[gdal-dev] OGRFree(pszSrcWKT) can crash Matlab

[jluis at ualg.pt]

Quoting Chaitanya kumar CH <chaitanya.ch at gmail.com>:

> Joaquim,
>
> You did change pszSrcWkt (
> http://code.google.com/p/mirone/source/browse/trunk/mex/gdalwarp_mex.cpp#385
> ).
>
> You should remove line 385 and move the line 386 to 330.


Ok, thank you guys.

Joaquim


>
> On Sat, Jan 23, 2010 at 11:17 PM, <jluis at ualg.pt> wrote:
>
>> Quoting Even Rouault <even.rouault at mines-paris.org>:
>>
>>  Joaquim,
>>>
>>> you don't provide enough code to really see what's wrong, but my guess is
>>> that you misuse GDAL API.
>>>
>>
>> Even
>>
>> Quite possible the first and I was trying to summarize what happens
>>
>>
>>
>>> If you get pszSrcWKT, with
>>>
>>> pszSrcWKT = GDALGetProjectionRef(hDS);
>>>
>>
>>
>> No, I got it from oSrcSRS.exportToWkt( &pszSrcWKT );
>>
>> See lines 325 & 329 of
>>
>> code.google.com/p/mirone/source/browse/trunk/mex/gdalwarp_mex.cpp
>>
>> Joaquim
>>
>>
>>
>>> The pszSrcWKT string is owned by the hDS object, so as soon as you destroy
>>> the hDS object with GDALClose(), the pointer becomes invalid and points to
>>> arbitrary data.
>>>
>>> For the same reason, you should never free the string returned by
>>> GDALGetProjectionRef() with OGRFree() or any other free() call.
>>>
>>> See http://gdal.org/classGDALDataset.html :
>>>
>>> GDALDataset::GetProjectionRef() :
>>>
>>> returns:
>>> a pointer to an internal reference string. It should not be altered, freed
>>> or expected to last for long.
>>>
>>> Best regards,
>>>
>>> Even
>>>
>>> Le Saturday 23 January 2010 16:04:58 jluis at ualg.pt, vous avez écrit :
>>>
>>>> Hi,
>>>>
>>>> I have a serious crash in one of my Matlab MEXs caused by a call to
>>>> OGRFree() with an argument of size one (possibly the  null terminator).
>>>> This happens on OSX SL (64 bits) but not on Windows (where I still
>>>> build at 32 bits)
>>>>
>>>> Here is, as far as I could trace it, what is happening (SVN version of
>>>> the same age as GDAL1.7RC1)
>>>>
>>>> mexPrintf("---- %s\n%s\n", pszSrcWKT,pszDstWKT);
>>>>
>>>> That prints
>>>>
>>>> ---- GEOGCS["WGS 84",DATUM["WGS_1984",SPHEROID["WGS
>>>>
>>>> 84",6378137,298.257223563,AUTHORITY["EPSG","7030"]],TOWGS84[0,0,0,0,0,0,0],
>>>>
>>>> AUTHORITY["EPSG","6326"]],PRIMEM["Greenwich",0,AUTHORITY["EPSG","8901"]],UNI
>>>>
>>>> T["degree",0.0174532925199433,AUTHORITY["EPSG","9108"]],AUTHORITY["EPSG","43
>>>> 26"]] PROJCS["UTM Zone 29, Northern Hemisphere",GEOGCS["WGS
>>>> 84",DATUM["WGS_1984",SPHEROID["WGS
>>>>
>>>> 84",6378137,298.257223563,AUTHORITY["EPSG","7030"]],TOWGS84[0,0,0,0,0,0,0],
>>>>
>>>> AUTHORITY["EPSG","6326"]],PRIMEM["Greenwich",0,AUTHORITY["EPSG","8901"]],UNI
>>>>
>>>> T["degree",0.0174532925199433,AUTHORITY["EPSG","9108"]],AUTHORITY["EPSG","43
>>>>
>>>> 26"]],PROJECTION["Transverse_Mercator"],PARAMETER["latitude_of_origin",0],PA
>>>>
>>>> RAMETER["central_meridian",-9],PARAMETER["scale_factor",0.9996],PARAMETER["f
>>>> alse_easting",500000],PARAMETER["false_northing",0],UNIT["Meter",1]]
>>>>
>>>>
>>>> Next the code does:
>>>>
>>>>        GDALClose( hSrcDS );
>>>> mexPrintf("---- %s\n%s\n", pszSrcWKT,pszDstWKT);
>>>>
>>>> and now pszSrcWKT is empty. See
>>>>
>>>> ---- PROJCS["UTM Zone 29, Northern Hemisphere",GEOGCS["WGS
>>>> 84",DATUM["WGS_1984",SPHEROID["WGS
>>>>
>>>> 84",6378137,298.257223563,AUTHORITY["EPSG","7030"]],TOWGS84[0,0,0,0,0,0,0],
>>>>
>>>> AUTHORITY["EPSG","6326"]],PRIMEM["Greenwich",0,AUTHORITY["EPSG","8901"]],UNI
>>>>
>>>> T["degree",0.0174532925199433,AUTHORITY["EPSG","9108"]],AUTHORITY["EPSG","43
>>>>
>>>> 26"]],PROJECTION["Transverse_Mercator"],PARAMETER["latitude_of_origin",0],PA
>>>>
>>>> RAMETER["central_meridian",-9],PARAMETER["scale_factor",0.9996],PARAMETER["f
>>>> alse_easting",500000],PARAMETER["false_northing",0],UNIT["Meter",1]]
>>>>
>>>> If I make it print the size of pszSrcWKT  with strlen(pszSrcWKT) the
>>>> answer is 1 and finally the MEX (and the entire MATLAB) crash on the
>>>> call
>>>>
>>>> OGRFree(pszSrcWKT);
>>>>
>>>>
>>>> Joaquim Luis
>>>>
>>>> ----------------------------------------------------------------
>>>> This message was sent using IMP, the Internet Messaging Program.
>>>>
>>>> _______________________________________________
>>>> gdal-dev mailing list
>>>> gdal-dev at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>>>>
>>>
>>>
>>>
>>>
>>
>>
>> ----------------------------------------------------------------
>> This message was sent using IMP, the Internet Messaging Program.
>>
>>
>
>
> --
> Best regards,
> Chaitanya kumar CH.
> /tʃaɪθənjə/ /kʊmɑr/
> +91-9848167848
> 17.241582N 80.142635E
>



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.