Fortran - Relocation truncated to fit

Author

Relocation truncated to fit

Wong Yung

2006-04-07, 4:04 am

Hi

I hope people can help me with a problem I am having. I keep on getting
this error message when I try to compile my program:

ifort NRTYPE.F90 debug2.f90 factorial.f90 comp_2_arrays.f90
population_c.f90 pop_fluctuation2_c.f90 get_rho2.f90 -check all -o
BH_Ar6 BH_Ar6.f90 -L/home/wy/ARPACK/ -larpack_OPT -L$MKL -lmkl_lapack
-lmkl_em64t -lguide -lpthread

/tmp/ifortkvUiEu.o(.text+0x1dd5): In function `MAIN__':

BH_Ar6.f90: relocation truncated to fit: R_X86_64_PC32 against `.bss'

/tmp/ifortkvUiEu.o(.text+0x1dfe):BH_Ar6.f90: relocation truncated to
fit: R_X86_64_PC32 against `.bss'

There's more but it's on the same lines.

I have tried what some people have suggested about using -i-dynamic and
-mcmodel=large and it doesn't help. It does change the error message
though:

ifort NRTYPE.F90 debug2.f90 factorial.f90 comp_2_arrays.f90
population_c.f90 pop_fluctuation2_c.f90 get_rho2.f90 -check all
-mcmodel=large -i-dynamic -o BH_Ar6 BH_Ar6.f90 -L/home/wy/ARPACK/
-larpack_OPT -L$MKL -lmkl_lapack -lmkl_em64t -lguide -lpthread

/home/wy/ARPACK//libarpack_OPT.a(znaupd.o)(.text+0x59): In function
`znaupd_':

znaupd.f: relocation truncated to fit: R_X86_64_PC32 against symbol
`debug_' defined in COMMON section in
/home/wy/ARPACK//libarpack_OPT.a(znaupd.o)

/home/wy/ARPACK//libarpack_OPT.a(znaupd.o)(.text+0x70):znaupd.f:
relocation truncated to fit: R_X86_64_PC32 against `.bss'

And more on the same lines.

I am running SUSE 9.2 Professional with Intel Fortran compiler:

Version 9.0 Build 20050430

My system is a dual Opteron.

Thanks in advance for any help!

glen herrmannsfeldt

2006-04-07, 4:04 am

Wong Yung wrote:

> I hope people can help me with a problem I am having. I keep on getting
> this error message when I try to compile my program:

> ifort NRTYPE.F90 debug2.f90 factorial.f90 comp_2_arrays.f90
> population_c.f90 pop_fluctuation2_c.f90 get_rho2.f90 -check all -o
> BH_Ar6 BH_Ar6.f90 -L/home/wy/ARPACK/ -larpack_OPT -L$MKL -lmkl_lapack
> -lmkl_em64t -lguide -lpthread

> /tmp/ifortkvUiEu.o(.text+0x1dd5): In function `MAIN__':

> BH_Ar6.f90: relocation truncated to fit: R_X86_64_PC32 against `.bss'

Usually this means that an address is too big for the available field.
In your case it looks like a 16 bit address and something bigger than
64K.

> /tmp/ifortkvUiEu.o(.text+0x1dfe):BH_Ar6.f90: relocation truncated to
> fit: R_X86_64_PC32 against `.bss'

> There's more but it's on the same lines.

> I have tried what some people have suggested about using -i-dynamic and
> -mcmodel=large and it doesn't help. It does change the error message
> though:

(snip)

In large model, each item must fit in 64K (for 16 bits) or 4G (for 32
bits), but you may have more than one such item.

You have names like X86_64, but it looks like a 16 bit compiler.
Otherwise, a 32 bit compiler and a program more than 4GB.

> I am running SUSE 9.2 Professional with Intel Fortran compiler:

> Version 9.0 Build 20050430

> My system is a dual Opteron.

-- glen

Greg Lindahl

2006-04-07, 7:02 pm

In article <1144386824.794452.142930@g10g2000cwb.googlegroups.com>,
Wong Yung <wongyung_peach@yahoo.com> wrote:

>I have tried what some people have suggested about using -i-dynamic and
>-mcmodel=large and it doesn't help. It does change the error message
>though:

You want -mcmodel=medium, not large. You can read about this option in the
gcc info pages

info gcc
then ^s to search, 'mcmodel' ...

-- greg

Wong Yung

2006-04-09, 10:02 pm

Hi, I tried the -mcmodel=medium option and get the same problems. In
fact the error messages are exactly the same, word for word as those
for -mcmodel=large.

Sponsored Links