[v8,00/50] x86emul: remaining AVX512 support

Message

Jan Beulich March 15, 2019, 10:30 a.m. UTC

This adds support for all AVX512* insns in SDM rev 069 as well as
for those from ISA extensions rev 035. Besides a few added patches
there are only minor changes from v7, perhaps apart from the
AVX512PF change to match the recently changed insn page. See
individual patches for details.

This goes on top of "x86emul: avoid speculative out of bounds
accesses", or else there's a conflict in at least the "gather" patch
here.

01: no need to set fault_suppression to false for VMOVNT*
02: support AVX512{F,BW,DQ} extract insns
03: support AVX512{F,BW,DQ} insert insns
04: basic AVX512F testing
05: support AVX512{F,BW,DQ} integer broadcast insns
06: basic AVX512VL testing
07: support AVX512{F,BW} zero- and sign-extending moves
08: support AVX512{F,BW} down conversion moves
09: support AVX512{F,BW} integer unpack insns
10: support AVX512{F,BW,_VBMI} full permute insns
11: support AVX512{F,BW} integer shuffle insns
12: support AVX512{BW,DQ} mask move insns
13: basic AVX512BW testing
14: basic AVX512DQ testing
15: support AVX512F move high/low insns
16: support AVX512F move duplicate insns
17: support AVX512{F,BW,VBMI} permute insns
18: support AVX512BW pack insns
19: support AVX512F floating-point conversion insns
20: support AVX512F legacy-equivalent packed int/FP conversion insns
21: support AVX512F legacy-equivalent scalar int/FP conversion insns
22: support AVX512DQ packed quad-int/FP conversion insns
23: support AVX512{F,DQ} uint-to-FP conversion insns
24: support AVX512{F,DQ} FP-to-uint conversion insns
25: support remaining AVX512F legacy-equivalent insns
26: support remaining AVX512BW legacy-equivalent insns
27: support AVX512{F,ER} reciprocal insns
28: support AVX512F floating point manipulation insns
29: support AVX512DQ floating point manipulation insns
30: support AVX512{F,_VBMI2} compress/expand insns
31: support remaining misc AVX512{F,BW} insns
32: support AVX512F gather insns
33: add high register S/G test cases
34: support AVX512F scatter insns
35: support AVX512PF insns
36: support AVX512CD insns
37: complete support of AVX512_VBMI insns
38: support of AVX512* population count insns
39: support of AVX512_IFMA insns
40: support remaining AVX512_VBMI2 insns
41: support AVX512_4FMAPS insns
42: support AVX512_4VNNIW insns
43: support AVX512_VNNI insns
44: support VPCLMULQDQ insns
45: support VAES insns
46: support GFNI insns
47: restore ordering within main switch statement
48: add an AES/VAES test case to the harness
49: add a SHA test case to the harness
50: add a PCLMUL/VPCLMUL test case to the harness

Jan