1 / 20

Understanding Unconditional Jumps and Conditional Branching in Computer Engineering**

In this lecture from ECE291 at the University of Illinois, Josh Potts explores the fundamentals of unconditional jumps and conditional branching in computer architecture. The discussion includes the mechanics of various jump instructions, such as short, near, and far jumps, and their specific use cases within instruction sets. Additionally, the lecture covers conditional jumps based on logic flags and the implications for program control flow, along with a deep dive into numerical comparison, including signed and unsigned integers.

vivian
Télécharger la présentation

Understanding Unconditional Jumps and Conditional Branching in Computer Engineering**

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. ECE291Computer Engineering IILecture 5 Josh Potts University of Illinois at Urbana- Champaign

  2. Outline • Unconditional jump • Conditional branching • Construction of loops ECE291

  3. Unconditional JumpJMP • Short jump2-byte instruction that allows jumps or branches to memory locations within +127 and -128 bytes from the memory location following the jump JMP SHORT Label • Near jump3-byte instruction that allows jumps or branches within +/- 32Kb from the instruction in the current code segment JMP Label • Far jump5-byte instruction that allows a jump to any memory location with in the entire memory space JMP Label • For 80386, 80486, the near jump is within +/-2G if the machine operates in the protected mode and +/-32K bytes if operates in the real mode OPCODE DISP OPCODE DISP low DISP high OPCODE IP low IP high CS low CS high ECE291

  4. Conditional Branching • Logic and arithmetic instructions set flags • Flags provide state information from previous instruction(s) • Using flags we can perform conditional jumping, i.e., transfer program execution to some different place within the program if condition was true • jump back or forward in your code to the location specified • instruction pointer (IP) gets updated (to point to the instruction to which execution will jump) if condition was false • continue execution at the following instruction • IP gets incremented as usual ECE291

  5. Conditional Branching (cont.) • Conditional jumps are always short jumps in the 8086-80286 • the range of the jump is +127 bytes and -128 bytes from the location following the conditional jump • In 80386, 80486 conditional jumps are either short or near jumps • Conditional jumps test: sign (S), zero (Z), carry (C), parity (P), and overflow (O) • Note: an FFh is above the 00h in the set of unsigned numbers an FFh (-1) is less than 00h for signed numbers when you compare unsigned FFh is above 00h, but signed FFh is less than 00h ECE291

  6. Numerical Comparison • CMP(comparison) compares A to B • a subtraction that only changes the flag bits • useful for checking the entire contents of a register or a memory location against another value • usually followed by a conditional jump instruction CMP AL, 10h ;compare with 10h (contents of AL does not change) JAE SUBER ;if 10h or above then jump to memory location SUBER • SUB (subtraction) calculates difference A - B • saves results to A and set flags ECE291

  7. Numerical ComparisonCondition Code Settings CMP Oprnd1, Oprnd2 Unsigned Operands Signed operands Z: equality/inequality Z: equality/inequality C: Oprnd1 < Oprnd2 (C=1) C: no meaning Oprnd1 >= Oprnd2 (C=0) S: no meaning S and O taken together O: no meaning If ((S=0) and (O=1)) or ((S=1) and (O=0)) then Oprnd1 < Oprnd2 If ((S=0) and (O=0)) or ((S=1) and (O=1)) then Oprnd1 >= Oprnd2 ECE291

  8. Comparing Signed Integers • Consider CMP AX,BX computed by the CPU • The Sign bit (S) will be set if the result of AX-BX has a 1 at the most significant bit of the result (i.e., 15th bit for 16-bit op) • The Overflow flag (O) will be set if the result of AX-BX produced a number that was out of range (-32768 - 32767 for 16-bit numbers) to be represented by an integer. • Difference in JS (jump on sign) and JL (jump less than) • The conditional jump JS looks at the sign bit (S) of the last compare (or subtraction). If S = = 1 then jump. • The conditional jump JL looks (SXOR O) of the last compare (or subtraction) • REGARDLESS of the value AX-BX, i.e., even if AX-BX causes overflow, the JL will be correctly executed ECE291

  9. Comparing Signed Integers (cont.) • JL is true if the condition: S xor O is met • JL is true for two conditions: • S=1, O=0: • (AX-BX) was negative and (AX-BX) did not overflow • Example (8-bit): (-5) - (2) = (-7) Result (-7) has the sign bit set Thus (-5) is less than (2). ECE291

  10. Comparing Signed Integers (cont.) • S=0, O=1: • Overflow!,Sign bit of the result is wrong! • Consider the following case: AX is a large negative number (-) BX is a positive number (+). The subtraction of (-) and (+) is the same as the addition of (-) and (-) The result causes negative overflow, and thus cannot be represented as a signed integer correctly (O=1). The result of AX-BX appears positive (S=0). • Example (8-bit): (-128) - (1) = (+127) • Result (+127) overflowed. Answer should have been (-129). • Result appears positive, but overflow occurred • Thus (-128) is less than (1), i.e., the condition is TRUE for executing JL ECE291

  11. AX BX AX BX BX AX -8 -8 -8 -7 -7 -7 -6 -6 -6 -5 -5 -5 -4 -4 -4 -3 -3 -3 -2 -2 -2 -1 -1 -1 0 0 0 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 Comparing Signed IntegersCMP AX, BX AX – BX = 2 – (-4) = 2 + 4 = 6 =0110 So s = 0, no overflow (o = 0) Therefore AX >= BX AX – BX = 6 – (-3) = 6 + 3 = 9 = 1001 So s = 1, overflow (o = 1) Therefore AX >= BX AX – BX = 2 – 4 = -2 = 1110 So s = 1, no overflow (o = 0) Therefore AX < BX ECE291

  12. Conditional Branching (cont.) • Terminology used to differentiate between jump instructions that use the carry flag and the overflow flag • Above/Below unsigned compare • Greater/Less signed (+/-) compare • Names of jump instructions J => Jump N => Not A/B G/L => Above/Below Greater/Less E => Equal ECE291

  13. Summary of Conditional Jump Instructions Command Description Condition JA=JNBE Jump if above C=0 & Z=0 Jump if not below or equal JBE=JNA Jump if below or equal C=1 | Z=1 JAE=JNB=JNC Jump if above or equal C=0 Jump if not below Jump if no carry JB=JNAE=JC Jump if below C=1 Jump if carry JE=JZ Jump if equal Z=1 Jump if Zero JNE=JNZ Jump if not equal Z=0 Jump if not zero JS Jump Sign (MSB=1) S=1 ECE291

  14. Summary of Conditional Jump Instructions Command Description Condition JNS Jump Not Sign (MSB=0) S=0 JO Jump if overflow set O=1 JNO Jump if no overflow O=0 JG=JNLE Jump if greater Jump if not less or equal S=O & Z=0 JGE=JNL Jump if greater or equal S=O Jump if not less JL=JNGE Jump if less S^O Jump if not greater or equal JLE=JNG Jump if less or equal S^O | Z=1 Jump if not greater JCXZ Jump if register CX=zero CX=0 ECE291

  15. Mapping High Level Branches into Linear Code CMP AX, BX JA true_label …. <False Processing> …. JMP done_label …. true_label: <True processing> …. done_label: <resume execution> ECE291

  16. Mapping High Level Branches into Linear Code (cont.) ECE291

  17. Mapping High Level Branches into Linear Code (cont.) • LOOP instruction • combination of a decrement CX and a conditional jump • LOOP decrements CX (ECX if in 32-bit mode) and if CX  0 it jumps to the address indicated by the label • if CX becomes a 0, the next sequential instruction executes ADDS PROC NEAR MOV CX, 100 ; load count MOV SI, OFFSET BLOCK1 MOV DI, OFFSET BLOCK2 Again: LODSW ;get Block1 data; AX = [SI]; SI = SI + 2 ADD AX, ES:[DI] ;add Block2 data STOSW ;store in Block2; [DI] = AX; DI = DI + 2 LOOP Again ;repeat 100 times RET ADDS ENDP ECE291

  18. ; if (J <= K) then ; L := L + 1 ; else L := L - 1 ; J, K, L are signed words MOV AX, J CMP AX, K JNEL DoElse INC L JMP ifDone DoElse: DEC L ifDone: ; while (J >= K) do begin ; J := J - 1; ; K := K + 1; ; L := J * K; ; end; WhlLoop: MOV AX, J CMP AX, K JNGE QuitLoop DEC J INC K MOV AX, J IMUL AX, K MOV L, AX JMP WhlLoop QuitLoop: Examples ECE291

  19. Example (LOOPNE) • The LOOPNE instruction is useful for controlling loops that stop on some condition or when the loop exceeds some number of iterations • Consider String1 that contains a sequence of characters that end with the byte containing zero • we want to convert those characters to upper case and copy them to String2 ….. String1 BYTE “This string contains lower case characters”, 0 String2 BYTE 128 dup (0) …….. ECE291

  20. Example (LOOPNE) LEA SI, String1 ;the same as use of OFFSET LEA DI, String2 MOV CX, 127 ;Max 127 chars to String2 StrLoop: LODSB ;get char from String1; AL =[SI]; SI = SI + 1 CMP AL, ‘a’ ;see if lower case JB NotLower ;chars are unsigned CMP AL, ‘z’ JA NotLower AND AL, 5Fh ;convert lower -> upper case; ;bit 6 must be 0 NotLower: STOSB ; [DI] = AL; DI = DI + 1 CMP AL, 0 ;see if zero terminator LOOPNE StrLoop ;quit if AL or CX = 0 ECE291

More Related