|
I have a solid guess, that it is a bug in compiler. Look at the next code:
int xmin=max(xl1, max(xl2, xl3));
It works fine in the debug version, but gives me some nonsense in the release one. At that very time this variant works properly in both versions:
int xmin=xl1;
if(xmin<xl2)
xmin=xl2;
if(xmin<xl3)
xmin=xl3;
Any idea how could it be?
|
|
|
|
|
a_matseevsky wrote: but gives me some nonsense in the release one What about providing some input/ouput in order to show us the nonsense?
Veni, vidi, vici.
|
|
|
|
|
OK, but I do not think, that it will make you happier.
As a matter of fact, there was two similar lines of code:
int xmin=max(il1, max(il2,il3));
int xmax=min(ir1, min(ir2,ir3));
The right variant was xmin=80, xmax=262.
Wrong variant, presenting only in release version, was the next:
xmin=76, xmax=38.
-- modified 16-Jan-14 20:37pm.
|
|
|
|
|
Actually it would make me happier, anyway you did not provide the input values.
Veni, vidi, vici.
|
|
|
|
|
Have you defined your own version of max() or are you using the one provided in some windows header? If the former, try commenting out that function and see if it still compiles in debug and/or release. If the latter, try defining your own function and see if it does compile in debug and/or release.
As an alternative to either of the above you can try and #define NOMINMAX before including windows.h . This will prevent the macro definitions for min() and max() .
These suggestions are based on the fact that MS provides macros for min() and max() in some Windows header, and that these macros might mess up code in unexpected ways. I know it did break std::valarray::min() and std::valarray::max() in VS 2003, and it did break std::min and std::max in later versions, until MS eventually fixed it.
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|
Thanks, boys, for yours comments, but I've just found the reason. And it did not make me happy. Really, great knowledge means great pain. Look at this small part of a stack frame:
_lrct$ = -212 ; size = 16
tv5476 = -204 ; size = 8
Do you see something interesting here? A RECT structure of size 16 bytes and something too close to it- in fact, variable tv5476 partially overlaps RECT. And what command
fst QWORD PTR tv5476[ebp]
does? It rewrites half of my RECT. I can only hope, that it is the bug in my own Visual Studio. I'll reinstall it- may be, it will help (once I met something like tis- a "new" operator refused to work at all and re installation resolved this situation)
|
|
|
|
|
I think you are reading that the wrong way round. The variable tv5476 is 8 bytes long starting at offset -204 , so it goes from -204 to -212. The variable _lrct$ is 16 bytes from -212 to -228. Your problem is much more likely to be a bug in your code that only shows up in the release version, and that is far from uncommon.
[edit] OK, my shoulders are broad enough to admit that I was wrong with that first statement. Lack of coffee/gin. [/edit]
Veni, vidi, abiit domum
modified 17-Jan-14 12:11pm.
|
|
|
|
|
Richard MacCutchan wrote: Your problem is much more likely to be a bug in your code that only shows up in the release version, and that is far from uncommon. Yes.
However his actual code, like his I/O values are 'top secret'.
Veni, vidi, vici.
|
|
|
|
|
Veni, vidi, abiit domum
|
|
|
|
|
You, boys, made my face red. Not because of me- you even do not know assembler. Command like
mov [ebp-4], eax
rewrites bytes from [ebp-4] UP to [ebp-1] .
Just as fst QWORD PTR [ebp-204] does.
RECT was allocated at [ebp-212], therefore its 16 bytes occupied addresses from ebp-212 to ebp-196.
So simple. And aforementioned fst overwrote half of them. And BTW, I saw the whole process under debugger. My recommendation- any boy, who pretended to be a pro, must know assembler. I know what the mainstream in programming is- to replace pros by cheap yesterdays scholars. And it really does not me happy.
|
|
|
|
|
Jolly good for you. You obviously won't need any help from us in the future then.
Veni, vidi, abiit domum
|
|
|
|
|
I need help. Question is, if any of you can give me some more or less valuable tip. When I asked my question the first time, I was not sure, that a bug really presents in Visual Studio. Now I know it. Problem is, what I have to do in such sorrowful situation. If someone else met such problem and fixed it, such person could share his experience. May be, this bug presents only in my own exemplar of Visual Studio- may be, it was damaged in some way and produces code with defects- in such case re-installation will solve this problem. Or such bug presents in each exemplar of Visual Studio 2005- in such case, I have to throw it away and buy 2008 or 2010. I know about a bug in Borland's compiler.
|
|
|
|
|
a_matseevsky wrote: I need help. Then you could start by showing the actual code that is going wrong, as raised in your original question. Also you need to show the exact values of the variables that cause the problems, and the actual and expected result values.
Veni, vidi, abiit domum
|
|
|
|
|
I did it. Problem was caused by incorrect allocation of local variables in stack frame. This is what I hardly can change- it is a bug in compiler. The only important fact is that one value was replaced by another. Exact values of variables here mean nothing.
|
|
|
|
|
If you continue to refuse to provide the information we have asked for, then there is no way we can offer any suggestions as to what may be wrong. If you are convinced that this is a compiler bug then you should collect all the information and send it to Microsoft.
Veni, vidi, abiit domum
|
|
|
|
|
I clearly demonstrated, what cased problem in particular. I can repeat it again.
There was two local variables:
_lrct$ = -212 ; size = 16
tv5476 = -204 ; size = 8
The next command rewrites 8 bytes of _lrct$
fst QWORD PTR tv5476[ebp]
That's all info. Which data in particular was replaced and by what- means nothing. The only important fact is that data was replaced with something else.
|
|
|
|
|
You have not clearly demonstrated anything. Unless we see the source code there is nothing we can suggest. And, as I said before, if you are convinced that this is a compiler bug, then you should be sending it to Microsoft.
Veni, vidi, abiit domum
|
|
|
|
|
This is not the C/C++ source code.
There is a reason Richard is asking for it: it is all too common among people working "close to the metal" to "over-optimize" their code in a way that is simply wrong, or leads the compiler to produce inieffective or incorrect code. I'm not saying you did that, but you wouldn't be the first nor would you be the last.
Anyway, we can't decide where the error or bug is without seeing the actual C/C++ source code. Nor can anyone provide additional suggestions or help if you don't at the very least offer us the same information that your compiler got.
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|
There are a whole lot of questions that remain unanswered. Either may deliver the true reason for the bug you're experiencing. Compiler bugs are certainly possible, but rather unlikely.
You should really look at the problem again and consider some or all of the following questions:
1. Where did you get the information that
- _lrct$ refers to the start address of the RECT struct you're referring to
- the RECT struct is really 16 bytes in size
- the offsets you show are in fact relative to the same base address
2. Did you derive from your observation of these addresses that part of your data is overwritten, or did you check the actual RECT structure to verify that?
3. What are the original declarations of the C/C++ symbols corresponding to the two addresses _lrct$ and tv5476?
4. How were the two objects allocated?
5. Are you sure that one of them (the RECT) hasn't been deallocated in the meantime? Note that optimizers may discard variables before the end of their lifetime as seen in code if they realize it is no longer used!
I'm sure I could think of more questions, but this could be much more productive if we could see the actual code ...
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|
|
Umm ... yes ... I kind of see it now
I think I'll go ... meditate or something
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|
Dear Stefan, it seems to me, that I have to remind you some theory. Local variables within some procedure may be addressed via ebp or esp. In my particular case they was addressed via ebp. It is much simpler to read such code, because ebp remains constant within a procedure. I some local variable is defined within some block, it becomes inaccessible, when eip leaves this particular block. Its place may be overwritten by something else, that's correct. But simultaneously debugger refuses to show this variable!!! But it showed it- therefore, lrct (16 byte RECT structure) was in its own block, was visible and accessible. But it was partially overwritten- 8 bytes was used as temporary storage for ST(0). That behavior cannot be taken as a normal one. And not lrct only- same fate waited h and maybe something else.
|
|
|
|
|
a_matseevsky wrote: within a procedure
This is just one of many bits of information you haven't really provided. Therefore the wide array of questions. I am well aware that many probably don't make a lot of sense to ask in your particular case if you see the exact code that the compiler used to produce the assembly.
But as repeatedly mentioned, you haven't provided that code, so all we can do is guess and poke in the dark.
In the meantime I've spotted the code that you posted in a different branch of this thread. It isn't the original code though, and doesn't provide sufficient insight to answer any of the questions I posted. You state that variant A doesn't work, but it doesn't use the max() function, so we have to assume that the assembler code generated from that doesn't match the bits you later found to be incorrect (if they are indeed that - we still do not know)
In the code you posted you use some variables that are only referenced over a couple of lines. They may not in fact be stored in the stack at all! As a result, the debugger will not show their contents in the optimized release code. Similarly, if you define a struct with data that is never referenced, the compiler may decide to optimize away the unneeded bits, reducing the size of the struct. You haven't posted the definition of the struct, nor where it's accessed, so it's impossible to tell if that is the case.
I could go on and easily bring up half a dozen or more other optimization techniques that you appear to be unaware of, and that will confound your ability to read useful information from the debugger alone within optimized code. But it's pointless.
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|
Stefan_Lang wrote:
In the meantime I've spotted the code that you posted in a different branch of this thread. It isn't the original code though, and doesn't provide sufficient insight to answer any of the questions I posted. You state that variant A doesn't work, but it doesn't use the max() function, so we have to assume that the assembler code generated from that doesn't match the bits you later found to be incorrect (if they are indeed that - we still do not know)
What do you mean, writing "it isn't the original code?" That I copied it from somewhere or what? I never stated, that variant A does not work. It is your idea. I asked, which works and which not (and why). You was not able to answer this question and preferred to ignore it.
In the code you posted you use some variables that are only referenced over a couple of lines. They may not in fact be stored in the stack at all! As a result, the debugger will not show their contents in the optimized release code. Similarly, if you define a struct with data that is never referenced, the compiler may decide to optimize away the unneeded bits, reducing the size of the struct. You haven't posted the definition of the struct, nor where it's accessed, so it's impossible to tell if that is the case.
All fields of the RECT structute was later used. If compiler decided that they are no more in use, it is just its bug. RECT structure is so well-known, that i did not placed its definition here. It contents 4 fields of type long- top, left, right and bottom. Just 16 bytes. I ran release version under debugger and I saw all assembler commands and order of their execution. Therefore, I know which variables was not stored in a stack and which temporary variables (which I did not declared!) was stored and where.
I could go on and easily bring up half a dozen or more other optimization techniques that you appear to be unaware of, and that will confound your ability to read useful information from the debugger alone within optimized code. But it's pointless.
Pointless is to repeat things like "there are more things on the Earth, on the heaven, than any dreamt of in our philosophy". What makes you think that I do not know about optimization? Stop walking around and answer to one simple question: why local variable, visible and accessible, was overwritten? You cannot name at least one more or less credible reason. If it is not visible, I couldn't get access to it and debugger wouldn't show it. Just this situation happens with variables, which are used in one-two lines of code. Debugger doesn't show them at all (worry not, I know such things).
|
|
|
|
|
You should be aware that asking us which optimized code would work and which wouldn't, without posting the full code with declarations and optimizer settings is ludicrous.
Therefore I decided to derive it based on you statements elsewhere: Variant B contains code that works by your own statement elsewhere. neither contains the max() funtion you refer to in your original question. So B is the working and A the not working - albeit not the original - version. Q. E. D.
As for what is pointless or not: advice is only pointless if it isn't heeded. You believe that the debugger shows you every information correctly? It doesn't. It can't. It's physically impossible.
I've dealt with optimizers 30 years ago: back then it wasn't to hard to anticipate what it would do, and in C you could often emulate pretty much the same without having to invoke the optimizer at all. meaning well optimized C code ran almost as well in debug mode as it did in release.
I've dealt with optimizers 20 years ago, and it got more tricky. Still, well optimized C-code often turned out to be near optimal.
I've dealt with optimizers 10 years ago, and the experience was very different: for one, the same optimizations in C code sometimes led to slower code, because it prevented the optimizer from performing extremely sophisticated optimizations that you'd never have thought of. Plus the size of the codebase made it impossible to optimize all of your code in that way anyway.
At that point I stopped trying to optimize my C/C++ code by hand.
Nowadays, when I check release code that doesn't do the same as debug code, I often find that half the variables aren't on stack, and some others get overwritten at unexpected moments. At least when I look at them through the debugger. It's the optimizer at work! What I usually do is pinpoint the location where everything goes south, then insert some print statements to ensure everything is as I expect it to be - and in 9 cases out of 10, something will not be as I expect it to be! And the debugger won't be able to show it!
That's why I keep telling you not to trust the debugger - and looking at the assembler code doesn't change that you're using the debugger.
GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)
|
|
|
|
|