What a frustating business this can be sometimes. I spent more than an hour building the case for the message below, testing scenarios, looking for coding errors, etc. Now, the crash has returned even with the Create_GEO and Destroy_GEO outside the loop. I know it worked fine once, but not now. Therefore, don't spend much effort thinking about this one until I have gathered some better information. Regards, DLH -----Original Message----- From: David Humphrey [mailto:[EMAIL PROTECTED]] Sent: Friday, April 20, 2001 10:56 AM To: '[EMAIL PROTECTED]' Subject: RE: [gxnet]: Floating Point Crash in Destroy_GEO Before I get off into posting the offending code, I have another clue. Perhaps this is documented somewhere. Is it a capital offense to create and destroy a session more than once during the execution of a program? Here's why I ask. I have a utility program that is converting grids from one format to another. Oasis is one of those destinations. Here's a schematic of the C++ code: loop over grids to be converted { read input data switch on type of output grid { case a:... case b:... case geosoft: create a session object use the session object to write a geosoft grid destroy the session object case z:... } } The program dies the second time it enters the "use the session object to write a geosoft grid" and tries to allocate a VV with CreateExt_VV (see code below). Note the printf's. The first time through, everything works fine. The second time through, it crashes and I never get the message "BGI_Grid::writeGrosoft: hcol = ...". If I move the create and destroy outside the loop so that the session object is created and destroyed only once, the problem does not arise. /*--------------------------------------------------------------------*/ void BGI_Grid::writeGeosoft ( void *pGeo, char *fileName) { // write a grid in native Geosoft binary format /*--------------------------------------------------------------------*/ printf ("BGI_Grid::writeGeosoft: pGeo = %d\n",pGeo); GX_HANDLE hGrd = CreateNewFile_IMG(pGeo,_l(GS_DOUBLE),_l(1),&nxs,&nys,fileName); printf ("BGI_Grid::writeGeosoft: hGrd = %d\n",hGrd); SetInfo_IMG(pGeo,&hGrd,&xinc,&yinc,&xmin,&ymin,&rotation); printf ("BGI_Grid::writeGeosoft: returned from SetInfo_IMG\n"); printf ("BGI_Grid::writeGeosoft: Grid parameters: %d %d %f %f %f %f\n",nxs,nys,xmin,ymin,xinc,yinc); GX_HANDLE hCol = CreateExt_VV(pGeo,_l(GS_DOUBLE),&nys); printf ("BGI_Grid::writeGeosoft: hCol = %d\n",hCol); for ( long nx=0; nx<nxs; nx++ ) { for ( long ny=0; ny<nys; ny++ ) { double zval = getval(nx,ny); if ( zval == null ) zval = -1.e32; SetReal_VV(pGeo,&hCol,&ny,&zval); } printf ("BGI_Grid.writeGeosoft: about to write column #%d...",nx); WriteX_IMG(pGeo,&hGrd,&nx,_l(0),&nys,&hCol); printf ("BGI_Grid.writeGeosoft: success\n"); } Destroy_VV(pGeo,&hCol); Destroy_IMG(pGeo,&hGrd); } Therefore, one of three things 1. One shouldn't create and destroy session objects more than once per execution of a program. 2. There's a bug somewhere, presumeable in the CreateExt_VV code, that keeps one from doing this. 3. There's somethings else in my code that is corrupting memory. In any case, I have either solved the problem or developed a workaround, so I'm happy...for the moment. Regards, DLH -----Original Message----- From: Stephen Cheesman [mailto:[EMAIL PROTECTED]] Sent: Thursday, April 19, 2001 9:16 AM To: '[EMAIL PROTECTED]' Cc: Tech (Geosoft Corporate) Subject: RE: [gxnet]: Floating Point Crash in Destroy_GEO David: At a quick glance, we can't give you a reason or work-around for the aborts you are getting in Destroy_GEO. It does not appear to be a true Floating-Point error. It may be a problem with with linking between DLLs in different versions, and it may be necessary to re-install one or more Oasis montaj components. I will forward this to our technical support. Any additional information (original code, files, etc) is always valuable in tracing down aborts. Stephen Cheesman Geosoft Inc. -----Original Message----- From: David Humphrey [mailto:[EMAIL PROTECTED]] Sent: Wednesday, April 18, 2001 11:49 AM To: '[EMAIL PROTECTED]' Subject: [gxnet]: Floating Point Crash in Destroy_GEO I'm working on a C++ program that accesses Oasis montaj functions through the GX Developer facilities. This program creates a session object with Create_GEO, creates a new grid file, writes values to it with the _VV and _IMG functions, and then tries to close the session object with Destroy_GEO. The contents of the grid are correct. Can anyone give me a clue why Destroy_GEO is crashing with a "Floating Point Crash"? The contents of the abort.log file appears below. Regards, David L. Humphrey Manager, Software Development Bell Geospace, Inc GEOSOFT PROGRAM TERMINATION Please FAX to Geosoft Tech Support at (Canada 416) 369 9599. or E-Mail to [EMAIL PROTECTED] . Also note what the program was doing at the time of failure. ------------------------------------------------------------ CURRENT COMPUTER PROFILE Operating System: Microsoft Windows NT 4.0 (Build 1381) CPU Type: Intel Pentium Number of CPUs: 2 Memory Utilization: 0% Total Physical Memory: 255.42 MB Available Physical Memory: 121.70 MB Total Virtual Memory: 2047.88 MB Available Virtual Memory: 2026.81 MB Maximum Paging Size: 493.99 MB Current Paging Size: 326.77 MB ------------------------------------------------------------ PROGRAM DETAILS Date and Time: Wed Apr 18 10:36:24 2001 Program Name: (Unknown) Version: Version 5.0.6 (KZ) ------------------------------------------------------------ Exception GEOCORE : The thread tried to read from or write to a virtual address for which it does not have appropriate access. Call Stack [Thread 0] (Current) pBindFunc_GXX //depot/rel/geo/5.0.6_rel/src/class/gx/gx.c#2 {3348} [Thread 1] SMPThread //depot/rel/geo/5.0.6_rel/src/class/sys/smpsys.c#1 {93} [Thread 2] SMPThread //depot/rel/geo/5.0.6_rel/src/class/sys/smpsys.c#1 {93} Error Stack (Error) ========= Floating Point Crash ========= ======================================== ------------- CPU Context ------------ ContextFlags : 1048817 Dr0 : 0 Dr1 : 0 Dr2 : 0 Dr3 : 0 Dr6 : 0 Dr7 : 0 SegGs : 0 SegFs : 131 SegEs : 50 SegDs : 50 Edi : 277225358 Esi : 15251971 Ebx : 0 Edx : 96 Ecx : 16 Eax : 15252035 Ebp : 4068104 Eip : 8341131 SegCs : 177 EFlags : 1056864 Esp : 2215753 SegSs : 50 ControlWord : 4294910199 StatusWord : 4294902786 TagWord : 4294967295 ErrorOffset : 7242437 ErrorSelector : 1085604017 DataOffset : 278288655 DataSelector : 4294901810 Cr0NpxState : 0 _______________________________________________________ More mailing list info http://www.geosoft.com/support/listserv/index.html List Archive http://www.mail-archive.com/gxnet@lists.geosoft.com _______________________________________________________ More mailing list info http://www.geosoft.com/support/listserv/index.html List Archive http://www.mail-archive.com/gxnet@lists.geosoft.com _______________________________________________________ More mailing list info http://www.geosoft.com/support/listserv/index.html List Archive http://www.mail-archive.com/gxnet@lists.geosoft.com _______________________________________________________ More mailing list info http://www.geosoft.com/support/listserv/index.html List Archive http://www.mail-archive.com/gxnet@lists.geosoft.com
