- developers-list - monetdb.org

[Monetdb-developers] suggestions on debugging/tracing
by mobigital-nbl＠yahoo.com 02 Mar '08

02 Mar '08

Hi, I am trying to investigate this bug (https://sourceforge.net/tracker/?func=detail&atid=482468&aid=1905467&group_…) which i submitted earlier. I was importing records with COPY TO a staging table in batches of 250K rows. Then copying them into final table. After completing 4,750,000 rows and importing a new batch thebelow exception occurs in GDKreallocmax at the last line of the snippetbelow: oldsize = GDK_MEM_BLKSIZE(s); with "s"=0x00000000100e0000" GDK_MEM_BLKSIZE macro tries to read 8 bytes prior to "s", which is 0x00000000100dfff8 and turns out to be unallocated. I am thinking the h->base of the heap is either free'ed at some point before being used or may be it is set to incorrect value? in any case i wanted to add some tracing to Heap to maintain some history of base change (and who did the change), so that when I hit another access violation during debugging I can use that to pinpoint why Heap base points to unallocated space. i wanted to see if anyone has already used something like that so that I can get a headstart quicker. if anyone have any suggestions on fixing it would be appreciated. Cheers.

1 0

Re: [Monetdb-developers] [Monetdb-checkins] MonetDB5/src/optimizer opt_replicator.mx, , 1.28, 1.29
by Milena.Ivanova＠cwi.nl 02 Mar '08

02 Mar '08

> Update of /cvsroot/monetdb/MonetDB5/src/optimizer > In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv3358 > > Modified Files: > opt_replicator.mx > Log Message: > defConstant already performs a search for the constant value. > > This was my first version too, but defConstant sets a pointer and does not copy the value. The next call finds whatever has left in memory and breaks. Milena > Index: opt_replicator.mx > =================================================================== > RCS file: /cvsroot/monetdb/MonetDB5/src/optimizer/opt_replicator.mx,v > retrieving revision 1.28 > retrieving revision 1.29 > diff -u -d -r1.28 -r1.29 > --- opt_replicator.mx 29 Feb 2008 15:43:20 -0000 1.28 > +++ opt_replicator.mx 2 Mar 2008 11:46:49 -0000 1.29 > @@ -438,20 +438,15 @@ > static int newReplica(MalStkPtr s, InstrPtr p, bit keep) > { > int i, j, k, c, bid, r; > - ValPtr v, v1; > + ValPtr v; > InstrPtr p1; > > assureSpace(); > p1 = copyInstruction(p); > for(i = 0; i< p->argc; i++){ > j = p->argv[i]; > - v = &s->stk[j]; > - c = fndConstant(repl, v); > - if ( c < 0 ){ > - v1 = VALnew(); > - VALcopy(v1,v); > - c = defConstant(repl, v1->vtype, v1); > - } > + v = &s->stk[j]; > + c = defConstant(repl, v->vtype, v); > setArg(p1,i,c); > > if ((i > 0) || > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Monetdb-checkins mailing list > Monetdb-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-checkins >

1 0

Re: [Monetdb-developers] [Monetdb-checkins] MonetDB5/src/optimizer opt_replicator.mx, , 1.23, 1.24
by Milena Ivanova 29 Feb '08

29 Feb '08

Romulo Goncalves wrote: > Update of /cvsroot/monetdb/MonetDB5/src/optimizer > In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv18911 > > Modified Files: > opt_replicator.mx > Log Message: > If you stop you should free the structure that you allocked in REPstart... > > Compilation works and also the replicator test... > > > Index: opt_replicator.mx > =================================================================== > RCS file: /cvsroot/monetdb/MonetDB5/src/optimizer/opt_replicator.mx,v > retrieving revision 1.23 > retrieving revision 1.24 > diff -u -d -r1.23 -r1.24 > --- opt_replicator.mx 29 Feb 2008 07:29:21 -0000 1.23 > +++ opt_replicator.mx 29 Feb 2008 13:13:20 -0000 1.24 > @@ -132,7 +132,7 @@ > opt_export str REPsemijoin(MalBlkPtr mb, MalStkPtr stk, InstrPtr pci); > opt_export str REPdump(int *ret); > opt_export str REPstart(MalBlkPtr mb, MalStkPtr stk, InstrPtr pci); > -opt_export str REPstop(void); > +opt_export str REPstop(MalStkPtr stk); > opt_export str REPsetRetain(int *ret, sht *p); > opt_export str REPsetReuse(int *ret, sht *p); > opt_export str REPsetCache(int *ret, sht *p); > @@ -242,8 +242,10 @@ > } > > > -str REPstop(void) > +str REPstop(MalStkPtr s) > { > + if( s->blk->replica ) > + GDKfree(((ReplicaRec*)s->blk->replica)->vmap); > replref--; > return MAL_SUCCEED; > } > In principle yes, but in this case I don't think so. The vmap is the map to replicated variables, that can be used next time the same query is executed, which is after rep.stop. Pls, revert Milena > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Monetdb-checkins mailing list > Monetdb-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-checkins >

1 0

Re: [Monetdb-developers] [Monetdb-checkins] MonetDB5/src/modules/mal tablet.mx, , 1.114, 1.115
by Romulo Goncalves 29 Feb '08

29 Feb '08

Niels Nes wrote: > Update of /cvsroot/monetdb/MonetDB5/src/modules/mal > In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv3888/src/modules/mal > > Modified Files: > tablet.mx > Log Message: > correctly check for null values > > > Index: tablet.mx > =================================================================== > RCS file: /cvsroot/monetdb/MonetDB5/src/modules/mal/tablet.mx,v > retrieving revision 1.114 > retrieving revision 1.115 > diff -u -d -r1.114 -r1.115 > --- tablet.mx 18 Feb 2008 19:29:07 -0000 1.114 > +++ tablet.mx 18 Feb 2008 19:57:37 -0000 1.115 > @@ -1264,9 +1264,8 @@ > e = s; > if ( s == NULL || > (s == e && fmt->nullstr[0] == 0) || Would not be better here to do only fmt->nullstr because if fmt->nullstr was initialized with NULL this if clause will crash the mserver... Romulo > - (!quote && strcasecmp(s,fmt->nullstr) == 0) || > - (quote && strncasecmp(s+1, fmt->nullstr, e-s) == 0) > - ) { > + (s != e && !quote && strcasecmp(s,fmt->nullstr) == 0) || > + (s != e && quote && strncasecmp(s+1, fmt->nullstr, e-s) == 0)) { > adt = fmt->data; > memcpy(adt, ATOMnilptr(fmt->adt), fmt->nillen); > } else { > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Monetdb-checkins mailing list > Monetdb-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-checkins

1 0

Re: [Monetdb-developers] load runtime incorrectly
by mobigital-nbl＠yahoo.com 28 Feb '08

28 Feb '08

FYI, this issue turned out to be due to the linking to non-debug MSVCR90.lib in: 1. iconv project. - had to make sure that MDFlags in iconv makefile also changed to /MDd when debug was requested. 2. pthreads - changed /MD to /MDd in flags to each .c compilation to .obj, also added /MDd in final linking and removed explicit linking with libcrt.lib - because /MDd already does it with debug crt. > additional info: > > it seems even after DEBUG build, my executable is looking for non-DEBUG version > of MSVCR90.dll > > --------------------------- > mserver5.exe - Unable To Locate Component > --------------------------- > This application has failed to start because MSVCR90.dll was not found. > Re-installing the application may fix this problem. > --------------------------- > > when i copy the set of MSVCR90.dll and manifest into debug directory, i get the > error shown below.(my prior message) > > here are my build commands that i use for all modules: > > nmake /nologo BITS64=1 NEED_MX=1 DEBUG=1 HAVE_ICONV=1 HAVE_MONETDB4=1 > HAVE_MONETDB5=1 HAVE_PYTHON=1 "prefix=%BUILDTOOLS_PREFIX%" > > nmake /nologo BITS64=1 NEED_MX=1 DEBUG=1 HAVE_ICONV=1 HAVE_MONETDB4=1 > HAVE_MONETDB5=1 HAVE_PYTHON=1 "prefix=%BUILDTOOLS_PREFIX%" install > > > > > Does anyone know why i may be getting this error: > > > > Runtime Error! > > > > Program: C:\monetdb\build\bin\mserver5.exe > > > > R6034 > > > > An application has made an attempt to load the C runtime library incorrectly. > > Please contact the application's support team for more information. > > > > ----------- > > also is there a 'clean' target for make? > > i don't see it anywhere. > > > >

1 0

[Monetdb-developers] FYI: reversed nightly testing order: Current, Stable
by Stefan Manegold 27 Feb '08

27 Feb '08

Dear all, just to keep you posted: As of tonight, nightly MonetDB testing will (again) first run the "Current" development head and then the "Stable" release branch. (With the release out for just over a week, now, "urgent" bug-fixing and TestWeb clean-up has settled, making early-morning "Stable" testing results less needed. Instead, increased feature development activities ask for early-morning "Current" testing results.) Best regards from A'dam, Stefan -- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |

1 0

Re: [Monetdb-developers] load runtime incorrectly
by mobigital-nbl＠yahoo.com 27 Feb '08

27 Feb '08

additional info: it seems even after DEBUG build, my executable is looking for non-DEBUG version of MSVCR90.dll --------------------------- mserver5.exe - Unable To Locate Component --------------------------- This application has failed to start because MSVCR90.dll was not found. Re-installing the application may fix this problem. --------------------------- when i copy the set of MSVCR90.dll and manifest into debug directory, i get the error shown below.(my prior message) here are my build commands that i use for all modules: nmake /nologo BITS64=1 NEED_MX=1 DEBUG=1 HAVE_ICONV=1 HAVE_MONETDB4=1 HAVE_MONETDB5=1 HAVE_PYTHON=1 "prefix=%BUILDTOOLS_PREFIX%" nmake /nologo BITS64=1 NEED_MX=1 DEBUG=1 HAVE_ICONV=1 HAVE_MONETDB4=1 HAVE_MONETDB5=1 HAVE_PYTHON=1 "prefix=%BUILDTOOLS_PREFIX%" install > > Does anyone know why i may be getting this error: > > Runtime Error! > > Program: C:\monetdb\build\bin\mserver5.exe > > R6034 > > An application has made an attempt to load the C runtime library incorrectly. > Please contact the application's support team for more information. > > ----------- > also is there a 'clean' target for make? > i don't see it anywhere. > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Monetdb-developers mailing list > Monetdb-developers(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-developers >

1 0

[Monetdb-developers] load runtime incorrectly
by mobigital-nbl＠yahoo.com 27 Feb '08

27 Feb '08

Does anyone know why i may be getting this error: Runtime Error! Program: C:\monetdb\build\bin\mserver5.exe R6034 An application has made an attempt to load the C runtime library incorrectly. Please contact the application's support team for more information. ----------- also is there a 'clean' target for make? i don't see it anywhere.

1 0

Re: [Monetdb-developers] iconv.dll DEBUG version
by mobigital-nbl＠yahoo.com 26 Feb '08

26 Feb '08

you're right the iconvd.lib and iconvd.dll were actually created, i just didn't see them. many thanks > >> Edit lib/Makefile.msvc and change -Feiconv.dll to -Feiconvd.dll, and > >> compile using > >> > >> nmake -f Makefile.msvc NO_NLS=1 DLL=1 MFLAGS=-MDd DEBUG=1 > >> PREFIX=C:\iconv-1.11.win64 > >> > >> This will fail to compile the programs, but should create the library. > > Did you read this last statement? Once you have the files iconvd.dll > and iconvd.lib, you don't need to go any further compiling the iconv > stuff. You only need these two files. Copy them to where you have your > iconv.dll and iconv.lib files and go back to compiling MonetDB. > > -- > Sjoerd Mullender

1 0

Re: [Monetdb-developers] pathfinder __func___ not declared
by mobigital-nbl＠yahoo.com 26 Feb '08

26 Feb '08

Is there a free version of intel compiler that will work? it looked to me that Intel compiler must be purchased? Stefan Manegold wrote: > Pathfinder requires C99 features that are not supported by the Microsoft > Visual C/C++ compiler; hence, on Windows, Pathfinder can only with the Intel > C/C++ compiler. See > http://monetdb.cwi.nl/projects/monetdb//MonetDB/Documentation/Windows-Insta… > for details. > > Stefan >

3 2