How does the indirect jump instruction jmp *.L4(,%rdi,8) work for a switch?

The * makes it indirect: the CPU computes the table-slot address .L4 + %rdi*8, reads the 8-byte target stored there, and jumps to that target.

* Indirect jump: compute the address, read the target stored there, then jump. *

A normal jmp .L2 jumps to a label whose address is fixed at assemble time. An indirect jump (jmp *…) jumps to an address it reads from memory at runtime — perfect when the destination depends on a variable.

jmp *.L4(,%rdi,8)

Breaking down the memory operand .L4(,%rdi,8):

• .L4 — base address of the jump table
• %rdi — the switch value x (the index)
• 8 — scale factor: each entry is an 8-byte (64-bit) address on x86_64
• Effective address = .L4 + %rdi*8

The crucial *: without it, you'd jump into the table data itself. With it, you jump to the address the table contains. This is also how function-pointer calls (call *%rax) and vtables work.

Go deeper:

• Branch table / jump table (Wikipedia) — jump tables and the indirect jump that reads them.
• Addressing mode (Wikipedia) — the base+index*scale addressing mode being computed.