On Fri, Oct 16, 2020 at 01:44:15PM +0200, Mattia Rizzolo wrote: > > I'll work on testing a backport of the bullseye > > version on buster to test this. > > If you'd like, I can provide you with a .deb of the backport, for easier > testing.
The backport was trivial to build, and I've been running it for >24 hours with no issues so far. So things look promising for this being fixed upstream by now. > > Given the amount of code churn, I'm > > skeptical that we'll be able to isolate an easily backportable fix, but > > we'll see. > > Indeed, and I'm not really _that_ interested in doing that work myself, > so if you see that 1.87 doesn't segfaults at most I'd probably upload > 1.87 to buster-backports rather than figuring out the proper fix for the > "main" buster release. Unless somebody can isolate the fix :) > Probably going through the upstream git repository might help with that, > fwiw. It looks like there are enough enhancements (performance improvements, new features, etc) in newer versions that a backport makes some sense. AIUI, though I don't see where this is documented, a backports upload is not intended to take the place of a proper stable update when it comes to fixing bugs in stable versions. However, I'm also not particularly interested in investing a bunch of time to isolate and backport the relevant upstream change. Another option might be to try and convince the stable release managers to allow a new upstream version. Given the risk of data loss presented by this bug, it might not be too difficult. noah