I've recently come across a very strange segfault while developing my application. Basically, if I add another variable to one of my structs, a segfault is caused upon execution, for no apparent reason. Removing this variable immediately solves the problem. The struct is as follows:
typedef struct Note {
char cNote;
unsigned int uiDuration;
unsigned int uiVelocity;
};
As soon as I add a
long lStartTime;
variable anywhere in the struct, the code compiles as usual but will throw a segmentation fault. GDB's backtrace is lost somewhere in some obscure WIN methods that I don't even use.
Any ideas?
Thanks!
I see several possible explanations:
Something somewhere assumes that the struct is of a certain size. Changing the size break things.
You might have a memory bug of some sort, which is brought to the surface by you changing the layout of things in memory. Try a tool like valgrind or Purify.
You are changing the struct in a header file, but are failing to rebuild all source files that use the struct.
Related
I'm noticing that whenever I am declaring or initializing int c = 0, I seem to get a segmentation fault. It is only when I comment out that particular line of my code that it works fine. Here is a portion of my code:
int a = 0;
int b = 0;
int c = 0;
struct alphabet{
int first;
char word[30];
};
I've tested my code with just a and b as my only integers. It only began giving me that the segmentation fault once I tried initializing c later on. I just don't understand why this would even happen. Can someone help me?
Thank you in advance!
Since you have posted a very short peace of code, it is hard to give you a concrete advise. Hovewer, let's try as for the 'average' case. All your three int variables are located at the stack frame. Thus, adding more variables like c you may move the location of the rest stack variables which may be defined later. In particular, if you instantiate an instance of your structure alphabet which has an array field word. Imagine what happens if you underrun it (out of baundary access to the left baundary of the array). In the case if this underrun overlaps the area where c is located and you write to it by initializing c, then potentially string operation on word may crash. So, my advise would be do a code review starting down from the this function. Focus is on using uninitialized variables, out of boundary access and hanging pointers in respect to stack variables. A good practice would also be to use some static code analysis tool. For more concrete hints you need to extend your question to give more details as is recommended in the comments above.
I have a problem with my code, it seems to fail in a weird way in release mode, debug mode works just fine.
I am out of ideas and clues. I would like to get a direction for what should i do.
So here is the problematic code:
static inline void FindPoint(Line *First,AABB *Second,point *Pt,int direction)
{
std::vector<point*> pntvec; ... }
When i call function FindPoint it is all fine and everything is set well.
But when i reach the vector initialization the First pointer is reset to the same address as the vector.
That causes unexpected behavior.
So i thought the problem was in the commonalities. Both Line and the vector are using the point structure.
Yet i was not able to find anything unusual.
here are the structures
typedef struct
{
double x;
double y;
}point;
typedef struct
{
double incline;
double y;
double x;
double c;
point start,finish;
}Line;
And here is the initialization of the data sent to parameter First:
Line *objvec = new Line;
so what could be the problem can any one tell?
Thank you in advance
Edit:
A better look at the code
http://pastebin.com/rki7EdM6
You're using pointers more heavily than you should for these simple types. Pointers and dynamic allocation are the likely sources of your undefined behavior. So a simple first step is to refactor and clean up your code, preferring stack variables and pass-by-reference for simple types. In all likelihood, the undefined behavior will go away, or the error will be easier to spot.
[Caveat: I can't get to pastebin, so I can't see your full code. Prefer simple examples right here on the site]
For example, your function signature should probably be this (I'm guessing about which parameter is the output)
static inline point FindPoint(const Line &First, const AABB &Second,int direction)
And
Line *objvec = new Line;
should just be
Line objvec;
etc, etc.
std::vector<point> pntvec;
is probably the most important. A vector of pointers is ok, but should be avoided if possible. One of the nice parts about standard containers is the auto-cleanup when the variable goes out of scope. You don't get that with a vector of pointers, since the vector doesn't know to call delete on every element. You have to do it yourself.
"so what could be the problem can any one tell?"
If such a situation occurs, the most probable reason is, you have hit undefined behavior in your code somewhere else.
I can't tell any more from the context and code you have posted. But using a tool like valgrind is likely to do this right on your service.
I'm working on this VST convolution plugin (Windows 7 64bit, VS2010) and I decided to try the Intel c++ compiler. I was in the process of optimizing the algorithm so I had a backup project in case of any screw ups and one I was doing experiments on. Both projects would compile and run with no problems. After installing the Intel compiler though the project I was experimenting on would cause a heap corruption error, so I start debugging to track down the problem but I can't find the line of code that causes it since the heap corruption error is not triggered during execution but after the termination of the DLL (there are also no access violations showed by the debugger).
At this point I start cutting out parts of the code to see if I can isolate the problem and I discover (obviously) that it was the class I was eperimenting on. Now here comes the weird part: I can change the code inside the methods but as soon as I add a variable to the backup class (the one that works fine), even an int, I get the heap corruption error, just a decleared and never referenced variable is enough.
This is the class CRTConvolver:
class CRTConvolver
{
public:
CRTConvolver();
~CRTConvolver();
bool Init(float* Imp, unsigned ImpLen, unsigned DataLen);
void doConv(float* input);
Buff Output;
int debug_test;
private:
void ZeroVars();
int Order(int sampleFrames);
template <class T> void swap ( T& a, T& b );
Buff *Ir_FFT,*Input_FFT,Output2,Tmp,Prev,Last;
float *Tail;
unsigned nBlocks,BlockLen,Bl_Indx;
IppsFFTSpec_R_32f* spec;
};
that "int debug_test;" makes the difference between a perfectly working VST module and a program that crashes on initialization from Cubase.
always for debugging purposes here are destr and constr:
CRTConvolver::CRTConvolver()
{
//IppStatus status=ippInit();
//ZeroVars();
}
CRTConvolver::~CRTConvolver()
{
//Init(NULL,NULL,NULL);
}
Here is what class Buff looks like:
class Buff {
public:
Buff();
Buff(unsigned len);
~Buff();
float* buff;
unsigned long length;
private:
void Init(unsigned long len);
void flush();
friend class CRTConvolver;
}
Buff::Buff()
{
length=NULL;
buff=NULL;
}
Buff::~Buff()
{
// flush();
}
basically this class if created and destructed does absolutely nothing, it just contains the length and buff variables. If I also bypass those two variable initializations the heap error goes away.
The software crashes on simple construction and subsequent destruction of the class CRTConvolver even though all it does is nothing, this is the part that really doesn't make sense to me...
As a side note, I create my CRTConvolver class like this:
ConvEng = new CRTConvolver[NCHANNELS];
If I declare it like this instead:
CRTConvolver ConvEng[NCHANNELS];
I get a stack corruption error around variable ConvEng.
If I switch back to Microsoft compiler the situation stays the same even when compiling and running the exact same version that could run without errors before....
I can't stress enough the fact that before installing the Intel compiler everything was running just fine, is it possible that something went wrong or there's an incompatibility somewhere ?
I'm really running out of ideas here, I hope someone will be able to help.
thanks
Going to guess, since the problem is most likely undefined behavior, but in some other place in your code:
Obey the rule of three. You should have a copy constructor and assignment operator. If you're using std containers, or making copies or assignments, without these you run into trouble if you delete memory inside the destructor.
It looks to me that the CRTConvolver default constructor (used in creating an array) is writing to memory it doesn't own. If the Intel compiler has different class layout rules (or data alignment rules), it might be unmasking a bug that was benign under the Microsoft compiler's rules.
Does the CRTConvolver class contain any instances of the Buff class?
Updated to respond to code update:
The CRTConvolver class contains four instances of Buff, so I suspect that is where the problem lies. It could be a version mismatch -- the CRTConvolver class thinks that Buff is smaller than it really is. I suggest you recompile everything and get back to us.
Okay, I have a struct, TextBlock, that simulates moving blocks of text around the screen. Here's the header:
struct TextBlock
{
RECT textArea;
RECT rectArea;
double whatBlock;
double x;
double y;
double angle;
double speed;
double width;
double height;
char *word;
bool stuck;
};
When it's like this, everything works perfectly fine. The problem comes when I add another member that I need. The way it works is that I have two arrays of TextBlocks. The first is for moving ones, the second is for ones that don't move, signifying where the moving ones need to go. The words are all randomized from a sentence to a jumble, so this data member will be set (commented out) to the index of which static block belongs to the moving one so I know when it's in the right place.
int whatBlock;
After creating this, I go through all of the created objects and set
tb[i][j].whatBlock = 0; //same area as other data members being set, moving text
stb[i][j].whatBlock = 0; //static text block
When I try to run this, without doing anything else to the data member, it comes up with an error:
The instruction at [address] referenced memory at [different address]. The memory could not be "written".
Note that if I don't try to modify it, and just create the data member, it works.
At this point of almost being done and having tons of these kinds of problems, I'm getting a bit fed up with this program >.> Any help at all on this would be greatly appreciated.
EDIT: This issue is now fixed. I replied to the accepted answer with the explanation, but it poses another problem, even if it doesn't affect this program.
Force a rebuild of everything. You may have an object file that is out-of-date with respect to the header file that defines TextBlock
If that doesn't fix it, run your program under a debugger and see what the faulting instruction is. Either that will allow you to fix the program, or you can ask again with mroe informatin.
Without you posting more code, I can only tell that you likely have a memory corruption bug in your program - i.e. you are reading or writing beyond the end of allocated memory.
If you post more code, I'll edit this answer accordingly.
I can't really give advices since we cannot access the complete source code.
Anyway, I can suggest you that it may be not in the struct TextBlock where really hides the bug. For instance, every access to TextBlock's member means that you are accessing to this hidden variable.
If this pointer is corrupted, you may experience problem where you don't expect it, leading you to search in the wrong places.
I have a piece of templated code that is never run, but is compiled. When I remove it, another part of my program breaks.
First off, I'm a bit at a loss as to how to ask this question. So I'm going to try throwing lots of information at the problem.
Ok, so, I went to completely redesign my test project for my experimental core library thingy. I use a lot of template shenanigans in the library. When I removed the "user" code, the tests gave me a memory allocation error. After quite a bit of experimenting, I narrowed it down to this bit of code (out of a couple hundred lines):
void VOODOO(components::switchBoard &board) {
board.addComponent<using_allegro::keyInputs<'w'> >();
}
Fundementally, what's weirding me out is that it appears that the act of compiling this function (and the template function it then uses, and the template functions those then use...), makes this bug not appear. This code is not being run. Similar code (the same, but for different key vals) occurs elsewhere, but is within Boost TDD code.
I realize I certainly haven't given enough information for you to solve it for me; I tried, but it more-or-less spirals into most of the code base. I think I'm most looking for "here's what the problem could be", "here's where to look", etc. There's something that's happening during compile because of this line, but I don't know enough about that step to begin looking.
Sooo, how can a (presumably) compilied, but never actually run, bit of templated code, when removed, cause another part of code to fail?
Error:
Unhandled exceptionat 0x6fe731ea (msvcr90d.dll) in Switchboard.exe:
0xC0000005: Access violation reading location 0xcdcdcdc1.
Callstack:
operator delete(void * pUser Data)
allocator< class name related to key inputs callbacks >::deallocate
vector< same class >::_Insert_n(...)
vector< " " >::insert(...)
vector<" ">::push_back(...)
It looks like maybe the vector isn't valid, because _MyFirst and similar data members are showing values of 0xcdcdcdcd in the debugger. But the vector is a member variable...
Update: The vector isn't valid because it's never made. I'm getting a channel ID value stomp, which is making me treat one type of channel as another.
Update:
Searching through with the debugger again, it appears that my method for giving each "channel" it's own, unique ID isn't giving me a unique ID:
inline static const char channel<template args>::idFunction() {
return reinterpret_cast<char>(&channel<CHANNEL_IDENTIFY>::idFunction);
};
Update2: These two are giving the same:
slaveChannel<switchboard, ALLEGRO_BITMAP*, entityInfo<ALLEGRO_BITMAP*>
slaveChannel<key<c>, char, push<char>
Sooo, having another compiled channel type changing things makes sense, because it shifts around the values of the idFunctions? But why are there two idFunctions with the same value?
you seem to be returning address of the function as a character? that looks weird. char has much smaller bit count than pointer, so it's highly possible you get same values. that could reason why changing code layout fixes/breaks your program
As a general answer (though aaa's comment alludes to this): When something like this affects whether a bug occurs, it's either because (a) you're wrong and it is being run, or (b) the way that the inclusion of that code happens to affect your code, data, and memory layout in the compiled program causes a heisenbug to change from visible to hidden.
The latter generally occurs when something involves undefined behavior. Sometimes a bogus pointer value will cause you to stomp on a bit of your code (which might or might not be important depending on the code layout), or sometimes a bogus write will stomp on a value in your data stack that might or might not be a pointer that's used later, or so forth.
As a simple example, supposing you have a stack that looks like:
float data[10];
int never_used;
int *important pointer;
And then you erroneously write
data[10] = 0;
Then, assuming that stack got allocated in linear order, you'll stomp on never_used, and the bug will be harmless. However, if you remove never_used (or change something so the compiler knows it can remove it for you -- maybe you remove a never-called function call that would use it), then it will stomp on important_pointer instead, and you'll now get a segfault when you dereference it.