I have program that uses std::string, but memmove the std::string` instances.
It worked fine until gcc 5.1.
However this no longer works as of gcc 5.3. I think developers finally did SSO with internal pointer.
I will definitely fix that, but is there easy way to fix it with some define or pragma?
Code looks similar to this:
// MyClass have std::string inside
MyClass *a = malloc(MAX * sizeof(MyClass));
// ...
// placement new on a[0]
// ...
memmove(&a[1], &a[0], sizeof(MyClass));
// ...
process(a[1]);
This is old code, please do not comment about malloc usage.
I will refactor or switch to std::vector, but I want the code to work until I do so.
You are experiencing effects of undefined behavior, but I think you know this. You cannot rely on the effects of byte-wise copying non-POD resp. not trivially copyable types, and the compiler is free to change that behavior.
I think it may be possible to define a safe overload for memmove with your class as arguments and use the copy-constructor inside it. I don't know if that is strictly legal, but you seem to be using the C-function instead of the C++ version in namespace std, so at least you are not changing namespace std which is not allowed.
void memmove(MyClass* a, MyClass* b, size_t)
{
*a = *b;
}
Strictly speaking, I think this is still undefined behavior because 17.6.4.3 of the C++ standard specifies that
If a program declares or defines a name in a context where it is
reserved, other than as explicitly allowed by this Clause, its
behavior is undefined.
In addition, all names in C library are reserved names and shall not be used by the program (17.6.4.3.2). Practically, I think this will work.
You may need to compile with -fno-builtin to prevent gcc from replacing memmove globally. If it is illegal to overwrite the function, you can replace it dynamically with LD_PRELOAD.
This is a hack solution! Your code may still not work because the compiler makes the assumption that, when you memmove it is a POD/TriviallyCopyable object and uses that for some optimisation, e.g. by assuming that after the memmove, both objects are represented by identical bytes. This is broken when you re-implement memmove with the copy-constructor.
Related
How do I declare an external MFC function that has LPCTSTR tokens?
I have the following C function in a library I wish to use:
DLL_PUBLIC void output( LPCTSTR format, ... )
Where DLL_PUBLIC is resolved to
__declspec(dllimport) void output( LPCTSTR format, ...)
In Rust, I was able to get things to build with:
use winapi::um::winnt::CHAR;
type LPCTSTR = *const CHAR;
#[link(name="mylib", kind="static")]
extern {
#[link_name = "output"]
fn output( format:LPCTSTR, ...);
}
I'm not sure this is the best approach, but it seems to get me part way down. Although the module is declared in a DLL module, the symbol decorations in the native DLL are such that it does not have _imp prepended to it in the binary. So I find that "static" seems to look for the correct module, although I still have not been able to get it to link.
The dumpbin output for the module (this target function) in "mylib.dll" is:
31 ?output##YAXPEBDZZ (void __cdecl output(char const *,...))
This is assuming that what you are trying to accomplish here is to link Rust code against a custom DLL implementation. If that is the case, then things are looking good.
First things first, though, you'll need to do some sanity cleanup. LPCTSTR is not a type. It is a preprocessor symbol that either expands to LPCSTR (aka char const*) or LPCWSTR (aka wchar_t const*).
When the library gets built the compiler commits to either one of those, and that decision is made for all eternity. Clients, on the other hand, that #inlcude the header are still free to choose, and you have no control over this. Lucky you, if you're using C++ linkage, and have the linker save you. But we aren't using C++ linkage.
The first order of action is to change the C function signature using an explicit type, so that clients and implementation always agree. I will be using char const* here.
C++ library
Building the library is fairly straight forward. The following is a bare-bones C++ library implementation that simply outputs a formatted string to STDOUT.
dll.cpp:
#include <stdio.h>
#include <stdarg.h>
extern "C" __declspec(dllexport) void output(char const* format, ...)
{
va_list argptr{};
va_start(argptr, format);
vprintf(format, argptr);
va_end(argptr);
}
The following changes to the original code are required:
extern "C": This is requesting C linkage, controlling how symbols are decorated as seen by the linker. It's the only reasonable choice when planning to cross language boundaries.
__declspec(dllexport): This is telling the compiler to inform the linker to export the symbol. C and C++ clients will use a declaration with a corresponding __declspec(dllimport) directive.
char const*: See above.
This is all that's required to build the library. With MSVC the target architecture is implied by the toolchain used. Open up a Visual Studio command prompt that matches the architecture eventually used by Rust's toolchain, and run the following command:
cl.exe /LD dll.cpp
This produces, among other artifacts, dll.dll and dll.lib. The latter being the import library that needs to be discoverable by Rust. Copying it to the Rust client's crate's root directory is sufficient.
Consuming the library from Rust
Let's start from scratch here and make a new binary crate:
cargo new --bin client
Since we don't need any other dependencies, the default Cargo.toml can remain unchanged. As a sanity check you can cargo run to verify that everything is properly set up.
If that all went down well it's time to import the only public symbol exported by dll.dll. Add the following to src/main.rs:
#[link(name = "dll", kind = "dylib")]
extern "C" {
pub fn output(format: *const u8, ...);
}
And that's all there is to it. Again, a few details are important here, namely:
name = "dll": Specifies the import library. The .lib extension is implied, and must not be appended.
kind = "dylib": We're importing from a dynamic link library. This is the default and can be omitted, though I'm keeping it for posterity.
extern "C": As in the C++ code this controls name decoration and the calling convention. For variadic functions the C calling convention (__cdecl) is required.
*const u8: This is Rust's native type that corresponds to char const* in C and C++. Using type aliases (whether those provided by the winapi crate or otherwise) is not required. It wouldn't hurt either, but let's just keep this simple.
With that everything is set up and we can take this out for a spin. Replace the default generated fn main() with the following code in src/main.rs:
fn main() {
unsafe { output("Hello, world!\0".as_ptr()) };
}
and there you have it. cargo running this produces the famous output:
Hello, world!
So, all is fine, right? Well, no, not really. Actually, nothing is fine. You could have just as well written, compiled, and executed the following:
fn main() {
unsafe { output(b"Make sure this has reasons to crash: %f".as_ptr(), "💩") };
}
which produces the following output for me:
Make sure this has reasons to crash: 0.000000💩
though any other observable behavior is possible, too. After all, the behavior is undefined. There are two bugs: 1 The format specifier doesn't match the argument, and 2 the format string isn't NUL terminated.
Either one can be fixed, trivially even, though you have opted out of Rust's safety guarantees. Rust can't help you detect either issue, and when control reaches the library implementation, it cannot detect this either. It will just do what it was asked to do, subverting each and every one of Rust's safety guarantees.
Remarks
A few words of caution: Getting developers interested in Rust is great, and I will do my best to try whenever I get a chance to. Getting Rust-curious developers excited about Rust is often just a natural progression.
Though I will say that trying to get developers excited about Rust by starting out with unsafe Rust isn't going to be successful. It's eventually going to provoke a response like: "Look, ma, a steep learning curve with absolutely no benefit whatsoever, who could possibly resist?!" (I'm exaggerating, I know).
If your ultimate goal is to establish Rust as a superior alternative to C (and in part C++), don't start by evaluating how not to benefit from Rust. Specifically, trying to import a variadic function (the unsafest language construct in all of C++) and exposing it as an unsafe function to Rust is almost guaranteed to be the beginning of a lost battle.
Now, this may read bad as it is already, but this isn't over yet. In an attempt to make your C++ code accessible from Rust, things have gotten worse! With a C++ compiler and static code analysis tools (assuming the format string is known at compile time, and the tools understand the semantics), the tooling can and frequently will warn about mismatches. That option is now gone, forever, and there's not even a base level of protection.
If you absolutely want to make some sort of logging available to Rust, export a function from the library that takes a single char const*, use Rust's format! macro, and provide a variadic wrapper to C and C++ clients.
Consider the following code:
typedef struct _sMYSTRUCT_BASE
{
int b_a;
int b_b;
int b_c;
} sMYSTRUCT_BASE;
typedef struct _sMYSTRUCT
{
sMYSTRUCT_BASE base;
int a;
int b;
} sMYSTRUCT;
Private const sMYSTRUCT mystruct_init =
{
0,
1,
3,
4
};
I am looking for a way to generate an error (compile-, or runtime) to indicated that the structure initialization hasn't explicitly 'touched' all structure members.
There are 5 integers in the structure, but 'mystruct_init' only have 4 values.
I know that last member (mystruct_init.b) will be zero, but I need some kind of warning/error to inform the programmer about the mistake.
This has to work on a very old compiler (maybe not even ansi-c compliant).
Modern compilers are capable of producing such a warning...in gcc, it's turned on with -Wmissing-field-initializers (which warns about initializers that exist but do not initialize all members, but not about structs with no initializer expression; these can at least sometimes be caught by turning on -Wuninitialized, which will warn you if it sees you reading a potentially uninitialized value, at least if you read it in the same function the variable was declared in).
If your very old compiler happens to supply such a warning, you could of course just turn it on, but that seems unlikely from your description.
Your best option, I think, if you want to do an exhaustive search for them, would be to see whether you can get the code to compile with some version of gcc -- it wouldn't have to compile well enough to actually run on your target platform in order to get the warnings. I can't guarantee that it will be able to compile your pre-ANSI C code, particularly if it widely uses compiler-specific extensions, but I can at least say that support for the legacy K&R syntax is still present in the modern C standard, so I wouldn't be surprised if your code compiles better than you might think.
If that works, then to consistently produce the warnings in your IDE, you could modify the build script so that it both compiles and links the code with the real compiler you're targeting, and also compiles it (but not necessarily links it) with gcc, just to generate additional warnings that can be picked up and displayed by the IDE.
The other option would be to see if you can find a compatible static analyzer that can perform such a check; I work on a tool called EnSoft Atlas that builds a data-flow graph which, together with a simple script, could be used to enforce initialization more thoroughly than the gcc warnings allow, by checking whether flow of uninitialized values to fields of structs occurs.
However, our support for C is still in beta. Atlas requires that Eclipse CDT (or JDT for Java) be able to parse your code, and the current C beta only fully supports modern strongly-typed struct initializers (i.e. struct foo f = (struct foo) {...} has fully connected data-flow, but support for the older initializer list syntax struct foo f = {...} was not implemented in our first pass), so I'm not sure it would be able to meet your needs at this time.
I came across an interesting error when I was trying to link to an MSVC-compiled library using MinGW while working in Qt Creator. The linker complained of a missing symbol that went like _imp_FunctionName. When I realized That it was due to a missing extern "C", and fixed it, I also ran the MSVC compiler with /FAcs to see what the symbols are. Turns out, it was __imp_FunctionName (which is also the way I've read on MSDN and quite a few guru bloggers' sites).
I'm thoroughly confused about how the MinGW linker complains about a symbol beginning with _imp, but is able to find it nicely although it begins with __imp. Can a deep compiler magician shed some light on this? I used Visual Studio 2010.
This is fairly straight-forward identifier decoration at work. The imp_ prefix is auto-generated by the compiler, it exports a function pointer that allows optimizing binding to DLL exports. By language rules, the imp_ is prefixed by a leading underscore, required since it lives in the global namespace and is generated by the implementation and doesn't otherwise appear in the source code. So you get _imp_.
Next thing that happens is that the compiler decorates identifiers to allow the linker to catch declaration mis-matches. Pretty important because the compiler cannot diagnose declaration mismatches across modules and diagnosing them yourself at runtime is very painful.
First there's C++ decoration, a very involved scheme that supports function overloads. It generates pretty bizarre looking names, usually including lots of ? and # characters with extra characters for the argument and return types so that overloads are unambiguous. Then there's decoration for C identifiers, they are based on the calling convention. A cdecl function has a single leading underscore, an stdcall function has a leading underscore and a trailing #n that permits diagnosing argument declaration mismatches before they imbalance the stack. The C decoration is absent in 64-bit code, there is (blessfully) only one calling convention.
So you got the linker error because you forgot to specify C linkage, the linker was asked to match the heavily decorated C++ name with the mildly decorated C name. You then fixed it with extern "C", now you got the single added underscore for cdecl, turning _imp_ into __imp_.
I have a dynamic link library written in VC++6. I wrote some code with VC++2005 which calls the native VC++6 library. Whenever I pass std::string to the native library, the result is always garbage. However, this does not happen if I pass other types like char *, int, etc. Any ideal what is causing this?
The following code illustrates this.
// VC++ 6 Code
class __declspec(dllexport) VC6
{
public:
VC6();
void DoSomething(const std::string &s);
}
VC6()::VC6() {}
void VC6::DoSomething(const std::string &s)
{
std::cout << s; // Resulting output on screen is garbage
}
// VC++ 2005 Code
void VC2005::DoSomething()
{
VC6 *vc6 = new VC6();
std::string s("Test String");
vc6->DoSomething(s);
delete vc6;
}
Classes such as std::string aren't necessarily defined the same way in every version of the runtime library (even though they have the same name), so you shouldn't mix the libraries this way. On the other hand, types such as int and char* are the same for a given platform so you can pass them.
In your example, it's better to pass the string as a (pointer,size) pair or simply as a null-terminated string.
Edit: forgot to mention the obvious solution of using the same compiler version. Do this if you want to pass objects around.
Don't do this. It doesn't work.
C++ does not define a fixed ABI, so you can't in general pass non-POD types between libraries or translation units compiled by different compilers.
In your case, VC6 and VC8 have different definitions of std::string (and the compilers may also insert different padding and other changes), and the result is garbage, and/or unpredictable behavior and crashes.
If you need to pass data to a VC6 DLL (a better option might be to recompile that code under a sane compiler), you have to stick to types where you can be sure it'll work. That means 1) POD types (either built-in primitives such as char*, or C structs containing only POD types), or COM objects.
You can write a wrapper DLL with a C interface. A C++/CLI wrapper may be needed if p-invoke alone can not handle the interop. Writing a COM server in ATL is probably a better choice to provide an object oriented interface in native code and to avoid writing another wrapper DLL in C++/CLI).
Visual C++ 2008 C runtime offers an operator 'offsetof', which is actually macro defined as this:
#define offsetof(s,m) (size_t)&reinterpret_cast<const volatile char&>((((s *)0)->m))
This allows you to calculate the offset of the member variable m within the class s.
What I don't understand in this declaration is:
Why are we casting m to anything at all and then dereferencing it? Wouldn't this have worked just as well:
&(((s*)0)->m)
?
What's the reason for choosing char reference (char&) as the cast target?
Why use volatile? Is there a danger of the compiler optimizing the loading of m? If so, in what exact way could that happen?
An offset is in bytes. So to get a number expressed in bytes, you have to cast the addresses to char, because that is the same size as a byte (on this platform).
The use of volatile is perhaps a cautious step to ensure that no compiler optimisations (either that exist now or may be added in the future) will change the precise meaning of the cast.
Update:
If we look at the macro definition:
(size_t)&reinterpret_cast<const volatile char&>((((s *)0)->m))
With the cast-to-char removed it would be:
(size_t)&((((s *)0)->m))
In other words, get the address of member m in an object at address zero, which does look okay at first glance. So there must be some way that this would potentially cause a problem.
One thing that springs to mind is that the operator & may be overloaded on whatever type m happens to be. If so, this macro would be executing arbitrary code on an "artificial" object that is somewhere quite close to address zero. This would probably cause an access violation.
This kind of abuse may be outside the applicability of offsetof, which is supposed to only be used with POD types. Perhaps the idea is that it is better to return a junk value instead of crashing.
(Update 2: As Steve pointed out in the comments, there would be no similar problem with operator ->)
offsetof is something to be very careful with in C++. It's a relic from C. These days we are supposed to use member pointers. That said, I believe that member pointers to data members are overdesigned and broken - I actually prefer offsetof.
Even so, offsetof is full of nasty surprises.
First, for your specific questions, I suspect the real issue is that they've adapted relative to the traditional C macro (which I thought was mandated in the C++ standard). They probably use reinterpret_cast for "it's C++!" reasons (so why the (size_t) cast?), and a char& rather than a char* to try to simplify the expression a little.
Casting to char looks redundant in this form, but probably isn't. (size_t) is not equivalent to reinterpret_cast, and if you try to cast pointers to other types into integers, you run into problems. I don't think the compiler even allows it, but to be honest, I'm suffering memory failure ATM.
The fact that char is a single byte type has some relevance in the traditional form, but that may only be why the cast is correct again. To be honest, I seem to remember casting to void*, then char*.
Incidentally, having gone to the trouble of using C++-specific stuff, they really should be using std::ptrdiff_t for the final cast.
Anyway, coming back to the nasty surprises...
VC++ and GCC probably won't use that macro. IIRC, they have a compiler intrinsic, depending on options.
The reason is to do what offsetof is intended to do, rather than what the macro does, which is reliable in C but not in C++. To understand this, consider what would happen if your struct uses multiple or virtual inheritance. In the macro, when you dereference a null pointer, you end up trying to access a virtual table pointer that isn't there at address zero, meaning that your app probably crashes.
For this reason, some compilers have an intrinsic that just uses the specified structs layout instead of trying to deduce a run-time type. But the C++ standard doesn't mandate or even suggest this - it's only there for C compatibility reasons. And you still have to be careful if you're working with class heirarchies, because as soon as you use multiple or virtual inheritance, you cannot assume that the layout of the derived class matches the layout of the base class - you have to ensure that the offset is valid for the exact run-time type, not just a particular base.
If you're working on a data structure library, maybe using single inheritance for nodes, but apps cannot see or use your nodes directly, offsetof works well. But strictly speaking, even then, there's a gotcha. If your data structure is in a template, the nodes may have fields with types from template parameters (the contained data type). If that isn't POD, technically your structs aren't POD either. And all the standard demands for offsetof is that it works for POD. In practice, it will work - your type hasn't gained a virtual table or anything just because it has a non-POD member - but you have no guarantees.
If you know the exact run-time type when you dereference using a field offset, you should be OK even with multiple and virtual inheritance, but ONLY if the compiler provides an intrinsic implementation of offsetof to derive that offset in the first place. My advice - don't do it.
Why use inheritance in a data structure library? Well, how about...
class node_base { ... };
class leaf_node : public node_base { ... };
class branch_node : public node_base { ... };
The fields in the node_base are automatically shared (with identical layout) in both the leaf and branch, avoiding a common error in C with accidentally different node layouts.
BTW - offsetof is avoidable with this kind of stuff. Even if you are using offsetof for some jobs, node_base can still have virtual methods and therefore a virtual table, so long as it isn't needed to dereference member variables. Therefore, node_base can have pure virtual getters, setters and other methods. Normally, that's exactly what you should do. Using offsetof (or member pointers) is a complication, and should only be used as an optimisation if you know you need it. If your data structure is in a disk file, for instance, you definitely don't need it - a few virtual call overheads will be insignificant compared with the disk access overheads, so any optimisation efforts should go into minimising disk accesses.
Hmmm - went off on a bit of a tangent there. Whoops.
char is guarenteed to be the smallest number of bits the architectural can "bite" (aka byte).
All pointers are actually numbers, so cast adress 0 to that type because it's the beginning.
Take the address of member starting from 0 (resulting into 0 + location_of_m).
Cast that back to size_t.
1) I also do not know why it is done in this way.
2) The char type is special in two ways.
No other type has weaker alignment restrictions than the char type. This is important for reinterpret cast between pointers and between expression and reference.
It is also the only type (together with its unsigned variant) for which the specification defines behavior in case the char is used to access stored value of variables of different type. I do not know if this applies to this specific situation.
3) I think that the volatile modifier is used to ensure that no compiler optimization will result in attempt to read the memory.
2 . What's the reason for choosing char reference (char&) as the cast target?
if type s has operator& overloaded then we can't get address using &s
so we reinterpret_cast the type s to primitive type char because primitive type char
doesn't have operator& overloaded
now we can get address from that
if in C then reinterpret_cast is not required
3 . Why use volatile? Is there a danger of the compiler optimizing the loading of m? If so, in what exact way could that happen?
here volatile is not relevant to compiler optimizing.
if type s have const or volatile or both qualifier(s) then
reinterpret_cast can't cast to char& because reinterpret_cast can't remove cv-qualifiers
so result is using <const volatile char&> for casting work from any combination