[Cscope-devel] [ cscope-Bugs-1756561 ] Segmentation Fault

[SourceForge.net <no...@so...>]

Bugs item #1756561, was opened at 2007-07-19 04:16
Message generated for change (Settings changed) made by broeker
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104664&aid=1756561&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: C lexer/parser
Group: None
>Status: Closed
>Resolution: Invalid
Priority: 5
Private: No
Submitted By: Peyman Taher (peyman_taher)
Assigned to: Hans-Bernhard Broeker (broeker)
Summary: Segmentation Fault

Initial Comment:
Here is the sequence of events that cause the segmentation fault:
I created the cscope.files file for Linux kernel version 2.6.21.5 directory tree and removed the files for all arch/ except for i386 and also all Documentation/ but I didn't remove the script/ directory. I traced the execution of cscope -b -q using strace in Debian Etch and here is the last few lines before segmentation fault:

########################################

open("scripts/Makefile.headersinst", O_RDONLY) = 7 
fcntl64(7, F_SETFD, FD_CLOEXEC)         = 0
fstat64(7, {st_mode=S_IFREG|0644, st_size=7007, ...}) = 0 
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7f12000
read(7, "# =============================="..., 8192) = 7007
read(7, "", 4096)                       = 0
write(6, "\"S.R>` \"S.Q|\nstrip \"S.R>  \"S.Q|\n"..., 4096) = 4096
write(3, "his\n \nadds\n\n\n49 #\250\n\t`$\n(\nobj\n\350\n\264"..., 4096) = 4096
access(" ;              \\
echo \"\\#ifndef $$STUBDEF\" ;                                          \\
echo \"\\#define $$STUBDEF\" ;                                          \\
echo \"\\#if $(ARCHDEF)\" ;                                             \\
if [ -r $(subst /$(_dst)/,/include/$(archasm)/,$@) ]; then              \\
echo \"\\#include <$(archasm)/$$FNAME", R_OK) = -1 ENOENT (No such file or directory)
access("/usr/include/ ;         \\
echo \"\\#ifndef $$STUBDEF\" ;                                          \\
echo \"\\#define $$STUBDEF\" ;                                          \\
echo \"\\#if $(ARCHDEF)\" ;                                             \\
if [ -r $(subst /$(_dst)/,/include/$(archasm)/,$@) ]; then              \\
echo \"\\#include <$(archasm)/$$FNAME", R_OK) = -1 ENOENT (No such file or directory)
access(" ;              \\
echo \"\\#ifndef $$STUBDEF\" ;                                          \\
echo \"\\#define $$STUBDEF\" ;                                          \\
echo \"\\#if $(ARCHDEF)\" ;                                             \\
if [ -r $(subst /$(_dst)/,/include/$(archasm)/,$@) ]; then              \\
echo \"\\#include <$(archasm)/$$FNAME>\" ;                      \\
else                                                                    \\
echo \"\\#error $(archasm)/$$FNAME does not exist in\"          \\http://sourceforge.net/tracker/?func=add&group_id=4664&atid=104664
\"the $(ARCH) architecture\" ;                  \\
fi ;                                                                    \\
echo \"\\#elif $(ALTARCHDEF)\" ;                                                \\
if [ -r $(subst /$(_dst)/,/include/$(altarchasm)/,$@) ]; then           \\
echo \"\\#include <$(altarchasm)/$$FNAME", R_OK) = -1 ENOENT (No such file or directory)
--- SIGSEGV (Segmentation fault) @ 0 (0) --- 
+++ killed by SIGSEGV +++ 
Process 5364 detached

############################################

So apparently segmentation fault occurs when cscope tries to parse the script/Makefile.headersinst file at line 135 of the file.

I have attachede the cscope.files file.

----------------------------------------------------------------------

Comment By: Peyman Taher (peyman_taher)
Date: 2007-07-20 09:00

Message:
Logged In: YES 
user_id=1848490
Originator: YES

I don't think when one buys a calculator expects it crashes whenever you
divide something by zero, right? This is called exception handling and a
program like cscope is supposed to handle it what so ever. So I guess I'll
follow what nhorman said or will check the source of the problem by my
own.
Thanks anyway.

----------------------------------------------------------------------

Comment By: Hans-Bernhard Broeker (broeker)
Date: 2007-07-19 22:01

Message:
Logged In: YES 
user_id=27517
Originator: NO

> I created the cscope.files file for Linux kernel version 2.6.21.5
directory
> tree and removed the files for all[...]

The root of your problem lies right here.  You created a cscope.files
without considering what should go into it.  What on earth is a _Makefile_
doing in your input list for cscope?  Is it really so hard to guess that
any 'find' command that generates a cscope.files list should include a
qualifier like '-name "*.[ch]"' to reduce it to actual C source files?

Maybe you'll be better off just using the script 'cscope-indexer' that
comes with the program as part of xcscope.el.

----------------------------------------------------------------------

Comment By: Neil Horman (nhorman)
Date: 2007-07-19 12:46

Message:
Logged In: YES 
user_id=827328
Originator: NO

I'm not quite sure what you've done, but the strace above looks completely
garbled with some other garbage.  What would be helpful in determing what
caused your core dump would be a backtrace.  why don't you build your
cscope binary with -g (if its not already), disable yor coredump rlim,
capture a core file when the SIGSEGV happens, load it into gdb, and post
the resultant backtrace.  That will tell us much more accurately whats gone
wrong here.  Thanks!

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104664&aid=1756561&group_id=4664