Intel and compatable CPU's Programming Information
prev: LODS/LODSB/LODSW/LODSD Load String Operand
next: LSL Load Segment Limit
LOOP/LOOPcond -- Loop Control with CX Counter
Opcode Instruction Clocks Description E2 cb LOOP rel8 11+m DEC count; jump short if count <> 0 E1 cb LOOPE rel8 11+m DEC count; jump short if count <> 0 and ZF=1 E1 cb LOOPZ rel8 11+m DEC count; jump short if count <> 0 and ZF=1 E0 cb LOOPNE rel8 11+m DEC count; jump short if count <> 0 and ZF=0 E0 cb LOOPNZ rel8 11+m DEC count; jump short if count <> 0 and ZF=0
IF AddressSize = 16 THEN CountReg is CX ELSE CountReg is ECX; FI; CountReg := CountReg - 1; IF instruction <> LOOP THEN IF (instruction = LOOPE) OR (instruction = LOOPZ) THEN BranchCond := (ZF = 1) AND (CountReg <> 0); FI; IF (instruction = LOOPNE) OR (instruction = LOOPNZ) THEN BranchCond := (ZF = 0) AND (CountReg <> 0); FI; FI; IF BranchCond THEN IF OperandSize = 16 THEN IP := IP + SignExtend(rel8); ELSE (* OperandSize = 32 *) EIP := EIP + SignExtend(rel8); FI; FI;
DescriptionLOOP decrements the count register without changing any of the flags. Conditions are then checked for the form of LOOP being used. If the conditions are met, a short jump is made to the label given by the operand to LOOP. If the address-size attribute is 16 bits, the CX register is used as the count register; otherwise the ECX register is used. The operand of LOOP must be in the range from 128 (decimal) bytes before the instruction to 127 bytes ahead of the instruction.
The LOOP instructions provide iteration control and combine loop index management with conditional branching. Use the LOOP instruction by loading an unsigned iteration count into the count register, then code the LOOP at the end of a series of instructions to be iterated. The destination of LOOP is a label that points to the beginning of the iteration.
Protected Mode Exceptions#GP(0) if the offset jumped to is beyond the limits of the current code segment
Real Address Mode ExceptionsNone
Virtual 8086 Mode ExceptionsNone