From: SourceForge.net <no...@so...> - 2005-08-22 11:47:00
|
Bugs item #1264946, was opened at 2005-08-20 18:23 Message generated for change (Comment added) made by broeker You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104664&aid=1264946&group_id=4664 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Submitted By: Jimi X (jimix) Assigned to: Hans-Bernhard Broeker (broeker) Summary: cscope SEGV when "Find this C symbol:" is -q is not used Initial Comment: Specifically running on LinuxPPC Failure occurs on 15.5 and 15.4 Reliable when running on large source trees like linux-2.6 ---------------------------------------------------------------------- >Comment By: Hans-Bernhard Broeker (broeker) Date: 2005-08-22 13:46 Message: Logged In: YES user_id=27517 > -o -name Make\* -o -name \*.make What good is that supposed to do? cscope is for C source, not for Makefiles. Please remove those from your cscope.files list and try again. Same goes for the assembly source files (*.[sS]). What *.G files would be is unclear to me at this time. And yes, we will need a backtrace. So please run this example in gdb, and show us 'backtrace' and 'info local' outputs from after the crash. ---------------------------------------------------------------------- Comment By: Neil Horman (nhorman) Date: 2005-08-22 13:15 Message: Logged In: YES user_id=827328 That works fine for me, using the cscope 15.5 rpm built in fc3. Perhaps you have a build issue, or an idosyncracy in your arch. I'd grab the ppc rpm for fc4 and try that on your system. Barring that, post the backtrace from the core of the segv, and perhaps we can glean whats gone awry. ---------------------------------------------------------------------- Comment By: Jimi X (jimix) Date: 2005-08-22 03:55 Message: Logged In: YES user_id=321037 Understood, tho it is enough to to this with any large source tree, here is a specific example: I grabbed the latest linux-2.6.13-rc6 from kernel.org Then: $ uname -a Linux yyy.xxx.xxx.xxx 2.6.12-rc2 #1 SMP Mon Apr 11 07:53:48 EDT 2005 ppc64 GNU/Linux $ cscope -V cscope: version 15.5 $ find . \( -name CVS -o -name SCCS \) -prune -o \( -name \*.[cCsShHlyG] -o -name Make\* -o -name \*.make \) -print > cscope.files $ cscope -b -k $ cscope -d -L -0 main Segmentation fault if I build the DB with -q ("cscope -b -k -q") I do not get the SEGV, I suppose I'm the last person using the slow DB for large projects :) ---------------------------------------------------------------------- Comment By: Neil Horman (nhorman) Date: 2005-08-21 19:46 Message: Logged In: YES user_id=827328 Theres a few other bugs that reference segfaults under various conditions on the bugs page. You should see if any of the patches attached to those bugs match up to your problem. It would be nice to get independent confirmation on some of those fixes so we could debate their inclusion ---------------------------------------------------------------------- Comment By: Hans-Bernhard Broeker (broeker) Date: 2005-08-21 16:29 Message: Logged In: YES user_id=27517 Searching for *which* C symbol? Using cscope exactly *how*? Does it happen only in the interactive curses-based interface, or also in single-shot (-L) and/or line-based (-l) mode? Does using -d, or -c make a difference? Do you use a cscope.files? If so, what's in it? You really have to be a lot more specific. People have been using cscope-15.5 on large source trees, obviously including the Linux kernel source, since forever, and while some problems did indeed surface, none of them were as grave and wide-spread as you claim to be seeing. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=104664&aid=1264946&group_id=4664 |