It works OK for AP stock. For AP opt app rebuild made it ~2% slower (may be in variance range) but works OK too. Hope it will help MB too (where original problem was), will see.
----- Original Message ----- From: "Raistmer" <[email protected]> To: "Rom Walton" <[email protected]> Cc: <[email protected]> Sent: Thursday, August 12, 2010 1:52 AM Subject: Re: [boinc_dev] Incorrect function when running under BOINC > OK, will check with stock first and then with opt build. > BTW, may such AP behavior result in damaged checkpoints (look alpha list > for > problem description) ? > > ----- Original Message ----- > From: "Rom Walton" <[email protected]> > To: "Raistmer" <[email protected]> > Cc: "David Anderson (BOINC)" <[email protected]>; > <[email protected]> > Sent: Thursday, August 12, 2010 1:36 AM > Subject: RE: [boinc_dev] Incorrect function when running under BOINC > > > Okay, the issue causing the 0x80000003 exception should now be fixed in > the trunk version of BOINC. > > The Astropulse client was reusing the same instance of an MFILE class to > write its current state. After it wrote it's state it closed the file > and released the MFILE buffer. The MFILE class didn't re-create the > buffer during the next attempt to write the current state. > > I have also fixed up the project files for Astropulse and SETI_BOINC so > that they use the BOINC based projects for libboinc_staticcrt, > libboincapi_staticcrt, and libgraphics. > > Anyway, checkout the stock project files and see if that also fixes the > CRT problem you sometimes see. > > ----- Rom > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of Rom Walton > Sent: Wednesday, August 11, 2010 1:22 PM > To: David Anderson (BOINC); [email protected] > Subject: Re: [boinc_dev] Incorrect function when running under BOINC > > Which would be a result of the CRT not being initialized. > > I've build a private drop of the ap_client with VS 2005. I'll update > the project files to 2008 and see what happens. > > ----- Rom > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of David Anderson > Sent: Wednesday, August 11, 2010 1:08 PM > To: [email protected] > Subject: Re: [boinc_dev] Incorrect function when running under BOINC > > The arg to msize() is the result of malloc(64K). > It looks like the memory-allocation system hasn't been > initialized. > > On 10-Aug-2010 10:42 PM, Rom Walton wrote: >> Hmmmmmm... >> >> According to the callstack msize is called with a null parameter. > Since >> it is a debug build I assume the CRT is asserting and the exception > code >> is 0x80000003. >> >> Using some of the stats before the callstack, how much virtual memory >> was in use before the crash? >> >> Which version of Visual Studio are you using? >> >> I'll try and repro. >> >> ----- Rom >> >> -----Original Message----- >> From: Raistmer [mailto:[email protected]] >> Sent: Tuesday, August 10, 2010 6:11 PM >> To: Rom Walton; [email protected] >> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >> >> Well, I getting exception with astropulse too now. All that was > changed >> - >> BOINC libs to link. >> Call stack and BPINC function that causes exception are: >> >> - Callstack - >> ChildEBP RetAddr Args to Child >> 000f8b5c 0050f297 00000000 00000000 00000000 00000000 >> ntdll!DbgBreakPoint+0x0 >> 000f8ba8 0042ea9c 00000000 00111274 0042e9ef 00000000 >> ap_5.05_win_x86_SSE3_r446_DEBUG!_msize+0x25 >> (f:\dd\vctools\crt_bld\self_x86\crt\src\msize.c:47) >> 000f8bb4 0042e9ef 00000000 00000015 00563f0c 00000014 >> ap_5.05_win_x86_SSE3_r446_DEBUG!realloc_aux+0x9 >> (d:\r\seti6\boinc\lib\mfile.cpp:61) >> 00111274 0042eadc 00534b20 0011129c 0011129c 0052dba1 >> ap_5.05_win_x86_SSE3_r446_DEBUG!MFILE::vprintf+0x1f >> (d:\r\seti6\boinc\lib\mfile.cpp:82) >> 0011128c 0040bfcc 00563f0c 00534b20 59543951 00563b68 >> ap_5.05_win_x86_SSE3_r446_DEBUG!MFILE::printf+0x10 >> (d:\r\seti6\boinc\lib\mfile.cpp:98) >> 001112d0 00413d72 5954392d 003d7780 00563b68 0012fd38 >> ap_5.05_win_x86_SSE3_r446_DEBUG!Astropulse::Outfile::write_header+0xb >> (d:\r\seti6\ap\client\ap_fileio.cpp:803) >> 001112f4 004076fb 016df8d8 02a9c000 00370000 00000000 >> ap_5.05_win_x86_SSE3_r446_DEBUG!Astropulse::Outfile::write_main+0x0 >> (d:\r\seti6\ap\client\ap_fileio.cpp:842) >> 00111334 00407f0a 595438b1 00000003 00000000 00000001 >> ap_5.05_win_x86_SSE3_r446_DEBUG!Astropulse::Client::do_checkpoint+0x0 >> (d:\r\seti6\ap\client\astropulse.h:648) >> 0012fd38 00409623 fffffffe 0012ff88 00000001 00000000 >> ap_5.05_win_x86_SSE3_r446_DEBUG!Astropulse::Science::mainloop+0x13 >> (d:\r\seti6\ap\client\ap_client_main.cpp:1680) >> 0012fd54 00409671 00000003 0012fd6c 00272df9 0012fd6c >> ap_5.05_win_x86_SSE3_r446_DEBUG!main+0x0 >> (d:\r\seti6\ap\client\ap_client_main.cpp:572) >> 0012fef8 0050ed56 00400000 00000000 00272def 0000000a >> ap_5.05_win_x86_SSE3_r446_DEBUG!WinMain+0x0 >> (d:\r\seti6\ap\client\ap_client_main.cpp:141) >> 0012ff88 76e0d0e9 7ffdf000 0012ffd4 773919bb 7ffdf000 >> ap_5.05_win_x86_SSE3_r446_DEBUG!__tmainCRTStartup+0x1b >> (f:\dd\vctools\crt_bld\self_x86\crt\src\crt0.c:263) >> 0012ff94 773919bb 7ffdf000 74288bf4 00000000 00000000 >> kernel32!BaseThreadInitThunk+0x0 >> (f:\dd\vctools\crt_bld\self_x86\crt\src\crt0.c:263) >> 0012ffd4 7739198e 0050edc1 7ffdf000 00000000 00000000 >> ntdll!RtlInitializeExceptionChain+0x0 >> (f:\dd\vctools\crt_bld\self_x86\crt\src\crt0.c:263) >> 0012ffec 00000000 0050edc1 7ffdf000 00000000 78746341 >> ntdll!RtlInitializeExceptionChain+0x0 >> (f:\dd\vctools\crt_bld\self_x86\crt\src\crt0.c:263) >> >> static inline char* realloc_aux(char* p, int len) { >> #ifdef _WIN32 >> if (_msize(p)>= len) return p; >> return (char*) realloc(p, len*2); >> #else >> return (char*) realloc(p, len); >> #endif >> } >> >> Any ideas why _msize(p) generates exception here? ... >> >> ----- Original Message ----- >> From: "Rom Walton"<[email protected]> >> To: "Raistmer"<[email protected]>;<[email protected]> >> Sent: Monday, August 02, 2010 4:06 AM >> Subject: RE: [boinc_dev] Incorrect function when running under BOINC >> >> >> Odd, assuming that your variant of the app is setup for static CRT >> compilation it should be working. >> >> The only time I've seen that error before is when a different >> function(Entry Point), from the default name, is specified to launch > the >> application. >> >> Basically when a non-standard entry point is used, the CRT cannot >> prepare the process for the likes of malloc et al. >> >> ----- Rom >> >> -----Original Message----- >> From: Raistmer [mailto:[email protected]] >> Sent: Sunday, August 01, 2010 5:59 PM >> To: Raistmer; Rom Walton; [email protected] >> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >> >> And now it fails with "CRT uninitialized" popup box even in offline >> mode. >> It's regress, not progress :/ >> What BOINC libs should be used? >> >> ----- Original Message ----- >> From: "Raistmer"<[email protected]> >> To: "Raistmer"<[email protected]>; "Rom Walton"<[email protected]>; >> <[email protected]> >> Sent: Monday, August 02, 2010 1:47 AM >> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >> >> >>> I linked vs libboinc_staticcrt.lib and libboincapi_staticcrt.lib >> compiled >>>from current trunk. >>> Now app not only fails, it summons exception data gathering window >> (under >>> Vista) and prompts to launch debugger. >>> In call stack place of failure is in malloc function now %) Same > BOINC >> API >>> call as before... >>> >>> ----- Original Message ----- >>> From: "Raistmer"<[email protected]> >>> To: "Rom Walton"<[email protected]>;<[email protected]> >>> Sent: Monday, August 02, 2010 1:17 AM >>> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >>> >>> >>>> It was 6.10.56 tag. >>>> Will try with trunk then. >>>> >>>> ----- Original Message ----- >>>> From: "Rom Walton"<[email protected]> >>>> To: "Raistmer"<[email protected]>;<[email protected]> >>>> Sent: Monday, August 02, 2010 12:26 AM >>>> Subject: RE: [boinc_dev] Incorrect function when running under BOINC >>>> >>>> >>>> How recent is the BOINC API you are building with? Trunk or some >> other >>>> branch? If some other branch you should switch to using trunk. >>>> >>>> We have had problems in the past with memset calls zeroing out >> complex >>>> data types like std::string and std::vector in the HOST_INFO >> structure. >>>> >>>> ----- Rom >>>> >>>> -----Original Message----- >>>> From: Raistmer [mailto:[email protected]] >>>> Sent: Sunday, August 01, 2010 4:05 PM >>>> To: Rom Walton; [email protected] >>>> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >>>> >>>> Yes, call stack presents: >>>> But it's inside BOINC API call... >>>> 0012ed40 004b8ed3 00000000 00000000 00000000 00000000 >>>> ntdll!DbgBreakPoint+0x0 >>>> >>>> 0012ed58 004e4f2c 00000000 0012ed88 0012ed78 004e4e53 >>>> MB_6.10_win_SSE3_ATI_r103!_invalid_parameter_noinfo+0xc >>>> (f:\dd\vctools\crt_bld\self_x86\crt\src\invarg.c:125) >>>> >>>> 0012ed68 004e4e53 00000000 0012ed88 0012ed90 004e4dc3 >>>> >> > MB_6.10_win_SSE3_ATI_r103!std::_Vector_const_iterator<_BOINC_MESSAGEMONI >>>> TORENTRY >>>> *,std::allocator<_BOINC_MESSAGEMONITORENTRY *> >::operator+=+0x14 >>>> (p:\bin\vs9\vc\include\vector:160) >>>> >>>> 0012ed78 004e4dc3 00000000 0012eda8 00000000 0170ff68 >>>> >> MB_6.10_win_SSE3_ATI_r103!std::_Vector_iterator<_BOINC_THREADLISTENTRY >>>> *,std::allocator<_BOINC_THREADLISTENTRY *> >::operator+=+0x0 >>>> (p:\bin\vs9\vc\include\vector:376) >>>> >>>> 0012ed90 004e48cd 0012eddc 00000000 00000000 005315c0 >>>> >> > MB_6.10_win_SSE3_ATI_r103!std::_Vector_iterator<_BOINC_MESSAGEMONITORENT >>>> RY >>>> *,std::allocator<_BOINC_MESSAGEMONITORENTRY *> >::operator++0xc >>>> (p:\bin\vs9\vc\include\vector:382) >>>> >>>> 0012edbc 004e477c 0012eddc 00000000 00000000 0012ee10 >>>> MB_6.10_win_SSE3_ATI_r103!std::vector<_BOINC_THREADLISTENTRY >>>> *,std::allocator<_BOINC_THREADLISTENTRY *> >::insert+0x1b >>>> (p:\bin\vs9\vc\include\vector:878) >>>> >>>> 0012edf4 004ef52d 0012ee10 6a5888c6 001b9860 0012f078 >>>> MB_6.10_win_SSE3_ATI_r103!std::vector<_BOINC_MESSAGEMONITORENTRY >>>> *,std::allocator<_BOINC_MESSAGEMONITORENTRY *> >::push_back+0x0 >>>> (p:\bin\vs9\vc\include\vector:824) >>>> >>>> 0012f24c 004eb0d6 0012f6b0 00530a30 20202020 706f633c >>>> MB_6.10_win_SSE3_ATI_r103!COPROCS::parse+0x0 >>>> (d:\r\seti6\boinc\lib\coproc.cpp:140) >>>> >>>> 0012f664 004e0762 0012f6b0 00000000 6a58806a 0042005c >>>> MB_6.10_win_SSE3_ATI_r103!HOST_INFO::parse+0x0 >>>> (d:\r\seti6\boinc\lib\hostinfo.cpp:111) >>>> >>>> 0012fae0 004f151a 00525ed8 005300e8 00525ed8 00525e98 >>>> MB_6.10_win_SSE3_ATI_r103!parse_init_data_file+0x0 >>>> (d:\r\seti6\boinc\lib\app_ipc.cpp:297) >>>> >>>> 0012faf8 0040d74b 00525eb8 00525e98 0170f7b0 00000000 >>>> MB_6.10_win_SSE3_ATI_r103!_boinc_parse_init_data_file+0x0 >>>> (d:\r\seti6\boinc\lib\app_ipc.cpp:297) >>>> >>>> ----- Original Message ----- >>>> From: "Rom Walton"<[email protected]> >>>> To: "Raistmer"<[email protected]>;<[email protected]> >>>> Sent: Sunday, August 01, 2010 11:28 PM >>>> Subject: RE: [boinc_dev] Incorrect function when running under BOINC >>>> >>>> >>>> Well 0x80000003 is what error code is returned when an application >>>> asserts, which is different than the original error code. >>>> >>>> Both of these lines are generated by the C Runtime Library and > should >> be >>>> properly filled out by rebuilding the application in the debug >>>> configuration. >>>> ERROR: Invalid parameter detected in function (null). File: (null) >> Line: >>>> 0? >>>> ERROR: Expression: (null)? >>>> >>>> However if the app did assert and returned 0x80000003 it should have >>>> also returned a crash report which should include the call stacks of >> all >>>> the threads which should give you the same information that the two >>>> ERROR lines above would give. >>>> >>>> ----- Rom >>>> >>>> -----Original Message----- >>>> From: Raistmer [mailto:[email protected]] >>>> Sent: Sunday, August 01, 2010 3:12 PM >>>> To: Raistmer; Rom Walton; [email protected] >>>> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >>>> >>>> Now I advanced further and got error in BOINC function call: >>>> >>>> Code: >>>> // Initialize BOINC >>>> // >>>> fprintf(stderr,"before boinc_parse_init_data_file\n"); >>>> boinc_parse_init_data_file(); >>>> fprintf(stderr,"before boinc_get_init_data\n"); >>>> >>>> STDERR: >>>> <![CDATA[ >>>> <message> >>>> ????????? ?????? ??????????? (0x80000003) - exit code -2147483645 >>>> (0x80000003) >>>> </message> >>>> <stderr_txt> >>>> after diagnstic init? >>>> before boinc_parse_init_data_file? >>>> ERROR: Invalid parameter detected in function (null). File: (null) >> Line: >>>> 0? >>>> ERROR: Expression: (null)? >>>> >>>> What could be wrong there? >>>> >>>> ----- Original Message ----- >>>> From: "Raistmer"<[email protected]> >>>> To: "Raistmer"<[email protected]>; "Rom Walton"<[email protected]>; >>>> <[email protected]> >>>> Sent: Saturday, July 24, 2010 10:31 PM >>>> Subject: Re: [boinc_dev] Incorrect function when running under BOINC >>>> >>>> >>>>> Well, i added few fprintfs just after init diagnostics, boinc init >>>> call >>>>> and so on. >>>>> there is nothing in stderr again. Probably it means very first > boinc >>>> api >>>>> call, init diagnostic, failed and caused exit(1) call. Any thoughts >>>> why it >>>>> could fail? >>>>> >>>>> ----- Original Message ----- >>>>> From: "Raistmer"<[email protected]> >>>>> To: "Rom Walton"<[email protected]>;<[email protected]> >>>>> Sent: Tuesday, July 20, 2010 12:12 PM >>>>> Subject: Re: [boinc_dev] Incorrect function when running under > BOINC >>>>> >>>>> >>>>>> Thanks! >>>>>> What I found: >>>>>> >>>>>> 0) >>>>>> if (err != CL_SUCCESS) { >>>>>> std::cerr<< "ERROR: "<< name<< " ("<< err<< ")"<< >>>>>> std::endl; >>>>>> exit(1); >>>>>> } >>>>>> Not th case, no "ERROR" in stderr. >>>>>> 1) >>>>>> fprintf(stderr, "bad arg: %s\n", argv[i]); >>>>>> usage(); >>>>>> exit(1); >>>>>> Not the case cause no "bad arg" in stderr >>>>>> >>>>>> 2) >>>>>> PowerSpectrum = (float*) calloc_a(NumPointsInChunk, >> sizeof(float), >>>>>> MEM_ALIGN); >>>>>> if (PowerSpectrum == NULL) { >>>>>> printf("Could not allocate Power Spectrum array in >>>>>> v_BaseLineSmooth()\n"); >>>>>> exit(1); >>>>>> } >>>>>> >>>>>> can't say if it's what I looking for cause additional output goes >> to >>>> >>>>>> stdout >>>>>> instead of stderr, need to rebuild to check... >>>>>> >>>>>> But if it's very place, why buffer allocation fails under BOINC >> while >>>> it >>>>>> goes OK w/o it? >>>>>> >>>>>> ----- Original Message ----- >>>>>> From: "Rom Walton"<[email protected]> >>>>>> To: "Raistmer"<[email protected]>;<[email protected]> >>>>>> Sent: Tuesday, July 20, 2010 11:47 AM >>>>>> Subject: RE: [boinc_dev] Incorrect function when running under >> BOINC >>>>>> >>>>>> >>>>>> Search the source code of the app for exit(1) or > exit(EXIT_FAILURE) >>>>>> >>>>>> ----- Rom >>>>>> >>>>>> -----Original Message----- >>>>>> From: [email protected] >>>>>> [mailto:[email protected]] On Behalf Of Raistmer >>>>>> Sent: Tuesday, July 20, 2010 3:41 AM >>>>>> To: [email protected] >>>>>> Subject: [boinc_dev] Incorrect function when running under BOINC >>>>>> >>>>>> When running offline app works OK, but being launched by BOINC it >>>> fails >>>>>> with >>>>>> "incorrect function exit code 0x1" >>>>>> Stderr: >>>>>> <stderr_out> >>>>>> <![CDATA[ >>>>>> <message> >>>>>> =ooo?y? ??yu?o . (0x1) - exit code 1 (0x1) >>>>>> </message> >>>>>> ]]> >>>>>> </stderr_out> >>>>>> >>>>>> No additional info available. >>>>>> Maybe there are some known possible reasons of such behavior? >> Someone >>>>>> encountered such situation? >>>>>> >>>>>> _______________________________________________ >>>>>> boinc_dev mailing list >>>>>> [email protected] >>>>>> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >>>>>> To unsubscribe, visit the above URL and >>>>>> (near bottom of page) enter your email address. >>>>>> >>>>>> _______________________________________________ >>>>>> boinc_dev mailing list >>>>>> [email protected] >>>>>> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >>>>>> To unsubscribe, visit the above URL and >>>>>> (near bottom of page) enter your email address. >>>>>> >>>>> >>>>> >>>> >>>> _______________________________________________ >>>> boinc_dev mailing list >>>> [email protected] >>>> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >>>> To unsubscribe, visit the above URL and >>>> (near bottom of page) enter your email address. >>>> >>> >>> >> _______________________________________________ >> boinc_dev mailing list >> [email protected] >> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >> To unsubscribe, visit the above URL and >> (near bottom of page) enter your email address. > _______________________________________________ > boinc_dev mailing list > [email protected] > http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev > To unsubscribe, visit the above URL and > (near bottom of page) enter your email address. > _______________________________________________ > boinc_dev mailing list > [email protected] > http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev > To unsubscribe, visit the above URL and > (near bottom of page) enter your email address. > > _______________________________________________ > boinc_dev mailing list > [email protected] > http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev > To unsubscribe, visit the above URL and > (near bottom of page) enter your email address. > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
