fixed minor bugs

[gitoleg]

forgotten parentheses in shift instructions

compare with bap.1.2.0:

  match st with
  | LSHIFT -> Bil.(Cast (LOW, !!bool_t, Var origDEST lsr (size - Var origCOUNT)))
  | RSHIFT | ARSHIFT ->
         Bil.(Cast (HIGH, !!bool_t, Var origDEST lsl (size - Var origCOUNT)))
  | _ -> failwith "impossible"

[ivg]

there are at least two more errors of the same kind, on line 860 and 862 (I'm pretty consistent, when I'm introducing bugs :D) Can you fix them also?

It's kind of disappointing that bap-veri was unable to find them. Did we just miss it from the output, or it didn't occur in a binary, or veri misses it due to a bug?

[ivg]

Since this PR and therefore, I would like that every change that affects instruction semantics, i.e., something that deals with BIL transformations or lifters, should be accompanied with the run of veri, that proves that we don't have any regressions. I think that we need to create a repository that will contain logs from bap-veri runs, so that the proof that we don't have a regression is encoded as a commit to this repo.

[gitoleg]

finally, your wish came true!
What our bil-tests said before:

XFAIL: veri-x86_64 for SHL
XFAIL: veri-x86_64 for SHLCL
XFAIL: veri-x86_64 for SHR
XFAIL: veri-x86_64 for SHRCL

And now those instructions have PASS status!
Anyway, will update testsuite to delete fail expectation