Bypass override of operator new in C++ -

is there way achieve bypass of override of operator new?

something this:

void* ::operator new( std::size_t size ) {     void *p = ( ::operator new( size ) );  // original, _not_ infinite recursion     // stuff p     return p;  } 

background: have legacy code, switched compile visual studio 2012. random crashes when malloc fails _heap_alloc sufficient memory blocks. (yes, code sprinkled small memory leaks , other bad behaviour over. unfortunately thorough cleanup not realistic, somewhere around 500 000 sloc.)

my current theory cause source files include header following overrides of operator new:

void* ::operator new( std::size_t size ) {     void* p = malloc( size );     if( p == null )         throw;     // set memory 0     memset( p, 0, size );     return p; };  void* ::operator new[]( size_t count ) throw(std::bad_alloc) {     // try allocate count bytes array     return (operator new(count)); } 

there no overrides of delete.

in essence, means application mixes allocation using malloc deallocation using delete instead of free.

first try q&d fix: introduce overrides of delete uses free. partly helps, because include order , linked libraries still messed up.

second try q&d fix: remove override. initialization of memory 0 unfortunately necessary. heritage old compiler used, did that, , coders assumed c++ that.

i aware new() take care of that, unfortunately not aware of method make use of without manually going through source code , updating it. won't poorly implemented classes assumes members nullified without doing in constructor.

therefore, third idea of q&d fix: using normal new in override, question asks about.

is there way achieve bypass of override of operator new?

none i'm aware of, why don't re-implement in custom function in same manner ms did in _heap_alloc-based implementation , add customization it?

one do-or-die allocation attempt anyway not ms does:

http://msdn.microsoft.com/en-us/library/vstudio/we2zys4d.aspx

the default behavior execute loop. within loop, function first attempts allocate requested storage

anyway- second n.m. put in comment

you need fix real memory bugs

that self-explaning; debugging purposes - pls have @ std::set_new_handler.

btw - there things find irritating code

• the explicit qualifications of operator
• not throwing std::bad_alloc re-throwing outside of catch-block

happy hacking!

regards, s.

p.s.:@all: cannot "vote up" question due limited exp here, recommend doing that. it's interesting topic , formulated.