out of scope pointers in threads
uche
thread gets executed, the pointer goes out of scope . How do I fix
it ?
int mywrite(char* id, int number_of_characters, char char_array)
{
HANDLE Producer;
DWORD ThId;
//global_char = char_array;
//create mutual exlusion for producer process to write into the
buffer
//create producer thread and start the function for inserting
characters
//mywriteTh is the entry point of the producer
ptr =new data;
ptr->character = char_array;
cout<<ptr->character<<endl;
ptr->id = id;
cout<<ptr->id<<endl;
Producer = (HANDLE) CreateThread (NULL, 0, mywriteTh,
reinterpret_cast<data> (ptr) , 0, &ThId); // i want to send the
pointer to this thread
cout<<ptr->id<<endl;
return 0;
}
DWORD WINAPI mywriteTh(data ptr)
{
//global_char is available here
DWORD tId = GetCurrentThreadId();
data *ptr_data = reinterpret_cast<data *>(ptr); // pointer is goes
out of scope
}
uche
please note: change data* to LPVOID ... HOWEVER, THIS DOESN'T SEEM TO
DO THE TRICK! I STILL GET A POINTER THAT IS OUT OF SCOPE!
James Kanze
> I am trying to send a pointer to the thread below; however,
> when the thread gets executed, the pointer goes out of scope .
> How do I fix it ?
What is "data"? Is it a data type, or a pointer to a data type?
You seem to use it both ways.
> int mywrite(char* id, int number_of_characters, char char_array)
> {
> HANDLE Producer;
> DWORD ThId;
> //global_char = char_array;
> //create mutual exlusion for producer process to write into the buffer
> //create producer thread and start the function for inserting characters
> //mywriteTh is the entry point of the producer
> ptr =new data;
OK, we have 'data' as a data type, and 'ptr' is, I suppose a
data*.
> ptr->character = char_array;
Again, what is char_array? Where does it come from?
> cout<<ptr->character<<endl;
> ptr->id = id;
> cout<<ptr->id<<endl;
> Producer = (HANDLE) CreateThread (NULL, 0, mywriteTh,
> reinterpret_cast<data> (ptr) , 0, &ThId); // i want to send the
> pointer to this thread
Anytime you need a reinterpret_cast for a function argument, you
should be asking yourself questions. If I understand the
interface description of CreateThread at MSDN (Microsoft seems
to go in a lot for obfuscated typenames), you don't need any
cast at all; just pass the pointer. (I'm guessing here that
LPVOID is a void*; where the L comes from, I don't know.)
> cout<<ptr->id<<endl;
> return 0;
> }
> DWORD WINAPI mywriteTh(data ptr)
And according to the documentation, this function must take a
void* (well, an LPVOID) as well, not a data. (The documentation
shows some __in as well. More obfuscation; I don't think it
means anything.)
> {
> //global_char is available here
> DWORD tId = GetCurrentThreadId();
> data *ptr_data = reinterpret_cast<data *>(ptr); // pointer is goes
> out of scope
Again, a reinterpret_cast is a no-no. You need a static_cast of
the void* to the exact type of the pointer that was converted to
void*, above.
> }
As for "going out of scope", of course the pointer goes out of
scope. But you're passing a copy of it to the thread, and the
memory it points to is still there.
--
James Kanze (GABI Software) email:james...@gmail.com
Conseils en informatique orientée objet/
Beratung in objektorientierter Datenverarbeitung
9 place Sémard, 78210 St.-Cyr-l'École, France, +33 (0)1 30 23 00 34
Pete Becker
> (I'm guessing here that
> LPVOID is a void*; where the L comes from, I don't know.)
The L stands for LONG. It's from the days when Windows still exposed
the Intel segmented architecture, so there were SHORT and LONG variants
of pointers.
--
Pete
Roundhouse Consulting, Ltd. (www.versatilecoding.com) Author of "The
Standard C++ Library Extensions: a Tutorial and Reference
(www.petebecker.com/tr1book)
Maic Schmidt
Your pointer seems to go "out of scope" because your procedure where you
declared the pointer ends,
and so its local variable stack is cleared.
Declare your data-pointer at class or global scope.
class SVDaemon
{
public:
void StartThread()
{
pointer = new MyData();
wndThread =
CreateThread(NULL,0,(LPTHREAD_START_ROUTINE)WindowThread,pointer,0,&threadId
);
}
private:
MyData* pointer;
DWORD threadId;
HANDLE wndThread;
static DWORD WindowThread(MyData* data);
};
Greetz
Maic
"uche" <uraniu...@hotmail.com> schrieb im Newsbeitrag
news:c201d729-47c7-449f...@a29g2000pra.googlegroups.com...
James Kanze
> On 2008-11-28 14:47:04 -0500, James Kanze <james.ka...@gmail.com> said:
> > (I'm guessing here that
> > LPVOID is a void*; where the L comes from, I don't know.)
> The L stands for LONG. It's from the days when Windows still
> exposed the Intel segmented architecture, so there were SHORT
> and LONG variants of pointers.
Interesting example of why you shouldn't use Hungarian
notation:-).
Hendrik Schober
>
> [LPVOID]
>
> Interesting example of why you shouldn't use Hungarian
> notation:-).
But this isn't HN. ('lpvMyPtr' would be.)
Schobi
Pete Becker
Well, yes, but Microsoft certainly promoted it as Hungarian Notation,
despite Charles Simonyi's disclaimers.
Hendrik Schober
> On 2008-11-30 06:53:48 -0500, Hendrik Schober <spam...@gmx.de> said:
>
>> James Kanze wrote:
>>> [LPVOID]
>>>
>>> Interesting example of why you shouldn't use Hungarian
>>> notation:-).
>> But this isn't HN. ('lpvMyPtr' would be.)
>>
>
> Well, yes, but Microsoft certainly promoted it as Hungarian Notation,
> despite Charles Simonyi's disclaimers.
I thought I knew the difference between Charles' idea (i.e
prefixing with meaning) and its application at MS (prefixing
with type). I hadn't thought that HN -- in the MS sense of
the word -- applied to types.
Schobi