I'm trying to shift eax right by 2, currently it is 0x037f and I want to get it to 0x0003.
I'm using Cygwin, it compiles fine but when I open it with gdb, once I get to line 7 I get the title error.
I've tried changing the syntax to (%eax, 2, 0) (I want to shift it left by 2 and fill with 0's) I also tried with just a 2, it has a value of 8 because I don't know if its shifting by bytes or bits.
asm(" push %%eax \n"
" push %%ebx \n"
" push %[nIn] \n"
" fstcw %[trueCWIn] \n" //store FPU CW into trueCW
" mov %[trueCWOut], %%eax \n" //store old FPU CW into tempVar
" mov %%eax, %[tempVarIn] \n" //store old FPU CW into tempVar
" shrd %%eax,8 \n" //shift FPU CW right to bytes
" add %[roundModeOut], %%eax \n" //adding rounding modifier
//shift left 2 bytes
//add in restore mask (0x007F)
" frndint \n" //do rounding calculation
//store result into n
" fldcw %[trueCWIn] \n" //restoring the FPU CW to normal
" pop %%ebx \n"
" pop %%eax \n"
: [trueCWOut] "=m" (trueCW),
[tempCWOut] "=m" (tempCW),
[maskOut] "=m" (mask),
[tempVarOut] "=m" (tempVar),
[roundModeOut] "=m" (roundMode),
[nOut] "=m" (n)
: [trueCWIn] "m" (trueCW),
[tempCWIn] "m" (tempCW),
[maskIn] "m" (mask),
[tempVarIn] "m" (tempVar),
[roundModeIn] "m" (roundMode),
[nIn] "m" (n)
:"eax", "ebx"
);
I found the answer, it was a simple syntax problem.
shr $8, %%eax
Before, eax was 0x037F. This shifted eax 8 bits, which then made it 0x0003.