Skip to content

Segfault on imported DMPlex meshes with essential BCs #285

Description

@jcgraciosa

`tests/test_1012_stokesImportedDMPlex.py::test_stokes_essential_bc_imported_dmplex` segfaults (SIGSEGV, exit 139) as of 2026-06-24.

Confirmed universal across 3 independent builds (release `amr-dev` petsc4py 3.25.2, daily-dev `amr-dev` petsc4py 3.25.0, plain `dev` petsc4py 3.24.4) — not AMR-specific, not a build artifact.

This is the regression test for PR #86 (`gthyagi/codex/fix_stokes_essential_bc_imported_dmplex`), merged 2026-03-20 — after `v3.0.1`. Repro: gmsh mesh → write to file → `PETSc.DMPlex().createFromFile()` → `uw.discretisation.Mesh` wrapping with `markVertices=True, useMultipleTags=True, useRegions=True` → Stokes LU solve. Crashes as a hard segfault, no Python traceback.

Currently skipped in `release/v3.1.0` to unblock v3.1.0 validation. Needs a gdb/lldb backtrace to actually triage.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions