Consider the following code:
// A non-copyable, non-movable aggregate
struct Strange
{
const int & i;
char & c;
};
class Container
{
private:
int my_i;
char my_c;
Strange thing;
public:
// Valid, because both `my_i´ and `my_c´ are non-const
// objects to which both references can be bound.
explicit
Container
( )
noexcept
: thing{ my_i , my_c }
{ }
// How could this be implemented?
auto &
operator=
( const Container & that )
noexcept
{
this->my_i = that->my_i;
this->my_c = that->my_c;
// What to do with `thing´?
return *this;
}
};
Possible solutions
Dynamically allocate the Strange object
class Container
{
private:
int my_i;
char my_c;
Strange * thing;
public:
// Note that it isn't exception safe.
explicit
Container
( )
: thing(new Strange{ my_i , my_c })
{ }
auto &
operator=
( const Container & that )
noexcept
{
this->my_i = that->my_i;
this->my_c = that->my_c;
delete this->thing;
this->thing = new Strange { this->my_i , this->my_c };
return *this;
}
};
Concerns:
Not efficient.
Not safe: allocation may fail and throw.
Dangerous: great care has to be taken not to leak memory.
Using a smart pointer (i.e. std::unique_ptr) would solve only the last point, aside from making the code more readable.
Use placement new
class Container
{
private:
int my_i;
char my_c;
Strange thing;
public:
explicit
Container
( )
noexcept
: thing{ my_i , my_c }
{ }
auto &
operator=
( const Container & that )
noexcept
{
this->my_i = that.my_i;
this->my_c = that.my_c;
// Placement new is exception safe, and so is
// construction of `Strange´.
this->thing.~Strange();
new(&this->thing) Strange { this->my_i , this->my_c };
return *this;
}
};
Concerns:
Will Strange's destructor free the memory taken by thing?
I think that, just as constructors, destructors are not responsible for memory management. Moreover, my code seems to work fine. However, I would like clarification on this.
What about memory alignment?
My guess is that, since it replaces an existing object of the same type, memory would already be aligned. Is this correct?
Would Container's destructor take care of destructing thing?
Questions
Aside from corroborating and/or refuting the concerns I explained above, I would like to know if there are other alternatives. If so, please, give an example implementation.
This question arose when working on a class that ought to offer an interface similar to that of std::unordered_map. Instead of reimplementing it, my class encapsulates such container, and just acts as a proxy for most methods: its iterators envelop those provided by the map, and its pair is an aggregate structure with appropiately named members (which are references to the actual data), represented in the examples provided as Strange. Since iterators are required to return references and pointers to the actual data, my custom iterators contain a pair. The problem was modifying it (when incrementing or assigning the iterator). I acknowledge that this probably isn't a good idea, and that those references will impact performance, but am anyway interested in the matter.
Edit
I have just realized that, instead of returning references and pointers to a member custom pair pointing to the actual data (that of the encapsulated map) from my custom iterator, I could return in-place constructed custom pairs (i.e. Strange objects). Often, we do not see that we are in a cave, and instead of exiting it, continue forward :). Excuse me for the noise, I'll mark the question as "Closed".
(If we're talking about moving objects and using the auto keyword, you should add the c++11 tag to your question).
I'm not sure I really understand your question; the example you gave doesn't seem well thought to me; it would be much better to use pointers in Strange.
For example, this compiles and works absolutely fine, and is functionally equivalent to what you want to do I think.
struct Strange
{
Strange()
: i(nullptr), c(nullptr) {}
Strange( const int *_i, const char *_c )
: i(_i), c(_c) {}
const int *i;
const char *c;
};
class Container
{
int my_i;
char my_c;
Strange thing;
public:
Container()
: thing(&my_i,&my_c)
{ }
Container( int i, char c )
: my_i(i), my_c(c), thing(&my_i,&my_c)
{ }
Container( int i, char c, const Strange& s )
: my_i(i), my_c(c), thing(s) // use default copy-constructor
{ }
Container &
operator=
( const Container & that )
{
my_i = that.my_i;
my_c = that.my_c;
thing = that.thing;
return *this;
}
};
int main()
{
Container a(12,24);
Container b(25,42);
b = a;
}
Note that referencing memory within objects is generally dangerous.
For example, using memcpy on this would be a catastrophy.
(Compiled with clang and g++)
Related
if we have a code like this :
void g(int *n){
int m=n[1];
n= null ;
}
void main(){
int ∗ p = (int∗)malloc(10 ∗ sizeof (int));
int * q= p;
g(p);
}
so we know if we overload malloc, calloc , realloc ,new ,free and delete functions we can track first pointer that create or delete with this functions(p pointer in above example ) but how can i find and track other pointers and functions that using this allocated memory ? ( q pointer and g function in above example ) .should i overload Assignment statement and function call ? if yes how ? in other words i want to know live objects and last used time and location of an allocated memory too .
i want to implement an custom memory leak detection tools so i need to find all objects and pointer that using an allocated memory before report it that's leak or not .
What you're talking about is called reference counting. Searching stackoverflow for this topic gives you these results:
What is a smart pointer and when should I use one?
How does a reference-counting smart pointer's reference counting work?
what exactly reference counting in c++ means?,
The standard template library already has a std::shared_ptr which does that.
We need to keep track of the lifecycle of the resource, this is possible by controlling the creation, the copying and the deletion. If you do it without using a class, you'll end up with some functions, a state variable and a global variable. This is not very effective in keeping the concept of the shared resource in the focus of the user's mind. One tends to forget that one is using a shared resource because one sees a naked pointer and tends to use it like one, which would disregard the provided shared-pointer functionality.
If you were to encapsulate the functionality in a class, you should want to implement the concept for all types i.e. you should want to use templates. one way would be this:
#include <vector>
#include <stdexcept>
#include <cstdlib>
#ifndef SHARED_PTR_H
#define SHARED_PTR_H
template<class Tres_t>
class shared_pointer
{
// members
Tres_t* data_;
static std::vector<std::size_t> ref_counts_;
std::size_t ref_index_ = {};
public:
// lifecycle
shared_pointer() = delete;
shared_pointer(const std::size_t& size)
: data_{nullptr}
{
data_ = static_cast<Tres_t*>(std::malloc(size * sizeof(Tres_t)));
ref_counts_.push_back(1);
ref_index_ = ref_counts_.size() - 1;
}
shared_pointer(const shared_pointer& rhs)
: data_{rhs.data_}, ref_index_{rhs.ref_index_}
{
if (ref_index_ >= ref_counts_.size())
{
throw std::runtime_error("shared_pointer_ctor_index_error");
}
++ref_counts_[ref_index_];
}
shared_pointer(shared_pointer&& rhs)
: data_{rhs.data_}, ref_index_{rhs.ref_index_} {}
shared_pointer& operator=(const shared_pointer& rhs)
{
data_ = rhs.data_;
ref_index_ = rhs.ref_index_;
if (ref_index_ >= ref_counts_.size())
{
throw std::runtime_error("shared_pointer_ctor_index_error");
}
++ref_counts_[ref_index_];
}
shared_pointer& operator=(shared_pointer&& rhs)
{
data_ = rhs.data_;
ref_index_ = rhs.ref_index_;
}
~shared_pointer()
{
if (ref_counts_[ref_index_] == 0)
{
std::logic_error("shared_point_dtor_reference_counting_error");
}
--ref_counts_[ref_index_];
if (ref_counts_[ref_index_] == 0)
{
std::free(data_);
}
}
// main functionality
Tres_t* data()
{
return data_;
}
const Tres_t* data() const
{
return data_;
}
};
template<class Tres_t>
std::vector<std::size_t> shared_pointer<Tres_t>::ref_counts_ = {};
#endif
I've tested this code only rudimentarily so I'll leave it to you to test and improve it.
Say I have a class:
class A
{
private:
const int * const v;
public:
A();
}
I want v to be allocated in the initialization list, and I think I can define the following constructor:
A::A():v((int*)malloc(10*sizeof(int))){}
However, what about v has to be allocated in a non-standard way like the following:
cudaMalloc(&v,10*sizeof(int));
Note cudaMalloc is a CUDA API to allocate GPU memory.
(Ignoring the bigger-picture matters of overall design, exception safety etc. and focusing on the question in its most narrow scope)
Abandon the idea of doing in the initializer list and do it in the constructor body instead
A::A() : v(NULL)
{
cudaMalloc(&v, 10 * sizeof(int));
}
Or, alternatively, wrap the allocation function into your own function that returns the pointer
void *wrapped_cudaMalloc(size_t size)
{
void *m = NULL;
cudaMalloc(&m, size);
return m;
}
...
A::A() : v(wrapped_cudaMalloc(10 * sizeof(int)))
{}
Just for the sake of completeness, there's also an ugly convoluted way to do it in the initializer list without creating any wrappers by exploiting the properties of , operator
A::A() : v((cudaMalloc(&v, 10 * sizeof(int)), v))
{}
Note the additional pair of () around the comma-expression, which is needed to satisfy the initialization syntax (otherwise , will be treated as argument separator instead of comma operator).
In addition to AndreyT's excellent post (and his creative use of the comma operator), you could also wrap things up thusly:
class cudaMallocedInt
{
private:
int *v;
public:
cudaMallocedInt(int n)
{
cudaMalloc(&v, n * sizeof(int));
}
cudaMallocedInt(const cudaMallocedInt &o)
{
// Do whatever is appropriate here. Probably some sort of dance.
}
~cudaMallocedInt()
{
// Remember to cudaFree or whatever
}
operator int*()
{
return v;
}
};
class A
{
private:
cudaMallocedInt v;
public:
A()
: v(10)
{
}
...
};
Update: As Johnsyweb pointed out in comments, please be sure to adhere to the Rule of Three to make sure things don't go boom and you waste your weekends tracking down hard to debug errors instead of having fun!
How can I take ownership of std::string char data without copying and withoug keeping source std::string object? (I want to use moving semantics but between different types.)
I use the C++11 Clang compiler and Boost.
Basically I want to do something equivalent to this:
{
std::string s(“Possibly very long user string”);
const char* mine = s.c_str();
// 'mine' will be passed along,
pass(mine);
//Made-up call
s.release_data();
// 's' should not release data, but it should properly destroy itself otherwise.
}
To clarify, I do need to get rid of std::string: further down the road. The code deals with both string and binary data and should handle it in the same format. And I do want the data from std::string, because that comes from another code layer that works with std::string.
To give more perspective where I run into wanting to do so: for example I have an asynchronous socket wrapper that should be able to take both std::string and binary data from user for writing. Both "API" write versions (taking std::string or row binary data) internally resolve to the same (binary) write. I need to avoid any copying as the string may be long.
WriteId write( std::unique_ptr< std::string > strToWrite )
{
// Convert std::string data to contiguous byte storage
// that will be further passed along to other
// functions (also with the moving semantics).
// strToWrite.c_str() would be a solution to my problem
// if I could tell strToWrite to simply give up its
// ownership. Is there a way?
unique_ptr<std::vector<char> > dataToWrite= ??
//
scheduleWrite( dataToWrite );
}
void scheduledWrite( std::unique_ptr< std::vecor<char> > data)
{
…
}
std::unique_ptr in this example to illustrate ownership transfer: any other approach with the same semantics is fine to me.
I am wondering about solutions to this specific case (with std::string char buffer) and this sort of problem with strings, streams and similar general: tips to approach moving buffers around between string, stream, std containers and buffer types.
I would also appreciated tips and links with C++ design approaches and specific techniques when it comes to passing buffer data around between different API's/types without copying. I mention but not using streams because I'm shaky on that subject.
How can I take ownership of std::string char data without copying and withoug keeping source std::string object ? (I want to use moving semantics but between different types)
You cannot do this safely.
For a specific implementation, and in some circumstances, you could do something awful like use aliasing to modify private member variables inside the string to trick the string into thinking it no longer owns a buffer. But even if you're willing to try this it won't always work. E.g. consider the small string optimization where a string does not have a pointer to some external buffer holding the data, the data is inside the string object itself.
If you want to avoid copying you could consider changing the interface to scheduledWrite. One possibility is something like:
template<typename Container>
void scheduledWrite(Container data)
{
// requires data[i], data.size(), and &data[n] == &data[0] + n for n [0,size)
…
}
// move resources from object owned by a unique_ptr
WriteId write( std::unique_ptr< std::vector<char> > vecToWrite)
{
scheduleWrite(std::move(*vecToWrite));
}
WriteId write( std::unique_ptr< std::string > strToWrite)
{
scheduleWrite(std::move(*strToWrite));
}
// move resources from object passed by value (callers also have to take care to avoid copies)
WriteId write(std::string strToWrite)
{
scheduleWrite(std::move(strToWrite));
}
// assume ownership of raw pointer
// requires data to have been allocated with new char[]
WriteId write(char const *data,size_t size) // you could also accept an allocator or deallocation function and make ptr_adapter deal with it
{
struct ptr_adapter {
std::unique_ptr<char const []> ptr;
size_t m_size;
char const &operator[] (size_t i) { return ptr[i]; }
size_t size() { return m_size; }
};
scheduleWrite(ptr_adapter{data,size});
}
This class take ownership of a string using move semantics and shared_ptr:
struct charbuffer
{
charbuffer()
{}
charbuffer(size_t n, char c)
: _data(std::make_shared<std::string>(n, c))
{}
explicit charbuffer(std::string&& str)
: _data(std::make_shared<std::string>(str))
{}
charbuffer(const charbuffer& other)
: _data(other._data)
{}
charbuffer(charbuffer&& other)
{
swap(other);
}
charbuffer& operator=(charbuffer other)
{
swap(other);
return *this;
}
void swap(charbuffer& other)
{
using std::swap;
swap(_data, other._data);
}
char& operator[](int i)
{
return (*_data)[i];
}
char operator[](int i) const
{
return (*_data)[i];
}
size_t size() const
{
return _data->size();
}
bool valid() const
{
return _data;
}
private:
std::shared_ptr<std::string> _data;
};
Example usage:
std::string s("possibly very long user string");
charbuffer cb(std::move(s)); // s is empty now
// use charbuffer...
You could use polymorphism to resolve this. The base type is the interface to your unified data buffer implementation. Then you would have two derived classes. One for std::string as the source, and the other uses your own data representation.
struct MyData {
virtual void * data () = 0;
virtual const void * data () const = 0;
virtual unsigned len () const = 0;
virtual ~MyData () {}
};
struct MyStringData : public MyData {
std::string data_src_;
//...
};
struct MyBufferData : public MyData {
MyBuffer data_src_;
//...
};
When using arrays you can do something like
class SomeClass
{
public:
int* LockMember( size_t& numInts );
private:
int* member;
size_t numInts;
};
int* SomeClass::LockMember( size_t& out_numInts )
{
out_numInts = numInts - 1;
return member + 1;
}
To return an array offset by some amount so as to prevent someone from modifying some part of contingeous memory, or, atleast, show some intent that this part of contingeous memory of the object should remain untouched.
Since I use vectors everywhere, I am wondering if there was some way to accomplish the same sort of thing:
class SomeClass
{
public:
std::vector<int> LockMember( void );
private:
std::vector<int> member;
};
std::vector<int> SomeClass::LockMember( void )
{
// somehow make a vector with its beginning iterator pointing to member.begin() + 1
// have a size smaller by one, still the same end iterator. The vector must be
// pointing to the same data as in this class as it needs to be modifiable.
return magicOffsetVector;
}
With the commented part replaced by real code. Any ideas?
If I understand you correctly: You want some memory with two parts: At the beginning you want something that can't be touched, and after that you want something that is open for use by client code.
You could do something along the following code. This will give the client code a copy to play with. This does mean you would have to do a lot of copying, though.
class SomeClass
{
public:
std::vector<int> getMember( void ) const;
void setMember(std::vector<int> newContent);
private:
std::vector<int> member;
size_t magicOffset;
};
// Read restricted part
std::vector<int> SomeClass::getMember( void ) const
{
return vector<int>(member.begin() + magicOffset, member.end());
}
// Assign to restricted part
void SomeClass::setMember(const std::vector<int>& v)
{
std::copy(v.begin(), v.end(), member.begin() + magicOffset);
}
In order to avoid the copying, it is possible that you could allocate memory for two vectors, one for the protected part and one for the unprotected part, and use placement new to put both vectors into that memory, thus ensuring that they are in contiguous memory. And then give the client code more or less free access to the public part of the vector. However, there's still the thing with bookkeeping variables in vector, and basically this would be an awful hack that's just waiting to blow up.
However, if you only need access to the unrestricted part on a per-element basis, you could just do range-checking on the arguments, i.e.:
int getElement(size_t idx)
{
idx += magicOffset;
if (idx > member.size() || idx < 0) throw std::out_of_range("Illegal index");
return member[idx];
}
And then either provide a setElement, or return int&.
This is a homework assignment. The Field container was the assignment from a week ago, and now I'm supposed to use the Field container to act as a dynamic array for a struct NumPair which holds two char * like so:
struct NumPair
{
char *pFirst, *pSecond;
int count;
NumPair( char *pfirst = "", char *psecond = "", int count = 0)
: pFirst(strdup(pfirst)), pSecond(strdup(psecond)), count(count)
{ }
NumPair( const NumPair& np )
: count(np.count), pFirst(strdup(np.pFirst)), pSecond(strdup(np.pSecond))
{ }
NumPair& operator=( const NumPair& np )
{
if(this != &np)
{
pFirst = strdup(np.pFirst);
pSecond = strdup(np.pSecond);
count = np.count;
}
return *this;
}
and the Field container
Field<NumPair> dict_;
The homework requires the use of char *, and not string, so that we can get better with all this low-level stuff. I've already had some question about char to wchar_t conversions, etc.
Now I have a question as to whether or not I'm destructing the NumPair properly. The scenario is as follows:
1) Field destructor gets called
template <class T>
Field<T>::~Field()
{
delete[] v_;
}
2) Delete calls the destructor of every element NumPair in v_;
~NumPair()
{
free(pFirst);
free(pSecond);
}
Is this okay? I haven't really read too many articles about mixing and matching elements created on the heap and free-store as we wish. I figure as long as I don't use delete on an improper malloc'ed element, I should be fine.
However, I don't know the entire intricacies of the delete command, so I'm wondering whether or not this is valid design, and what I could do to make it better.
Also, of course this isn't. I'm getting an error of the type:
This may be due to a corruption of the heap and points to dbgheap
extern "C" _CRTIMP int __cdecl _CrtIsValidHeapPointer(
const void * pUserData
)
{
if (!pUserData)
return FALSE;
if (!_CrtIsValidPointer(pHdr(pUserData), sizeof(_CrtMemBlockHeader), FALSE))
return FALSE;
return HeapValidate( _crtheap, 0, pHdr(pUserData) ); // Here
}
Again, how could I improve this without the use of string?
FIELD CTOR/Copy Ctor/Assignment
template <class T>
Field<T>::Field()
: v_(0), vused_(0), vsize_(0)
{ }
template <class T>
Field<T>::Field(size_t n, const T &val)
: v_(0), vused_(n), vsize_(0)
{
if(n > 0)
{
vsize_ = 1;
while(vsize_ < n)
vsize_ <<= 1;
v_ = new T[vsize_];
std::fill(v_, (v_ + vused_), val);
}
}
template <class T>
Field<T>::Field(const Field<T> &other)
: v_(new T[other.vsize_]), vsize_(other.vsize_), vused_(other.vused_)
{
std::copy(other.v_, (other.v_ + other.vused_), v_);
}
template <class T>
Field<T>& Field<T>::operator =(const Field<T> &other)
{
this->v_ = other.v_;
this->vused_ = other.vused_;
this->vsize_ = other.vsize_;
return *this;
}
FIELD MEMBERS
T *v_;
size_t vsize_;
size_t vused_;
Your copy constructor (of Field<>) seems OK, but the operator= is problematic.
Not only does it leak memory (what happens to the original v_?), but after that, two instances of Field<> hold a pointer to the same block of memory, and the one that is destructed first will invalidate the others v_ - and you can't even tell whether that has happened.
It's not always easy to decide how to deal with operator= - some think that implicit move semantics are okay, but the rest of us see how that played out with the majority of people, with std::auto_ptr. Probably the easiest solution is to disable copying altogether, and use explicit functions for moving ownership.
Your string handling in NumPair looks ok (strdup + free) and your Field container delete[] looks okay but it's hard to say because you don't show what v_ is.
eq mentions in a comment that you should also beware of how you are copying NumPairs. By default, C++ will give you an implicit member-wise copy constructor. This is where a RAII type like std::string makes your life easier: Your std::string containing struct can be copied without any special handling on your part and memory referenced in the string will be taken care of by the string's copy. If you duplicate your NumPair (by assigning it or returning it from a function for example) then the destruction of the temporary will free your strings out from under you.
Your copy constructor for Field just copies the pointers in v_. If you have two copies of a Field, all of the NumPairs in v_ will be deleted when the first Field goes out of scope, and then deleted again when the second one does.