Correct reference counting and handling of NULL pointers in new OCaml bindings. #2
Conversation
The Z3 context and its reference counters are stored in a structure which is allocated by the C layer outside the OCaml heap, whenever a Z3 context is created. The structure and its Z3 context are disposed, once the last reference counter reaches zero. Reference counters are decremented by C-level finalizers. The OCaml representations for a Z3 context wrap only a pointer to the corresponding structure.
This patch refactors the update_api script and the z3.ml implementation to properly translate between OCaml options and NULL pointers. Some unifications and simplifications (avoidance of unnecessary value allocation) in the script that creates the native bindings.
|
Looks good to me, thanks for the fixes! Was it really necessary to introduce another pointer indirection for the refcount custom blocks? I remember putting extra effort into custom blocks that get copied, so from my point of view this was (or should have been) working. |
|
Hi Christoph, I suspect that the critical part was the context_of part of the C-layer (the macro Here, the custom block that holds the Hence, the finalizer The patch allocates exactly one |
(this leads to a not-implemented-yet in final_check_eh() due to missing code surrounding free variable production)
fix build of obj_ref_hashtable
…junctions (Z3Prover#6779) After introducing the rewriter.sort_disjunctions option (Z3Prover#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*) () Z3Prover#23 0x00005555555ee6f6 in main () (gdb) ```
This PR contains two commits.
The first corrects a bug in the reference counting mechanism used in the C layer of the new OCaml bindings. Instead of storing the reference counters directly in a custom block that holds the context (and that might become copied), the patch allocates memory to store the reference counters and only stores a pointer to this memory block in the context's custom blocks.
The second commit refactors parts of the update_api script (only the OCaml parts are affected) to translate properly between OCaml options and NULL pointers. Moreover, it avoids some duplicate allocations for OCaml values and adds some missing allocations.