GNU bug report logs - #5847
ETAGS: Segmentation fault, because of incorrect scope presumption

Previous Next

Full log

Message #31 received at 5847-done <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: agrambot <at> gmail.com
Cc: gosselmeyer <at> googlemail.com, 5847-done <at> debbugs.gnu.org
Subject: Re: bug#5847: [Hubert Gosselmeyer] Re: bug#5847: ETAGS: Segmentation
 fault, because of incorrect scope presumption
Date: Wed, 31 Aug 2016 18:59:20 +0300

> Date: Sat, 20 Aug 2016 19:27:21 +0300
> From: Eli Zaretskii <eliz <at> gnu.org>
> Cc: 5847 <at> debbugs.gnu.org, Hubert Gosselmeyer <gosselmeyer <at> googlemail.com>
> 
> > Date: Sat, 20 Aug 2016 18:55:36 +0300
> > From: Eli Zaretskii <eliz <at> gnu.org>
> > Cc: 5847 <at> debbugs.gnu.org
> > 
> > That file blows up the runtime stack because of recursive calls to
> > put_entries when etags writes out the tree of nodes it collected.
> > Since recovery from stack overflow is inherently OS dependent and
> > unreliable, I don't see what we can do in this case.  Even if we do
> > detect the upcoming stack overflow, the only thing we can do is print
> > an error message and quit.
> 
> Actually, one way of avoiding stack overflow would be to use some BFS
> algorithm that allocates memory off the heap.  Patches welcome.

I've done that now and committed the changes on the master branch.
The resulting etags successfully tagged a dozen of large source trees
together, whereas the original version crashed quite quickly.  The
test file posted here also no longer causes a crash.

The non-recursive version is somewhat slower, so patches to speed it
up are welcome.

I'm closing this bug.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.