[GRASS-user] v.net.iso - segmentation fault error

[Markus Metz]

On Wed, Nov 16, 2016 at 7:14 PM, Eduardo Corbelle Rico <
eduardo.corbelle at gmx.net> wrote:
>
> O Mér, 16-11-2016 ás 15:22 +0100, Markus Metz escribiu:
> >
> > On Wed, Nov 16, 2016 at 11:44 AM, Eduardo Corbelle Rico <eduardo.corb
> > elle at gmx.net> wrote:
> > >
> > > O Xov, 27-10-2016 ás 14:56 +0200, Markus Metz escribiu:
> > > > On Thu, Oct 27, 2016 at 11:16 AM, Eduardo Corbelle Rico
> > > > <eduardo.corbelle at gmx.net> wrote:
> > > > >
> > > > > O Mér, 26-10-2016 ás 23:18 +0200, Markus Metz escribiu:
> > > > > > On Thu, Oct 13, 2016 at 9:55 AM, Eduardo Corbelle Rico
> > > > > > <eduardo.corbelle at gmx.net> wrote:
> > > > > > >
> > > > > > > Dear all,
> > > > > > >
> > > > > > > I am unable to use v.net.iso because of a "segmentation
> > fault"
> > > > > > > error. I
> > > > > > >  have a script that used to work correctly in GRASS 6.4 (32
> > > > > > > bit)
> > > > > > > but
> > > > > > > produces this error after I switched to GRASS 7.0.4 (64
> > bit).
> > > > > > >
> > > > > > > If I try to follow both examples shown in https://grass.osg
> > eo.o
> > > > > > > rg/g
> > > > > > > rass
> > > > > > > 70/manuals/v.net.iso.html (using the Spearfish dataset for
> > > > > > > GRASS 7)
> > > > > > > the
> > > > > > > same error appears:
> > > > > > >
> > > > > > > > Building graph...
> > > > > > > > Registering arcs...
> > > > > > > > Segmentation fault
> > > > > >
> > > > > > The segmentation fault must happen in the vector libraries. I
> > > > > > tested
> > > > > > on Linux and can not reproduce the segmentation fault.
> > Valgrind
> > > > > > also
> > > > > > does not show anything that could cause a segmentation fault.
> > Can
> > > > > > you
> > > > > > provide a gdb backtrace?
> > > > > >
> > > > > > Markus M
> > > > >
> > > > > Thank you Markus,
> > > > >
> > > > > My system is debian Stretch, and the following lines describe
> > my
> > > > > attempt at generating a gdb backtrace. Please let me know if I
> > > > > should
> > > > > do otherwise.
> > > >
> > > > So far so good. The module crashes in a library function that
> > comes
> > > > from libavl without modifications, but I can still not see why it
> > > > could cause a segmentation fault. Can you recompile with the
> > compiler
> > > > debugging option -g? gdb will then be able to provide more
> > > > information:
> > > >
> > > > make distclean
> > > > CFLAGS="-g" ./configure <configure options>
> > > > make
> > > >
> > > > optionally make install
> > > >
> > > > Then start GRASS and use gdb for v.net.iso as before.
> > > >
> > > > Markus M
> > > >
> > >
> > > Dear Markus,
> > >
> > > Following those instructions does not seem to have changed the
> > > resulting backtrace info (copied below). Maybe I missed something
> > along
> > > the way?
> > >
> > > ---------------------
> >
> >
> > Is this something Debian specific that debugging symbols must be
> > explicitly activated? Are you using GCC or another compiler?
> >
> > Alternatively, you can also try valgrind:
> >
> > CMD="v.net.iso input=myroads_net output=myroads_net_iso
> > center_cats=1-100000 costs=1000,2000,5000 --o"
> > valgrind --tool=memcheck --log-file=v.net.iso.vg.log $CMD
> >
> > Markus M
> >
>
> You were right: it was my fault not having grass-core-dbgsym and grass-
> dev-dbgsym packages installed.

You do not need these packages when you compile from source.

> I copied below the results of the
> debugging commands:
>
> ================
> GRASS 7.0.5 (spearfish60_grass7):~ > gdb `which v.net.iso`
> GNU gdb (Debian 7.11.1-2+b1) 7.11.1
> Copyright (C) 2016 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl
> .html>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law.  Type "show
> copying"
> and "show warranty" for details.
> This GDB was configured as "x86_64-linux-gnu".
> Type "show configuration" for configuration details.
> For bug reporting instructions, please see:
> <http://www.gnu.org/software/gdb/bugs/>.
> Find the GDB manual and other documentation resources online at:
> <http://www.gnu.org/software/gdb/documentation/>.
> For help, type "help".
> Type "apropos word" to search for commands related to "word"...
> Reading symbols from /usr/lib/grass70/bin/v.net.iso...Reading symbols
> from /usr/lib/debug/.build-
> id/5a/060fc6c3da41153d82601e3fa37ba6d88e9c6c.debug...done.
> done.
>
>
> (gdb) run input=myroads_net output=myroads_net_iso center_cats=1-100000
> costs=1000,2000,5000

This seems to be the GRASS version installed with your package manager
because make install installs by default to /usr/local/:
> Starting program: /usr/lib/grass70/bin/v.net.iso input=myroads_net
> output=myroads_net_iso center_cats=1-100000 costs=1000,2000,5000
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-
> gnu/libthread_db.so.1".
> Building graph...
> Registering arcs...
>
> Program received signal SIGSEGV, Segmentation fault.
> tavl_probe (tree=tree at entry=0x55555599b6c0, item=item at entry=0x55555599f
> 7b0)
>     at tavl.c:146
> 146     tavl.c: Non hai tal ficheiro ou directorio.
> (gdb) bt full
> #0  tavl_probe (tree=tree at entry=0x55555599b6c0, item=item at entry=0x55555
> 599f7b0)
>     at tavl.c:146
>         y = 0x0
>         z = 0x55555599b6c0
>         p = 0x0

At this stage (tavl.c:146), p should be equal to z, and p has been accessed
previously, i.e. the segmentation fault should occur earlier. Therefore I
suspect changes in tavl.c introduced when creating the package. You would
need to remove all GRASS 7 packages from your system, then compile from
source again.

Markus M
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-user/attachments/20161117/e568d933/attachment-0001.html>