1 /*===-- umodsi3.S - 32-bit unsigned integer modulus -----------------------===// 2 * 3 * The LLVM Compiler Infrastructure 4 * 5 * This file is dual licensed under the MIT and the University of Illinois Open 6 * Source Licenses. See LICENSE.TXT for details. 7 * 8 *===----------------------------------------------------------------------===// 9 * 10 * This file implements the __umodsi3 (32-bit unsigned integer modulus) 11 * function for the ARM architecture. A naive digit-by-digit computation is 12 * employed for simplicity. 13 * 14 *===----------------------------------------------------------------------===*/ 15 16 #include "../assembly.h" 17 18 #define a r0 19 #define b r1 20 #define r r2 21 #define i r3 22 23 .syntax unified 24 .align 3 25 DEFINE_COMPILERRT_FUNCTION(__umodsi3) 26 #if __ARM_ARCH_7S__ 27 tst r1, r1 28 beq LOCAL_LABEL(divzero) 29 udiv r2, r0, r1 30 mls r0, r2, r1, r0 31 bx lr 32 LOCAL_LABEL(divzero): 33 mov r0, #0 34 bx lr 35 #else 36 // We use a simple digit by digit algorithm; before we get into the actual 37 // divide loop, we must calculate the left-shift amount necessary to align 38 // the MSB of the divisor with that of the dividend. 39 clz r2, a 40 tst b, b // detect b == 0 41 clz r3, b 42 bxeq lr // return a if b == 0 43 subs i, r3, r2 44 bxlt lr // return a if MSB(a) < MSB(b) 45 46 LOCAL_LABEL(mainLoop): 47 // This loop basically implements the following: 48 // 49 // do { 50 // if (a >= b << i) { 51 // a -= b << i; 52 // if (a == 0) break; 53 // } 54 // } while (--i) 55 // 56 // Note that this does not perform the final iteration (i == 0); by doing it 57 // this way, we can merge the two branches which is a substantial win for 58 // such a tight loop on current ARM architectures. 59 subs r, a, b, lsl i 60 movhs a, r 61 subsne i, i, #1 62 bhi LOCAL_LABEL(mainLoop) 63 64 // Do the final test subtraction and update of remainder (i == 0), as it is 65 // not performed in the main loop. 66 subs r, a, b 67 movhs a, r 68 bx lr 69 #endif 70
