I have been getting some segfaults recently, too. I seem to come from doing a lot of changes in parameterized cells. Maybe something like deleting a parameterized cell and replacing it after changing parameters loses some important information that doesn't get recorded by the "undo" function, so that when you try to "undo" past that change, the database gets corrupted. But that's just a guess after having it happen several times in a row to me, and I didn't get the chance to debug it. Of course when I ran the debugger and repeated the steps, it wouldn't crash. I have been working on many different layouts over the past two weeks, and the segfaulting happened on just one of them. All the others, I had no problems at all. Your problem may not be related to that at all, but I thought I'd mention in in case it is.

FYI, magic does save work periodically, and

magic -r

should read back the last saved file and put you back within striking distance of where you were when the crash occurred. However, the last time I suffered a crash, the backup file was nowhere to be seen and I lost a lot of work (another unreproducible result). Always good to do "writeall" reasonably often.