Debugging with IDB

Objective

Background of this item: when the model does not run as expected, the calculations will need to be checked step by step. Debuggers such as IDB help to identify what are the values taken by some variables at some point, check that a subroutine of interest is actually called during the execution of the code (and how many times it happens),... This might be very useful when one would like to speed up the identification of the variable which is the first to take weird values (particularly after a significant modification of the code) and the snap of code which cause that trouble. An basic alternative to the use of debuggers is added WRITE statements to the code.

IDB on obelix

Authors: P. Brender
Last revision: J. Ghattas (2019/05/02)

Boundary conditions for using a debugger

Of course, getting a functional version of the code after a modification of one of the routine of ORCHIDEE continues to require a few steps and the debugger we present only helps to speed up the second one :

1. Getting a version of the code which can be compiled. The first errors displayed by the compiler before crashing should be of some help to solve that issue.
2. Once the code including the modification can be compiled properly, it may often happen that some of the variables take aberrant values, even for runs in offline mode on one point. You are likely to be interested by this tutorial if you are used to proceed to tedious cycles of :
   • addition within the subroutine of interest of lines such as "PRINT *, 'MY_VAR='my_var
   • compilation of the code
   • screening of the standard output of the executed code
3. Check that the introduction of the new feature doesn't lead to weird behaviour for runs at the global scales and/or coupled with the GCM.

Step by step manual

Activate the debug flags when compiling the orchidee_ol executable

The goal of this step is to include marks within the executable which let the debugger identify which line of the source code is processed at every step of the execution. Follow the explanations for compiling ORCHIDEE, XIOS and IOIPSL in debug mode at following page How to get starting debugging.

Launch the executable within the debugger

Start the executable from the folder of your choice within which you have added a run.def file with all the options of your choice and information about the location of the forcing files, the restart files,...

In the present example, we will run the code unto the first execution of line 278 of sechiba.f90, print the value of all the variables defined at that point, continue until the next execution of that line, print again the value of lai_max and then finish the execution of orchidee_ol.

First, we copy the newly compiled executable (see above) in the execution folder. Check that all the restart and forcing files are correctly referred to in the run.def and make sure that all restart files of the previous run have been removed from the source folder.

user@computer:~/my_execution_folder>cp ~/my_orchidee_install/bin/orchidee_ol .
user@computer:~/my_execution_folder>ls -l
orchidee_ol   run.def driver_start.nc sechiba_start.nc stomate_start.nc

Then launch the executable within the environment of the debugger. If you are within the lab, it might be nicer to use the graphical interface environment. Type :

user@computer:~/my_execution_folder>idb ./orchidee_ol
Intel(R) Debugger for applications running on Intel(R) 64
Reading symbols from ~/my_execution_folder/orchidee_ol...done.

From outside the lab, it is more efficient to use only the command line :

user@computer:~/my_execution_folder>idbc ./orchidee_ol

We set a first breakpoint (line of the code at which we want the execution to halt) and run the executable to that point. Here we added a condition on a variable (kjit) to stop only when the time-step kjit==5800 is reached.

(idb)break sechiba.f90:278 if (kjit == 5800)  
(idb)run

Then we have a look at all the variable defined within the present subroutine

(idb)print lai_max

...continue the execution of the code step by step :

(idb)step

Look at the evolution of the variable after each stop in the iteration :

(idb)display lai_max

...look at the code around the current position

(idb)list

Now we can change the condition on 'kjit' for the first breakpoint (the only defined as can be checked by running info b or info break). Then, we continue the execution of the code until the new condition is matched :

(idb)cond 1 (kjit==5900)
(idb)cont

We could also simply remove this condition and continue the execution of the code until the next execution of the line 278 of sechiba.f90 :

(idb)cond 1
(idb)cont

show the list of breakpoints

(idb)info breakpoints

...remove all breakpoints to finish the execution of the code and quit the debugger

(idb)delete break
(idb)cont
(idb)quit
user@computer:~/my_execution_folder>
user@computer:~/my_execution_folder>ls -l
orchidee_ol   run.def driver_start.nc sechiba_start.nc stomate_start.nc driver_restart.nc sechiba_restart.nc stomate_restart.nc out_orchidee.txt

...and one may start again from the beginning after having removed the restart files

user@computer:~/my_execution_folder>rm -f *restart.nc
user@computer:~/my_execution_folder>idb ./orcdidee_ol

Rather than leaving the debugger and coming again, we could have rm the restart files from inside and start again the execution from the beginning :

(idb)shell rm -f *restart.nc
(idb)run

Note also that shortcuts do exist for most of the commands (r for run, s for step, b for breakpoint,...). Using it further speeds things a little bit.

Other elements of syntax

1. In line help
   • general menu

     (idb)help
     

   • on a precise command

     (idb)help info
     

   • or subcommand

     (idb)help info breakpoints
     

2. List
   • all the available commands

     (idb)complete
     

   • all the commands starting with a defined subcommand

     (idb)complete info
     

   • local variables defined at different level of the call stack (for exemple in dim2_driver, which is always the outermost (frame -1) or the inner one, for instance sechiba.f90 if we have just stopped at the breakpoint defined above.

     (idb)bt
     (idb)frame 1
     (idb)info locals
     (idb)frame 0
     (idb)info locals
     

3. Magic
   • Change the value of one variable and continue the run afterwards to check if it solve all the problems.

     (idb)set variable fluxsens(1)=1
     

4. Reference sheet
   • Quick reference sheet​. Also available on the net ​http://www.scribd.com/doc/3589/gdb-quickref

Totalview

Totalview is GUI debugger that works very will with mpi/omp binaries. You can find it installed in big HPCs such as Curie or ADA.

Curie

SPMD

In order to run totalview you need to make available to you like this:

module load totalview

In order to run you simulation in interactive mode so you can use the debugger interface:

ccc_mprun -n 16 -p standard -A <your project id> -d tv ./orchidee_ol 

This call ask for 16 processos in the standard queue. -d tv selects totalview as a debugger.

When the startup window shows up, select "Enable memory debugging".

MPMD

The script below defines how to run an MPI-OpenMP with totalview (eg: coupled LMDZ + Orchidee) program:

#!/bin/bash
#MSUB -r mictgrm_test # Request name
#MSUB -n 64 # Number of tasks to use
#MSUB -c 2 # Number of tasks to use
#MSUB -T 4000 # Elapsed time limit in seconds
#MSUB -o orchid_%I.o # Standard output. %I is the job id
#MSUB -e orchid_%I.e # Error output. %I is the job id
#MSUB -Q normal 
#MSUB -D
#MSUB -X
#MSUB -A gen6328
#MSUB -q standard


set -x
module unload netcdf hdf5
module load netcdf/4.3.3.1_hdf5_parallel
module load hdf5/1.8.9_parallel
module load totalview

#module load ddt
#unset SLURM_SPANK_AUKS

# enable core dump file in case of error
ulimit -c unlimited

export KMP_STACKSIZE=3g
export KMP_LIBRARY=turnaround
export MKL_SERIAL=YES
OMP_NUM_THREADS=2

cat << END > pp.conf
4    ./xios.x
60   totalview ./lmdz.x
END

mpirun -tv -n 4 ./xios.x : -n 60 ./lmdz.x

In this specific case, coupled lmdz + Orchidee runs 60 MPI procs and 4 XIOS procs with 2 OpenMP threads. In total, it requires 128 procs.

When the startup window shows up, select "Enable memory debugging".

Notes

Totalview is not able to debug when the binary is compiled with -p flag (only for profiling purposes). For that reason, it needs to be removed from the compilation.

If you compile orchidee with the makeorchidee_fcm tool, make sure to remove it from arch.fcm file:

arch.fcm:

%DEBUG_FFLAGS        -fpe0 -p -O0 -g -traceback -fp-stack-check -ftrapuv -check bounds -check all

to

%DEBUG_FFLAGS        -fpe0 -O0 -g -traceback -fp-stack-check -ftrapuv -check bounds -check all

Make sure this module is not loaded when compiling the source code AND running the executable. Unload by

module unload gnu/4.8.1

Allinea DDT (Orchidee + XIOS) on Curie

DDT is a parallel debugger found in Curie Hpc. How to run an Orchidee Parallel simulation with XIOS in server mode.

1- Load ddt and run in interative mode

module load ddt 
ddt &

2- Select Run (ddt main window)

3- Properly setup the simulation The total number of processors will be the sum of Orchidee and XIOS.

4- Click "Play" to start the simulation

5- For more info

​Youtube DDT tutorial video

Alternative debuggers

Open source debugger

GNU debugger, a free and open source alternative to idb :

• gdb

A nice list of reference for using gdb and idb (in -gdb mode, the default, as explained in the syntax paragraph):

But gdb is not working too well on code compiled with ifort, use idbc in these situations.

Other Graphical User Interfaces

Beware, unfortunately these tools are a bit "buggy" themselves, so the execution might crash more often with them than when launching the executable through the command line debugger presented above.

• ddd

Syntax

By default, idb actually use the very same syntax as gdb but may also be used with an other syntax, when launched with the -dbx option. The two are pretty similar. Simply beware of that when you copy snapshot on forums,...

Ex of diff :

idbc -gdb
(idb)set variable fluxsens(1)=1

idbc -dbx
(idb)assign fluxsens(1)=1

Other documents

• ​A presentation of the debug tools available at IDRIS (should also be of some use for those using mercure)
• ​A course about the creation of makefiles (or to understand how the compilation step is handled)