add support for multibyte nop sequences
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index 5ae708d..da2eabd 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -2390,6 +2390,36 @@
#ifdef XBYAK_UNDEF_JNL
#undef jnl
#endif
+
+ void nop(int size = 1)
+ {
+ /*
+ AMD and Intel seem to agree on the same sequences for up to 9 bytes:
+ https://support.amd.com/TechDocs/55723_SOG_Fam_17h_Processors_3.00.pdf
+ https://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+ */
+ static const uint8_t nopSeq[9][9] = {
+ {0x90},
+ {0x66, 0x90},
+ {0x0F, 0x1F, 0x00},
+ {0x0F, 0x1F, 0x40, 0x00},
+ {0x0F, 0x1F, 0x44, 0x00, 0x00},
+ {0x66, 0x0F, 0x1F, 0x44, 0x00, 0x00},
+ {0x0F, 0x1F, 0x80, 0x00, 0x00, 0x00, 0x00},
+ {0x0F, 0x1F, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00},
+ {0x66, 0x0F, 0x1F, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00},
+ };
+ static const int numSeq = sizeof(nopSeq) / sizeof(nopSeq[0]);
+ while (size) {
+ int len = size > numSeq ? numSeq : size;
+ const uint8_t *seq = nopSeq[len - 1];
+ for (int i = 0; i < len; i++) {
+ db(seq[i]);
+ }
+ size -= len;
+ }
+ }
+
#ifndef XBYAK_DONT_READ_LIST
#include "xbyak_mnemonic.h"
void align(int x = 16)
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index 95014a0..ea7c6a3 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -497,7 +497,6 @@
void mulx(const Reg32e& r1, const Reg32e& r2, const Operand& op) { opGpr(r1, r2, op, T_F2 | T_0F38, 0xf6, true); }
void mwait() { db(0x0F); db(0x01); db(0xC9); }
void neg(const Operand& op) { opR_ModM(op, 0, 3, 0xF6); }
-void nop() { db(0x90); }
void not_(const Operand& op) { opR_ModM(op, 0, 2, 0xF6); }
void or_(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x08, 1); }
void or_(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x08); }
