[CIG-SHORT] Problem running user compiled version of Relax
Sylvain Barbot
sylbar.vainbot at gmail.com
Sun Jul 5 04:28:33 PDT 2015
Dear Jeff,
Thank you for pointing out this problem. This was indeed an error
introduced recently. A pre-stress array was not allocated properly
when pre-stress is not used. Sorry about that.
I have fixed the problem now and pushed to the repository. The binary
versions on geodynamics.org and on fink are not affected. Only the
development version had this problem.
You can get a clean version with
> git pull
Let me know if you run into any other problem.
Best wishes,
Sylvain
On Tue, Jun 30, 2015 at 4:32 AM, Sylvain Barbot
<sylbar.vainbot at gmail.com> wrote:
> Hi Jeff,
>
> Since the error involves stressstruc, can you send the full output message
> and/or the input file so we can reproduce the problem?
>
> Thanks,
> S.
>
>
> On Monday, June 29, 2015, Falgout, Jeff <jfalgout at usgs.gov> wrote:
>>
>> Hi Sylvain -
>>
>> As soon as I hit send, I figured out my mistake for getting a useful
>> segfault:
>>
>> $ FCFLAGS="-check arg_temp_created -gen-interfaces -warn interfaces
>> -fp-stack-check -check bounds -g -traceback" ./waf configure
>> --proj-dir=/usr/bin --gmt-dir=/usr/bin --gmt-incdir=/usr/include
>> --gmt-libdir=/usr/lib64 --mkl-libdir=/opt/intel/mkl/lib/intel64/
>> --mkl-incdir=/opt/intel/mkl/include --check-c-compiler=icc
>> --check-fortran-compiler=ifort --check-cxx-compiler=icpc
>>
>> # number of surface loads
>> 0
>> # max sampling size (hor.,vert.): 1.60E+1 2.13E+0
>> #
>> ----------------------------------------------------------------------------
>> forrtl: severe (408): fort: (2): Subscript #1 of the array STRESSSTRUC has
>> value 1 which is greater than the upper bound of -1
>>
>> Image PC Routine Line Source
>> relax 00000000006B9C60 Unknown Unknown
>> Unknown
>> relax 0000000000418795 MAIN__ 292
>> relax.f90
>> relax 0000000000410D5E Unknown Unknown
>> Unknown
>> libc.so.6 000000314E01ED5D Unknown Unknown
>> Unknown
>> relax 0000000000410C69 Unknown Unknown
>> Unknown
>>
>>
>> Jeff
>>
>> On Mon, Jun 29, 2015 at 2:17 PM, Falgout, Jeff <jfalgout at usgs.gov> wrote:
>>>
>>> Hi Sylvain -
>>>
>>> Here is more info on the problem we ran into with the git version from 29
>>> June 2015 (fingers crossed line doesn't kill me)
>>>
>>> $ ifort --version
>>> ifort (IFORT) 15.0.0 20140723
>>> Copyright (C) 1985-2014 Intel Corporation. All rights reserved.
>>>
>>>
>>> $ FCFLAGS="-check arg_temp_created -gen-interfaces -warn interfaces
>>> -fp-stack-check -check bounds -g -traceback"
>>>
>>> $ ./waf distclean
>>> 'distclean' finished successfully 0.086s
>>>
>>> $ ./waf configure --proj-dir=/usr/bin --gmt-dir=/usr/bin
>>> --gmt-incdir=/usr/include --gmt-libdir=/usr/lib64
>>> --mkl-libdir=/opt/intel/mkl/lib/intel64/ --gmt-libdir=/usr/lib64
>>> --mkl-libdir=/opt/intel/mkl/lib/intel64/ --check-fortran-compiler=ifort
>>> --check-cxx-compiler=icpc
>>> Setting top to :
>>> /home/jfalgout/olboyd/Tools/Relax/relax-git/relax
>>> Setting out to :
>>> /home/jfalgout/olboyd/Tools/Relax/relax-git/relax/build
>>> Checking for 'gcc' (c compiler) : /usr/bin/gcc
>>> Checking for 'ifort' (fortran compiler) : /opt/intel/bin/ifort
>>> Checking for 'icpc' (c++ compiler) : /opt/intel/bin/icpc
>>> Compiling a simple fortran app : yes
>>> Checking for header proj_api.h : yes
>>> Checking for gmt.h in '/usr/include' : yes
>>> Checking for openmp flag -fopenmp : yes
>>> Checking for MKL : yes
>>> Checking preprocessor option -cpp : yes
>>> Checking length option : yes
>>> 'configure' finished successfully 35.914s
>>>
>>>
>>> $ ./waf build
>>> Waf: Entering directory
>>> `/home/jfalgout/olboyd/Tools/Relax/relax-git/relax/build'
>>> [ 3/20] fc: src/ctfft.f -> build/src/ctfft.f.1.o
>>> [ 3/20] fc: src/types.f90 -> build/src/types.f90.1.o build/types.mod
>>> [ 3/20] fc: src/okada/dc3d.f -> build/src/okada/dc3d.f.1.o
>>> [ 4/20] c: src/writevtk.c -> build/src/writevtk.c.1.o
>>> [ 5/20] c: src/writegrd4.2.c -> build/src/writegrd4.2.c.1.o
>>> [ 8/20] c: src/proj.c -> build/src/proj.c.1.o
>>> [ 9/20] fc: src/getdata.f -> build/src/getdata.f.1.o
>>> [ 9/20] fc: src/getopt_m.f90 -> build/src/getopt_m.f90.1.o
>>> build/getopt_m.mod
>>> [ 9/20] fc: src/mkl_dfti.f90 -> build/src/mkl_dfti.f90.1.o
>>> build/mkl_dft_type.mod build/mkl_dfti.mod
>>> [10/20] c: src/papi_prof.c -> build/src/papi_prof.c.1.o
>>> [11/20] fc: src/fourier.f90 -> build/src/fourier.f90.1.o
>>> build/fourier.mod
>>> [12/20] fc: src/green.f90 -> build/src/green.f90.1.o build/green.mod
>>> [13/20] fc: src/elastic3d.f90 -> build/src/elastic3d.f90.1.o
>>> build/elastic3d.mod
>>> [14/20] fc: src/okada/green_space.f90 ->
>>> build/src/okada/green_space.f90.1.o build/green_space.mod
>>> [15/20] fc: src/viscoelastic3d.f90 -> build/src/viscoelastic3d.f90.1.o
>>> build/viscoelastic3d.mod
>>> [16/20] fc: src/friction3d.f90 -> build/src/friction3d.f90.1.o
>>> build/friction3d.mod
>>> [17/20] fc: src/export.f90 -> build/src/export.f90.1.o build/export.mod
>>> [18/20] fc: src/input.f90 -> build/src/input.f90.1.o build/input.mod
>>> /home/jfalgout/olboyd/Tools/Relax/relax-git/relax/src/input.f90(323):
>>> remark #8293: Recommended relationship between field width 'W', the number
>>> of fractional digits 'D' and the number of exponential digits 'E' in this
>>> edit descriptor is 'W>=D+E+5'.
>>> PRINT '("# output every ",I3.3," steps, dt scaled by ",ES7.2E1)',
>>> in%skip,in%tscale
>>>
>>> -------------------------------------------------------------------------------^
>>> [19/20] fc: src/relax.f90 -> build/src/relax.f90.1.o
>>> /home/jfalgout/olboyd/Tools/Relax/relax-git/relax/src/relax.f90(980):
>>> remark #8293: Recommended relationship between field width 'W', the number
>>> of fractional digits 'D' and the number of exponential digits 'E' in this
>>> edit descriptor is 'W>=D+E+5'.
>>> 1101 FORMAT (I3.3,"*",ES9.2E2,3ES9.2E2,ES9.2E2,"/",ES7.2E1,2ES9.2E2)
>>>
>>> -------------------------------------------------------------------------------^
>>> /home/jfalgout/olboyd/Tools/Relax/relax-git/relax/src/relax.f90(979):
>>> remark #8293: Recommended relationship between field width 'W', the number
>>> of fractional digits 'D' and the number of exponential digits 'E' in this
>>> edit descriptor is 'W>=D+E+5'.
>>> 1100 FORMAT (I3.3," ",ES9.2E2,3ES9.2E2,ES9.2E2,"/",ES7.2E1,2ES9.2E2)
>>>
>>> -------------------------------------------------------------------------------^
>>> [20/20] fcprogram: build/src/relax.f90.1.o build/src/ctfft.f.1.o
>>> build/src/types.f90.1.o build/src/fourier.f90.1.o build/src/green.f90.1.o
>>> build/src/okada/green_space.f90.1.o build/src/okada/dc3d.f.1.o
>>> build/src/elastic3d.f90.1.o build/src/friction3d.f90.1.o
>>> build/src/viscoelastic3d.f90.1.o build/src/writevtk.c.1.o
>>> build/src/writegrd4.2.c.1.o build/src/proj.c.1.o build/src/export.f90.1.o
>>> build/src/getdata.f.1.o build/src/getopt_m.f90.1.o build/src/input.f90.1.o
>>> build/src/mkl_dfti.f90.1.o build/src/papi_prof.c.1.o -> build/relax
>>> Waf: Leaving directory
>>> `/home/jfalgout/olboyd/Tools/Relax/relax-git/relax/build'
>>>
>>> 'build' finished successfully 17.861s
>>>
>>>
>>> (The arrow is pointing to the 7 in ES7.2E1 in the 3 warnings)
>>>
>>>
>>> When I run the problem, we get a segfault as follows:
>>>
>>> # number of inter-seismic strike-slip segments
>>> 0
>>> # number of inter-seismic tensile segments
>>> 0
>>> # number of events
>>> 1
>>> # number of coseismic strike-slip segments
>>> 2
>>> #
>>> ----------------------------------------------------------------------------
>>> # n slip xs ys zs length width strike dip
>>> rake
>>> #
>>> ----------------------------------------------------------------------------
>>> 001 5.00E+0 1.89E+1 -1.05E+1 5.10E+0 4.00E+1 1.87E+1 173.2 31.9
>>> 90.0
>>> 002 6.63E+0 1.91E+1 -8.94E+0 0.00E+0 4.00E+1 5.33E+0 173.2 73.0
>>> 90.0
>>> #
>>> ----------------------------------------------------------------------------
>>> # number of coseismic tensile segments
>>> 0
>>> # number of coseismic dilatation point sources
>>> 0
>>> # number of surface loads
>>> 0
>>> # max sampling size (hor.,vert.): 1.60E+1 2.13E+0
>>> #
>>> ----------------------------------------------------------------------------
>>> forrtl: severe (174): SIGSEGV, segmentation fault occurred
>>> Image PC Routine Line
>>> Source
>>> relax 00000000005AFBC1 Unknown Unknown
>>> Unknown
>>> relax 00000000005AE317 Unknown Unknown
>>> Unknown
>>> relax 000000000055EDE4 Unknown Unknown
>>> Unknown
>>> relax 000000000055EBF6 Unknown Unknown
>>> Unknown
>>> relax 000000000050D7D4 Unknown Unknown
>>> Unknown
>>> relax 000000000051197D Unknown Unknown
>>> Unknown
>>> libpthread.so.0 000000314EC0F710 Unknown Unknown
>>> Unknown
>>> relax 0000000000454B07 Unknown Unknown
>>> Unknown
>>> relax 0000000000414431 Unknown Unknown
>>> Unknown
>>> relax 0000000000410CBE Unknown Unknown
>>> Unknown
>>> libc.so.6 000000314E01ED5D Unknown Unknown
>>> Unknown
>>> relax 0000000000410BC9 Unknown Unknown
>>> Unknown
>>>
>>>
>>>
>>> At one point I had the Segfault spitting out a little more info, but it's
>>> not doing that now ... Let me know if there is anything you want me to do
>>> differently or any debug info I can provide.
>>>
>>> Thanks for you help,
>>>
>>> Jeff
>>>
>>>
>>> On Wed, Jun 24, 2015 at 4:34 PM, Oliver Boyd <olboyd at usgs.gov> wrote:
>>> >
>>> > We didn’t fix the problem. We just didn't use the files that came out
>>> > of the git repository, but rather used those in the source tar ball from the
>>> > CIG website.
>>> >
>>> > From: Sylvain Barbot <sylbar.vainbot at gmail.com>
>>> > Reply-To: <cig-short at geodynamics.org>
>>> > Date: Wednesday, June 24, 2015 3:44 PM
>>> > To: "cig-short at geodynamics.org" <cig-short at geodynamics.org>
>>> > Cc: Walter Landry <wlandry at caltech.edu>
>>> > Subject: Re: [CIG-SHORT] Problem running user compiled version of Relax
>>> >
>>> > Hi Oliver,
>>> >
>>> > Can you and Jeff let us know how you fixed the problem? I'd like to
>>> > know if and how I can improve the code.
>>> >
>>> > Thanks,
>>> > Sylvain
>>> >
>>> > On Wednesday, June 24, 2015, Oliver Boyd <olboyd at usgs.gov> wrote:
>>> >>
>>> >> Jeff Falgout at the USGS helped me to get it working. Looks like the
>>> >> libraries are fine, but there may be a problem with the git
>>> >> repository.
>>> >> When I build from the source tarball on the CIG website, it works
>>> >> fine.
>>> >>
>>> >> On 6/18/15 1:32 AM, "Walter Landry" <wlandry at caltech.edu> wrote:
>>> >>
>>> >> >Oliver Boyd <olboyd at usgs.gov> wrote:
>>> >> >> forrtl: severe (174): SIGSEGV, segmentation fault occurred
>>> >> >> Image PC Routine Line
>>> >> >>Source
>>> >> >> relax 00000000005AF571 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 00000000005ADCC7 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 000000000055E794 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 000000000055E5A6 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 000000000050D184 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 000000000051132D Unknown Unknown
>>> >> >>Unknown
>>> >> >> libpthread.so.0 0000003F77C0F710 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 0000000000454AB7 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 00000000004143E1 Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 0000000000410C6E Unknown Unknown
>>> >> >>Unknown
>>> >> >> libc.so.6 0000003F7701ED5D Unknown Unknown
>>> >> >>Unknown
>>> >> >> relax 0000000000410B79 Unknown Unknown
>>> >> >>Unknown
>>> >> >> srun: error: compute87: task 0: Exited with exit code 174
>>> >> >
>>> >> >It looks like you might be linking against the wrong library at run
>>> >> >time. In your slurm script, run 'ldd' on the executable and make
>>> >> > sure
>>> >> >it links against the right libraries in the right places. If that
>>> >> >does not work, we can try running it in a debugger.
>>> >> >
>>> >> >Cheers,
>>> >> >Walter Landry
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>> --
>>
>> Jeff Falgout
>> US Geological Survey
>> Core Science Analytics, Synthesis, and Libraries
>> Denver Federal Center
>> Building 810, Mail Stop 302
>> Denver, CO 80225
>> jfalgout at usgs.gov
>> Phone: 303-202-4261
>> Cell: 303-648-1767
>>
>
More information about the CIG-SHORT
mailing list