www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Some asm help for the 'thiscall' calling convention?

reply Andrej Mitrovic <none none.none> writes:
I'm in the same situation as the person who posted about this 5 years ago:
http://www.digitalmars.com/d/archives/digitalmars/D/learn/thiscall_calling_convention_4943.html

This isn't so much relevant to COM as it is to this ASIO implementation. The
issue is that only Microsoft compilers can use the 'thiscall' calling
convention which is used with ASIO, while for other compilers your best options
are to either use inline assembly when calling functions, or to compile a DLL
wrapper using a Microsoft compiler (there's OpenAsio which does exactly that).

I could use the OpenAsio DLL wrapper, but I'd like to see if this could be done
with inline asm instead.

Here's what one call looks like for a C++ Borland compiler (see the
Resolver::init wrapper function):
http://codepad.org/rArgvPZC

In D, I can call a function like IASIO.init() without issues, and some other
functions such as "getDriverVersion" which take no parameters will work. But
trying to use functions which take parameters will fail with an access
violation, probably because D uses stdcall for COM methods, while these ASIO
COM methods need to be called with 'thiscall' convention.

I've attempted to translate this to D's inline asm but I get back access
violations. Here's my attempt:
http://codepad.org/gFLJ9RJd

I admit I barely know much asm but I thought I'd give it a shot. Calling
"IASIO.init()" works for me, of course. But calling any IASIO methods which
take parameters will fail since parameters are passed differently in the
'thiscall' convention.

Anyone know how to translate that asm block to D so it's valid?
Apr 23 2011
next sibling parent reply Simon <s.d.hammett gmail.com> writes:
On 24/04/2011 02:23, Andrej Mitrovic wrote:

 I'm in the same situation as the person who posted about this 5 years ago:
http://www.digitalmars.com/d/archives/digitalmars/D/learn/thiscall_calling_convention_4943.html

 This isn't so much relevant to COM as it is to this ASIO implementation. The
issue is that only Microsoft compilers can use the 'thiscall' calling
convention which is used with ASIO, while for other compilers your best options
are to either use inline assembly when calling functions, or to compile a DLL
wrapper using a Microsoft compiler (there's OpenAsio which does exactly that).

 I could use the OpenAsio DLL wrapper, but I'd like to see if this could be
done with inline asm instead.

 Here's what one call looks like for a C++ Borland compiler (see the
Resolver::init wrapper function):
 http://codepad.org/rArgvPZC

 In D, I can call a function like IASIO.init() without issues, and some other
functions such as "getDriverVersion" which take no parameters will work. But
trying to use functions which take parameters will fail with an access
violation, probably because D uses stdcall for COM methods, while these ASIO
COM methods need to be called with 'thiscall' convention.

 I've attempted to translate this to D's inline asm but I get back access
violations. Here's my attempt:
 http://codepad.org/gFLJ9RJd

 I admit I barely know much asm but I thought I'd give it a shot. Calling
"IASIO.init()" works for me, of course. But calling any IASIO methods which
take parameters will fail since parameters are passed differently in the
'thiscall' convention.

 Anyone know how to translate that asm block to D so it's valid?


IASIO is an abstract data type; it's a COM interface.
you can't declare it as a member of your class as you have done:

Your code:

	class Resolver
	{
		IASIO that_;

Borland code:

	class Resolver
	{
	    IASIO* that_;

Notice the *

In your code the struct will be 4 bytes long, while the real struct will 
be a lot bigger.

You can never declare any COM interface as a real structure in your code.

-- 
My enormous talent is exceeded only by my outrageous laziness.
http://www.ssTk.co.uk
Apr 23 2011
next sibling parent Andrej Mitrovic <andrej.mitrovich gmail.com> writes:
Then how come I can create an instance with CoCreateInstance without
the call failing and returning "S_OK" which means the call succeeded,
and I can also call void functions like IASIO.controlPanel() and get
back this:
http://imgur.com/v4Uct
Apr 23 2011
prev sibling parent reply Andrej Mitrovic <andrej.mitrovich gmail.com> writes:
And by calling IASIO.controlPanel(), I mean calling it on the instance.

E.g.:
class Foo
{
    IASIO bar; // can call bar.controlPanel();
}
Apr 23 2011
parent Simon <s.d.hammett gmail.com> writes:
On 24/04/2011 03:25, Andrej Mitrovic wrote:

 And by calling IASIO.controlPanel(), I mean calling it on the instance.

 E.g.:
 class Foo
 {
      IASIO bar; // can call bar.controlPanel();
 }


Not sure how D handles interfaces. If bar is implicitly a pointer
then the bit where you do:

  void *this_ = &that_;

means you are loading a pointer to a pointer. Try losing the &

-- 
My enormous talent is exceeded only by my outrageous laziness.
http://www.ssTk.co.uk
Apr 24 2011
prev sibling parent reply Kagamin <spam here.lot> writes:
Andrej Mitrovic Wrote:


 But trying to use functions which take parameters will fail with an access
violation, probably because D uses stdcall for COM methods, while these ASIO
COM methods need to be called with 'thiscall' convention.


COM uses stdcall convention. Everything else is not COM.
Apr 24 2011
next sibling parent Andrej Mitrovic <andrej.mitrovich gmail.com> writes:
Nevermind guys I'll just use OpenAsio.
Apr 24 2011
prev sibling parent reply Adam Sansier <Adam.Sansier gmail.com> writes:
On Sunday, 24 April 2011 at 22:09:24 UTC, Kagamin wrote:

 Andrej Mitrovic Wrote:


 But trying to use functions which take parameters will fail 
 with an access violation, probably because D uses stdcall for 
 COM methods, while these ASIO COM methods need to be called 
 with 'thiscall' convention.


 COM uses stdcall convention. Everything else is not COM.


Ignore the fool, mark your functions extern(C++) and it will 
work. IUnknown uses extern(Windows) while ASIO uses extern(C++) 
for some reason. It may not be COM in the eyes of God... I mean 
Kagamin, though.
Jul 13 2016
parent reply flamencofantasy <flamencofantasy gmail.com> writes:
On Wednesday, 13 July 2016 at 20:39:00 UTC, Adam Sansier wrote:

 On Sunday, 24 April 2011 at 22:09:24 UTC, Kagamin wrote:

 Andrej Mitrovic Wrote:


 But trying to use functions which take parameters will fail 
 with an access violation, probably because D uses stdcall for 
 COM methods, while these ASIO COM methods need to be called 
 with 'thiscall' convention.


 COM uses stdcall convention. Everything else is not COM.


 Ignore the fool, mark your functions extern(C++) and it will 
 work. IUnknown uses extern(Windows) while ASIO uses extern(C++) 
 for some reason. It may not be COM in the eyes of God... I mean 
 Kagamin, though.


You revived this thread solely to insert a personal attack???
Jul 13 2016
parent reply Adam Sansier <Adam.Sansier gmail.com> writes:
On Wednesday, 13 July 2016 at 22:09:05 UTC, flamencofantasy wrote:

 On Wednesday, 13 July 2016 at 20:39:00 UTC, Adam Sansier wrote:

 On Sunday, 24 April 2011 at 22:09:24 UTC, Kagamin wrote:

 Andrej Mitrovic Wrote:


 But trying to use functions which take parameters will fail 
 with an access violation, probably because D uses stdcall 
 for COM methods, while these ASIO COM methods need to be 
 called with 'thiscall' convention.


 COM uses stdcall convention. Everything else is not COM.


 Ignore the fool, mark your functions extern(C++) and it will 
 work. IUnknown uses extern(Windows) while ASIO uses 
 extern(C++) for some reason. It may not be COM in the eyes of 
 God... I mean Kagamin, though.


 You revived this thread solely to insert a personal attack???


Um, no, I revived it so that people searching for answers 
wouldn't be led astray by idiots who pretend to know everything. 
That's not a personal attack. It's simply pointing out that there 
is a solution. If it happens to point out that someone makes some 
blanket statement is an idiot, that's just icing on the cake. We 
have to point out the idiots so either 1. they stop making such 
blanket statements and acting like they are correct. 2. People 
quickly learn not to listen to them. 3. People know their 
statements are wrong.

I wouldn't be in this position if they didn't put me here. I'm 
sorry if I won't sit idle by and let them spread their ignorance. 
You may see it as a minor thing, but that your ignorance. I'm 
sorry if that offends you, are you one of the idiots?

I'll be damned if I'm suppose to show humility and they aren't. 
It's a two way street.
Jul 13 2016
next sibling parent reply flamencofantasy <flamencofantasy gmail.com> writes:
On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 On Wednesday, 13 July 2016 at 22:09:05 UTC, flamencofantasy 
 wrote:

 On Wednesday, 13 July 2016 at 20:39:00 UTC, Adam Sansier wrote:

 On Sunday, 24 April 2011 at 22:09:24 UTC, Kagamin wrote:

 Andrej Mitrovic Wrote:


 But trying to use functions which take parameters will fail 
 with an access violation, probably because D uses stdcall 
 for COM methods, while these ASIO COM methods need to be 
 called with 'thiscall' convention.


 COM uses stdcall convention. Everything else is not COM.


 Ignore the fool, mark your functions extern(C++) and it will 
 work. IUnknown uses extern(Windows) while ASIO uses 
 extern(C++) for some reason. It may not be COM in the eyes of 
 God... I mean Kagamin, though.


 You revived this thread solely to insert a personal attack???


 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know 
 everything. That's not a personal attack. It's simply pointing 
 out that there is a solution. If it happens to point out that 
 someone makes some blanket statement is an idiot, that's just 
 icing on the cake. We have to point out the idiots so either 1. 
 they stop making such blanket statements and acting like they 
 are correct. 2. People quickly learn not to listen to them. 3. 
 People know their statements are wrong.

 I wouldn't be in this position if they didn't put me here. I'm 
 sorry if I won't sit idle by and let them spread their 
 ignorance. You may see it as a minor thing, but that your 
 ignorance. I'm sorry if that offends you, are you one of the 
 idiots?

 I'll be damned if I'm suppose to show humility and they aren't. 
 It's a two way street.


You have a nasty attitude and you constantly insult people on 
this forum. Whether you call yourself Adam Sansier or Hiemlick 
Hiemlicker or Joerg Joergonson makes no difference. I wonder why 
anyone is still willing to interact with you.
Jul 13 2016
parent reply ethgeh <ehehetrh qkgf.tk> writes:
On Wednesday, 13 July 2016 at 23:06:44 UTC, flamencofantasy wrote:

 On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 On Wednesday, 13 July 2016 at 22:09:05 UTC, flamencofantasy 
 wrote:

 On Wednesday, 13 July 2016 at 20:39:00 UTC, Adam Sansier 
 wrote:

 [...]


 You revived this thread solely to insert a personal attack???


 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know 
 everything. That's not a personal attack. It's simply pointing 
 out that there is a solution. If it happens to point out that 
 someone makes some blanket statement is an idiot, that's just 
 icing on the cake. We have to point out the idiots so either 
 1. they stop making such blanket statements and acting like 
 they are correct. 2. People quickly learn not to listen to 
 them. 3. People know their statements are wrong.

 I wouldn't be in this position if they didn't put me here. I'm 
 sorry if I won't sit idle by and let them spread their 
 ignorance. You may see it as a minor thing, but that your 
 ignorance. I'm sorry if that offends you, are you one of the 
 idiots?

 I'll be damned if I'm suppose to show humility and they 
 aren't. It's a two way street.


 You have a nasty attitude and you constantly insult people on 
 this forum. Whether you call yourself Adam Sansier or Hiemlick 
 Hiemlicker or Joerg Joergonson makes no difference. I wonder 
 why anyone is still willing to interact with you.


Are you sure he's the same ppl ? I had a conversation with 
"Hiemlick Hiemlicker" a month ago and the guy didn't seem to be 
as biased as this one. That said if you search "Joerg Joergonson" 
there's a match for a related Topic (COM & asio). My guess is 
that this guy is an addicted, most of the time the conversations 
are OK but when he's high he sucks a lot.
Jul 13 2016
parent Adam Sansier <Adam.Sansier gmail.com> writes:
On Thursday, 14 July 2016 at 00:51:16 UTC, ethgeh wrote:

 On Wednesday, 13 July 2016 at 23:06:44 UTC, flamencofantasy 
 wrote:

 On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 On Wednesday, 13 July 2016 at 22:09:05 UTC, flamencofantasy 
 wrote:

 On Wednesday, 13 July 2016 at 20:39:00 UTC, Adam Sansier 
 wrote:

 [...]


 You revived this thread solely to insert a personal attack???


 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know 
 everything. That's not a personal attack. It's simply 
 pointing out that there is a solution. If it happens to point 
 out that someone makes some blanket statement is an idiot, 
 that's just icing on the cake. We have to point out the 
 idiots so either 1. they stop making such blanket statements 
 and acting like they are correct. 2. People quickly learn not 
 to listen to them. 3. People know their statements are wrong.

 I wouldn't be in this position if they didn't put me here. 
 I'm sorry if I won't sit idle by and let them spread their 
 ignorance. You may see it as a minor thing, but that your 
 ignorance. I'm sorry if that offends you, are you one of the 
 idiots?

 I'll be damned if I'm suppose to show humility and they 
 aren't. It's a two way street.


 You have a nasty attitude and you constantly insult people on 
 this forum. Whether you call yourself Adam Sansier or Hiemlick 
 Hiemlicker or Joerg Joergonson makes no difference. I wonder 
 why anyone is still willing to interact with you.


 Are you sure he's the same ppl ? I had a conversation with 
 "Hiemlick Hiemlicker" a month ago and the guy didn't seem to be 
 as biased as this one. That said if you search "Joerg 
 Joergonson" there's a match for a related Topic (COM & asio). 
 My guess is that this guy is an addicted, most of the time the 
 conversations are OK but when he's high he sucks a lot.


Well, I'm not sure who all these people are. I figure 
flamencofantasy and ethgeh are alias too!

Regardless, I treat people like they treat others.  I may seem I 
like go overboard, it is possible, but it's only because of the 
butterfly effect! ;)  I'd rather make it painfully obvious to 
them that they way they approach people is wrong than ignore 
it(then things only fester, look at the world today).

Hopefully Make, Kagamin, Ketmar, and any others I've offended 
will grow up a little and realize the way they approach helping 
people is not necessarily the best it can be. Making absolute 
statements rarely help(mostly because they don't even know 
absolute what they are talking about, they only make absolute 
assumptions). Also, in some cases, they quickly ignore the 
persons own experience and write them off in an extremely 
arrogant way(Ketmar is good at this). Even if that is not 
arrogant, it is wrong. Every persons experiences are valid, even 
if they are not as "intelligent" as the next guy(not everyone 
gets to or wants to spend 100K hours of their life sitting in 
front of a computer screen).

Sometimes people just need the shit slapped out of them, simple 
as that, they'll get over it.
Jul 16 2016
prev sibling parent reply Kagamin <spam here.lot> writes:
On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know everything.


My word is not COM specification of course, there's the official 
documentation and tons of books about COM, what one prefers, they 
all say the same thing, one doesn't need to trust me on that. 
This one is a particularly good read: 
https://www.amazon.com/Inside-Microsoft-Programming-Dale-Rog
rson/dp/1572313498/ explains all fundamentals of COM.
Jul 14 2016
next sibling parent reply Adam Sansier <Adam.Sansier gmail.com> writes:
On Thursday, 14 July 2016 at 14:01:29 UTC, Kagamin wrote:

 On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know 
 everything.


 My word is not COM specification of course, there's the 
 official documentation and tons of books about COM, what one 
 prefers, they all say the same thing, one doesn't need to trust 
 me on that. This one is a particularly good read: 
 https://www.amazon.com/Inside-Microsoft-Programming-Dale-Rog
rson/dp/1572313498/ explains all fundamentals of COM.


That isn't the point. The point is you make absolute claims that 
are obviously not true. It is a personality deficiency you have, 
not about trust.

It's like if you claim all prime numbers are odd, state it as 
absolute fact, then when someone says "Hey, I found a prime 
number that is even" you automatically assume they must be wrong. 
Instead of looking at it as objectively. All it does is lead to 
confusion.

Unless you know something to be absolute fact, don't state it as 
such then don't pretend it to be such... Your still doing that by 
trying to show "proof" that you are right.

Yet you are wrong. Regardless what the asio standard does right 
or wrong, it uses COM, correct? Is asio not well established? Yes 
it is. Hence it proves that not all COM is as you think it is. 
Just accept it, learn that there are no absolutes and that you 
can be wrong, and the world would be a better place.
Jul 16 2016
parent Kagamin <spam here.lot> writes:
On Saturday, 16 July 2016 at 20:31:25 UTC, Adam Sansier wrote:

 Yet you are wrong. Regardless what the asio standard does right 
 or wrong, it uses COM, correct? Is asio not well established? 
 Yes it is. Hence it proves that not all COM is as you think it 
 is.


The component must comply with the COM standard to provide a COM 
interface, just using COM somehow somewhere is not enough for it. 
The absolute fact is the COM specification, not my words. If you 
don't like what the COM specification says, argue with it, not 
with me. I suppose ASIO is whatever compiled on the developer's 
machine, he didn't bother about COM.
Jul 18 2016
prev sibling parent reply Andrew Godfrey <X y.com> writes:
On Thursday, 14 July 2016 at 14:01:29 UTC, Kagamin wrote:

 On Wednesday, 13 July 2016 at 22:30:51 UTC, Adam Sansier wrote:

 Um, no, I revived it so that people searching for answers 
 wouldn't be led astray by idiots who pretend to know 
 everything.


 My word is not COM specification of course, there's the 
 official documentation and tons of books about COM, what one 
 prefers, they all say the same thing, one doesn't need to trust 
 me on that. This one is a particularly good read: 
 https://www.amazon.com/Inside-Microsoft-Programming-Dale-Rog
rson/dp/1572313498/ explains all fundamentals of COM.


COM is a model; in practice people pick the parts they need, and 
often still call it "COM". I see it used all the time between 
components that are built with the same compiler build, and so 
can be lax about calling convention. I've also seen cases with 
QueryInterface but lacking AddRef and Release, and I've seen the 
reverse. All of these were called "COM" by many people. So even 
if there's some ISO standard saying that COM must include all 
these elements, saying "everything else is not COM" would hinder 
communication.
Jul 16 2016
parent Kagamin <spam here.lot> writes:
On Saturday, 16 July 2016 at 21:16:29 UTC, Andrew Godfrey wrote:

 COM is a model; in practice people pick the parts they need, 
 and often still call it "COM".


No need to rename what has a name: 
https://en.wikipedia.org/wiki/Interface-based_programming
Jul 18 2016