View text source at Wikipedia
The FLAGS register is the status register that contains the current state of an x86 CPU. The size and meanings of the flag bits are architecture dependent. It usually reflects the result of arithmetic operations as well as information about restrictions placed on the CPU operation at the current time. Some of those restrictions may include preventing some interrupts from triggering, prohibition of execution of a class of "privileged" instructions. Additional status flags may bypass memory mapping and define what action the CPU should take on arithmetic overflow.
The carry, parity, auxiliary carry (or half carry), zero and sign flags are included in many architectures (many modern (RISC) architectures do not have flags, such as carry, and even if they do use flags, then half carry is rare, since BCD math no longer common, and it even has limited support on long mode on x86-64).
In the i286 architecture, the register is 16 bits wide. Its successors, the EFLAGS and RFLAGS registers (in modern x86-64), are 32 bits and 64 bits wide, respectively. The wider registers retain compatibility with their smaller predecessors.
Intel x86 FLAGS register[1] | ||||||
---|---|---|---|---|---|---|
Bit # | Mask | Abbreviation | Description | Category | =1 | =0 |
FLAGS | ||||||
0 | 0x0001 | CF | Carry flag | Status | CY (Carry) | NC (No Carry) |
1 | 0x0002 | — | Reserved, always 1 in EFLAGS[2][3] | — | ||
2 | 0x0004 | PF | Parity flag | Status | PE (Parity Even) | PO (Parity Odd) |
3 | 0x0008 | — | Reserved[3] | — | ||
4 | 0x0010 | AF | Auxiliary Carry flag[4] | Status | AC (Auxiliary Carry) | NA (No Auxiliary Carry) |
5 | 0x0020 | — | Reserved[3] | — | ||
6 | 0x0040 | ZF | Zero flag | Status | ZR (Zero) | NZ (Not Zero) |
7 | 0x0080 | SF | Sign flag | Status | NG (Negative) | PL (Positive) |
8 | 0x0100 | TF | Trap flag (single step) | Control | ||
9 | 0x0200 | IF | Interrupt enable flag | Control | EI (Enable Interrupt) | DI (Disable Interrupt) |
10 | 0x0400 | DF | Direction flag | Control | DN (Down) | UP (Up) |
11 | 0x0800 | OF | Overflow flag | Status | OV (Overflow) | NV (Not Overflow) |
12–13 | 0x3000 | IOPL | I/O privilege level (286+ only), always all-1s on 8086 and 186 |
System | ||
14 | 0x4000 | NT | Nested task flag (286+ only), always 1 on 8086 and 186 |
System | ||
15 | 0x8000 | MD | Mode flag (NEC V-series only),[5] reserved on all Intel CPUs. Always 1 on 8086/186, 0 on 286 and later. |
Control | (NEC only) Native Mode (186 compatible) |
(NEC only) Emulation Mode (8080 compatible) |
EFLAGS | ||||||
16 | 0x0001 0000 | RF | Resume flag (386+ only) | System | ||
17 | 0x0002 0000 | VM | Virtual 8086 mode flag (386+ only) | System | ||
18 | 0x0004 0000 | AC | Alignment Check (486+, ring 3), SMAP Access Check (Broadwell+, ring 0-2) |
System | ||
19 | 0x0008 0000 | VIF | Virtual interrupt flag (Pentium+) | System | ||
20 | 0x0010 0000 | VIP | Virtual interrupt pending (Pentium+) | System | ||
21 | 0x0020 0000 | ID | Able to use CPUID instruction (Pentium+) | System | ||
22–29 | 0x3FC0 0000 | — | Reserved | — | ||
30 | 0x4000 0000 | (none) | AES key schedule loaded flag[6] (CPUs with VIA PadLock only) |
System | ||
31 | 0x8000 0000 | AI | Alternate Instruction Set enabled (VIA C5XL processors only)[7] |
System | ||
RFLAGS | ||||||
32‑63 | 0xFFFF FFFF… …0000 0000 |
— | Reserved | — |
Note: The mask column in the table is the AND bitmask (as hexadecimal value) to query the flag(s) within FLAGS register value.
All FLAGS registers contain the condition codes, flag bits that let the results of one machine-language instruction affect another instruction. Arithmetic and logical instructions set some or all of the flags, and conditional jump instructions take variable action based on the value of certain flags. For example, jz
(Jump if Zero), jc
(Jump if Carry), and jo
(Jump if Overflow) depend on specific flags. Other conditional jumps test combinations of several flags.
FLAGS registers can be moved from or to the stack. This is part of the job of saving and restoring CPU context, against a routine such as an interrupt service routine whose changes to registers should not be seen by the calling code. Here are the relevant instructions:
In 64-bit mode, PUSHF/POPF and PUSHFQ/POPFQ are available but PUSHFD/POPFD are not.[8]: 4–349, 4–432
The lower 8 bits of the FLAGS register is also open to direct load/store manipulation by SAHF and LAHF (load/store AH into flags).
The ability to push and pop FLAGS registers lets a program manipulate information in the FLAGS in ways for which machine-language instructions do not exist. For example, the cld
and std
instructions clear and set the direction flag (DF), respectively; but there is no instruction to complement DF. This can be achieved with the following assembly code:
; This is 8086 code, with 16-bit registers pushed onto the stack,
; and the flags register is only 16 bits with this CPU.
pushf ; Use the stack to transfer the FLAGS
pop ax ; … into the AX register
push ax ; and copy them back onto the stack for storage
xor ax, 400h ; Toggle (invert, ‘complement’) the DF only; other bits are unchanged
push ax ; Use the stack again to move the modified value
popf ; … into the FLAGS register
; Insert here the code that required the DF flag to be complemented
popf ; Restore the original value of the FLAGS
By manipulating the FLAGS register, a program can determine the model of the installed processor. For example, the alignment flag can only be changed on the 486 and above. If the program tries to modify this flag and senses that the modification did not persist, the processor is earlier than the 486.
Starting with the Intel Pentium, the CPUID instruction reports the processor model. However, the above method remains useful to distinguish between earlier models.