[GRASS-user] m.eigensystem segfaults under Jaunty (beta)

GRASS GRASS user
1

Nikos:
m.eigensystem, compiled against latest grass6_dev _AND_
grass-6.4.0RC4 source code, segfaults under Ubuntu-J 64-bit
Beta.

It works fine with Ubuntu Intrepid! So, something in Jaunty
doesn't like m.eigensystem (fortran code?)!

Hamish:
> > any chance of debugging? (gdb, etc)
> > how far does it get?

perhaps you could compile into the fortran code some informative
hints, e.g. put scatter some of these before and after reading
the input values, etc:

----8<----
PRINT *,"Made it past normalizing the data"
----8<----

(you have recompiled everything from scratch since the upgrade,
right? ie you don't have any binaries trying to access library
adresses which have since changed)

Hamish

2

Nikos:

> m.eigensystem, compiled against latest grass6_dev _AND_
> grass-6.4.0RC4 source code, segfaults under Ubuntu-J 64-bit
> Beta.
>
> It works fine with Ubuntu Intrepid! So, something in Jaunty
> doesn't like m.eigensystem (fortran code?)!

Hamish:

> > > any chance of debugging? (gdb, etc)
> > > how far does it get?

Hamish:

perhaps you could compile into the fortran code some informative
hints, e.g. put scatter some of these before and after reading
the input values, etc:

----8<----
       PRINT *,"Made it past normalizing the data"
----8<----

(you have recompiled everything from scratch since the upgrade,
right? ie you don't have any binaries trying to access library
adresses which have since changed)

OK, maybe I'll try that also. Yes, everything from scratch -- not a
single binary...

...wait, there is FWTools. but this shouldn't have any influence in
grass' source code compilation, right?

Nikos

3

Nikos:

> m.eigensystem, compiled against latest grass6_dev _AND_
> grass-6.4.0RC4 source code, segfaults under Ubuntu-J 64-bit
> Beta.

Hamish:

perhaps you could compile into the fortran code some informative
hints, e.g. put scatter some of these before and after reading
the input values, etc:

----8<----
       PRINT *,"Made it past normalizing the data"
----8<----

So I did. Randomely I've introduced in main.f some PRINT lines (I
changed the text -- details below).

I get an error when I try to compile:

# attempt to compile after changes
make MODULE_TOPDIR=/geo/osgeo/src/grass6_devel

Makefile:20: warning: overriding commands for target
`/geo/osgeo/src/grass6_devel/dist.x86_64-unknown-linux-gnu/bin/m.eigensystem'
/geo/osgeo/src/grass6_devel//include/Make/Module.make:15: warning:
ignoring old commands for target
`/geo/osgeo/src/grass6_devel/dist.x86_64-unknown-linux-gnu/bin/m.eigensystem'
Makefile:27: warning: overriding commands for target `clean'
/geo/osgeo/src/grass6_devel//include/Make/Rules.make:72: warning:
ignoring old commands for target `clean'
f77 -c -o main.o main.f
   MAIN eigen:
Error on line 43: Declaration error declaration among executables
Error on line 44: Declaration error declaration among executables
Error on line 45: Declaration error declaration among executables
/usr/bin/f77: aborting compilation
make: *** [main.o] Error 25
# # # # # #

Probably I am not doing it right. Any help to (finally) debug this?

Nikos
---

svn diff
Index: main.f

--- main.f (revision 36664)
+++ main.f (working copy)
@@ -39,6 +39,7 @@
C of the eigen value (E).

C
+ PRINT *,"Starting the program..."
        DOUBLE PRECISION A(30,30),W1,W2,WR(30),WI(30),Z(30,30),FV1(30)
        DOUBLE PRECISION SUM1,SUM2
        INTEGER IV1(30),N
@@ -51,9 +52,10 @@
C PRINT *,"N=",N

C
+ PRINT *,"Before reading the matrix"
C read the matrix
        READ(5,*,END=1000) (( A(I,J),J=1,N),I=1,N)
-
+ PRINT *,"Read the matrix"
C
C run the real-general eigen subroutine
        CALL RG(30,N,A,WR,WI,1,Z,IV1,FV1,IERR)
@@ -72,6 +74,7 @@
        WRITE(6,100) "E",WR(I),WI(I), W1/SUM1 * 100.0

C
+ PRINT *,"Normalising the data"
C Normalize the eigenvectors before printing
        SUM2 = 0.0
        DO 6 J=1,N
@@ -113,8 +116,8 @@
    36 CONTINUE

     2 CONTINUE
+ PRINT *,"Made it past normalizing the data"

-
C
        CALL EXIT(0)
  1000 PRINT *,"Incomplete input file"