3 - rework eflags optimization (will be a consequence of TCG port)
4 - SVM: rework the implementation: simplify code, move most intercept
5 tests as dynamic, correct segment access, verify exception safety,
6 remove most of the added CPU state.
7 - arpl eflags computation is invalid
8 - x86_64: fxsave/fxrestore intel/amd differences
9 - x86_64: lcall/ljmp intel/amd differences ?
10 - x86_64: cmpxchgl intel/amd differences ?
11 - x86_64: cmovl bug intel/amd differences ?
12 - x86: monitor invalid
13 - better code fetch (different exception handling + CS.limit support)
14 - user/kernel PUSHL/POPL in helper.c
15 - add missing cpuid tests
16 - return UD exception if LOCK prefix incorrectly used
17 - test ldt limit < 7 ?
18 - fix some 16 bit sp push/pop overflow (pusha/popa, lcall lret)
19 - full support of segment limit/rights
20 - full x87 exception support
21 - improve x87 bit exactness (use bochs code ?)
23 Optimizations/Features:
26 - evaluate x87 stack pointer statically
27 - find a way to avoid translating several time the same TB if CR0.TS
29 - move kqemu support outside target-i386.