Tamas, I believe this was happening because of a known issue allocating DLL objects on the heap. I have made the stream objects into member variables to work around this problem on the trunk (r22052). Let me know if your tests continue to fail.
Kirk. From: Tamas Szekeres [mailto:szeker...@gmail.com] Sent: dimanche 27 mars 2011 10:06 To: Kirk McKelvey Cc: Even Rouault; gdal-dev Subject: Re: RE : [gdal-dev] GDAL autotest (mrsid_1) crashing on Win64 I did some investigations related to the Win64 problem, but I have no clues how to get further with this problem. I've used Unified_DSDK_8.0_win64-vc9 and the MSVC 2008x64 compiler and even the following code is causing heap corruption when running the mrsid autotest: LTIOFileStream *ltiofs = new LTIOFileStream(); if ( ltiofs ) delete ltiofs; // heap corruption here HEAP[python.exe]: Invalid Address specified to RtlFreeHeap( 00000000025E0000, 0000000001E97610 ) Windows has triggered a breakpoint in python.exe. This issue seems to be related to the changes r21974<http://trac.osgeo.org/gdal/changeset/21974> Best regards, Tamas 2011/3/23 Kirk McKelvey <kmckel...@lizardtech.com<mailto:kmckel...@lizardtech.com>> Strange, nothing happened to the driver code in that range. I will bring up a 64-bit build and see if I can debug it. Am I correct that your 32-bit builds are passing this test? ________________________________________ De : gdal-dev-boun...@lists.osgeo.org<mailto:gdal-dev-boun...@lists.osgeo.org> [gdal-dev-boun...@lists.osgeo.org<mailto:gdal-dev-boun...@lists.osgeo.org>] de la part de Tamas Szekeres [szeker...@gmail.com<mailto:szeker...@gmail.com>] Date d'envoi : mercredi 23 mars 2011 08:36 À : Even Rouault Cc : gdal-dev Objet : Re: [gdal-dev] GDAL autotest (mrsid_1) crashing on Win64 Even, According to the buildlog at http://vbkto.dyndns.org/sdk/ r21980 was still OK, and r21988 was already wrong. Best regards, Tamas 2011/3/23 Even Rouault <even.roua...@mines-paris.org<mailto:even.roua...@mines-paris.org><mailto:even.roua...@mines-paris.org<mailto:even.roua...@mines-paris.org>>> Selon Tamas Szekeres <szeker...@gmail.com<mailto:szeker...@gmail.com><mailto:szeker...@gmail.com<mailto:szeker...@gmail.com>>>: Tamas, Indeed there have been recent changes. You might want to drop a note about that to Kirk in ticket http://trac.osgeo.org/gdal/ticket/4002 . After latest fixes from Kirk and me, I no longer see problems under Linux however. I don't know which version of the GeoDSDK you are using. But, at least under Linux, there's an ABI incompatibility of GeoSDSK v7 with latest libgeotiff (due to the addition of new fields in geo_normalize.h) that cause crashes. But the libgeotiff change is a bit older than the MrSID driver one, so it would be surprising that Kirk changes have an impact on that. Under Linux, using version 8 solves that issue. If you revert only the changes of the MrSID driver (to what it was before r21974), what happens ? Best regards, Even > Hi All, > > I'm experiencing permanent crash in the mrsid_1 autotest on the Win64 builds > of http://vbkto.dyndns.org/sdk/. This problem may be related due to a recent > change (in 2-3 days). > > Any ideas about the reason? > > Best regards, > > Tamas >
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/gdal-dev