After introducing the rewriter.sort_disjunctions option (#6774), I
noticed a segfault in a Z3 run that was working fine for me before the
PR.
I traced the difference to a slight discrepancy between the first patch
I submitted and the one we ended up merging: my first version would skip
sorting the disjuncts in mk_nflat_core, but still return BR_DONE, while
the patch in master returns BR_FAILED instead.
This patch fixes that problem, and it makes slightly more sense to me to
return a BR_DONE since, if `s` is true, some disjunct (e.g. a `false`
or a repeat) might have been simplified away. However I don't fully
understand this code.
... and I can't say I understand why the segfault happens. Perhaps that
is a separate issue?
This is the file to reproduce:
https://gist.github.com/mtzguido/b7360c74d3d2e42d89f1bd9149ad26f6
Here's a stack trace of the failure, mk_nflat_or_core is not involved.
```
(gdb) where
#0 0x0000555555b98497 in smt::context::get_lit_assignment(unsigned int) const ()
#1 0x0000555555b984cb in smt::context::get_assignment(sat::literal) const ()
#2 0x0000555555b98504 in smt::context::get_assignment(unsigned int) const ()
#3 0x0000555555ca83b8 in smt::context::get_assignment_core(expr*) const ()
#4 0x0000555555c9af5a in smt::context::get_assignment(expr*) const ()
#5 0x0000555555d7bd1d in (anonymous namespace)::has_child_assigned_to(smt::context&, app*, lbool, expr*&, unsigned int) ()
#6 0x0000555555d7c413 in (anonymous namespace)::rel_case_split_queue::next_case_split_core(ptr_vector<expr>&, unsigned int&, unsigned int&, lbool&) ()
#7 0x0000555555d7c589 in (anonymous namespace)::rel_case_split_queue::next_case_split(unsigned int&, lbool&) ()
#8 0x0000555555c9c1b7 in smt::context::decide() ()
#9 0x0000555555ca39fd in smt::context::bounded_search() ()
#10 0x0000555555ca30c2 in smt::context::search() ()
#11 0x0000555555ca273d in smt::context::check(unsigned int, expr* const*, bool) ()
#12 0x0000555555cb166a in smt::kernel::check(unsigned int, expr* const*) ()
#13 0x0000555555cb9695 in (anonymous namespace)::smt_solver::check_sat_core2(unsigned int, expr* const*) ()
#14 0x00005555560dc0c6 in solver_na2as::check_sat_core(unsigned int, expr* const*) ()
#15 0x00005555560d73f3 in combined_solver::check_sat_core(unsigned int, expr* const*) ()
#16 0x00005555560d34e3 in solver::check_sat(unsigned int, expr* const*) ()
#17 0x0000555556097b26 in cmd_context::check_sat(unsigned int, expr* const*) ()
#18 0x0000555556082ff0 in smt2::parser::parse_check_sat() ()
#19 0x0000555556084dc0 in smt2::parser::parse_cmd() ()
#20 0x00005555560861b6 in smt2::parser::operator()() ()
#21 0x00005555560757e6 in parse_smt2_commands(cmd_context&, std::basic_istream<char, std::char_traits<char> >&, bool, params_ref const&, char const*) ()
#22 0x00005555555e8f68 in read_smtlib2_commands(char const*) ()
#23 0x00005555555ee6f6 in main ()
(gdb)
```
* patterns: add option for pattern decomposition (pi.decompose_patterns)
True by default, retaining current behavior.
* rewriter: add option for sorting of disjunctions (rewriter.sort_disjunctions)
True by default, retaining current behavior.
* use heap to track infeasible columns
* fix the formatting
Signed-off-by: Lev Nachmanson <levnach@hotmail.com>
---------
Signed-off-by: Lev Nachmanson <levnach@hotmail.com>
check-assumptions with compound formulas create fresh proxy variables both during compilation to internal format and for the assumptions. These fresh variables may occur in lemmas that are created during search. The lemmas are garbage for future check-sats, but the solver needs to be allowed to invoke GC. Adding a GC call before a check-sat with assumptions allows removing some lemmas every time a new assumptions are used. Eager GC when using assumptions is used elsewhere, for example in cube&conquer scenarios where lemmas learned from one set of assumptions are less likely to be useful for other assumptions.
With the GC invocation memory grows at a lesser pace. However, it is not entirely free of memory increases. To avoid memory bloat, have the solver pre-compile the assumptions by defining them as propositional variables, add assertions that the propositional variables are equivalent to the compound formulas and use the propositional variables as assumptions. The same propositional variables come with no extra overhead when invoking check-assumptions. The lemmas are then over the same fixed vocabulary. It is generally a good idea to recycle useful lemmas during the enumeration pass.
destructive equality resolution uses an occurs check function that is only safe for quantifier-free formulas. In the special case where a bound variable is Boolean and occurs on a side of an equality the other side cannot have a quantifier.
