I want to load double from the stack:

fldq 12(%ebp)

however, the compiler curses: Error: invalid instruction suffix for `fld '

What is the problem?

  • Did you read the fldq documentation? - nick_n_a
  • @nick_n_a, no, I did not read. Although I was searching in the internet, I didn’t really find anything. And should it be fundamentally different from fldl and fldt ? These two work the same ... - Jin X
  • With fstq and fstpq the same story. He still does n’t know fstt (but fstpt works fine) ... - Jin X
  • I figured out with fldq , fstq , fstpq ! It turns out that fldl , fstl , fstpl generate instructions with qword ptr , and to generate dword ptr you need to write flds , fsts , fstps . Here it is a tupnyak! AT & T zhzhot easy! As for fstt , the fst instruction, it turns out, does not work at all with tbyte ptr (according to the Intel specification - looked in the directory), but ftsp works! Here's a parsley :) - Jin X

0