Segmentation fault - invalid memory reference

[bruceyoung01]

I know segmentation fault is the most difficult issue to specify where the problem is. Here I got a segmentation fault while running wrf.exe. I am using wrfchem3.9

d01 2014-07-25_00:00:00 calling inc/HALO_CUP_G3_IN_inline.inc
d01 2014-07-25_00:00:00 in kf_eta_cps

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x2AAAAEB5A287
#1 0x2AAAAEB5A88E
#2 0x3C07C3250F
#3 0x3A85E09 in __module_cu_kfeta_MOD_kf_eta_para
#4 0x3A9DA93 in __module_cu_kfeta_MOD_kf_eta_cps
#5 0x3247623 in __module_cumulus_driver_MOD_cumulus_driver
#6 0x2A77D0E in __module_first_rk_step_part1_MOD_first_rk_step_part1
#7 0x1F05C33 in solve_em_
#8 0x1D0E391 in solve_interface_
#9 0x493912 in __module_integrate_MOD_integrate
#10 0x40B7D3 in __module_wrf_top_MOD_wrf_run


I have no idea what causes this error. Here I attached the error log file and namelist.input.

Thank you
Zhifeng

 

[madhavhmk]

Did you try increasing stacksize to unlimited?
limit stacksize unlimited (in csh)

Also I think if you are specifying start and end times you dont have to give a seperate run_hours as 6. Make it zero and try again. It may help.

 

[bruceyoung01]

Thank you for your reply. Since I am using bash, I just tried
ulimit -s unlimited

And I also changed run_hours as zero. Unfortunately, it did not work.
From the error, it seems it's related to the cumulus scheme. I am not sure whether it is.

 

[Ming Chen]

I have moved this post to wrf-chem support. Hope someone can get back to you soon.

 

[kwthomas]

Hi...

Any chance this is being run on BRIDGES? There is a known problem where setting the stacksize to
unlimited doesn't work. Staff people are working on it.

You can try building a debug version. That will allow you to get a traceback with line numbers.

 

[bruceyoung01]

What is BRIDGES? Would you please explain it? Thank you

 

[kwthomas]

BRIDGES is a supercomputer at the Pittsburgh Supercomputing Center (PSC).

 

[bruceyoung01]

No, I am not using it. I am using the cluster from our university.

 

[RobWaters]

I am now also receiving this error.

I have two identical configurations, except geographic location. kf_eta_cps seems to only be failing in one.

Any solution found so far?

 

[RobWaters]

For me this was because i was missing LANDN from my GFS requests which was causing soil temperatures of absolute zero on one side of the island i was modelling.