Jump Instruction in MIPS

• Using J instruction, we cannot jump anywhere in the memory space

• We can specify a range given in 26 bits

• This 26 bit address is taken as a word address

• So effectively in word it will be 28 bits

• Hence jump within a segment will be 2 power 28 i.e 256MB

• With 32 bits we can get 2 power 32 i.e 4GB of memory

  In a 32-bit memory address system, each memory address is represented using 32 bits. This allows for a total of 2^32 unique memory addresses, which equals 4 gigabytes (GB) of addressable memory space.

  To ensure that memory accesses are aligned to 4-byte boundaries, the last two bits of the memory address are set to zero

• 4GB space is divided into segments of size 256MB each

• Say for example if the current instruction is in segment 2, then we can jump to any instruction within that segment 2.
• An exception for this includes whenever there is a jump for the last word in the segment
• Because as the initial value PC=PC+4, jump will occur to the next segment

Jump Instruction in RISC V

There are 2 types of jump instructions in RISC V

1) Jump and link (jal)

• For conditional statements like if else and for looping statements like while loop we use branch instructions
• But, for function calls we don't use branch instructions
• We may be jumping to some far off location using function calls

Hence we will be using Jump and link (jal) which is a UJ format instruction

So how it works?

Lets us assume that the 20 bits stored in the instruction be: 10000100000110011001

using these 20 bits we need to create a 32 bit immediate address

The LSB 8 bits are 10011001 are placed in position [19:12] of the 32 bit immediate address

next 1 bit i.e 1 is placed in position 11

next 10 bits i.e 0000100000 are placed in position [10:1]

and the last bit i.e 1 is placed in position 20

This is the way how we will create a 32 bit immediate address in the case of jal instruction from the 20 bits that are stored in the instruction

The opcode for jal instruction is: 1101111

Functionality of jal instruction

Example: jal x5, Imm[20:1]

• Initially the value of PC will be 0
• Here it computes the return address PC+4 and store it in register x5 (i.e x5 <- PC+4)
• Then PC gets changed to PC=PC+40 so that we can move from jal to I5 (which is the function call)
• Function gets executed in I5 and the return address will take the value of x5. PC <-x5 (i.e PC=4)

2) Jump and link register (jalr)

This is a special instruction which belongs to I-format which is mainly used to implement the return address.

The return address ret is a sudo instruction.

What is sudo instruction? Sudo instruction is not a actual instruction, It is implemented by some other actual instruction i.e jalr

The opcode for jalr instruction is: 1100111

Here the immediate instruction will be 12 bits There is no multiplication by 2 (no shifting of bits)

• Writes PC+4 to register destination rd (i.e rd <- PC+4)
• Sets PC=[contents of rs]+ {32 bit sign extended Immediate address}

Example: jalr x0, ra, 0

• Here at first x0 <- PC+4 (i.e the return address will be zero)
• PC=[contents of ra] + 0 (i.e PC <- ra)

Jalr instruction works similar to jal instruction but the only difference would be

jal	jalr
It follows UJ format Instruction	It follows I format Instruction
opcode= 1101111	opcode = 1100111
It stores return address (PC+4) in its destination register and PC is PC+ offset, where offset =(sign extended to 32 bits)[31:21]imm[20:1]0	It's functionality is it writes PC+4 to rd and it sets PC= [rs1]+imm (sign extended to 32 bits)
Multiplication by 2 (logical shift)	No logical shift

Jump Instructions In ARM