Skip to content

XRSTOR

Jump to bottom
Henk-Jan Lebbink edited this page Jun 4, 2026 · 16 revisions

XRSTOR — Restore Processor Extended States

Opcode / Instruction Op/ En 64/32 bit Mode Support CPUID Feature Flag Description
NP 0F AE /5 XRSTOR mem M V/V XSAVE Restore state components specified by EDX:EAX from mem.
NP REX.W + 0F AE /5 XRSTOR64 mem M V/N.E. XSAVE Restore state components specified by EDX:EAX from mem.

Instruction Operand Encoding

Op/En Operand 1 Operand 2 Operand 3 Operand 4
M ModRM:r/m (r) N/A N/A N/A

Description

Performs a full or partial restore of processor state components from the XSAVE area located at the memory address specified by the source operand. The implicit EDX:EAX register pair specifies a 64-bit instruction mask. The specific state components restored correspond to the bits set in the requested-feature bitmap (RFBM), which is the logical-AND of EDX:EAX and XCR0. ® The format of the XSAVE area is detailed in Section 13.4, “XSAVE Area,” of Intel 64 and IA-32 Architectures Software Developer’s Manual, Volume 1. Like FXRSTOR and FXSAVE, the memory format used for x87 state depends ® on a REX.W prefix; see Section 13.5.1, “x87 State” of Intel 64 and IA-32 Architectures Software Developer’s Manual, Volume 1. ® Section 13.8, “Operation of XRSTOR,” of Intel 64 and IA-32 Architectures Software Developer’s Manual, Volume 1 provides a detailed description of the operation of the XRSTOR instruction. The following items provide a high-level outline:

  • Execution of XRSTOR may take one of two forms: standard and compacted. Bit 63 of the XCOMP_BV field in the XSAVE header determines which form is used: value 0 specifies the standard form, while value 1 specifies the compacted form.

  • 1 If RFBM[i] = 0, XRSTOR does not update state component i.

  • If RFBM[i] = 1 and bit i is clear in the XSTATE_BV field in the XSAVE header, XRSTOR initializes state component i.

  • If RFBM[i]= 1 and XSTATE_BV[i] = 1, XRSTOR loads state component i from the XSAVE area.

  • The standard form of XRSTOR treats MXCSR (which is part of state component 1 — SSE) differently from the XMM registers. If either form attempts to load MXCSR with an illegal value, a general-protection exception (#GP) occurs.

  • XRSTOR loads the internal value XRSTOR_INFO, which may be used to optimize a subsequent execution of XSAVEOPT or XSAVES.

  • Immediately following an execution of XRSTOR, the processor tracks as in-use (not in initial configuration) any state component i for which RFBM[i] = 1 and XSTATE_BV[i] = 1; it tracks as modified any state component i for which RFBM[i]= 0.

Use of a source operand not aligned to 64-byte boundary (for 64-bit and 32-bit modes) results in a general-protec-tion (#GP) exception. In 64-bit mode, the upper 32 bits of RDX and RAX are ignored. ® See Section 13.6, “Processor Tracking of XSAVE-Managed State,” of Intel 64 and IA-32 Architectures Software Developer’s Manual, Volume 1 for discussion of the bitmaps XINUSE and XMODIFIED and of the quantity XRSTOR_INFO.

  1. There is an exception if RFBM[1] = 0 and RFBM[2] = 1. In this case, the standard form of XRSTOR will load MXCSR from memory, even though MXCSR is part of state component 1 — SSE. The compacted form of XRSTOR does not make this exception.

Operation

RFBM ← XCR0 AND EDX:EAX;
                            /* bitwise logical AND */
COMPMASK ← XCOMP_BV field from XSAVE header;
RSTORMASK ← XSTATE_BV field from XSAVE header;
IF COMPMASK[63] = 0
    THEN
        /* Standard form of XRSTOR */
        TO_BE_RESTORED ← RFBM AND RSTORMASK;
        TO_BE_INITIALIZED ← RFBM AND NOT RSTORMASK;
        IF TO_BE_RESTORED[0] = 1
            THEN
                XINUSE[0] ← 1;
                load x87 state from legacy region of XSAVE area;
        ELSIF TO_BE_INITIALIZED[0] = 1
            THEN
                XINUSE[0] ← 0;
                initialize x87 state;
        FI;
        IF RFBM[1] = 1 OR RFBM[2] = 1
            THEN load MXCSR from legacy region of XSAVE area;
        FI;
        IF TO_BE_RESTORED[1] = 1
            THEN
                XINUSE[1] ← 1;
                load XMM registers from legacy region of XSAVE area; // this step does not load MXCSR
        ELSIF TO_BE_INITIALIZED[1] = 1
            THEN
                XINUSE[1] ← 0;
                set all XMM registers to 0; // this step does not initialize MXCSR
        FI;
        FOR i ← 2 TO 62
            IF TO_BE_RESTORED[i] = 1
                THEN
                    XINUSE[i] ← 1;
                    load XSAVE state component i at offset n from base of XSAVE area;
                        // n enumerated by CPUID.0DH.i:EBX)
            ELSIF TO_BE_INITIALIZED[i] = 1
                THEN
                    XINUSE[i] ← 0;
                    initialize XSAVE state component i;
            FI;
        ENDFOR;
    ELSE
        /* Compacted form of XRSTOR */
        IF CPUID.0DH.01H:EAX.XSAVEC[1] = 0
            THEN
                    /* compacted form not supported */
                #GP(0);
        FI;
        FORMAT = COMPMASK AND 7FFFFFFF_FFFFFFFFH;
        RESTORE_FEATURES = FORMAT AND RFBM;
        TO_BE_RESTORED ← RESTORE_FEATURES AND RSTORMASK;
        FORCE_INIT ← RFBM AND NOT FORMAT;
        TO_BE_INITIALIZED = (RFBM AND NOT RSTORMASK) OR FORCE_INIT;
        IF TO_BE_RESTORED[0] = 1
            THEN
                XINUSE[0] ← 1;
                load x87 state from legacy region of XSAVE area;
        ELSIF TO_BE_INITIALIZED[0] = 1
            THEN
                XINUSE[0] ← 0;
                initialize x87 state;
        FI;
        IF TO_BE_RESTORED[1] = 1
            THEN
                XINUSE[1] ← 1;
                load SSE state from legacy region of XSAVE area; // this step loads the XMM registers and MXCSR
        ELSIF TO_BE_INITIALIZED[1] = 1
            THEN
                set all XMM registers to 0;
                XINUSE[1] ← 0;
                MXCSR ← 1F80H;
        FI;
        NEXT_FEATURE_OFFSET = 576;
                            // Legacy area and XSAVE header consume 576 bytes
        FOR i ← 2 TO 62
            IF FORMAT[i] = 1
                THEN
                    IF TO_BE_RESTORED[i] = 1
                        THEN
                            XINUSE[i] ← 1;
                            load XSAVE state component i at offset NEXT_FEATURE_OFFSET from base of XSAVE area;
                    FI;
NEXT_FEATURE_OFFSET = NEXT_FEATURE_OFFSET + n (n enumerated by CPUID.0DH.i:EAX);
            FI;
            IF TO_BE_INITIALIZED[i] = 1
                THEN
                    XINUSE[i] ← 0;
                    initialize XSAVE state component i;
            FI;
        ENDFOR;
FI;
XMODIFIED ← NOT RFBM;
IF in VMX non-root operation
    THEN VMXNR ← 1;
    ELSE VMXNR ← 0;
FI;
LAXA ← linear address of XSAVE area;
XRSTOR_INFO ← CPL,VMXNR,LAXA,COMPMASK;

Flags Affected

None.

Intel C/C++ Compiler Intrinsic Equivalent

XRSTOR void _xrstor( void * , unsigned __int64);
XRSTOR void _xrstor64( void * , unsigned __int64);

Protected Mode Exceptions

#GP(0) If a memory operand effective address is outside the CS, DS, ES, FS, or GS segment limit. If a memory operand is not aligned on a 64-byte boundary, regardless of segment. If bit 63 of the XCOMP_BV field of the XSAVE header is 1 and CPUID.0DH.01H:EAX.XSAVEC[1] = 0. If the standard form is executed and a bit in XCR0 is 0 and the corresponding bit in the XSTATE_BV field of the XSAVE header is 1. If the standard form is executed and bytes 23:8 of the XSAVE header are not all zero. If the compacted form is executed and a bit in XCR0 is 0 and the corresponding bit in the XCOMP_BV field of the XSAVE header is 1. If the compacted form is executed and a bit in the XCOMP_BV field in the XSAVE header is 0 and the corresponding bit in the XSTATE_BV field is 1. If the compacted form is executed and bytes 63:16 of the XSAVE header are not all zero. If attempting to write any reserved bits of the MXCSR register with 1.

#SS(0) If a memory operand effective address is outside the SS segment limit.

#PF(fault-code) If a page fault occurs.

#NM If CR0.TS[bit 3] = 1.

#UD If CPUID.01H:ECX.XSAVE[26] = 0. If CR4.OSXSAVE[bit 18] = 0. If the LOCK prefix is used.

#AC If this exception is disabled a general protection exception (\#GP) is signaled if the memory operand is not aligned on a 64-byte boundary, as described above. If the alignment check exception (\#AC) is enabled (and the CPL is 3), signaling of \#AC is not guaranteed and may vary with implementation, as follows. In all implementations where

#AC is not signaled, a general protection exception is signaled in its place. In addition, the width of the alignment check may also vary with implementation. For instance, for a given implementation, an alignment check exception might be signaled for a 2-byte misalignment, whereas a general protec-tion exception might be signaled for all other misalignments (4-, 8-, or 16-byte misalignments).

Real-Address Mode Exceptions

#GP If a memory operand is not aligned on a 64-byte boundary, regardless of segment. If any part of the operand lies outside the effective address space from 0 to FFFFH. If bit 63 of the XCOMP_BV field of the XSAVE header is 1 and CPUID.0DH.01H:EAX.XSAVEC[1] = 0. If the standard form is executed and a bit in XCR0 is 0 and the corresponding bit in the XSTATE_BV field of the XSAVE header is 1. If the standard form is executed and bytes 23:8 of the XSAVE header are not all zero. If the compacted form is executed and a bit in XCR0 is 0 and the corresponding bit in the XCOMP_BV field of the XSAVE header is 1. If the compacted form is executed and a bit in the XCOMP_BV field in the XSAVE header is 0 and the corresponding bit in the XSTATE_BV field is 1. If the compacted form is executed and bytes 63:16 of the XSAVE header are not all zero. If attempting to write any reserved bits of the MXCSR register with 1.

#NM If CR0.TS[bit 3] = 1.

#UD If CPUID.01H:ECX.XSAVE[26] = 0. If CR4.OSXSAVE[bit 18] = 0. If the LOCK prefix is used.

Virtual-8086 Mode Exceptions

Same exceptions as in protected mode.

Compatibility Mode Exceptions

Same exceptions as in protected mode.

64-Bit Mode Exceptions

#GP(0) If a memory address is in a non-canonical form. If a memory operand is not aligned on a 64-byte boundary, regardless of segment. If bit 63 of the XCOMP_BV field of the XSAVE header is 1 and CPUID.0DH.01H:EAX.XSAVEC[1] = 0. If the standard form is executed and a bit in XCR0 is 0 and the corresponding bit in the XSTATE_BV field of the XSAVE header is 1. If the standard form is executed and bytes 23:8 of the XSAVE header are not all zero. If the compacted form is executed and a bit in XCR0 is 0 and the corresponding bit in the XCOMP_BV field of the XSAVE header is 1. If the compacted form is executed and a bit in the XCOMP_BV field in the XSAVE header is 0 and the corresponding bit in the XSTATE_BV field is 1. If the compacted form is executed and bytes 63:16 of the XSAVE header are not all zero. If attempting to write any reserved bits of the MXCSR register with 1.

#SS(0) If a memory address referencing the SS segment is in a non-canonical form.

#PF(fault-code) If a page fault occurs.

#NM If CR0.TS[bit 3] = 1.

#UD If CPUID.01H:ECX[26, “XSAVE”] = 0. If CR4.OSXSAVE[bit 18] = 0. If the LOCK prefix is used.

#AC If this exception is disabled a general protection exception (\#GP) is signaled if the memory operand is not aligned on a 64-byte boundary, as described above. If the alignment check exception (\#AC) is enabled (and the CPL is 3), signaling of \#AC is not guaranteed and may vary with implementation, as follows. In all implementations where

#AC is not signaled, a general protection exception is signaled in its place. In addition, the width of the alignment check may also vary with implementation. For instance, for a given implementation, an alignment check exception might be signaled for a 2-byte misalignment, whereas a general protec-tion exception might be signaled for all other misalignments (4-, 8-, or 16-byte misalignments).

Source: Intel® 64 and IA-32 Architectures Software Developer's Manual, Combined Volumes (Order Number 325462-091US, March 2026)
Generated: 7-6-2026

Clone this wiki locally