Jump to content
  • Advertisement
Sign in to follow this  
Endar

Bugs you expected to throw a compiler error ...

This topic is 2554 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I just made a mistake and saw something in C, using Visual Studio 2008 C++ Express, which was so dumb (at least to my mind), that I thought it would throw a compiler error, or at least a warning, but doesn't (at least at the standard error/warning level).


// set the free block mem to a released value
memset( ((u8*)pFree) + sizeof(MemHeapFree),
MEMHEAP_RELEASED_MEM,
pFree->uSize - sizeof(sizeof(MemHeapFree)) );


See it? See the sizeof(sizeof(expression))? I just had a massive WTF moment when I saw that while stepping through the code.

So, this thread is for any other things in C, C++ or any other languages that you think other people would find strange, amusing or unbelievable.

Oh, and for anyone that's interested, the sizeof(sizeof(x)) comes out to 4, which I guess is expected once you know the compiler actually accepts it.

Share this post


Link to post
Share on other sites
Advertisement
Okay, I understand your point and I see the mistake, but IMO letting the compiler flag this as an error would be stupid.
What if I want to know the size of whatever the type is that sizeof () returns?

That's not a common thing to do, but letting the compiler parse for mistakes like that would be horribly slow - and even then, the compiler shouldn't error on something that's perfectly valid. I could accept a warning though, but it still comes down to the same problem - parsing all sorts of potential mistakes like this would only make the compiler slower.

Share this post


Link to post
Share on other sites
The compiler can and should only check for lexical, syntactic or semantic errors that cannot be translated. Your example is perfectly valid source code.

I guess what you're looking for is a static code analyzer.
Here's a list of tools that do that, but I'm not sure any of these will the detect the mistake in your example...
http://en.wikipedia.org/wiki/List_of_tools_for_static_code_analysis

Share this post


Link to post
Share on other sites
Sizeof(sizeof(x)) only evaluates to 4 on a 32-bit system. This looks like it is part of a memory management system and a pointer to the data is stored in a data structure. Doing it this way would ensure there is always enough memory allocated to store the pointer without hard-coding it. Does it make a bit more sense now?

Share this post


Link to post
Share on other sites
There is nothing wrong in that code.

size_t is defined as result of sizeof operation. While somewhat unusual, it's perfectly legit to query the sizeof(sizeof(x)) to determine its size. While not strictly defined, size_t is commonly size of register on that machine or size of addressable space.

Above code says:

Set pFree-uSize bytes at a specific offset to specific value, but leave last sizeof(size_t) bytes intact. perhaps the last 4 are used for some specific purpose.

In C++, the above code might look like std::fill(pFree.begin() + sizeof(MemFree), pFree.end() - sizeof(FreeContainerType::size_t), X);


Bugs like this are annoying, but causes are quite complex. Pointer arithmetic is idiomatic to C and trying to report every such slightly uncommon use of language features would result in a flood of false positives.


I'm not really sure if a C idiom exists to avoid mistakes like that. Especially since it's really not a mistake.

Share this post


Link to post
Share on other sites
There is nothing wrong in that code.
The author knows it's valid. It's not wrong, except for the part where it didn't do what the author intended, meaning it caused a bug (and in a memory allocator a bug is quite a wrong thing).
So, this thread is for any other things in C, C++ or any other languages that you think other people would find strange, amusing or unbelievable.
Did you all miss this part? ^^^
Too busy bikeshedding in Parkinson's committee? I heard you like bike sheds (any colour will do)...

Share this post


Link to post
Share on other sites

[quote name='Antheus' timestamp='1306070656' post='4814217']There is nothing wrong in that code.
The author knows it's valid. It's not wrong, except for the part where it didn't do what the author intended, meaning it caused a bug (and in a memory allocator a bug is quite a wrong thing).
So, this thread is for any other things in C, C++ or any other languages that you think other people would find strange, amusing or unbelievable.
Did you all miss this part? ^^^
Too busy bikeshedding in Parkinson's committee? I heard you like bike sheds (any colour will do)...
[/quote]

Yes yes yes! :)

Guys, you are great :D I do appreciate the objective analysis of the code, but yes, I know it is part of a memory management system. In fact is it my own memory management system that I was fixing, in part because of a bug (not the one I posted) that was in that very function call. And I'm not looking for a static code analyzer, although I do appreciate the offer. I also completely understand why that expression returns 4 on a 32-bit system. I'm also not suggesting that this and the other million similar kinds of bugs that people write into their code every day should be picked up by the compiler, and although I immediately wondered how it had gotten past the compiler, after a half second of looking at it, I understood why it had.

As Hodgman pointed out, after the initial "What the heck?" moment, I just found this funny and was looking to add a bit more humor to the forum. So, please, whatever you've got, I'd definitely like to see, in whatever language or form that you feel like sharing it in. :D

Share this post


Link to post
Share on other sites
In that case, I consider the functions sprintf, strcpy, strcmp etc without a length parameter to be some of the silliest features of C :-) the fact that nobody considered it to ever be a security problem, and that so many programs use these functions in network code and other security sensitive code without a second thought... Now that is daft indeed :-) (yeah, modern compilers sometimes warn about unsafe functions such as these but jees, it took them how many decades to realise?) Edited by braindigitalis

Share this post


Link to post
Share on other sites

In that case, I consider the functions sprintf, strcpy, strcmp etc without a length parameter to be some of the silliest features of C :-) the fact that nobody considered it to ever be a security problem, and that so many programs use these functions in network code and other security sensitive code without a second thought... Now that is daft indeed :-) (yeah, modern compilers sometimes warn about unsafe functions such as these but jees, it took them how many decades to realise?)



I agree, that was kinda strange.

Come on guys and gals, surely more people have something they wish to share with the rest of the class?

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

Participate in the game development conversation and more when you create an account on GameDev.net!

Sign me up!