[Gdal-dev] HDF EOS segmentation fault

[Jan Cermak]

Frank Warmerdam wrote:
> Jan Cermak wrote:
>> Hi,
>>
>> When I read in an hdfeos dataset from a MYD06 file I get a segfault of 
>> the following type:
>> *** glibc detected *** python: double free or corruption (fasttop): 
>> 0x081b8d90 ***
>> This occurs either at one of the steps following the data read or at 
>> the termination of the program. There does not seem to be a pattern. 
>> Here is some example code:
>>
>> #!/usr/bin/env python
>> import gdal
>> DSname='HDF4_EOS:EOS_SWATH:"/home/jancer/data/tmp/MODIS/MYD06_L2.A2006220.1315.005.2006222153137.hdf":mod06:Cloud_Optical_Thickness' 
>>
>> DSobj = gdal.Open(DSname)
>> data = DSobj.ReadAsArray()
>> data.shape
>> halfdata = data/2.
>>
>> There is a sample file at
>> ftp://ladsweb.nascom.nasa.gov/allData/5/MYD06_L2/2006/220/MYD06_L2.A2006220.1315.005.2006222153137.hdf 
>>
> 
> Jan,
> 
> I believe you are experiencing one of several bugs that have been fixed
> in the last six months or so in GDAL's HDF driver.  Can you update to
> either a 1.4 branch snapshot or a trunk snapshot from:
> 
>  http://download.osgeo.org/gdal/daily
> 
> and see if the problem still persists? (or perhaps grab a recent FWTools
> build).
> 
> Best regards,


Hi Frank,

actually that was what I suspected when the problem first occurred. So I 
updated to gdal 1.4.2 (1007/06/27) a week ago. This didn't make any 
difference though. I just tried the latest nightly build but this gives 
me an import error, so I reverted to 1.4.2

By the way, I only get the segfault for MYD06 files. MYD021km etc. work 
fine.

Any suggestions?
Jan