vecstable Sdiff src/hotspot/cpu/x86

src/hotspot/cpu/x86/assembler_x86.hpp

rev 60516 : manual merge with default

 603     EVEX_8bit  = 0,
 604     EVEX_16bit = 1,
 605     EVEX_32bit = 2,
 606     EVEX_64bit = 3,
 607     EVEX_NObit = 4
 608   };
 609 
 610   enum WhichOperand {
 611     // input to locate_operand, and format code for relocations
 612     imm_operand  = 0,            // embedded 32-bit|64-bit immediate operand
 613     disp32_operand = 1,          // embedded 32-bit displacement or address
 614     call32_operand = 2,          // embedded 32-bit self-relative displacement
 615 #ifndef _LP64
 616     _WhichOperand_limit = 3
 617 #else
 618      narrow_oop_operand = 3,     // embedded 32-bit immediate narrow oop
 619     _WhichOperand_limit = 4
 620 #endif
 621   };
 622 

 623   enum ComparisonPredicate {
 624     eq = 0,
 625     lt = 1,
 626     le = 2,
 627     _false = 3,
 628     neq = 4,
 629     nlt = 5,
 630     nle = 6,
 631     _true = 7
 632   };
 633 













































 634   //---<  calculate length of instruction  >---
 635   // As instruction size can't be found out easily on x86/x64,
 636   // we just use '4' for len and maxlen.
 637   // instruction must start at passed address
 638   static unsigned int instr_len(unsigned char *instr) { return 4; }
 639 
 640   //---<  longest instructions  >---
 641   // Max instruction length is not specified in architecture documentation.
 642   // We could use a "safe enough" estimate (15), but just default to
 643   // instruction length guess from above.
 644   static unsigned int instr_maxlen() { return 4; }
 645 
 646   // NOTE: The general philopsophy of the declarations here is that 64bit versions
 647   // of instructions are freely declared without the need for wrapping them an ifdef.
 648   // (Some dangerous instructions are ifdef's out of inappropriate jvm's.)
 649   // In the .cpp file the implementations are wrapped so that they are dropped out
 650   // of the resulting jvm. This is done mostly to keep the footprint of MINIMAL
 651   // to the size it was prior to merging up the 32bit and 64bit assemblers.
 652   //
 653   // This does mean you'll get a linker/runtime error if you use a 64bit only instruction

 933   void rep_stos();
 934   void rep_stosb();
 935   void repne_scan();
 936 #ifdef _LP64
 937   void repne_scanl();
 938 #endif
 939 
 940   // Vanilla instructions in lexical order
 941 
 942   void adcl(Address dst, int32_t imm32);
 943   void adcl(Address dst, Register src);
 944   void adcl(Register dst, int32_t imm32);
 945   void adcl(Register dst, Address src);
 946   void adcl(Register dst, Register src);
 947 
 948   void adcq(Register dst, int32_t imm32);
 949   void adcq(Register dst, Address src);
 950   void adcq(Register dst, Register src);
 951 
 952   void addb(Address dst, int imm8);

 953   void addw(Address dst, int imm16);
 954 
 955   void addl(Address dst, int32_t imm32);
 956   void addl(Address dst, Register src);
 957   void addl(Register dst, int32_t imm32);
 958   void addl(Register dst, Address src);
 959   void addl(Register dst, Register src);
 960 
 961   void addq(Address dst, int32_t imm32);
 962   void addq(Address dst, Register src);
 963   void addq(Register dst, int32_t imm32);
 964   void addq(Register dst, Address src);
 965   void addq(Register dst, Register src);
 966 
 967 #ifdef _LP64
 968  //Add Unsigned Integers with Carry Flag
 969   void adcxq(Register dst, Register src);
 970 
 971  //Add Unsigned Integers with Overflow Flag
 972   void adoxq(Register dst, Register src);

 983 
 984   // Add Scalar Single-Precision Floating-Point Values
 985   void addss(XMMRegister dst, Address src);
 986   void addss(XMMRegister dst, XMMRegister src);
 987 
 988   // AES instructions
 989   void aesdec(XMMRegister dst, Address src);
 990   void aesdec(XMMRegister dst, XMMRegister src);
 991   void aesdeclast(XMMRegister dst, Address src);
 992   void aesdeclast(XMMRegister dst, XMMRegister src);
 993   void aesenc(XMMRegister dst, Address src);
 994   void aesenc(XMMRegister dst, XMMRegister src);
 995   void aesenclast(XMMRegister dst, Address src);
 996   void aesenclast(XMMRegister dst, XMMRegister src);
 997   // Vector AES instructions
 998   void vaesenc(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
 999   void vaesenclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1000   void vaesdec(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1001   void vaesdeclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1002 


1003   void andl(Address  dst, int32_t imm32);
1004   void andl(Register dst, int32_t imm32);
1005   void andl(Register dst, Address src);
1006   void andl(Register dst, Register src);
1007 
1008   void andq(Address  dst, int32_t imm32);
1009   void andq(Register dst, int32_t imm32);
1010   void andq(Register dst, Address src);
1011   void andq(Register dst, Register src);
1012 
1013   // BMI instructions
1014   void andnl(Register dst, Register src1, Register src2);
1015   void andnl(Register dst, Register src1, Address src2);
1016   void andnq(Register dst, Register src1, Register src2);
1017   void andnq(Register dst, Register src1, Address src2);
1018 
1019   void blsil(Register dst, Register src);
1020   void blsil(Register dst, Address src);
1021   void blsiq(Register dst, Register src);
1022   void blsiq(Register dst, Address src);

1108   void crc32(Register crc, Address adr, int8_t sizeInBytes);
1109 
1110   // Convert Scalar Double-Precision Floating-Point Value to Scalar Single-Precision Floating-Point Value
1111   void cvtsd2ss(XMMRegister dst, XMMRegister src);
1112   void cvtsd2ss(XMMRegister dst, Address src);
1113 
1114   // Convert Doubleword Integer to Scalar Double-Precision Floating-Point Value
1115   void cvtsi2sdl(XMMRegister dst, Register src);
1116   void cvtsi2sdl(XMMRegister dst, Address src);
1117   void cvtsi2sdq(XMMRegister dst, Register src);
1118   void cvtsi2sdq(XMMRegister dst, Address src);
1119 
1120   // Convert Doubleword Integer to Scalar Single-Precision Floating-Point Value
1121   void cvtsi2ssl(XMMRegister dst, Register src);
1122   void cvtsi2ssl(XMMRegister dst, Address src);
1123   void cvtsi2ssq(XMMRegister dst, Register src);
1124   void cvtsi2ssq(XMMRegister dst, Address src);
1125 
1126   // Convert Packed Signed Doubleword Integers to Packed Double-Precision Floating-Point Value
1127   void cvtdq2pd(XMMRegister dst, XMMRegister src);

1128 
1129   // Convert Packed Signed Doubleword Integers to Packed Single-Precision Floating-Point Value
1130   void cvtdq2ps(XMMRegister dst, XMMRegister src);

1131 
1132   // Convert Scalar Single-Precision Floating-Point Value to Scalar Double-Precision Floating-Point Value
1133   void cvtss2sd(XMMRegister dst, XMMRegister src);
1134   void cvtss2sd(XMMRegister dst, Address src);
1135 
1136   // Convert with Truncation Scalar Double-Precision Floating-Point Value to Doubleword Integer
1137   void cvttsd2sil(Register dst, Address src);
1138   void cvttsd2sil(Register dst, XMMRegister src);
1139   void cvttsd2siq(Register dst, Address src);
1140   void cvttsd2siq(Register dst, XMMRegister src);
1141 
1142   // Convert with Truncation Scalar Single-Precision Floating-Point Value to Doubleword Integer
1143   void cvttss2sil(Register dst, XMMRegister src);
1144   void cvttss2siq(Register dst, XMMRegister src);
1145 

1146   void cvttpd2dq(XMMRegister dst, XMMRegister src);
1147 
















1148   //Abs of packed Integer values
1149   void pabsb(XMMRegister dst, XMMRegister src);
1150   void pabsw(XMMRegister dst, XMMRegister src);
1151   void pabsd(XMMRegister dst, XMMRegister src);
1152   void vpabsb(XMMRegister dst, XMMRegister src, int vector_len);
1153   void vpabsw(XMMRegister dst, XMMRegister src, int vector_len);
1154   void vpabsd(XMMRegister dst, XMMRegister src, int vector_len);
1155   void evpabsq(XMMRegister dst, XMMRegister src, int vector_len);
1156 
1157   // Divide Scalar Double-Precision Floating-Point Values
1158   void divsd(XMMRegister dst, Address src);
1159   void divsd(XMMRegister dst, XMMRegister src);
1160 
1161   // Divide Scalar Single-Precision Floating-Point Values
1162   void divss(XMMRegister dst, Address src);
1163   void divss(XMMRegister dst, XMMRegister src);
1164 
1165 
1166 #ifndef _LP64
1167  private:

1487 
1488   // Move Double Quadword
1489   void movdq(XMMRegister dst, Register src);
1490   void movdq(Register dst, XMMRegister src);
1491 
1492   // Move Aligned Double Quadword
1493   void movdqa(XMMRegister dst, XMMRegister src);
1494   void movdqa(XMMRegister dst, Address src);
1495 
1496   // Move Unaligned Double Quadword
1497   void movdqu(Address     dst, XMMRegister src);
1498   void movdqu(XMMRegister dst, Address src);
1499   void movdqu(XMMRegister dst, XMMRegister src);
1500 
1501   // Move Unaligned 256bit Vector
1502   void vmovdqu(Address dst, XMMRegister src);
1503   void vmovdqu(XMMRegister dst, Address src);
1504   void vmovdqu(XMMRegister dst, XMMRegister src);
1505 
1506    // Move Unaligned 512bit Vector
1507   void evmovdqub(Address dst, XMMRegister src, int vector_len);
1508   void evmovdqub(XMMRegister dst, Address src, int vector_len);
1509   void evmovdqub(XMMRegister dst, XMMRegister src, int vector_len);
1510   void evmovdqub(XMMRegister dst, KRegister mask, Address src, int vector_len);
1511   void evmovdquw(Address dst, XMMRegister src, int vector_len);
1512   void evmovdquw(Address dst, KRegister mask, XMMRegister src, int vector_len);
1513   void evmovdquw(XMMRegister dst, Address src, int vector_len);
1514   void evmovdquw(XMMRegister dst, KRegister mask, Address src, int vector_len);
1515   void evmovdqul(Address dst, XMMRegister src, int vector_len);
1516   void evmovdqul(XMMRegister dst, Address src, int vector_len);
1517   void evmovdqul(XMMRegister dst, XMMRegister src, int vector_len);



1518   void evmovdquq(Address dst, XMMRegister src, int vector_len);
1519   void evmovdquq(XMMRegister dst, Address src, int vector_len);
1520   void evmovdquq(XMMRegister dst, XMMRegister src, int vector_len);



1521 
1522   // Move lower 64bit to high 64bit in 128bit register
1523   void movlhps(XMMRegister dst, XMMRegister src);
1524 
1525   void movl(Register dst, int32_t imm32);
1526   void movl(Address dst, int32_t imm32);
1527   void movl(Register dst, Register src);
1528   void movl(Register dst, Address src);
1529   void movl(Address dst, Register src);
1530 
1531   // These dummies prevent using movl from converting a zero (like NULL) into Register
1532   // by giving the compiler two choices it can't resolve
1533 
1534   void movl(Address  dst, void* junk);
1535   void movl(Register dst, void* junk);
1536 
1537 #ifdef _LP64
1538   void movq(Register dst, Register src);
1539   void movq(Register dst, Address src);
1540   void movq(Address  dst, Register src);
1541 
1542   // These dummies prevent using movq from converting a zero (like NULL) into Register
1543   // by giving the compiler two choices it can't resolve
1544 
1545   void movq(Address  dst, void* dummy);
1546   void movq(Register dst, void* dummy);
1547 #endif
1548 
1549   // Move Quadword
1550   void movq(Address     dst, XMMRegister src);
1551   void movq(XMMRegister dst, Address src);



1552 
1553   void movsbl(Register dst, Address src);
1554   void movsbl(Register dst, Register src);
1555 
1556 #ifdef _LP64
1557   void movsbq(Register dst, Address src);
1558   void movsbq(Register dst, Register src);
1559 
1560   // Move signed 32bit immediate to 64bit extending sign
1561   void movslq(Address  dst, int32_t imm64);
1562   void movslq(Register dst, int32_t imm64);
1563 
1564   void movslq(Register dst, Address src);
1565   void movslq(Register dst, Register src);
1566   void movslq(Register dst, void* src); // Dummy declaration to cause NULL to be ambiguous
1567 #endif
1568 
1569   void movswl(Register dst, Address src);
1570   void movswl(Register dst, Register src);
1571

1612   void mulss(XMMRegister dst, Address src);
1613   void mulss(XMMRegister dst, XMMRegister src);
1614 
1615   void negl(Register dst);
1616 
1617 #ifdef _LP64
1618   void negq(Register dst);
1619 #endif
1620 
1621   void nop(int i = 1);
1622 
1623   void notl(Register dst);
1624 
1625 #ifdef _LP64
1626   void notq(Register dst);
1627 
1628   void btsq(Address dst, int imm8);
1629   void btrq(Address dst, int imm8);
1630 #endif
1631 


1632   void orl(Address dst, int32_t imm32);
1633   void orl(Register dst, int32_t imm32);
1634   void orl(Register dst, Address src);
1635   void orl(Register dst, Register src);
1636   void orl(Address dst, Register src);
1637 
1638   void orb(Address dst, int imm8);
1639 
1640   void orq(Address dst, int32_t imm32);
1641   void orq(Register dst, int32_t imm32);
1642   void orq(Register dst, Address src);
1643   void orq(Register dst, Register src);
1644 






1645   // Pack with unsigned saturation
1646   void packuswb(XMMRegister dst, XMMRegister src);
1647   void packuswb(XMMRegister dst, Address src);

1648   void vpackuswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);

1649 
1650   // Pemutation of 64bit words
1651   void vpermq(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1652   void vpermq(XMMRegister dst, XMMRegister src, int imm8);
1653   void vpermq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);




1654   void vperm2i128(XMMRegister dst,  XMMRegister nds, XMMRegister src, int imm8);
1655   void vperm2f128(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8);



1656   void evpermi2q(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1657 
1658   void pause();
1659 
1660   // Undefined Instruction
1661   void ud2();
1662 
1663   // SSE4.2 string instructions
1664   void pcmpestri(XMMRegister xmm1, XMMRegister xmm2, int imm8);
1665   void pcmpestri(XMMRegister xmm1, Address src, int imm8);
1666 
1667   void pcmpeqb(XMMRegister dst, XMMRegister src);


1668   void vpcmpeqb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1669   void evpcmpeqb(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1670   void evpcmpeqb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1671   void evpcmpeqb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1672 

1673   void evpcmpgtb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1674   void evpcmpgtb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1675 
1676   void evpcmpuw(KRegister kdst, XMMRegister nds, XMMRegister src, ComparisonPredicate vcc, int vector_len);
1677   void evpcmpuw(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src, ComparisonPredicate of, int vector_len);
1678   void evpcmpuw(KRegister kdst, XMMRegister nds, Address src, ComparisonPredicate vcc, int vector_len);
1679 
1680   void pcmpeqw(XMMRegister dst, XMMRegister src);
1681   void vpcmpeqw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1682   void evpcmpeqw(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1683   void evpcmpeqw(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1684 


1685   void pcmpeqd(XMMRegister dst, XMMRegister src);
1686   void vpcmpeqd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1687   void evpcmpeqd(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1688   void evpcmpeqd(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1689 
1690   void pcmpeqq(XMMRegister dst, XMMRegister src);

1691   void vpcmpeqq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1692   void evpcmpeqq(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1693   void evpcmpeqq(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1694 



1695   void pmovmskb(Register dst, XMMRegister src);
1696   void vpmovmskb(Register dst, XMMRegister src);
1697 
1698   // SSE 4.1 extract
1699   void pextrd(Register dst, XMMRegister src, int imm8);
1700   void pextrq(Register dst, XMMRegister src, int imm8);
1701   void pextrd(Address dst, XMMRegister src, int imm8);
1702   void pextrq(Address dst, XMMRegister src, int imm8);

1703   void pextrb(Address dst, XMMRegister src, int imm8);
1704   // SSE 2 extract
1705   void pextrw(Register dst, XMMRegister src, int imm8);
1706   void pextrw(Address dst, XMMRegister src, int imm8);
1707 
1708   // SSE 4.1 insert
1709   void pinsrd(XMMRegister dst, Register src, int imm8);
1710   void pinsrq(XMMRegister dst, Register src, int imm8);

1711   void pinsrd(XMMRegister dst, Address src, int imm8);
1712   void pinsrq(XMMRegister dst, Address src, int imm8);
1713   void pinsrb(XMMRegister dst, Address src, int imm8);

1714   // SSE 2 insert
1715   void pinsrw(XMMRegister dst, Register src, int imm8);
1716   void pinsrw(XMMRegister dst, Address src, int imm8);
1717 
1718   // SSE4.1 packed move







1719   void pmovzxbw(XMMRegister dst, XMMRegister src);
1720   void pmovzxbw(XMMRegister dst, Address src);
1721 
1722   void vpmovzxbw( XMMRegister dst, Address src, int vector_len);

1723   void vpmovzxbw(XMMRegister dst, XMMRegister src, int vector_len);



1724   void evpmovzxbw(XMMRegister dst, KRegister mask, Address src, int vector_len);
1725 












1726   void evpmovwb(Address dst, XMMRegister src, int vector_len);
1727   void evpmovwb(Address dst, KRegister mask, XMMRegister src, int vector_len);
1728 
1729   void vpmovzxwd(XMMRegister dst, XMMRegister src, int vector_len);
1730 
1731   void evpmovdb(Address dst, XMMRegister src, int vector_len);
1732 
1733   // Sign extend moves
1734   void pmovsxbw(XMMRegister dst, XMMRegister src);
1735   void vpmovsxbw(XMMRegister dst, XMMRegister src, int vector_len);
1736 
1737   // Multiply add
1738   void pmaddwd(XMMRegister dst, XMMRegister src);
1739   void vpmaddwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1740   // Multiply add accumulate
1741   void evpdpwssd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1742 
1743 #ifndef _LP64 // no 32bit push/pop on amd64
1744   void popl(Address dst);
1745 #endif
1746 
1747 #ifdef _LP64
1748   void popq(Address dst);
1749 #endif
1750 
1751   void popcntl(Register dst, Address src);
1752   void popcntl(Register dst, Register src);
1753 
1754   void vpopcntd(XMMRegister dst, XMMRegister src, int vector_len);
1755 
1756 #ifdef _LP64

1760 
1761   // Prefetches (SSE, SSE2, 3DNOW only)
1762 
1763   void prefetchnta(Address src);
1764   void prefetchr(Address src);
1765   void prefetcht0(Address src);
1766   void prefetcht1(Address src);
1767   void prefetcht2(Address src);
1768   void prefetchw(Address src);
1769 
1770   // Shuffle Bytes
1771   void pshufb(XMMRegister dst, XMMRegister src);
1772   void pshufb(XMMRegister dst, Address src);
1773   void vpshufb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1774 
1775   // Shuffle Packed Doublewords
1776   void pshufd(XMMRegister dst, XMMRegister src, int mode);
1777   void pshufd(XMMRegister dst, Address src,     int mode);
1778   void vpshufd(XMMRegister dst, XMMRegister src, int mode, int vector_len);
1779 
1780   // Shuffle Packed Low Words

1781   void pshuflw(XMMRegister dst, XMMRegister src, int mode);
1782   void pshuflw(XMMRegister dst, Address src,     int mode);
1783 






1784   // Shuffle packed values at 128 bit granularity
1785   void evshufi64x2(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
1786 
1787   // Shift Right by bytes Logical DoubleQuadword Immediate
1788   void psrldq(XMMRegister dst, int shift);
1789   // Shift Left by bytes Logical DoubleQuadword Immediate
1790   void pslldq(XMMRegister dst, int shift);
1791 
1792   // Logical Compare 128bit
1793   void ptest(XMMRegister dst, XMMRegister src);
1794   void ptest(XMMRegister dst, Address src);
1795   // Logical Compare 256bit
1796   void vptest(XMMRegister dst, XMMRegister src);
1797   void vptest(XMMRegister dst, Address src);
1798 



1799   // Interleave Low Bytes
1800   void punpcklbw(XMMRegister dst, XMMRegister src);
1801   void punpcklbw(XMMRegister dst, Address src);
1802 
1803   // Interleave Low Doublewords
1804   void punpckldq(XMMRegister dst, XMMRegister src);
1805   void punpckldq(XMMRegister dst, Address src);
1806 
1807   // Interleave Low Quadwords
1808   void punpcklqdq(XMMRegister dst, XMMRegister src);
1809 
1810 #ifndef _LP64 // no 32bit push/pop on amd64
1811   void pushl(Address src);
1812 #endif
1813 
1814   void pushq(Address src);
1815 
1816   void rcll(Register dst, int imm8);
1817 
1818   void rclq(Register dst, int imm8);

1841   void sarq(Register dst, int imm8);
1842   void sarq(Register dst);
1843 
1844   void sbbl(Address dst, int32_t imm32);
1845   void sbbl(Register dst, int32_t imm32);
1846   void sbbl(Register dst, Address src);
1847   void sbbl(Register dst, Register src);
1848 
1849   void sbbq(Address dst, int32_t imm32);
1850   void sbbq(Register dst, int32_t imm32);
1851   void sbbq(Register dst, Address src);
1852   void sbbq(Register dst, Register src);
1853 
1854   void setb(Condition cc, Register dst);
1855 
1856   void palignr(XMMRegister dst, XMMRegister src, int imm8);
1857   void vpalignr(XMMRegister dst, XMMRegister src1, XMMRegister src2, int imm8, int vector_len);
1858   void evalignq(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
1859 
1860   void pblendw(XMMRegister dst, XMMRegister src, int imm8);

1861 
1862   void sha1rnds4(XMMRegister dst, XMMRegister src, int imm8);
1863   void sha1nexte(XMMRegister dst, XMMRegister src);
1864   void sha1msg1(XMMRegister dst, XMMRegister src);
1865   void sha1msg2(XMMRegister dst, XMMRegister src);
1866   // xmm0 is implicit additional source to the following instruction.
1867   void sha256rnds2(XMMRegister dst, XMMRegister src);
1868   void sha256msg1(XMMRegister dst, XMMRegister src);
1869   void sha256msg2(XMMRegister dst, XMMRegister src);
1870 
1871   void shldl(Register dst, Register src);
1872   void shldl(Register dst, Register src, int8_t imm8);
1873   void shrdl(Register dst, Register src);
1874   void shrdl(Register dst, Register src, int8_t imm8);
1875 
1876   void shll(Register dst, int imm8);
1877   void shll(Register dst);
1878 
1879   void shlq(Register dst, int imm8);
1880   void shlq(Register dst);

1959   void xbegin(Label& abort, relocInfo::relocType rtype = relocInfo::none);
1960 
1961   void xchgb(Register reg, Address adr);
1962   void xchgw(Register reg, Address adr);
1963   void xchgl(Register reg, Address adr);
1964   void xchgl(Register dst, Register src);
1965 
1966   void xchgq(Register reg, Address adr);
1967   void xchgq(Register dst, Register src);
1968 
1969   void xend();
1970 
1971   // Get Value of Extended Control Register
1972   void xgetbv();
1973 
1974   void xorl(Register dst, int32_t imm32);
1975   void xorl(Register dst, Address src);
1976   void xorl(Register dst, Register src);
1977 
1978   void xorb(Register dst, Address src);

1979 
1980   void xorq(Register dst, Address src);
1981   void xorq(Register dst, Register src);
1982 
1983   void set_byte_if_not_zero(Register dst); // sets reg to 1 if not zero, otherwise 0
1984 
1985   // AVX 3-operands scalar instructions (encoded with VEX prefix)
1986 
1987   void vaddsd(XMMRegister dst, XMMRegister nds, Address src);
1988   void vaddsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
1989   void vaddss(XMMRegister dst, XMMRegister nds, Address src);
1990   void vaddss(XMMRegister dst, XMMRegister nds, XMMRegister src);
1991   void vdivsd(XMMRegister dst, XMMRegister nds, Address src);
1992   void vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
1993   void vdivss(XMMRegister dst, XMMRegister nds, Address src);
1994   void vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src);
1995   void vfmadd231sd(XMMRegister dst, XMMRegister nds, XMMRegister src);
1996   void vfmadd231ss(XMMRegister dst, XMMRegister nds, XMMRegister src);
1997   void vmulsd(XMMRegister dst, XMMRegister nds, Address src);
1998   void vmulsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
1999   void vmulss(XMMRegister dst, XMMRegister nds, Address src);
2000   void vmulss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2001   void vsubsd(XMMRegister dst, XMMRegister nds, Address src);
2002   void vsubsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2003   void vsubss(XMMRegister dst, XMMRegister nds, Address src);
2004   void vsubss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2005 
2006   void vmaxss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2007   void vmaxsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2008   void vminss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2009   void vminsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2010 
2011   void shlxl(Register dst, Register src1, Register src2);
2012   void shlxq(Register dst, Register src1, Register src2);
2013 
2014   //====================VECTOR ARITHMETIC=====================================


2015 
2016   // Add Packed Floating-Point Values
2017   void addpd(XMMRegister dst, XMMRegister src);
2018   void addpd(XMMRegister dst, Address src);
2019   void addps(XMMRegister dst, XMMRegister src);
2020   void vaddpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2021   void vaddps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2022   void vaddpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2023   void vaddps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2024 
2025   // Subtract Packed Floating-Point Values
2026   void subpd(XMMRegister dst, XMMRegister src);
2027   void subps(XMMRegister dst, XMMRegister src);
2028   void vsubpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2029   void vsubps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2030   void vsubpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2031   void vsubps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2032 
2033   // Multiply Packed Floating-Point Values
2034   void mulpd(XMMRegister dst, XMMRegister src);

2104   void vpaddd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2105   void vpaddq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2106 
2107   // Sub packed integers
2108   void psubb(XMMRegister dst, XMMRegister src);
2109   void psubw(XMMRegister dst, XMMRegister src);
2110   void psubd(XMMRegister dst, XMMRegister src);
2111   void psubq(XMMRegister dst, XMMRegister src);
2112   void vpsubb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2113   void vpsubw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2114   void vpsubd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2115   void vpsubq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2116   void vpsubb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2117   void vpsubw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2118   void vpsubd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2119   void vpsubq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2120 
2121   // Multiply packed integers (only shorts and ints)
2122   void pmullw(XMMRegister dst, XMMRegister src);
2123   void pmulld(XMMRegister dst, XMMRegister src);

2124   void vpmullw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2125   void vpmulld(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2126   void vpmullq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);

2127   void vpmullw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2128   void vpmulld(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2129   void vpmullq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2130 


























2131   // Shift left packed integers
2132   void psllw(XMMRegister dst, int shift);
2133   void pslld(XMMRegister dst, int shift);
2134   void psllq(XMMRegister dst, int shift);
2135   void psllw(XMMRegister dst, XMMRegister shift);
2136   void pslld(XMMRegister dst, XMMRegister shift);
2137   void psllq(XMMRegister dst, XMMRegister shift);
2138   void vpsllw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2139   void vpslld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2140   void vpsllq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2141   void vpsllw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2142   void vpslld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2143   void vpsllq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2144   void vpslldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2145 
2146   // Logical shift right packed integers
2147   void psrlw(XMMRegister dst, int shift);
2148   void psrld(XMMRegister dst, int shift);
2149   void psrlq(XMMRegister dst, int shift);
2150   void psrlw(XMMRegister dst, XMMRegister shift);

2152   void psrlq(XMMRegister dst, XMMRegister shift);
2153   void vpsrlw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2154   void vpsrld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2155   void vpsrlq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2156   void vpsrlw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2157   void vpsrld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2158   void vpsrlq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2159   void vpsrldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2160   void evpsrlvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2161   void evpsllvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2162 
2163   // Arithmetic shift right packed integers (only shorts and ints, no instructions for longs)
2164   void psraw(XMMRegister dst, int shift);
2165   void psrad(XMMRegister dst, int shift);
2166   void psraw(XMMRegister dst, XMMRegister shift);
2167   void psrad(XMMRegister dst, XMMRegister shift);
2168   void vpsraw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2169   void vpsrad(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2170   void vpsraw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2171   void vpsrad(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);

2172   void evpsraq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2173   void evpsraq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2174 












2175   void vpshldvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2176   void vpshrdvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2177 
2178   // And packed integers
2179   void pand(XMMRegister dst, XMMRegister src);
2180   void vpand(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2181   void vpand(XMMRegister dst, XMMRegister nds, Address src, int vector_len);

2182   void vpandq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2183 
2184   // Andn packed integers
2185   void pandn(XMMRegister dst, XMMRegister src);
2186   void vpandn(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2187 
2188   // Or packed integers
2189   void por(XMMRegister dst, XMMRegister src);
2190   void vpor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2191   void vpor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2192   void vporq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2193 



2194   // Xor packed integers
2195   void pxor(XMMRegister dst, XMMRegister src);
2196   void vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2197   void vpxor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);


2198   void evpxorq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2199   void evpxorq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2200 
2201 
2202   // vinserti forms
2203   void vinserti128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2204   void vinserti128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2205   void vinserti32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2206   void vinserti32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2207   void vinserti64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2208 
2209   // vinsertf forms
2210   void vinsertf128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2211   void vinsertf128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2212   void vinsertf32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2213   void vinsertf32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2214   void vinsertf64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2215   void vinsertf64x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2216 
2217   // vextracti forms
2218   void vextracti128(XMMRegister dst, XMMRegister src, uint8_t imm8);
2219   void vextracti128(Address dst, XMMRegister src, uint8_t imm8);
2220   void vextracti32x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2221   void vextracti32x4(Address dst, XMMRegister src, uint8_t imm8);

2240   void vpbroadcastd(XMMRegister dst, XMMRegister src, int vector_len);
2241   void vpbroadcastd(XMMRegister dst, Address src, int vector_len);
2242   void vpbroadcastq(XMMRegister dst, XMMRegister src, int vector_len);
2243   void vpbroadcastq(XMMRegister dst, Address src, int vector_len);
2244 
2245   void evbroadcasti64x2(XMMRegister dst, XMMRegister src, int vector_len);
2246   void evbroadcasti64x2(XMMRegister dst, Address src, int vector_len);
2247 
2248   // scalar single/double precision replicate
2249   void vbroadcastss(XMMRegister dst, XMMRegister src, int vector_len);
2250   void vbroadcastss(XMMRegister dst, Address src, int vector_len);
2251   void vbroadcastsd(XMMRegister dst, XMMRegister src, int vector_len);
2252   void vbroadcastsd(XMMRegister dst, Address src, int vector_len);
2253 
2254   // gpr sourced byte/word/dword/qword replicate
2255   void evpbroadcastb(XMMRegister dst, Register src, int vector_len);
2256   void evpbroadcastw(XMMRegister dst, Register src, int vector_len);
2257   void evpbroadcastd(XMMRegister dst, Register src, int vector_len);
2258   void evpbroadcastq(XMMRegister dst, Register src, int vector_len);
2259 
2260   void evpgatherdd(XMMRegister dst, KRegister k1, Address src, int vector_len);














2261 
2262   // Carry-Less Multiplication Quadword
2263   void pclmulqdq(XMMRegister dst, XMMRegister src, int mask);
2264   void vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask);
2265   void evpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask, int vector_len);
2266   // AVX instruction which is used to clear upper 128 bits of YMM registers and
2267   // to avoid transaction penalty between AVX and SSE states. There is no
2268   // penalty if legacy SSE instructions are encoded using VEX prefix because
2269   // they always clear upper 128 bits. It should be used before calling
2270   // runtime code and native libraries.
2271   void vzeroupper();
2272 
2273   // AVX support for vectorized conditional move (float/double). The following two instructions used only coupled.
2274   void cmppd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cop, int vector_len);
2275   void blendvpd(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len);
2276   void cmpps(XMMRegister dst, XMMRegister nds, XMMRegister src, int cop, int vector_len);
2277   void blendvps(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len);





































2278   void vpblendd(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
2279 





2280  protected:
2281   // Next instructions require address alignment 16 bytes SSE mode.
2282   // They should be called only from corresponding MacroAssembler instructions.
2283   void andpd(XMMRegister dst, Address src);
2284   void andps(XMMRegister dst, Address src);
2285   void xorpd(XMMRegister dst, Address src);
2286   void xorps(XMMRegister dst, Address src);
2287 
2288 };
2289 
2290 // The Intel x86/Amd64 Assembler attributes: All fields enclosed here are to guide encoding level decisions.
2291 // Specific set functions are for specialized use, else defaults or whatever was supplied to object construction
2292 // are applied.
2293 class InstructionAttr {
2294 public:
2295   InstructionAttr(
2296     int vector_len,     // The length of vector to be applied in encoding - for both AVX and EVEX
2297     bool rex_vex_w,     // Width of data: if 32-bits or less, false, else if 64-bit or specially defined, true
2298     bool legacy_mode,   // Details if either this instruction is conditionally encoded to AVX or earlier if true else possibly EVEX
2299     bool no_reg_mask,   // when true, k0 is used when EVEX encoding is chosen, else embedded_opmask_register_specifier is used

2355   int  get_embedded_opmask_register_specifier(void) const { return _embedded_opmask_register_specifier; }
2356 
2357   // Set the vector len manually
2358   void set_vector_len(int vector_len) { _avx_vector_len = vector_len; }
2359 
2360   // Set revert rex_vex_w for avx encoding
2361   void set_rex_vex_w_reverted(void) { _rex_vex_w_reverted = true; }
2362 
2363   // Set rex_vex_w based on state
2364   void set_rex_vex_w(bool state) { _rex_vex_w = state; }
2365 
2366   // Set the instruction to be encoded in AVX mode
2367   void set_is_legacy_mode(void) { _legacy_mode = true; }
2368 
2369   // Set the current instuction to be encoded as an EVEX instuction
2370   void set_is_evex_instruction(void) { _is_evex_instruction = true; }
2371 
2372   // Internal encoding data used in compressed immediate offset programming
2373   void set_evex_encoding(int value) { _evex_encoding = value; }
2374 
2375   // Set the Evex.Z field to be used to clear all non directed XMM/YMM/ZMM components

2376   void reset_is_clear_context(void) { _is_clear_context = false; }
2377 
2378   // Map back to current asembler so that we can manage object level assocation
2379   void set_current_assembler(Assembler *current_assembler) { _current_assembler = current_assembler; }
2380 
2381   // Address modifiers used for compressed displacement calculation
2382   void set_address_attributes(int tuple_type, int input_size_in_bits) {
2383     if (VM_Version::supports_evex()) {
2384       _tuple_type = tuple_type;
2385       _input_size_in_bits = input_size_in_bits;
2386     }
2387   }
2388 
2389   // Set embedded opmask register specifier.
2390   void set_embedded_opmask_register_specifier(KRegister mask) {
2391     _embedded_opmask_register_specifier = (*mask).encoding() & 0x7;
2392   }
2393 
2394 };
2395

 603     EVEX_8bit  = 0,
 604     EVEX_16bit = 1,
 605     EVEX_32bit = 2,
 606     EVEX_64bit = 3,
 607     EVEX_NObit = 4
 608   };
 609 
 610   enum WhichOperand {
 611     // input to locate_operand, and format code for relocations
 612     imm_operand  = 0,            // embedded 32-bit|64-bit immediate operand
 613     disp32_operand = 1,          // embedded 32-bit displacement or address
 614     call32_operand = 2,          // embedded 32-bit self-relative displacement
 615 #ifndef _LP64
 616     _WhichOperand_limit = 3
 617 #else
 618      narrow_oop_operand = 3,     // embedded 32-bit immediate narrow oop
 619     _WhichOperand_limit = 4
 620 #endif
 621   };
 622 
 623   // Comparison predicates for integral types & FP types when using SSE
 624   enum ComparisonPredicate {
 625     eq = 0,
 626     lt = 1,
 627     le = 2,
 628     _false = 3,
 629     neq = 4,
 630     nlt = 5,
 631     nle = 6,
 632     _true = 7
 633   };
 634 
 635   // Comparison predicates for FP types when using AVX
 636   // O means ordered. U is unordered. When using ordered, any NaN comparison is false. Otherwise, it is true.
 637   // S means signaling. Q means non-signaling. When signaling is true, instruction signals #IA on NaN.
 638   enum ComparisonPredicateFP {
 639     EQ_OQ = 0,
 640     LT_OS = 1,
 641     LE_OS = 2,
 642     UNORD_Q = 3,
 643     NEQ_UQ = 4,
 644     NLT_US = 5,
 645     NLE_US = 6,
 646     ORD_Q = 7,
 647     EQ_UQ = 8,
 648     NGE_US = 9,
 649     NGT_US = 0xA,
 650     FALSE_OQ = 0XB,
 651     NEQ_OQ = 0xC,
 652     GE_OS = 0xD,
 653     GT_OS = 0xE,
 654     TRUE_UQ = 0xF,
 655     EQ_OS = 0x10,
 656     LT_OQ = 0x11,
 657     LE_OQ = 0x12,
 658     UNORD_S = 0x13,
 659     NEQ_US = 0x14,
 660     NLT_UQ = 0x15,
 661     NLE_UQ = 0x16,
 662     ORD_S = 0x17,
 663     EQ_US = 0x18,
 664     NGE_UQ = 0x19,
 665     NGT_UQ = 0x1A,
 666     FALSE_OS = 0x1B,
 667     NEQ_OS = 0x1C,
 668     GE_OQ = 0x1D,
 669     GT_OQ = 0x1E,
 670     TRUE_US =0x1F
 671   };
 672 
 673   enum Width {
 674     B = 0,
 675     W = 1,
 676     D = 2,
 677     Q = 3
 678   };
 679 
 680   //---<  calculate length of instruction  >---
 681   // As instruction size can't be found out easily on x86/x64,
 682   // we just use '4' for len and maxlen.
 683   // instruction must start at passed address
 684   static unsigned int instr_len(unsigned char *instr) { return 4; }
 685 
 686   //---<  longest instructions  >---
 687   // Max instruction length is not specified in architecture documentation.
 688   // We could use a "safe enough" estimate (15), but just default to
 689   // instruction length guess from above.
 690   static unsigned int instr_maxlen() { return 4; }
 691 
 692   // NOTE: The general philopsophy of the declarations here is that 64bit versions
 693   // of instructions are freely declared without the need for wrapping them an ifdef.
 694   // (Some dangerous instructions are ifdef's out of inappropriate jvm's.)
 695   // In the .cpp file the implementations are wrapped so that they are dropped out
 696   // of the resulting jvm. This is done mostly to keep the footprint of MINIMAL
 697   // to the size it was prior to merging up the 32bit and 64bit assemblers.
 698   //
 699   // This does mean you'll get a linker/runtime error if you use a 64bit only instruction

 979   void rep_stos();
 980   void rep_stosb();
 981   void repne_scan();
 982 #ifdef _LP64
 983   void repne_scanl();
 984 #endif
 985 
 986   // Vanilla instructions in lexical order
 987 
 988   void adcl(Address dst, int32_t imm32);
 989   void adcl(Address dst, Register src);
 990   void adcl(Register dst, int32_t imm32);
 991   void adcl(Register dst, Address src);
 992   void adcl(Register dst, Register src);
 993 
 994   void adcq(Register dst, int32_t imm32);
 995   void adcq(Register dst, Address src);
 996   void adcq(Register dst, Register src);
 997 
 998   void addb(Address dst, int imm8);
 999   void addw(Register dst, Register src);
1000   void addw(Address dst, int imm16);
1001 
1002   void addl(Address dst, int32_t imm32);
1003   void addl(Address dst, Register src);
1004   void addl(Register dst, int32_t imm32);
1005   void addl(Register dst, Address src);
1006   void addl(Register dst, Register src);
1007 
1008   void addq(Address dst, int32_t imm32);
1009   void addq(Address dst, Register src);
1010   void addq(Register dst, int32_t imm32);
1011   void addq(Register dst, Address src);
1012   void addq(Register dst, Register src);
1013 
1014 #ifdef _LP64
1015  //Add Unsigned Integers with Carry Flag
1016   void adcxq(Register dst, Register src);
1017 
1018  //Add Unsigned Integers with Overflow Flag
1019   void adoxq(Register dst, Register src);

1030 
1031   // Add Scalar Single-Precision Floating-Point Values
1032   void addss(XMMRegister dst, Address src);
1033   void addss(XMMRegister dst, XMMRegister src);
1034 
1035   // AES instructions
1036   void aesdec(XMMRegister dst, Address src);
1037   void aesdec(XMMRegister dst, XMMRegister src);
1038   void aesdeclast(XMMRegister dst, Address src);
1039   void aesdeclast(XMMRegister dst, XMMRegister src);
1040   void aesenc(XMMRegister dst, Address src);
1041   void aesenc(XMMRegister dst, XMMRegister src);
1042   void aesenclast(XMMRegister dst, Address src);
1043   void aesenclast(XMMRegister dst, XMMRegister src);
1044   // Vector AES instructions
1045   void vaesenc(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1046   void vaesenclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1047   void vaesdec(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1048   void vaesdeclast(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1049 
1050   void andw(Register dst, Register src);
1051 
1052   void andl(Address  dst, int32_t imm32);
1053   void andl(Register dst, int32_t imm32);
1054   void andl(Register dst, Address src);
1055   void andl(Register dst, Register src);
1056 
1057   void andq(Address  dst, int32_t imm32);
1058   void andq(Register dst, int32_t imm32);
1059   void andq(Register dst, Address src);
1060   void andq(Register dst, Register src);
1061 
1062   // BMI instructions
1063   void andnl(Register dst, Register src1, Register src2);
1064   void andnl(Register dst, Register src1, Address src2);
1065   void andnq(Register dst, Register src1, Register src2);
1066   void andnq(Register dst, Register src1, Address src2);
1067 
1068   void blsil(Register dst, Register src);
1069   void blsil(Register dst, Address src);
1070   void blsiq(Register dst, Register src);
1071   void blsiq(Register dst, Address src);

1157   void crc32(Register crc, Address adr, int8_t sizeInBytes);
1158 
1159   // Convert Scalar Double-Precision Floating-Point Value to Scalar Single-Precision Floating-Point Value
1160   void cvtsd2ss(XMMRegister dst, XMMRegister src);
1161   void cvtsd2ss(XMMRegister dst, Address src);
1162 
1163   // Convert Doubleword Integer to Scalar Double-Precision Floating-Point Value
1164   void cvtsi2sdl(XMMRegister dst, Register src);
1165   void cvtsi2sdl(XMMRegister dst, Address src);
1166   void cvtsi2sdq(XMMRegister dst, Register src);
1167   void cvtsi2sdq(XMMRegister dst, Address src);
1168 
1169   // Convert Doubleword Integer to Scalar Single-Precision Floating-Point Value
1170   void cvtsi2ssl(XMMRegister dst, Register src);
1171   void cvtsi2ssl(XMMRegister dst, Address src);
1172   void cvtsi2ssq(XMMRegister dst, Register src);
1173   void cvtsi2ssq(XMMRegister dst, Address src);
1174 
1175   // Convert Packed Signed Doubleword Integers to Packed Double-Precision Floating-Point Value
1176   void cvtdq2pd(XMMRegister dst, XMMRegister src);
1177   void vcvtdq2pd(XMMRegister dst, XMMRegister src, int vector_len);
1178 
1179   // Convert Packed Signed Doubleword Integers to Packed Single-Precision Floating-Point Value
1180   void cvtdq2ps(XMMRegister dst, XMMRegister src);
1181   void vcvtdq2ps(XMMRegister dst, XMMRegister src, int vector_len);
1182 
1183   // Convert Scalar Single-Precision Floating-Point Value to Scalar Double-Precision Floating-Point Value
1184   void cvtss2sd(XMMRegister dst, XMMRegister src);
1185   void cvtss2sd(XMMRegister dst, Address src);
1186 
1187   // Convert with Truncation Scalar Double-Precision Floating-Point Value to Doubleword Integer
1188   void cvttsd2sil(Register dst, Address src);
1189   void cvttsd2sil(Register dst, XMMRegister src);
1190   void cvttsd2siq(Register dst, Address src);
1191   void cvttsd2siq(Register dst, XMMRegister src);
1192 
1193   // Convert with Truncation Scalar Single-Precision Floating-Point Value to Doubleword Integer
1194   void cvttss2sil(Register dst, XMMRegister src);
1195   void cvttss2siq(Register dst, XMMRegister src);
1196 
1197   // Convert vector double to int
1198   void cvttpd2dq(XMMRegister dst, XMMRegister src);
1199 
1200   // Convert vector float and double
1201   void vcvtps2pd(XMMRegister dst, XMMRegister src, int vector_len);
1202   void vcvtpd2ps(XMMRegister dst, XMMRegister src, int vector_len);
1203 
1204   // Convert vector long to vector FP
1205   void evcvtqq2ps(XMMRegister dst, XMMRegister src, int vector_len);
1206   void evcvtqq2pd(XMMRegister dst, XMMRegister src, int vector_len);
1207 
1208   // Evex casts with truncation
1209   void evpmovwb(XMMRegister dst, XMMRegister src, int vector_len);
1210   void evpmovdw(XMMRegister dst, XMMRegister src, int vector_len);
1211   void evpmovdb(XMMRegister dst, XMMRegister src, int vector_len);
1212   void evpmovqd(XMMRegister dst, XMMRegister src, int vector_len);
1213   void evpmovqb(XMMRegister dst, XMMRegister src, int vector_len);
1214   void evpmovqw(XMMRegister dst, XMMRegister src, int vector_len);
1215 
1216   //Abs of packed Integer values
1217   void pabsb(XMMRegister dst, XMMRegister src);
1218   void pabsw(XMMRegister dst, XMMRegister src);
1219   void pabsd(XMMRegister dst, XMMRegister src);
1220   void vpabsb(XMMRegister dst, XMMRegister src, int vector_len);
1221   void vpabsw(XMMRegister dst, XMMRegister src, int vector_len);
1222   void vpabsd(XMMRegister dst, XMMRegister src, int vector_len);
1223   void evpabsq(XMMRegister dst, XMMRegister src, int vector_len);
1224 
1225   // Divide Scalar Double-Precision Floating-Point Values
1226   void divsd(XMMRegister dst, Address src);
1227   void divsd(XMMRegister dst, XMMRegister src);
1228 
1229   // Divide Scalar Single-Precision Floating-Point Values
1230   void divss(XMMRegister dst, Address src);
1231   void divss(XMMRegister dst, XMMRegister src);
1232 
1233 
1234 #ifndef _LP64
1235  private:

1555 
1556   // Move Double Quadword
1557   void movdq(XMMRegister dst, Register src);
1558   void movdq(Register dst, XMMRegister src);
1559 
1560   // Move Aligned Double Quadword
1561   void movdqa(XMMRegister dst, XMMRegister src);
1562   void movdqa(XMMRegister dst, Address src);
1563 
1564   // Move Unaligned Double Quadword
1565   void movdqu(Address     dst, XMMRegister src);
1566   void movdqu(XMMRegister dst, Address src);
1567   void movdqu(XMMRegister dst, XMMRegister src);
1568 
1569   // Move Unaligned 256bit Vector
1570   void vmovdqu(Address dst, XMMRegister src);
1571   void vmovdqu(XMMRegister dst, Address src);
1572   void vmovdqu(XMMRegister dst, XMMRegister src);
1573 
1574    // Move Unaligned 512bit Vector
1575   void evmovdqub(Address dst, XMMRegister src, bool merge, int vector_len);
1576   void evmovdqub(XMMRegister dst, Address src, bool merge, int vector_len);
1577   void evmovdqub(XMMRegister dst, XMMRegister src, bool merge, int vector_len);
1578   void evmovdqub(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1579   void evmovdquw(Address dst, XMMRegister src, bool merge, int vector_len);
1580   void evmovdquw(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1581   void evmovdquw(XMMRegister dst, Address src, bool merge, int vector_len);
1582   void evmovdquw(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1583   void evmovdqul(Address dst, XMMRegister src, int vector_len);
1584   void evmovdqul(XMMRegister dst, Address src, int vector_len);
1585   void evmovdqul(XMMRegister dst, XMMRegister src, int vector_len);
1586   void evmovdqul(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1587   void evmovdqul(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1588   void evmovdqul(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1589   void evmovdquq(Address dst, XMMRegister src, int vector_len);
1590   void evmovdquq(XMMRegister dst, Address src, int vector_len);
1591   void evmovdquq(XMMRegister dst, XMMRegister src, int vector_len);
1592   void evmovdquq(Address dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1593   void evmovdquq(XMMRegister dst, KRegister mask, Address src, bool merge, int vector_len);
1594   void evmovdquq(XMMRegister dst, KRegister mask, XMMRegister src, bool merge, int vector_len);
1595 
1596   // Move lower 64bit to high 64bit in 128bit register
1597   void movlhps(XMMRegister dst, XMMRegister src);
1598 
1599   void movl(Register dst, int32_t imm32);
1600   void movl(Address dst, int32_t imm32);
1601   void movl(Register dst, Register src);
1602   void movl(Register dst, Address src);
1603   void movl(Address dst, Register src);
1604 
1605   // These dummies prevent using movl from converting a zero (like NULL) into Register
1606   // by giving the compiler two choices it can't resolve
1607 
1608   void movl(Address  dst, void* junk);
1609   void movl(Register dst, void* junk);
1610 
1611 #ifdef _LP64
1612   void movq(Register dst, Register src);
1613   void movq(Register dst, Address src);
1614   void movq(Address  dst, Register src);
1615 
1616   // These dummies prevent using movq from converting a zero (like NULL) into Register
1617   // by giving the compiler two choices it can't resolve
1618 
1619   void movq(Address  dst, void* dummy);
1620   void movq(Register dst, void* dummy);
1621 #endif
1622 
1623   // Move Quadword
1624   void movq(Address     dst, XMMRegister src);
1625   void movq(XMMRegister dst, Address src);
1626   void movq(XMMRegister dst, XMMRegister src);
1627   void movq(Register dst, XMMRegister src);
1628   void movq(XMMRegister dst, Register src);
1629 
1630   void movsbl(Register dst, Address src);
1631   void movsbl(Register dst, Register src);
1632 
1633 #ifdef _LP64
1634   void movsbq(Register dst, Address src);
1635   void movsbq(Register dst, Register src);
1636 
1637   // Move signed 32bit immediate to 64bit extending sign
1638   void movslq(Address  dst, int32_t imm64);
1639   void movslq(Register dst, int32_t imm64);
1640 
1641   void movslq(Register dst, Address src);
1642   void movslq(Register dst, Register src);
1643   void movslq(Register dst, void* src); // Dummy declaration to cause NULL to be ambiguous
1644 #endif
1645 
1646   void movswl(Register dst, Address src);
1647   void movswl(Register dst, Register src);
1648

1689   void mulss(XMMRegister dst, Address src);
1690   void mulss(XMMRegister dst, XMMRegister src);
1691 
1692   void negl(Register dst);
1693 
1694 #ifdef _LP64
1695   void negq(Register dst);
1696 #endif
1697 
1698   void nop(int i = 1);
1699 
1700   void notl(Register dst);
1701 
1702 #ifdef _LP64
1703   void notq(Register dst);
1704 
1705   void btsq(Address dst, int imm8);
1706   void btrq(Address dst, int imm8);
1707 #endif
1708 
1709   void orw(Register dst, Register src);
1710 
1711   void orl(Address dst, int32_t imm32);
1712   void orl(Register dst, int32_t imm32);
1713   void orl(Register dst, Address src);
1714   void orl(Register dst, Register src);
1715   void orl(Address dst, Register src);
1716 
1717   void orb(Address dst, int imm8);
1718 
1719   void orq(Address dst, int32_t imm32);
1720   void orq(Register dst, int32_t imm32);
1721   void orq(Register dst, Address src);
1722   void orq(Register dst, Register src);
1723 
1724   // Pack with signed saturation
1725   void packsswb(XMMRegister dst, XMMRegister src);
1726   void vpacksswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1727   void packssdw(XMMRegister dst, XMMRegister src);
1728   void vpackssdw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1729 
1730   // Pack with unsigned saturation
1731   void packuswb(XMMRegister dst, XMMRegister src);
1732   void packuswb(XMMRegister dst, Address src);
1733   void packusdw(XMMRegister dst, XMMRegister src);
1734   void vpackuswb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1735   void vpackusdw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1736 
1737   // Permutations
1738   void vpermq(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1739   void vpermq(XMMRegister dst, XMMRegister src, int imm8);
1740   void vpermq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1741   void vpermb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1742   void vpermw(XMMRegister dst,  XMMRegister nds, XMMRegister src, int vector_len);
1743   void vpermd(XMMRegister dst,  XMMRegister nds, Address src, int vector_len);
1744   void vpermd(XMMRegister dst,  XMMRegister nds, XMMRegister src, int vector_len);
1745   void vperm2i128(XMMRegister dst,  XMMRegister nds, XMMRegister src, int imm8);
1746   void vperm2f128(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8);
1747   void vpermilps(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1748   void vpermilpd(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1749   void vpermpd(XMMRegister dst, XMMRegister src, int imm8, int vector_len);
1750   void evpermi2q(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1751 
1752   void pause();
1753 
1754   // Undefined Instruction
1755   void ud2();
1756 
1757   // SSE4.2 string instructions
1758   void pcmpestri(XMMRegister xmm1, XMMRegister xmm2, int imm8);
1759   void pcmpestri(XMMRegister xmm1, Address src, int imm8);
1760 
1761   void pcmpeqb(XMMRegister dst, XMMRegister src);
1762   void vpcmpCCbwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cond_encoding, int vector_len);
1763 
1764   void vpcmpeqb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1765   void evpcmpeqb(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1766   void evpcmpeqb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1767   void evpcmpeqb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1768 
1769   void vpcmpgtb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1770   void evpcmpgtb(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1771   void evpcmpgtb(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1772 
1773   void evpcmpuw(KRegister kdst, XMMRegister nds, XMMRegister src, ComparisonPredicate vcc, int vector_len);
1774   void evpcmpuw(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src, ComparisonPredicate of, int vector_len);
1775   void evpcmpuw(KRegister kdst, XMMRegister nds, Address src, ComparisonPredicate vcc, int vector_len);
1776 
1777   void pcmpeqw(XMMRegister dst, XMMRegister src);
1778   void vpcmpeqw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1779   void evpcmpeqw(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1780   void evpcmpeqw(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1781 
1782   void vpcmpgtw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1783 
1784   void pcmpeqd(XMMRegister dst, XMMRegister src);
1785   void vpcmpeqd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1786   void evpcmpeqd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src, int vector_len);
1787   void evpcmpeqd(KRegister kdst, KRegister mask, XMMRegister nds, Address src, int vector_len);
1788 
1789   void pcmpeqq(XMMRegister dst, XMMRegister src);
1790   void vpcmpCCq(XMMRegister dst, XMMRegister nds, XMMRegister src, int cond_encoding, int vector_len);
1791   void vpcmpeqq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1792   void evpcmpeqq(KRegister kdst, XMMRegister nds, XMMRegister src, int vector_len);
1793   void evpcmpeqq(KRegister kdst, XMMRegister nds, Address src, int vector_len);
1794 
1795   void pcmpgtq(XMMRegister dst, XMMRegister src);
1796   void vpcmpgtq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1797 
1798   void pmovmskb(Register dst, XMMRegister src);
1799   void vpmovmskb(Register dst, XMMRegister src);
1800 
1801   // SSE 4.1 extract
1802   void pextrd(Register dst, XMMRegister src, int imm8);
1803   void pextrq(Register dst, XMMRegister src, int imm8);
1804   void pextrd(Address dst, XMMRegister src, int imm8);
1805   void pextrq(Address dst, XMMRegister src, int imm8);
1806   void pextrb(Register dst, XMMRegister src, int imm8);
1807   void pextrb(Address dst, XMMRegister src, int imm8);
1808   // SSE 2 extract
1809   void pextrw(Register dst, XMMRegister src, int imm8);
1810   void pextrw(Address dst, XMMRegister src, int imm8);
1811 
1812   // SSE 4.1 insert
1813   void pinsrd(XMMRegister dst, Register src, int imm8);
1814   void pinsrq(XMMRegister dst, Register src, int imm8);
1815   void pinsrb(XMMRegister dst, Register src, int imm8);
1816   void pinsrd(XMMRegister dst, Address src, int imm8);
1817   void pinsrq(XMMRegister dst, Address src, int imm8);
1818   void pinsrb(XMMRegister dst, Address src, int imm8);
1819   void insertps(XMMRegister dst, XMMRegister src, int imm8);
1820   // SSE 2 insert
1821   void pinsrw(XMMRegister dst, Register src, int imm8);
1822   void pinsrw(XMMRegister dst, Address src, int imm8);
1823 
1824   // AVX insert
1825   void vpinsrd(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1826   void vpinsrb(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1827   void vpinsrq(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1828   void vpinsrw(XMMRegister dst, XMMRegister nds, Register src, int imm8);
1829   void vinsertps(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8);
1830 
1831   // Zero extend moves
1832   void pmovzxbw(XMMRegister dst, XMMRegister src);
1833   void pmovzxbw(XMMRegister dst, Address src);
1834   void pmovzxbd(XMMRegister dst, XMMRegister src);
1835   void vpmovzxbw( XMMRegister dst, Address src, int vector_len);
1836   void pmovzxdq(XMMRegister dst, XMMRegister src);
1837   void vpmovzxbw(XMMRegister dst, XMMRegister src, int vector_len);
1838   void vpmovzxdq(XMMRegister dst, XMMRegister src, int vector_len);
1839   void vpmovzxbd(XMMRegister dst, XMMRegister src, int vector_len);
1840   void vpmovzxbq(XMMRegister dst, XMMRegister src, int vector_len);
1841   void evpmovzxbw(XMMRegister dst, KRegister mask, Address src, int vector_len);
1842 
1843   // Sign extend moves
1844   void pmovsxbd(XMMRegister dst, XMMRegister src);
1845   void pmovsxbq(XMMRegister dst, XMMRegister src);
1846   void pmovsxbw(XMMRegister dst, XMMRegister src);
1847   void pmovsxwd(XMMRegister dst, XMMRegister src);
1848   void vpmovsxbd(XMMRegister dst, XMMRegister src, int vector_len);
1849   void vpmovsxbq(XMMRegister dst, XMMRegister src, int vector_len);
1850   void vpmovsxbw(XMMRegister dst, XMMRegister src, int vector_len);
1851   void vpmovsxwd(XMMRegister dst, XMMRegister src, int vector_len);
1852   void vpmovsxwq(XMMRegister dst, XMMRegister src, int vector_len);
1853   void vpmovsxdq(XMMRegister dst, XMMRegister src, int vector_len);
1854 
1855   void evpmovwb(Address dst, XMMRegister src, int vector_len);
1856   void evpmovwb(Address dst, KRegister mask, XMMRegister src, int vector_len);
1857 
1858   void vpmovzxwd(XMMRegister dst, XMMRegister src, int vector_len);
1859 
1860   void evpmovdb(Address dst, XMMRegister src, int vector_len);
1861 




1862   // Multiply add
1863   void pmaddwd(XMMRegister dst, XMMRegister src);
1864   void vpmaddwd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1865   // Multiply add accumulate
1866   void evpdpwssd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1867 
1868 #ifndef _LP64 // no 32bit push/pop on amd64
1869   void popl(Address dst);
1870 #endif
1871 
1872 #ifdef _LP64
1873   void popq(Address dst);
1874 #endif
1875 
1876   void popcntl(Register dst, Address src);
1877   void popcntl(Register dst, Register src);
1878 
1879   void vpopcntd(XMMRegister dst, XMMRegister src, int vector_len);
1880 
1881 #ifdef _LP64

1885 
1886   // Prefetches (SSE, SSE2, 3DNOW only)
1887 
1888   void prefetchnta(Address src);
1889   void prefetchr(Address src);
1890   void prefetcht0(Address src);
1891   void prefetcht1(Address src);
1892   void prefetcht2(Address src);
1893   void prefetchw(Address src);
1894 
1895   // Shuffle Bytes
1896   void pshufb(XMMRegister dst, XMMRegister src);
1897   void pshufb(XMMRegister dst, Address src);
1898   void vpshufb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
1899 
1900   // Shuffle Packed Doublewords
1901   void pshufd(XMMRegister dst, XMMRegister src, int mode);
1902   void pshufd(XMMRegister dst, Address src,     int mode);
1903   void vpshufd(XMMRegister dst, XMMRegister src, int mode, int vector_len);
1904 
1905   // Shuffle Packed High/Low Words
1906   void pshufhw(XMMRegister dst, XMMRegister src, int mode);
1907   void pshuflw(XMMRegister dst, XMMRegister src, int mode);
1908   void pshuflw(XMMRegister dst, Address src,     int mode);
1909 
1910   //shuffle floats and doubles
1911   void pshufps(XMMRegister, XMMRegister, int);
1912   void pshufpd(XMMRegister, XMMRegister, int);
1913   void vpshufps(XMMRegister, XMMRegister, XMMRegister, int, int);
1914   void vpshufpd(XMMRegister, XMMRegister, XMMRegister, int, int);
1915 
1916   // Shuffle packed values at 128 bit granularity
1917   void evshufi64x2(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
1918 
1919   // Shift Right by bytes Logical DoubleQuadword Immediate
1920   void psrldq(XMMRegister dst, int shift);
1921   // Shift Left by bytes Logical DoubleQuadword Immediate
1922   void pslldq(XMMRegister dst, int shift);
1923 
1924   // Logical Compare 128bit
1925   void ptest(XMMRegister dst, XMMRegister src);
1926   void ptest(XMMRegister dst, Address src);
1927   // Logical Compare 256bit
1928   void vptest(XMMRegister dst, XMMRegister src);
1929   void vptest(XMMRegister dst, Address src);
1930 
1931   // Vector compare
1932   void vptest(XMMRegister dst, XMMRegister src, int vector_len);
1933 
1934   // Interleave Low Bytes
1935   void punpcklbw(XMMRegister dst, XMMRegister src);
1936   void punpcklbw(XMMRegister dst, Address src);
1937 
1938   // Interleave Low Doublewords
1939   void punpckldq(XMMRegister dst, XMMRegister src);
1940   void punpckldq(XMMRegister dst, Address src);
1941 
1942   // Interleave Low Quadwords
1943   void punpcklqdq(XMMRegister dst, XMMRegister src);
1944 
1945 #ifndef _LP64 // no 32bit push/pop on amd64
1946   void pushl(Address src);
1947 #endif
1948 
1949   void pushq(Address src);
1950 
1951   void rcll(Register dst, int imm8);
1952 
1953   void rclq(Register dst, int imm8);

1976   void sarq(Register dst, int imm8);
1977   void sarq(Register dst);
1978 
1979   void sbbl(Address dst, int32_t imm32);
1980   void sbbl(Register dst, int32_t imm32);
1981   void sbbl(Register dst, Address src);
1982   void sbbl(Register dst, Register src);
1983 
1984   void sbbq(Address dst, int32_t imm32);
1985   void sbbq(Register dst, int32_t imm32);
1986   void sbbq(Register dst, Address src);
1987   void sbbq(Register dst, Register src);
1988 
1989   void setb(Condition cc, Register dst);
1990 
1991   void palignr(XMMRegister dst, XMMRegister src, int imm8);
1992   void vpalignr(XMMRegister dst, XMMRegister src1, XMMRegister src2, int imm8, int vector_len);
1993   void evalignq(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
1994 
1995   void pblendw(XMMRegister dst, XMMRegister src, int imm8);
1996   void vblendps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int imm8, int vector_len);
1997 
1998   void sha1rnds4(XMMRegister dst, XMMRegister src, int imm8);
1999   void sha1nexte(XMMRegister dst, XMMRegister src);
2000   void sha1msg1(XMMRegister dst, XMMRegister src);
2001   void sha1msg2(XMMRegister dst, XMMRegister src);
2002   // xmm0 is implicit additional source to the following instruction.
2003   void sha256rnds2(XMMRegister dst, XMMRegister src);
2004   void sha256msg1(XMMRegister dst, XMMRegister src);
2005   void sha256msg2(XMMRegister dst, XMMRegister src);
2006 
2007   void shldl(Register dst, Register src);
2008   void shldl(Register dst, Register src, int8_t imm8);
2009   void shrdl(Register dst, Register src);
2010   void shrdl(Register dst, Register src, int8_t imm8);
2011 
2012   void shll(Register dst, int imm8);
2013   void shll(Register dst);
2014 
2015   void shlq(Register dst, int imm8);
2016   void shlq(Register dst);

2095   void xbegin(Label& abort, relocInfo::relocType rtype = relocInfo::none);
2096 
2097   void xchgb(Register reg, Address adr);
2098   void xchgw(Register reg, Address adr);
2099   void xchgl(Register reg, Address adr);
2100   void xchgl(Register dst, Register src);
2101 
2102   void xchgq(Register reg, Address adr);
2103   void xchgq(Register dst, Register src);
2104 
2105   void xend();
2106 
2107   // Get Value of Extended Control Register
2108   void xgetbv();
2109 
2110   void xorl(Register dst, int32_t imm32);
2111   void xorl(Register dst, Address src);
2112   void xorl(Register dst, Register src);
2113 
2114   void xorb(Register dst, Address src);
2115   void xorw(Register dst, Register src);
2116 
2117   void xorq(Register dst, Address src);
2118   void xorq(Register dst, Register src);
2119 
2120   void set_byte_if_not_zero(Register dst); // sets reg to 1 if not zero, otherwise 0
2121 
2122   // AVX 3-operands scalar instructions (encoded with VEX prefix)
2123 
2124   void vaddsd(XMMRegister dst, XMMRegister nds, Address src);
2125   void vaddsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2126   void vaddss(XMMRegister dst, XMMRegister nds, Address src);
2127   void vaddss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2128   void vdivsd(XMMRegister dst, XMMRegister nds, Address src);
2129   void vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2130   void vdivss(XMMRegister dst, XMMRegister nds, Address src);
2131   void vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2132   void vfmadd231sd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2133   void vfmadd231ss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2134   void vmulsd(XMMRegister dst, XMMRegister nds, Address src);
2135   void vmulsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2136   void vmulss(XMMRegister dst, XMMRegister nds, Address src);
2137   void vmulss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2138   void vsubsd(XMMRegister dst, XMMRegister nds, Address src);
2139   void vsubsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2140   void vsubss(XMMRegister dst, XMMRegister nds, Address src);
2141   void vsubss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2142 
2143   void vmaxss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2144   void vmaxsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2145   void vminss(XMMRegister dst, XMMRegister nds, XMMRegister src);
2146   void vminsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
2147 
2148   void shlxl(Register dst, Register src1, Register src2);
2149   void shlxq(Register dst, Register src1, Register src2);
2150 
2151   //====================VECTOR ARITHMETIC=====================================
2152   void evpmovd2m(KRegister kdst, XMMRegister src, int vector_len);
2153   void evpmovq2m(KRegister kdst, XMMRegister src, int vector_len);
2154 
2155   // Add Packed Floating-Point Values
2156   void addpd(XMMRegister dst, XMMRegister src);
2157   void addpd(XMMRegister dst, Address src);
2158   void addps(XMMRegister dst, XMMRegister src);
2159   void vaddpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2160   void vaddps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2161   void vaddpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2162   void vaddps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2163 
2164   // Subtract Packed Floating-Point Values
2165   void subpd(XMMRegister dst, XMMRegister src);
2166   void subps(XMMRegister dst, XMMRegister src);
2167   void vsubpd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2168   void vsubps(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2169   void vsubpd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2170   void vsubps(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2171 
2172   // Multiply Packed Floating-Point Values
2173   void mulpd(XMMRegister dst, XMMRegister src);

2243   void vpaddd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2244   void vpaddq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2245 
2246   // Sub packed integers
2247   void psubb(XMMRegister dst, XMMRegister src);
2248   void psubw(XMMRegister dst, XMMRegister src);
2249   void psubd(XMMRegister dst, XMMRegister src);
2250   void psubq(XMMRegister dst, XMMRegister src);
2251   void vpsubb(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2252   void vpsubw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2253   void vpsubd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2254   void vpsubq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2255   void vpsubb(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2256   void vpsubw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2257   void vpsubd(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2258   void vpsubq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2259 
2260   // Multiply packed integers (only shorts and ints)
2261   void pmullw(XMMRegister dst, XMMRegister src);
2262   void pmulld(XMMRegister dst, XMMRegister src);
2263   void pmuludq(XMMRegister dst, XMMRegister src);
2264   void vpmullw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2265   void vpmulld(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2266   void vpmullq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2267   void vpmuludq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2268   void vpmullw(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2269   void vpmulld(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2270   void vpmullq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2271 
2272   // Minimum of packed integers
2273   void pminsb(XMMRegister dst, XMMRegister src);
2274   void vpminsb(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2275   void pminsw(XMMRegister dst, XMMRegister src);
2276   void vpminsw(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2277   void pminsd(XMMRegister dst, XMMRegister src);
2278   void vpminsd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2279   void vpminsq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2280   void minps(XMMRegister dst, XMMRegister src);
2281   void vminps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2282   void minpd(XMMRegister dst, XMMRegister src);
2283   void vminpd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2284 
2285   // Maximum of packed integers
2286   void pmaxsb(XMMRegister dst, XMMRegister src);
2287   void vpmaxsb(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2288   void pmaxsw(XMMRegister dst, XMMRegister src);
2289   void vpmaxsw(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2290   void pmaxsd(XMMRegister dst, XMMRegister src);
2291   void vpmaxsd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2292   void vpmaxsq(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2293   void maxps(XMMRegister dst, XMMRegister src);
2294   void vmaxps(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2295   void maxpd(XMMRegister dst, XMMRegister src);
2296   void vmaxpd(XMMRegister dst, XMMRegister src1, XMMRegister src2, int vector_len);
2297 
2298   // Shift left packed integers
2299   void psllw(XMMRegister dst, int shift);
2300   void pslld(XMMRegister dst, int shift);
2301   void psllq(XMMRegister dst, int shift);
2302   void psllw(XMMRegister dst, XMMRegister shift);
2303   void pslld(XMMRegister dst, XMMRegister shift);
2304   void psllq(XMMRegister dst, XMMRegister shift);
2305   void vpsllw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2306   void vpslld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2307   void vpsllq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2308   void vpsllw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2309   void vpslld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2310   void vpsllq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2311   void vpslldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2312 
2313   // Logical shift right packed integers
2314   void psrlw(XMMRegister dst, int shift);
2315   void psrld(XMMRegister dst, int shift);
2316   void psrlq(XMMRegister dst, int shift);
2317   void psrlw(XMMRegister dst, XMMRegister shift);

2319   void psrlq(XMMRegister dst, XMMRegister shift);
2320   void vpsrlw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2321   void vpsrld(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2322   void vpsrlq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2323   void vpsrlw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2324   void vpsrld(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2325   void vpsrlq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2326   void vpsrldq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2327   void evpsrlvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2328   void evpsllvw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2329 
2330   // Arithmetic shift right packed integers (only shorts and ints, no instructions for longs)
2331   void psraw(XMMRegister dst, int shift);
2332   void psrad(XMMRegister dst, int shift);
2333   void psraw(XMMRegister dst, XMMRegister shift);
2334   void psrad(XMMRegister dst, XMMRegister shift);
2335   void vpsraw(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2336   void vpsrad(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2337   void vpsraw(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2338   void vpsrad(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2339   void evpsravw(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2340   void evpsraq(XMMRegister dst, XMMRegister src, int shift, int vector_len);
2341   void evpsraq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2342 
2343   // Variable shift left packed integers
2344   void vpsllvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2345   void vpsllvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2346 
2347   // Variable shift right packed integers
2348   void vpsrlvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2349   void vpsrlvq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2350 
2351   // Variable shift right arithmetic packed integers
2352   void vpsravd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2353   void evpsravq(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2354 
2355   void vpshldvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2356   void vpshrdvd(XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
2357 
2358   // And packed integers
2359   void pand(XMMRegister dst, XMMRegister src);
2360   void vpand(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2361   void vpand(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2362   void evpandd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2363   void vpandq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2364 
2365   // Andn packed integers
2366   void pandn(XMMRegister dst, XMMRegister src);
2367   void vpandn(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2368 
2369   // Or packed integers
2370   void por(XMMRegister dst, XMMRegister src);
2371   void vpor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2372   void vpor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2373   void vporq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2374 
2375   void evpord(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2376   void evpord(XMMRegister dst, KRegister mask, XMMRegister nds, Address src, bool merge, int vector_len);
2377 
2378   // Xor packed integers
2379   void pxor(XMMRegister dst, XMMRegister src);
2380   void vpxor(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2381   void vpxor(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2382   void vpxorq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2383   void evpxord(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2384   void evpxorq(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2385   void evpxorq(XMMRegister dst, XMMRegister nds, Address src, int vector_len);
2386 

2387   // vinserti forms
2388   void vinserti128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2389   void vinserti128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2390   void vinserti32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2391   void vinserti32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2392   void vinserti64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2393 
2394   // vinsertf forms
2395   void vinsertf128(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2396   void vinsertf128(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2397   void vinsertf32x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2398   void vinsertf32x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2399   void vinsertf64x4(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8);
2400   void vinsertf64x4(XMMRegister dst, XMMRegister nds, Address src, uint8_t imm8);
2401 
2402   // vextracti forms
2403   void vextracti128(XMMRegister dst, XMMRegister src, uint8_t imm8);
2404   void vextracti128(Address dst, XMMRegister src, uint8_t imm8);
2405   void vextracti32x4(XMMRegister dst, XMMRegister src, uint8_t imm8);
2406   void vextracti32x4(Address dst, XMMRegister src, uint8_t imm8);

2425   void vpbroadcastd(XMMRegister dst, XMMRegister src, int vector_len);
2426   void vpbroadcastd(XMMRegister dst, Address src, int vector_len);
2427   void vpbroadcastq(XMMRegister dst, XMMRegister src, int vector_len);
2428   void vpbroadcastq(XMMRegister dst, Address src, int vector_len);
2429 
2430   void evbroadcasti64x2(XMMRegister dst, XMMRegister src, int vector_len);
2431   void evbroadcasti64x2(XMMRegister dst, Address src, int vector_len);
2432 
2433   // scalar single/double precision replicate
2434   void vbroadcastss(XMMRegister dst, XMMRegister src, int vector_len);
2435   void vbroadcastss(XMMRegister dst, Address src, int vector_len);
2436   void vbroadcastsd(XMMRegister dst, XMMRegister src, int vector_len);
2437   void vbroadcastsd(XMMRegister dst, Address src, int vector_len);
2438 
2439   // gpr sourced byte/word/dword/qword replicate
2440   void evpbroadcastb(XMMRegister dst, Register src, int vector_len);
2441   void evpbroadcastw(XMMRegister dst, Register src, int vector_len);
2442   void evpbroadcastd(XMMRegister dst, Register src, int vector_len);
2443   void evpbroadcastq(XMMRegister dst, Register src, int vector_len);
2444 
2445   // Gather AVX2 and AVX3
2446   void vpgatherdd(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2447   void vpgatherdq(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2448   void vgatherdpd(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2449   void vgatherdps(XMMRegister dst, Address src, XMMRegister mask, int vector_len);
2450   void evpgatherdd(XMMRegister dst, KRegister mask, Address src, int vector_len);
2451   void evpgatherdq(XMMRegister dst, KRegister mask, Address src, int vector_len);
2452   void evgatherdpd(XMMRegister dst, KRegister mask, Address src, int vector_len);
2453   void evgatherdps(XMMRegister dst, KRegister mask, Address src, int vector_len);
2454 
2455   //Scatter AVX3 only
2456   void evpscatterdd(Address dst, KRegister mask, XMMRegister src, int vector_len);
2457   void evpscatterdq(Address dst, KRegister mask, XMMRegister src, int vector_len);
2458   void evscatterdps(Address dst, KRegister mask, XMMRegister src, int vector_len);
2459   void evscatterdpd(Address dst, KRegister mask, XMMRegister src, int vector_len);
2460 
2461   // Carry-Less Multiplication Quadword
2462   void pclmulqdq(XMMRegister dst, XMMRegister src, int mask);
2463   void vpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask);
2464   void evpclmulqdq(XMMRegister dst, XMMRegister nds, XMMRegister src, int mask, int vector_len);
2465   // AVX instruction which is used to clear upper 128 bits of YMM registers and
2466   // to avoid transaction penalty between AVX and SSE states. There is no
2467   // penalty if legacy SSE instructions are encoded using VEX prefix because
2468   // they always clear upper 128 bits. It should be used before calling
2469   // runtime code and native libraries.
2470   void vzeroupper();
2471 
2472   // Vector double compares
2473   void vcmppd(XMMRegister dst, XMMRegister nds, XMMRegister src, int cop, int vector_len);
2474   void evcmppd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2475                ComparisonPredicateFP comparison, int vector_len);
2476 
2477   // Vector float compares
2478   void vcmpps(XMMRegister dst, XMMRegister nds, XMMRegister src, int comparison, int vector_len);
2479   void evcmpps(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2480                ComparisonPredicateFP comparison, int vector_len);
2481 
2482   // Vector integer compares
2483   void vpcmpgtd(XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
2484   void evpcmpd(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2485                int comparison, int vector_len);
2486   void evpcmpd(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2487                int comparison, int vector_len);
2488 
2489   // Vector long compares
2490   void evpcmpq(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2491                int comparison, int vector_len);
2492   void evpcmpq(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2493                int comparison, int vector_len);
2494 
2495   // Vector byte compares
2496   void evpcmpb(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2497                int comparison, int vector_len);
2498   void evpcmpb(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2499                int comparison, int vector_len);
2500 
2501   // Vector short compares
2502   void evpcmpw(KRegister kdst, KRegister mask, XMMRegister nds, XMMRegister src,
2503                int comparison, int vector_len);
2504   void evpcmpw(KRegister kdst, KRegister mask, XMMRegister nds, Address src,
2505                int comparison, int vector_len);
2506 
2507   // Vector blends
2508   void blendvps(XMMRegister dst, XMMRegister src);
2509   void blendvpd(XMMRegister dst, XMMRegister src);
2510   void pblendvb(XMMRegister dst, XMMRegister src);
2511   void vblendvps(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len);
2512   void vblendvpd(XMMRegister dst, XMMRegister nds, XMMRegister src1, XMMRegister src2, int vector_len);
2513   void vpblendvb(XMMRegister dst, XMMRegister nds, XMMRegister src, XMMRegister mask, int vector_len);
2514   void vpblendd(XMMRegister dst, XMMRegister nds, XMMRegister src, int imm8, int vector_len);
2515   void evblendmpd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2516   void evblendmps(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2517   void evpblendmb(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2518   void evpblendmw(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2519   void evpblendmd(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2520   void evpblendmq(XMMRegister dst, KRegister mask, XMMRegister nds, XMMRegister src, bool merge, int vector_len);
2521  protected:
2522   // Next instructions require address alignment 16 bytes SSE mode.
2523   // They should be called only from corresponding MacroAssembler instructions.
2524   void andpd(XMMRegister dst, Address src);
2525   void andps(XMMRegister dst, Address src);
2526   void xorpd(XMMRegister dst, Address src);
2527   void xorps(XMMRegister dst, Address src);
2528 
2529 };
2530 
2531 // The Intel x86/Amd64 Assembler attributes: All fields enclosed here are to guide encoding level decisions.
2532 // Specific set functions are for specialized use, else defaults or whatever was supplied to object construction
2533 // are applied.
2534 class InstructionAttr {
2535 public:
2536   InstructionAttr(
2537     int vector_len,     // The length of vector to be applied in encoding - for both AVX and EVEX
2538     bool rex_vex_w,     // Width of data: if 32-bits or less, false, else if 64-bit or specially defined, true
2539     bool legacy_mode,   // Details if either this instruction is conditionally encoded to AVX or earlier if true else possibly EVEX
2540     bool no_reg_mask,   // when true, k0 is used when EVEX encoding is chosen, else embedded_opmask_register_specifier is used

2596   int  get_embedded_opmask_register_specifier(void) const { return _embedded_opmask_register_specifier; }
2597 
2598   // Set the vector len manually
2599   void set_vector_len(int vector_len) { _avx_vector_len = vector_len; }
2600 
2601   // Set revert rex_vex_w for avx encoding
2602   void set_rex_vex_w_reverted(void) { _rex_vex_w_reverted = true; }
2603 
2604   // Set rex_vex_w based on state
2605   void set_rex_vex_w(bool state) { _rex_vex_w = state; }
2606 
2607   // Set the instruction to be encoded in AVX mode
2608   void set_is_legacy_mode(void) { _legacy_mode = true; }
2609 
2610   // Set the current instuction to be encoded as an EVEX instuction
2611   void set_is_evex_instruction(void) { _is_evex_instruction = true; }
2612 
2613   // Internal encoding data used in compressed immediate offset programming
2614   void set_evex_encoding(int value) { _evex_encoding = value; }
2615 
2616   // When the Evex.Z field is set (true), it is used to clear all non directed XMM/YMM/ZMM components.
2617   // This method unsets it so that merge semantics are used instead.
2618   void reset_is_clear_context(void) { _is_clear_context = false; }
2619 
2620   // Map back to current asembler so that we can manage object level assocation
2621   void set_current_assembler(Assembler *current_assembler) { _current_assembler = current_assembler; }
2622 
2623   // Address modifiers used for compressed displacement calculation
2624   void set_address_attributes(int tuple_type, int input_size_in_bits) {
2625     if (VM_Version::supports_evex()) {
2626       _tuple_type = tuple_type;
2627       _input_size_in_bits = input_size_in_bits;
2628     }
2629   }
2630 
2631   // Set embedded opmask register specifier.
2632   void set_embedded_opmask_register_specifier(KRegister mask) {
2633     _embedded_opmask_register_specifier = (*mask).encoding() & 0x7;
2634   }
2635 
2636 };
2637

< prev index next >