm("moveq r0, #1 ");				// r0=1 if a=b
   1.190 +	asm("movlo r0, #0 ");				// r0=0 if a<b
   1.191 +	__JUMP(,lr);
   1.192 +	}
   1.193 +#endif
   1.194 +
   1.195 +#if defined(__GCC32__)
   1.196 +void __division_by_zero();
   1.197 +#define DIV_BY_ZERO " __division_by_zero "
   1.198 +#elif defined(__ARMCC__)
   1.199 +void __rt_div0 (void);
   1.200 +#define DIV_BY_ZERO " __cpp(__rt_div0) "
   1.201 +#endif
   1.202 +
   1.203 +void __division_by_zero()
   1.204 +	{
   1.205 +	RThread().RaiseException(EExcIntegerDivideByZero);
   1.206 +    }
   1.207 +
   1.208 +__NAKED__ unsigned long long __udivdi3(unsigned long long /*dividend*/, unsigned long long /*divisor*/)
   1.209 +//
   1.210 +// Dividend in r1:r0, divisor in r3:r2, Return quotient in r1:r0
   1.211 +//
   1.212 +	{
   1.213 +	asm("stmfd sp!, {r4-r7,lr} ");
   1.214 +	asm("bl UDiv01 ");					// do the division, result in r4,r5
   1.215 +	asm("mov r0, r4 ");
   1.216 +	asm("mov r1, r5 ");
   1.217 +	__POPRET("r4-r7,");
   1.218 +
   1.219 +	// Unsigned 64-bit division. Dividend in r0,r1, divisor in r2,r3
   1.220 +	// Quotient returned in r4,r5, Remainder in r3,r6
   1.221 +	// Registers r0-r7,r12 used, r8-r11 unmodified
   1.222 +	asm(".global UDiv01 ");
   1.223 +	asm("UDiv01: ");
   1.224 +	asm("movs r3, r3 ");				// check if divisor fits in 32 bits
   1.225 +	asm("bne udiv64a ");				// branch if not
   1.226 +	asm("movs r2, r2 ");				// check if divisor fits in 31 bits
   1.227 +	asm("bmi udiv64e ");				// branch if not
   1.228 +	asm("beq udiv64_divby0 ");			// if divisor=0, branch to error routine
   1.229 +
   1.230 +	// Divisor is <0x80000000
   1.231 +	// This means that a 32-bit accumulator is sufficient
   1.232 +	asm("mov r4, #0 ");					// use r3 as acc, result in r4, r5
   1.233 +	asm("mov r5, #0 ");
   1.234 +	asm("mov r6, #8 ");					// do 2 set of 32 iterations
   1.235 +	asm("udiv64b: ");
   1.236 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.237 +	asm("adcs r3, r3, r3 ");
   1.238 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.239 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.240 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.241 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.242 +	asm("adcs r3, r3, r3 ");
   1.243 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.244 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.245 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.246 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.247 +	asm("adcs r3, r3, r3 ");
   1.248 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.249 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.250 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.251 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.252 +	asm("adcs r3, r3, r3 ");
   1.253 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.254 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.255 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.256 +	asm("subs r6, r6, #1 ");			// loop
   1.257 +	asm("bne udiv64b ");
   1.258 +	asm("mov r6, #8 ");					// 2nd set of 32 iterations
   1.259 +	asm("udiv64c: ");
   1.260 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.261 +	asm("adcs r3, r3, r3 ");
   1.262 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.263 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.264 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.265 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.266 +	asm("adcs r3, r3, r3 ");
   1.267 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.268 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.269 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.270 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.271 +	asm("adcs r3, r3, r3 ");
   1.272 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.273 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.274 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.275 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.276 +	asm("adcs r3, r3, r3 ");
   1.277 +	asm("subs r3, r3, r2 ");			// subtract divisor from acc
   1.278 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.279 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.280 +	asm("subs r6, r6, #1 ");			// loop
   1.281 +	asm("bne udiv64c ");
   1.282 +	__JUMP(,lr);
   1.283 +
   1.284 +	// 2^31 <= Divisor < 2^32
   1.285 +	// Need 33-bit accumulator - use carry flag as 33rd bit
   1.286 +	asm("udiv64e: ");
   1.287 +	asm("mov r4, #0 ");					// use r3 as acc, result in r4, r5
   1.288 +	asm("mov r5, #0 ");
   1.289 +	asm("mov r6, #8 ");					// do 2 set of 32 iterations
   1.290 +	asm("udiv64f: ");
   1.291 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.292 +	asm("adcs r3, r3, r3 ");
   1.293 +	asm("subcs r3, r3, r2 ");
   1.294 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.295 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.296 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.297 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.298 +	asm("adcs r3, r3, r3 ");
   1.299 +	asm("subcs r3, r3, r2 ");
   1.300 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.301 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.302 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.303 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.304 +	asm("adcs r3, r3, r3 ");
   1.305 +	asm("subcs r3, r3, r2 ");
   1.306 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.307 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.308 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.309 +	asm("adds r1, r1, r1 ");			// shift dividend left into acc
   1.310 +	asm("adcs r3, r3, r3 ");
   1.311 +	asm("subcs r3, r3, r2 ");
   1.312 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.313 +	asm("adc r5, r5, r5 ");				// shift result bit left into quotient
   1.314 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.315 +	asm("subs r6, r6, #1 ");			// loop
   1.316 +	asm("bne udiv64f ");
   1.317 +	asm("mov r6, #8 ");					// 2nd set of 32 iterations
   1.318 +	asm("udiv64g: ");
   1.319 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.320 +	asm("adcs r3, r3, r3 ");
   1.321 +	asm("subcs r3, r3, r2 ");
   1.322 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.323 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.324 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.325 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.326 +	asm("adcs r3, r3, r3 ");
   1.327 +	asm("subcs r3, r3, r2 ");
   1.328 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.329 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.330 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.331 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.332 +	asm("adcs r3, r3, r3 ");
   1.333 +	asm("subcs r3, r3, r2 ");
   1.334 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.335 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.336 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.337 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.338 +	asm("adcs r3, r3, r3 ");
   1.339 +	asm("subcs r3, r3, r2 ");
   1.340 +	asm("subccs r3, r3, r2 ");			// subtract divisor from acc
   1.341 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.342 +	asm("addcc r3, r3, r2 ");			// if borrow, add back
   1.343 +	asm("subs r6, r6, #1 ");			// loop
   1.344 +	asm("bne udiv64g ");
   1.345 +	__JUMP(,lr);
   1.346 +	
   1.347 +	// Divisor >= 2^32, so quotient < 2^32
   1.348 +	// Use 64 bit accumulator, 32 bit quotient
   1.349 +	asm("udiv64a: ");
   1.350 +	asm("mov r4, #0 ");					// quotient in r4, use r1, r6 as accumulator
   1.351 +	asm("mov r6, #0 ");
   1.352 +	asm("mov r5, #8 ");					// do 32 iterations
   1.353 +	asm("udiv64d: ");
   1.354 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.355 +	asm("adcs r1, r1, r1 ");
   1.356 +	asm("adcs r6, r6, r6 ");
   1.357 +	asm("subs r7, r1, r2 ");			// subtract divisor from acc, result into r7,r12
   1.358 +	asm("sbcs r12, r6, r3 ");
   1.359 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.360 +	asm("movcs r1, r7 ");				// if no borrow, update acc
   1.361 +	asm("movcs r6, r12 ");
   1.362 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.363 +	asm("adcs r1, r1, r1 ");
   1.364 +	asm("adcs r6, r6, r6 ");
   1.365 +	asm("subs r7, r1, r2 ");			// subtract divisor from acc, result into r7,r12
   1.366 +	asm("sbcs r12, r6, r3 ");
   1.367 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.368 +	asm("movcs r1, r7 ");				// if no borrow, update acc
   1.369 +	asm("movcs r6, r12 ");
   1.370 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.371 +	asm("adcs r1, r1, r1 ");
   1.372 +	asm("adcs r6, r6, r6 ");
   1.373 +	asm("subs r7, r1, r2 ");			// subtract divisor from acc, result into r7,r12
   1.374 +	asm("sbcs r12, r6, r3 ");
   1.375 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.376 +	asm("movcs r1, r7 ");				// if no borrow, update acc
   1.377 +	asm("movcs r6, r12 ");
   1.378 +	asm("adds r0, r0, r0 ");			// shift dividend left into acc
   1.379 +	asm("adcs r1, r1, r1 ");
   1.380 +	asm("adcs r6, r6, r6 ");
   1.381 +	asm("subs r7, r1, r2 ");			// subtract divisor from acc, result into r7,r12
   1.382 +	asm("sbcs r12, r6, r3 ");
   1.383 +	asm("adc r4, r4, r4 ");				// shift result bit left into quotient
   1.384 +	asm("movcs r1, r7 ");				// if no borrow, update acc
   1.385 +	asm("movcs r6, r12 ");
   1.386 +	asm("subs r5, r5, #1 ");			// loop
   1.387 +	asm("bne udiv64d ");
   1.388 +	asm("mov r3, r1 ");					// remainder in r3,r6
   1.389 +	__JUMP(,lr);
   1.390 +
   1.391 +	asm("udiv64_divby0: ");
   1.392 +	asm("str lr, [sp, #-4]! ");
   1.393 +	asm("bl " DIV_BY_ZERO);
   1.394 +	__POPRET("");
   1.395 +	}
   1.396 +
   1.397 +}
   1.398 +