BorlandTalk.com Borland discussion newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Q: how to identify _pure_error_ culprit          BorlandTalk.com Forum Index -> C++ Builder (IDE) View previous topic :: View next topic   Author Message dhoke Guest Posted: Tue May 01, 2007 12:53 am    Post subject: Q: how to identify _pure_error_ culprit (bds2006) I have just integrated several basically inert protyped forms (wrt to form activity - many fields on some, field data processing/activities not implemented) into a project. A fresh open of the project, followed by F9 run has failed, yielding the following stack: :32702227 std::locale::id::id(this=:0066BA7C, _Val=0) :00652a71 ; _pure_error_ :32770960 _startup(mod_table=:0066C034) and displays the following code (fragment): class _CRTIMP2 locale : public _Locbase<int> { // nonmutable collection of facets that describe a locale public: typedef int category; // CLASS id class _CRTIMP2 id { // identifier stamp, unique for each distinct kind of facet public: id(size_t _Val = 0) : _Id(_Val) { // construct with specified stamp value } operator size_t() { // get stamp, with lazy allocation ========================== with current line: : _Id(_Val) I take it the clues tell me there is an attempt to call a pure virtual function, seemingly in the library initialization. I've examined all of the forms, and as stated, they are (mostly) empty (not in terms of fields, but in terms of any actual processing, or auxiliary data/classes), I believe with nothing but standard B/CG form controls, with nothing in their "domain" involving virtual functions that I can spot. How does one identify who/what is the actual culprit? Thanks. Back to top dhoke Guest Posted: Tue May 01, 2007 1:10 am    Post subject: Re: how to identify _pure_error_ culprit I have just removed all of the recently added forms, adding them back one at a time and running after each one was added. They are not added in the same order they were originally (when all added at once.) The error does not occur until after I have added the final form, so I presume I have the culprit form. However, this form has no members that are not standard B/CG tool palette items, and has no code except for an empty constructor. ***Any reasonable guesses, or certainties, as to why this error is occurring? Thanks. "dhoke" <dhoke.nojunk@east-shore.com> wrote in message news:46364996$1 (AT) newsgroups (DOT) borland.com... Quote: (bds2006) I have just integrated several basically inert protyped forms (wrt to form activity - many fields on some, field data processing/activities not implemented) into a project. A fresh open of the project, followed by F9 run has failed, yielding the following stack: :32702227 std::locale::id::id(this=:0066BA7C, _Val=0) :00652a71 ; _pure_error_ :32770960 _startup(mod_table=:0066C034) and displays the following code (fragment): class _CRTIMP2 locale : public _Locbase<int { // nonmutable collection of facets that describe a locale public: typedef int category; // CLASS id class _CRTIMP2 id { // identifier stamp, unique for each distinct kind of facet public: id(size_t _Val = 0) : _Id(_Val) { // construct with specified stamp value } operator size_t() { // get stamp, with lazy allocation ========================== with current line: : _Id(_Val) I take it the clues tell me there is an attempt to call a pure virtual function, seemingly in the library initialization. I've examined all of the forms, and as stated, they are (mostly) empty (not in terms of fields, but in terms of any actual processing, or auxiliary data/classes), I believe with nothing but standard B/CG form controls, with nothing in their "domain" involving virtual functions that I can spot. How does one identify who/what is the actual culprit? Thanks. Back to top dhoke Guest Posted: Tue May 01, 2007 1:29 am    Post subject: Re: how to identify _pure_error_ culprit I have found that the "this" addresss passed in as the address of the local being initialized, appears to point into a jump table, a piece of which follows (we appear to be attempting to initialize a size_t _ID located @0x66b5f4 [in the shown instance]): 0066B5F4 FF25A0536A00 jmp dword ptr [$006a53a0] 0066B5FA FF25A4536A00 jmp dword ptr [$006a53a4] 0066B600 FF25A8536A00 jmp dword ptr [$006a53a8] 0066B606 FF25AC536A00 jmp dword ptr [$006a53ac] 0066B60C FF25B0536A00 jmp dword ptr [$006a53b0] 0066B612 FF25B4536A00 jmp dword ptr [$006a53b4] 0066B618 FF25B8536A00 jmp dword ptr [$006a53b8] 0066B61E FF25BC536A00 jmp dword ptr [$006a53bc] 0066B624 FF25C0536A00 jmp dword ptr [$006a53c0] 0066B62A FF25C4536A00 jmp dword ptr [$006a53c4] 0066B630 FF25C8536A00 jmp dword ptr [$006a53c8] 0066B636 FF25CC536A00 jmp dword ptr [$006a53cc] 0066B63C FF25D0536A00 jmp dword ptr [$006a53d0] 0066B642 FF25D4536A00 jmp dword ptr [$006a53d4] 0066B648 FF25D8536A00 jmp dword ptr [$006a53d8] 0066B64E FF25DC536A00 jmp dword ptr [$006a53dc] the stack view is similar to before (have finally gotten a slightly different address for "this"): :32702227 std::locale::id::id(this=:0066B5F4, _Val=0) :006525f5 ; _pure_error_ :32770960 _startup(mod_table=:0066C034) Ideas for solutions/workarounds still sought... Thanks. "dhoke" <dhoke.nojunk@east-shore.com> wrote in message news:46364d95$1 (AT) newsgroups (DOT) borland.com... Quote: I have just removed all of the recently added forms, adding them back one at a time and running after each one was added. They are not added in the same order they were originally (when all added at once.) The error does not occur until after I have added the final form, so I presume I have the culprit form. However, this form has no members that are not standard B/CG tool palette items, and has no code except for an empty constructor. ***Any reasonable guesses, or certainties, as to why this error is occurring? Thanks. "dhoke" <dhoke.nojunk@east-shore.com> wrote in message news:46364996$1 (AT) newsgroups (DOT) borland.com... (bds2006) I have just integrated several basically inert protyped forms (wrt to form activity - many fields on some, field data processing/activities not implemented) into a project. A fresh open of the project, followed by F9 run has failed, yielding the following stack: :32702227 std::locale::id::id(this=:0066BA7C, _Val=0) :00652a71 ; _pure_error_ :32770960 _startup(mod_table=:0066C034) and displays the following code (fragment): class _CRTIMP2 locale : public _Locbase<int { // nonmutable collection of facets that describe a locale public: typedef int category; // CLASS id class _CRTIMP2 id { // identifier stamp, unique for each distinct kind of facet public: id(size_t _Val = 0) : _Id(_Val) { // construct with specified stamp value } operator size_t() { // get stamp, with lazy allocation ========================== with current line: : _Id(_Val) I take it the clues tell me there is an attempt to call a pure virtual function, seemingly in the library initialization. I've examined all of the forms, and as stated, they are (mostly) empty (not in terms of fields, but in terms of any actual processing, or auxiliary data/classes), I believe with nothing but standard B/CG form controls, with nothing in their "domain" involving virtual functions that I can spot. How does one identify who/what is the actual culprit? Thanks. Back to top dhoke Guest Posted: Tue May 01, 2007 7:04 pm    Post subject: Re: how to identify _pure_error_ culprit Update (news is not good): SUMMARY It appears the linker has produced incorrect code. EXPLORATION DETAILS The "_pure_error_" shown in the stack is apparently the last public symbol in the module cpprtl\Source\misc\pureerr.cpp, which is linked in _before_ the module containing the offending code. (The actual routine name is apparently "pure_error()".) It was a misleading clue, apart from offering some positioning information when I identified it. The apparently incorrect code from the debugger CPU pane is: 006525E0 A034D16900 mov al,[$0069d134] 006525E5 84C0 test al,al 006525E7 7516 jnz $006525ff 006525E9 6A00 push $00 006525EB 68F4B56600 push $0066b5f4 006525F0 E889900100 call $0066b67e 006525F5 83C408 add esp,$08 006525F8 C60534D1690001 mov byte ptr [$0069d134],$01 006525FF C3 ret The offending code appears to be contained-in/linked-from a library I have indirect access to (no control over source), although it appears the code itself is somehow attributable to B/CG runtime library source contained in "BDS\4.0\include\dinkumware\iosfwd" (yes, it has no extension.) I have not yet spotted what in that source might cause the generation of this little stub of code - although it includes several headers which I haven't yet explored. Note that this library itself is the same binary instance in all of the build link attempts I ran, when I removed and re-added the new forms trying to "isolate" the error. The form that "causes" the error, is empty, and does not reference anything from the library, so I do not see how the forms presence can be a trigger other than by its size/location in the linked executable - nothing otherwise to do with the problem code. (I have not tried re-arranging the forms to see if the particular form really is insignificant.) (I note that in the map file, there are several apparently generated symbols of the form "cnnnn_n", and more than one of them apparently reference this particular code snippet. These symbols do not seem to have debugger visibility, as I have not been able to "goto" them by name with the goto address functionality.) A snippet of tdump output of the library containing the apparent code shows the following: ================================================== ; Module: c:\program files\borland\bds\4.0\include\dinkumware\iosfwd ; Disassembly of segments: _TEXT SEGMENT DWORD PUBLIC USE32 'CODE' ; SegID: 1 offsets: 0000 - 001F ; 00000000 A0 00 00 00 00 mov al,[#{1431}...] 00000005 84 C0 test al,al 00000007 75 16 jnz +0x16 00000009 6A 00 push 0x00 0000000B 68 00 00 00 00 push 0x00000000 00000010 E8 00 00 00 00 call std::locale::id::id(unsigned int) 00000015 83 C4 08 add esp,0x08 00000018 C6 05 00 00 00 00 01 mov byte ptr [#{1431}...],0x01 0000001F C3 ret ; _TEXT ENDS ; SegID: 1 offsets: 0000 - 001F _DATA SEGMENT DWORD PUBLIC USE32 'DATA' ; SegID: 2 offsets: 0000 - 012B ; ; the dump of this data segment has been omitted ; _DATA ENDS ; SegID: 2 offsets: 0000 - 012B ================================================== which appears to match the offending code from the debugger disassembly. The library code has zeroes (0) in the instruction values at 0x9 and 0xb, which are respectively 0 and 0x66b5f4 (at 0x6525e9 and 0x6525eb) in the linked executable code. The disassembly of memory at 0x66b5f4 appears thus in the debugger: 0066B5F4 FF25A0536A00 jmp dword ptr [$006a53a0] 0066B5FA FF25A4536A00 jmp dword ptr [$006a53a4] 0066B600 FF25A8536A00 jmp dword ptr [$006a53a8] 0066B606 FF25AC536A00 jmp dword ptr [$006a53ac] 0066B60C FF25B0536A00 jmp dword ptr [$006a53b0] 0066B612 FF25B4536A00 jmp dword ptr [$006a53b4] 0066B618 FF25B8536A00 jmp dword ptr [$006a53b8] 0066B61E FF25BC536A00 jmp dword ptr [$006a53bc] 0066B624 FF25C0536A00 jmp dword ptr [$006a53c0] 0066B62A FF25C4536A00 jmp dword ptr [$006a53c4] 0066B630 FF25C8536A00 jmp dword ptr [$006a53c8] 0066B636 FF25CC536A00 jmp dword ptr [$006a53cc] 0066B63C FF25D0536A00 jmp dword ptr [$006a53d0] 0066B642 FF25D4536A00 jmp dword ptr [$006a53d4] 0066B648 FF25D8536A00 jmp dword ptr [$006a53d8] 0066B64E FF25DC536A00 jmp dword ptr [$006a53dc] 0066B654 FF25E0536A00 jmp dword ptr [$006a53e0] 0066B65A FF25E4536A00 jmp dword ptr [$006a53e4] 0066B660 FF25E8536A00 jmp dword ptr [$006a53e8] 0066B666 FF25EC536A00 jmp dword ptr [$006a53ec] etc., which very much appears to be a jump table. Interestingly (at least to me), if one "follows" the jump at 0x66b5f4, the disassembly there appears as: 3277F2F8 0000 add [eax],al 3277F2FA 0000 add [eax],al 3277F2FC 019090900000 add [eax+$00009090],edx 3277F302 0000 add [eax],al 3277F304 019090900000 add [eax+$00009090],edx 3277F30A 0000 add [eax],al 3277F30C 019090900000 add [eax+$00009090],edx 3277F312 0000 add [eax],al 3277F314 019090900000 add [eax+$00009090],edx 3277F31A 0000 add [eax],al 3277F31C 019090900000 add [eax+$00009090],edx 3277F322 0000 add [eax],al etc., which looks considerably more like data... Is that the address the code should have had? The code reached from the call at 0x6525f0 is an indirect jump (in a jump table ), which apparently goes to a constructor for locale::id. That disassembled code is as follows: xlocale.64: id(size_t _Val = 0) 32702210 55 push ebp 32702211 8BEC mov ebp,esp 32702213 83C4DC add esp,-$24 32702216 53 push ebx 32702217 8B5D08 mov ebx,[ebp+$08] 3270221A B830F37732 mov eax,$3277f330 3270221F E8CCDA0400 call __InitExceptBlockLDTC(void *) xlocale.65: : _Id(_Val) 32702224 8B550C mov edx,[ebp+$0c] 32702227 8913 mov [ebx],edx xlocale.67: } 32702229 8B4DDC mov ecx,[ebp-$24] 3270222C 64890D00000000 mov fs:[$00000000],ecx 32702233 8BC3 mov eax,ebx xlocale.67: } 32702235 5B pop ebx 32702236 8BE5 mov esp,ebp 32702238 5D pop ebp 32702239 C3 ret At 0x32702217 This code picks up the bogus value 0x66b5f4, which was pushed back at addr. 0x6525eb. At 0x32702227, the code attempts to store to that address. (presumably because of DEP - and I think correctly so) the code generates an access violation trying to store the _Val constructor parameter into that location, which is code, not data. Conclusion - it appears to me that a piece of (B/CG [dinkumware] library?) generated code (generated in code that is part of a library we create) has been linked improperly - The 0x66b5f4 pushed at 0x6525eb should apparently be the address of a "locale::id" entity, and it is instead the address in the middle of a executable code jumptable. (There are many similar jump instructions preceding the particular address.) "dhoke" <dhoke.nojunk@east-shore.com> wrote in message news:46364996$1 (AT) newsgroups (DOT) borland.com... Quote: (bds2006) I have just integrated several basically inert protyped forms (wrt to form activity - many fields on some, field data processing/activities not implemented) into a project. A fresh open of the project, followed by F9 run has failed, yielding the following stack: :32702227 std::locale::id::id(this=:0066BA7C, _Val=0) :00652a71 ; _pure_error_ :32770960 _startup(mod_table=:0066C034) and displays the following code (fragment): class _CRTIMP2 locale : public _Locbase<int { // nonmutable collection of facets that describe a locale public: typedef int category; // CLASS id class _CRTIMP2 id { // identifier stamp, unique for each distinct kind of facet public: id(size_t _Val = 0) : _Id(_Val) { // construct with specified stamp value } operator size_t() { // get stamp, with lazy allocation ========================== with current line: : _Id(_Val) I take it the clues tell me there is an attempt to call a pure virtual function, seemingly in the library initialization. I've examined all of the forms, and as stated, they are (mostly) empty (not in terms of fields, but in terms of any actual processing, or auxiliary data/classes), I believe with nothing but standard B/CG form controls, with nothing in their "domain" involving virtual functions that I can spot. How does one identify who/what is the actual culprit? Thanks. Back to top dhoke Guest Posted: Tue May 01, 2007 8:13 pm    Post subject: Re: how to identify _pure_error_ culprit I have now tried re-arranging the forms, removing one at a time, re-adding. It seems that if I remove any one of the five recently added, the executable will startup without crashing. If I have all five, executable crashes before reaching WinMain(). "dhoke" <dhoke.nojunk@east-shore.com> wrote in message news:4637494a (AT) newsgroups (DOT) borland.com... Quote: Update (news is not good): (I have not tried re-arranging the forms to see if the particular form really is insignificant.) Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         BorlandTalk.com Forum Index -> C++ Builder (IDE) All times are GMT Page 1 of 1   Jump to: Select a forum C++ Builder Newsgroups----------------C++ BuilderC++ Builder (ActiveX)C++ Builder (Announcements)C++ Builder (Command Line Tools)C++ Builder DatabasesC++ Builder Databases (ADO)C++ Builder Databases (Desktop)C++ Builder Databases (InterBase Express)C++ Builder Databases (Multi-Tier)C++ Builder Databases (SQL Servers)C++ Builder (Graphics)C++ Builder (IDE)C++ Builder (Internet)C++ Builder (Internet Socket)C++ Builder (Internet Web)C++ Builder (Jobs)C++ Builder (Language)C++ Builder (Language C++)C++ Builder (Compatibility with MFC)C++ Builder (Native API)C++ Builder (Non-Technical)C++ Builder (Students)C++ Builder (Third Party Tools)C++ Builder (Upgrade)C++ Builder (VCL)C++ Builder (VCL Components Usage)C++ Builder (VCL Components Development)C++ Builder (VCL Usage)C++ Builder (WinAPI)C++ Builder Installation Delphi Newsgroups----------------Delphi (General)Delphi ActiveX Controls UsageDelphi ActiveX Controls DevelopmentDelphi As400Delphi DatabasesDelphi Databases (ADO)Delphi Databases (dbExpress)Delphi Databases (Desktop)Delphi Databases (InterBase Express)Delphi Databases (Multi-Tier)Delphi Databases (SQL Servers)Delphi DeploymentDelphi GraphicsDelphi IDEDelphi InternationalizationDelphi InternetDelphi Internet Isapi-WebBrokerDelphi Internet WebSnapDelphi Internet WinsockDelphi JobsDelphi Language BASMDelphi Language ObjectPascalDelphi Model Driven Architecture (General)Delphi Model Driven Architecture (Third Party)Delphi MultimediaDelphi Native APIDelphi Non-TechnicalDelphi ObjectPascalDelphi OLE AutomationDelphi OO designDelphi OpenToolsAPIDelphi Reporting-ChartingDelphi RTLDelphi StudentsDelphi TeamSourceDelphi Thirdparty ToolsDelphi Thirdparty Tools (General)Delphi Thirdparty Tools (IntraWeb)Delphi Thirdparty Tools (RaveReports)Delphi UpgradeDelphi VCLDelphi VCL ComponentsDelphi VCL Components UsingDelphi VCL Components WritingDelphi WebServices SOAPDelphi WinAPIDelphi XMLDelphi Installationcomp.lang.pascal.delphi.components.misccomp.lang.pascal.delphi.components.usagecomp.lang.pascal.delphi.components.writingcomp.lang.pascal.delphi.databasescomp.lang.pascal.delphi.misccomp.sources.delphi JBuilder Newsgroups----------------JBuilderJBuilder AnnouncementsJBuilder Applet IssuesJBuilder As400JBuilder CompilerJBuilder CORBA/RMIJBuilder DatabasesJBuilder DB SwingJBuilder DebuggerJBuilder DeploymentJBuilder DocumentationJBuilder EnterpriseJBuilder HandheldJBuilder IDEJBuilder JavaJBuilder JavaAPIJBuilder Java LanguageJBuilder Java SwingJBuilder Java BeansJBuilder Java Beans UsageJBuilder Java Beans DevelopmentJBuilder JBCLJBuilder JDataStoreJBuilder JobsJBuilder LinuxJBuilder MacintoshJBuilder & MobileSetJBuilder MultiLingual AppsJBuilder Non-TechnicalJBuilder OpenToolsJBuilder Servlets (JSP)JBuilder StudentsJBuilder Team DevelopmentJBuilder Thirdparty ToolsJBuilder UpgradeJBuilder Installation InterBase Newsgroups----------------InterBaseInterBase Bug DiscussionInterBase GeneralInterBase IBConsoleInterBase OpenSourceInterBase ReplicationInterBase SQLInterBase ThirdpartyInterBase Installation Miscellaneous Borland Newsgroups----------------AppCenterAppServerEnteraVisiBroker  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.