NetCDF developers,
Recently the NCL group got a dataset from user which cause
segmentation fault.
Here are the some info about this dataset (from out ncl script):
> cat b.ncl
setfileoption("nc", "usenewhlfs", True)
f=addfile("vp.850.nc","r")
print(f)
> ncl b.ncl
Copyright (C) 1995-2012 - All Rights Reserved
University Corporation for Atmospheric Research
NCAR Command Language Version 6.1.0
The use of this software is governed by a License Agreement.
See http://www.ncl.ucar.edu/ for more details.
Variable: f
Type: file
filename: vp.850
path: vp.850.nc
attributes:
description : ERA Interim vp at 850 hPa.
references : Simmons et al. 2006
source : ECMWF:
http://data-portal.ecmwf.int/data/d/interim_daily/
creation_date : Thu Jan 26 17:59:46 UTC 2012
source_file : /ecmwf/interim//*/u&v*.grib
dimensions:
time = 33349 // unlimited
lat = 121
lon = 240
chunk dimensions:
time = 1 // unlimited
lat = 121
lon = 240
variables:
time: <double> [ 33349 <time | unlimited> ]
Chunking Info: [ 1 <time | unlimited> ]
attributes:
last_time : 2011-10-30 00:00:00
first_time : 1989-01-01 00:00:00
delta_t : 0000-00-00 06:00:00
long_name : initial time
units : hours since 1800-01-01 00:00:00
lat: <float> [ 121 <lat> ]
Chunking Info: [ 121 <lat> ]
attributes:
actual_range : ( -90.000000, 90.000000 )
delta_y : 1.500000
coordinate_defines : center
mapping : cylindrical_equidistant_projection_grid
grid_resolution : 1.5_degrees
long_name : latitude
units : degrees_north
lon: <float> [ 240 <lon> ]
Chunking Info: [ 240 <lon> ]
attributes:
actual_range : ( -180.000000, 178.500000 )
delta_x : 1.500000
coordinate_defines : center
mapping : cylindrical_equidistant_projection_grid
grid_resolution : 1.5_degrees
long_name : longitude
units : degrees_east
vp: <float> [ 33349 <time | unlimited>, 121 <lat>, 240 <lon> ]
Chunking Info: [ 1 <time | unlimited>, 121 <lat>, 240 <lon> ]
attributes:
_FillValue : -999.900024
mapping : cylindrical_equidistant_projection_grid
grid_resolution : 1.5_degrees
pressure_level : 7310034281957426002
units : m^2s^-1
long_name : velocity_potential
By debugging with this dataset,
we are able to trace back the problem is from:
libsrc4/nc4hdf.c, in function nc4_get_var().
We made some change to this function and now NCL works.
The code we have changed are:
start from line: 1075:
The original:
if (!scalar && provide_fill)
{
void *filldata;
int real_data_size = 0;
int fill_len;
/* Skip past the real data we've already read. */
if (!no_read)
for (real_data_size = 1, d2 = 0; d2 < var->ndims; d2++)
real_data_size *= (count[d2] - start[d2]) * file_type_size;
to new:
if (!scalar && provide_fill)
{
void *filldata;
size_t real_data_size = 0;
size_t fill_len;
/* Skip past the real data we've already read. */
if (!no_read)
{
for (real_data_size = 1, d2 = 0; d2 < var->ndims; d2++)
real_data_size *= (count[d2] - start[d2]);
real_data_size *= file_type_size;
}
Regards,
Wei
huangwei@xxxxxxxx
VETS/CISL
National Center for Atmospheric Research
P.O. Box 3000 (1850 Table Mesa Dr.)
Boulder, CO 80307-3000 USA
(303) 497-8924
