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

Fri Nov 18 07:48:12 PST 2016

O Xov, 17-11-2016 ás 08:52 +0100, Markus Metz escribiu:
> 
> On Wed, Nov 16, 2016 at 7:14 PM, Eduardo Corbelle Rico <eduardo.corbe
> lle 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=0x55555
> 599f
> > 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=0x5
> 5555
> > 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

Ok. It seems that I was still running the version installed with the
package manager. When using the version compiled from source, the
segmentation error does not appear at all.

Thank you very much.

Eduardo