Is it always equivalent to a same-named 32-bit instruction?
Not always. Some C/Zc instructions compress common 32-bit operations, while others have dedicated stack-frame or table-jump semantics.
Indirect jump via table, save return address
CM.JALT (Zcmt) decodes only for index 32-255. It fetches an instruction-memory table entry at the JVT CSR base plus index<<2 on RV32 or index<<3 on RV64, writes pc+2 to x1 (ra), then writes the entry with bit 0 cleared to pc. Only jvt.mode=0 defines this behavior; a reserved mode makes the instruction reserved. It reuses c.fsdsp encodings, is incompatible with Zcd, and Zcmt depends on Zca and Zicsr.
Shows only official compressed fields, conditions, and architectural state; it does not model a pipeline, cache, or timing.
Zcmt table entries follow current data endianness and are fetched as instruction memory; JVT base must be at least 64-byte aligned.
CM.JALT uses index 32-255 to select a JVT entry, links pc+2 to ra, and jumps to the entry address with bit 0 cleared.
Understand this scenario with real code like «cm.jalt 32 # jump-and-link via table entry 32».
Understand this scenario with real code like «cm.jalt 32 # jump-and-link via table entry 32».
Understand this scenario with real code like «cm.jalt 32 # jump-and-link via table entry 32».
Not always. Some C/Zc instructions compress common 32-bit operations, while others have dedicated stack-frame or table-jump semantics.
Many 16-bit encodings can represent only a compressed register subset or fixed registers such as sp, ra, a0/a1.