#32293: import _scproxy and spatialite database backend segfaults
-----------------------------------------+------------------------
Reporter: Jon Janzen | Owner: nobody
Type: Uncategorized | Status: new
Component: Uncategorized | Version: 3.1
Severity: Normal | Keywords:
Triage Stage: Unreviewed | Has patch: 0
Needs documentation: 0 | Needs tests: 0
Patch needs improvement: 0 | Easy pickings: 0
UI/UX: 0 |
-----------------------------------------+------------------------
This bug started happening to me when I upgraded to macOS 11 Big Sur, but
this may be coincidence as I upgraded all of my homebrew packages at the
same time.
This bug was traced down from having the `graphene_django` module in
`INSTALLED_APPS`. This module imports a lot of things but one of which is
`urllib.request`, which in turn actually imports `_scproxy`.
=== Steps to reproduce:
1. `django-admin startproject crashdemo && cd crashdemo`
2. Edit `crashdemo/settings.py` to use the `spatialite` backend instead of
`sqlite3`
3. Add `import _scproxy` to `settings.py` (actual file doesn't matter, as
long as it's imported during django startup)
4. Run `python3 manage.py runserver --noreload` (`--noreload` not strictly
necessary, but it causes the segfault to occur in the main thread)
=== Expected outcome:
Django starts up normally.
=== Actual outcome:
{{{
Performing system checks...
System check identified no issues (0 silenced).
Segmentation fault: 11
}}}
=== My Environment:
macOS Big Sur 11.1 (x86-64)
sqlite3 is from homebrew
libspatialite/gdal are from homebrew
[email protected] from homebrew (issue still exists when building from latest
source)
django is 3.1.4 from pip (issue still exists when installing from latest
source)
I've set some debugging points inside the C source for the `_scproxy`
module, and it seems like the functions provided by the module aren't even
called leading up to the segfault. It seems like the shared object just
needs to be loaded? I'm not sure how to debug this issue further.
I've attached a macOS crash report and a Python traceback produced by
cpython-lldb's py-bt command that indicates that this segfault is
triggered when attempting to load the libspatialite extension into sqlite.
I've confirmed this by wrapping that call with `print` statements and
running without a debugger attached. The macOS crash report seems to point
to something within `libsqlite` as being the culprit, but I don't really
understand what I'm talking about here.
Since I imagine this is a highly environment-specific bug report (perhaps
just affecting my Mac), I'm happy to provide any additional information,
run any tests, or try potential solutions to help solve this issue.
--
Ticket URL: <https://code.djangoproject.com/ticket/32293>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
--
You received this message because you are subscribed to the Google Groups
"Django updates" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-updates/053.ff01437e16824eb065ebeadb23c9f37e%40djangoproject.com.
