movsxr

Movsxr is not an official x86-64 instruction. In many reference materials it appears not as a distinct, hardware-supported mnemonic but rather as a nickname, alias, or macro used in some assemblers or codebases to indicate a sign-extension move to a 64-bit destination register. Because it is not standardized, its exact meaning can vary between projects.

In actual x86-64 instruction sets, sign-extension moves are performed with the MOVSX family (move with sign-extend) and MOVSXD for a 32-bit source. MOVSXies extend from 8- or 16- or 32-bit sources to a 64-bit destination, preserving the sign of the source. MOVSXD specifically extends a 32-bit source to 64 bits. The hardware instructions are well-defined; any usage of movsxr in real code is typically a macro, a shorthand, or a compiler-internal alias that expands to MOVSX or MOVSXD as appropriate.

When movsxr appears in source text, it should be treated as an abstraction rather than a new opcode. Inspect the assembler or compiler documentation for that codebase to determine its expansion. A typical intent, if it maps to a sign-extension, is to take a smaller signed value (8, 16, or 32 bits) and produce a 64-bit signed result in a register.

Example intent (using standard instructions): sign-extend an 8-bit value to 64 bits, producing a value in RAX; or sign-extend a 32-bit value to 64 bits. In Intel syntax, this uses MOVSX rax, byte ptr [mem] or MOVSXD rax, dword ptr [mem], depending on the source width.

See also: MOVSX, MOVSXD, sign extension, x86-64 instruction set.