Board index » cppbuilder » access function in dll

2008-04-30 05:43:59 AM
cppbuilder78
I have a .dll which was intended for use in a VC++ app. I don't have vc++
but I have to asume that it compiles ok.
I need to use this .dll in my app. The thing is that it contains an
essential function that is not specifically exported (ie I run impdef on the
.dll and the func is not listed. I run implib and examine it (WinHex) and
there is no trace of the function name. I have tried loading the dll both
statically and dynamically but I can't get to the function. If I examine
the actual .dll I can see the required function name but only in one place,
whereas all the other function names occur twice (one in the export list and
once later).
So question is, does vc++ have some other mechanism to access this function,
and can I tap into it?
NB the function is clearly intended to be exported; it is published in the
.h file and is clearly doccumented, indeed it is essential to the
application.
Thanks
 

"Denville Longhurst" < XXXX@XXXXX.COM >wrote in message
you using a PE explorer tool?
it.
The DLL could have simply been compiled using the same .h file, where the .h
file contains declarations for its internal functions as well as its public
functions.
Gambit
 

"Denville Longhurst" < XXXX@XXXXX.COM >wrote in message
listing of what is exported.
Leo Havmøller.
 

{smallsort}

Thanks for that, it is as I thought.
The function is absolutely fundamental, the support documentation states
that it must be the first function called to get a handle to one or more
embedded devices attached to the network. All subsequent functions rely on
this handle. Without it no other function call can succeed. I can get to
all of the other functions in the normal ways.
I examined the .dll using a program called WinHex which provides a hex and
an ascii dump of any file. All of the functions within the dll are listed
twice, once in what is obviously the export list, and again later I presume
at the implementation. This particular function is only listed once, not in
the export list.
I think the dll supplied with this sdk must be broken, I just wanted to
check that MS hadn't added some new mechanism that I wasn't aware of.
Thanks again for your time.
Denville.
"Remy Lebeau (TeamB)" < XXXX@XXXXX.COM >wrote in message

Thanks for that, a very interesting tool.
No, my required function is not listed in the dll, so I guess they have
supplied an old or broken dll with the sdk.
Thanks again.
Denville.
"Leo Havmøller" < XXXX@XXXXX.COM >wrote in message

"Denville Longhurst" < XXXX@XXXXX.COM >wrote in message
likely goofed up when compiling it.
actual PE viewing tool instead. That will show you the actual exports table
(and imports, resources, etc).
listed in.
to memory addresses by the compiler. The exports table is a look-up table
that maps the exported name to the cooreponding address within the DLL.
this a Microsoft DLL, or a third-party DLL?
Gambit
 

I am certain now that, as you say, the dll writer goofed up. As well as
looking 'inside' the dll with WinHex I have used IMPDEF, IMPLIB and
DEPENDS.EXE, none of which include the function among those exported.
For info, the .dll comes from a firm called Digi, manufacturers of embedded
network devices, some of which I have programmed for special functionality
(all that is fine). As part of the SDK they supply a .dll containing
functions that a PC/Windows app can use to discover and list all units on
the network, and it is this dll that is missing the essential function. I
can see that what has happened is that the dll has recently been modified by
the addition of a new version of this function and that's where, somehow, it
got gooed up.
Unfortunately the sample source using the dll is writen for VC++ which I
don't have, but I bet if I did, it wouldn't compile. I was just a bit
worried that Microsoft might have sneaked some new access mechanism into
VC++.
Following your previous reply I had the confidence to put the manufacturer
on the spot and they are now (I hope) taking it seriously.
Tks again,
Denville.
"Remy Lebeau (TeamB)" < XXXX@XXXXX.COM >wrote in message

"Denville Longhurst" < XXXX@XXXXX.COM >wrote in message
update the documentation accordingly? Assuming the new function is exported
properly, that is.
Gambit
 

No, the problem is that the new function -isn't- exported and the old
function (which is) is depreciated and doesn't work any more.
Denville.
"Remy Lebeau (TeamB)" < XXXX@XXXXX.COM >wrote in message