Bug 408758 - memcheck internal error: Signal 10 (SIGBUS)
Summary: memcheck internal error: Signal 10 (SIGBUS)
Status: REPORTED
Alias: None
Product: valgrind
Classification: Developer tools
Component: memcheck (show other bugs)
Version: 3.14.0
Platform: Other macOS
: NOR normal
Target Milestone: ---
Assignee: Rhys Kidd
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-06-15 21:21 UTC by John Ralls
Modified: 2022-05-25 13:53 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments
Session transcript of crashed memcheck session (29.71 KB, text/plain)
2019-06-15 21:21 UTC, John Ralls
Details

Note You need to log in before you can comment on or make changes to this bug.
Description John Ralls 2019-06-15 21:21:42 UTC
Created attachment 120903 [details]
Session transcript of crashed memcheck session

MacOS 10.11.6 with the 2019 security updates, Apple LLVM version 7.3.0 (clang-703.0.31), running in a VMWare Fusion 8.5.10 VM provisioned with 4 cores and 8GB of memory on a late 2013 MacPro running MacOS 10.14.
Valgrind was built from the 3.14.0 source tarball and configured with --with-64bitonly.

The program under test was Gramps (https://github.com/gramps-project/gramps) running in  a gtk-osx (https://gitlab.gnome.org/GNOME/gtk-osx) build environment with recent HEADS of glib (https://gitlab.gnome.org/GNOME/glib) and gtk-3-24 (https://gitlab.gnome.org/GNOME/gtk). Gramps is a Python program, and Python was version 3.6.8 built in the same environment. Python, glib, gtk, and most dependencies were built with -O0 -g. 

All of the session output is in the attachment.
Comment 1 John Ralls 2019-06-15 21:24:21 UTC
The command line for the crashed session was:
valgrind --log-file=gramps-memory --suppressions=$HOME/Development/gtk-sources/valgrind-3.15.0/darwin15.supp --suppressions=/Users/john/Development/Gramps-Build/gramps-stable-git-Mavericks/src/Python-3.6.8/Misc/valgrind-python.supp --main-stacksize=67108864 $PREFIX/bin/python $PREFIX/bin/gramps

The elevated --main-stacksize was necessary to avoid an earlier crash; in that case valgrind mentioned the stack size (8MB) and suggested increasing it.