c - Getting absolute value from a binary int using bit arithmetics -

flt32 flt32_abs (flt32 x) {     int mask=x>>31;      printmask(mask,32);     puts("original");     printmask(x,32);      x=x^mask;      puts("after xor");      printmask(x,32);      x=x-mask;      puts("after x-mask");     printmask(x,32);     return x; } 

here's code, calling function on value -32 returning .125. i'm confused because it's pretty straight formula abs on bits, seem missing something. ideas?

is flt32 type floating point or fixed point numbers?

i suspect it's type fixed point arithmetic , not using correctly. let me explain it.

a fixed-point number uses, name says, fixed position decimal digit; means uses fixed number of bits decimal part. is, in fact, scaled integer.

i guess flt32 type using uses significant 24 bits whole part , least significant 8 bits decimal part; value real number of 32-bit representation value of same 32 bit representation integer, divided 256 (i.e. 2⁸).

for example, 32-bit number 0x00000020 interpreted integer 32. fixed-point number using 8 bits decimal part, value 0.125 (=32/256).

the code posted correct not using correctly.

the number -32 encoded fixed-point number using 8 decimal digits 0xffffe000 integer representation of -8192 (=-32*256). algorithm correctly produces 8192 0x00002000 (=32*256); 32 when interpreted fixed-point.

if pass -32 function without taking care encode fixed-point, correctly converts 32 , returns value. 32 (0x00000020) 0.125 (=1/8=32/256) when interpreted fixed-point (what assume function printmask() does).

how can test code correctly?

you have function creates fixed-point numbers integers. use correct representation of -32 , pass value flt32_abs() function.

in case don't have such function, easy write it. multiply integer 256 (or better, left-shift 8 bits) , that's all:

function int_to_fx32(int x) {     return x << 8; } 

the fixed-point libraries use macros such conversions because produce faster code. expressed macro, looks this:

#define int_to_fx32(x) ((x) << 8) 

now test:

fx32 negative = int_to_fx32(-32); fx32 positive = fx32_abs(negative); // should print 32 printmask(positive, 32);  // should print 8192 printf("%d", positive); // should print -8192 printf("%d", negative);  // should print 0.125 printmask(32, 32);