Related
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 4 years ago.
Improve this question
Is it ever acceptable to have a memory leak in your C or C++ application?
What if you allocate some memory and use it until the very last line of code in your application (for example, a global object's destructor)? As long as the memory consumption doesn't grow over time, is it OK to trust the OS to free your memory for you when your application terminates (on Windows, Mac, and Linux)? Would you even consider this a real memory leak if the memory was being used continuously until it was freed by the OS.
What if a third party library forced this situation on you? Would refuse to use that third party library no matter how great it otherwise might be?
I only see one practical disadvantage, and that is that these benign leaks will show up with memory leak detection tools as false positives.
No.
As professionals, the question we should not be asking ourselves is, "Is it ever OK to do this?" but rather "Is there ever a good reason to do this?" And "hunting down that memory leak is a pain" isn't a good reason.
I like to keep things simple. And the simple rule is that my program should have no memory leaks.
That makes my life simple, too. If I detect a memory leak, I eliminate it, rather than run through some elaborate decision tree structure to determine whether it's an "acceptable" memory leak.
It's similar to compiler warnings – will the warning be fatal to my particular application? Maybe not.
But it's ultimately a matter of professional discipline. Tolerating compiler warnings and tolerating memory leaks is a bad habit that will ultimately bite me in the rear.
To take things to an extreme, would it ever be acceptable for a surgeon to leave some piece of operating equipment inside a patient?
Although it is possible that a circumstance could arise where the cost/risk of removing that piece of equipment exceeds the cost/risk of leaving it in, and there could be circumstances where it was harmless, if I saw this question posted on SurgeonOverflow.com and saw any answer other than "no," it would seriously undermine my confidence in the medical profession.
–
If a third party library forced this situation on me, it would lead me to seriously suspect the overall quality of the library in question. It would be as if I test drove a car and found a couple loose washers and nuts in one of the cupholders – it may not be a big deal in itself, but it portrays a lack of commitment to quality, so I would consider alternatives.
I don't consider it to be a memory leak unless the amount of memory being "used" keeps growing. Having some unreleased memory, while not ideal, is not a big problem unless the amount of memory required keeps growing.
Let's get our definitions correct, first. A memory leak is when memory is dynamically allocated, eg with malloc(), and all references to the memory are lost without the corresponding free. An easy way to make one is like this:
#define BLK ((size_t)1024)
while(1){
void * vp = malloc(BLK);
}
Note that every time around the while(1) loop, 1024 (+overhead) bytes are allocated, and the new address assigned to vp; there's no remaining pointer to the previous malloc'ed blocks. This program is guaranteed to run until the heap runs out, and there's no way to recover any of the malloc'ed memory. Memory is "leaking" out of the heap, never to be seen again.
What you're describing, though, sound like
int main(){
void * vp = malloc(LOTS);
// Go do something useful
return 0;
}
You allocate the memory, work with it until the program terminates. This is not a memory leak; it doesn't impair the program, and all the memory will be scavenged up automagically when the program terminates.
Generally, you should avoid memory leaks. First, because like altitude above you and fuel back at the hangar, memory that has leaked and can't be recovered is useless; second, it's a lot easier to code correctly, not leaking memory, at the start than it is to find a memory leak later.
In theory no, in practise it depends.
It really depends on how much data the program is working on, how often the program is run and whether or not it is running constantly.
If I have a quick program that reads a small amount of data makes a calculation and exits, a small memory leak will never be noticed. Because the program is not running for very long and only uses a small amount of memory, the leak will be small and freed when the program exists.
On the other hand if I have a program that processes millions of records and runs for a long time, a small memory leak might bring down the machine given enough time.
As for third party libraries that have leaks, if they cause a problem either fix the library or find a better alternative. If it doesn't cause a problem, does it really matter?
Many people seem to be under the impression that once you free memory, it's instantly returned to the operating system and can be used by other programs.
This isn't true. Operating systems commonly manage memory in 4KiB pages. malloc and other sorts of memory management get pages from the OS and sub-manage them as they see fit. It's quite likely that free() will not return pages to the operating system, under the assumption that your program will malloc more memory later.
I'm not saying that free() never returns memory to the operating system. It can happen, particularly if you are freeing large stretches of memory. But there's no guarantee.
The important fact: If you don't free memory that you no longer need, further mallocs are guaranteed to consume even more memory. But if you free first, malloc might re-use the freed memory instead.
What does this mean in practice? It means that if you know your program isn't going to require any more memory from now on (for instance it's in the cleanup phase), freeing memory is not so important. However if the program might allocate more memory later, you should avoid memory leaks - particularly ones that can occur repeatedly.
Also see this comment for more details about why freeing memory just before termination is bad.
A commenter didn't seem to understand that calling free() does not automatically allow other programs to use the freed memory. But that's the entire point of this answer!
So, to convince people, I will demonstrate an example where free() does very little good. To make the math easy to follow, I will pretend that the OS manages memory in 4000 byte pages.
Suppose you allocate ten thousand 100-byte blocks (for simplicity I'll ignore the extra memory that would be required to manage these allocations). This consumes 1MB, or 250 pages. If you then free 9000 of these blocks at random, you're left with just 1000 blocks - but they're scattered all over the place. Statistically, about 5 of the pages will be empty. The other 245 will each have at least one allocated block in them. That amounts to 980KB of memory, that cannot possibly be reclaimed by the operating system - even though you now only have 100KB allocated!
On the other hand, you can now malloc() 9000 more blocks without increasing the amount of memory your program is tying up.
Even when free() could technically return memory to the OS, it may not do so. free() needs to achieve a balance between operating quickly and saving memory. And besides, a program that has already allocated a lot of memory and then freed it is likely to do so again. A web server needs to handle request after request after request - it makes sense to keep some "slack" memory available so you don't need to ask the OS for memory all the time.
There is nothing conceptually wrong with having the os clean up after the application is run.
It really depends on the application and how it will be run. Continually occurring leaks in an application that needs to run for weeks has to be taken care of, but a small tool that calculates a result without too high of a memory need should not be a problem.
There is a reason why many scripting language do not garbage collect cyclical references… for their usage patterns, it's not an actual problem and would thus be as much of a waste of resources as the wasted memory.
I believe the answer is no, never allow a memory leak, and I have a few reasons which I haven't seen explicitly stated. There are great technical answers here but I think the real answer hinges on more social/human reasons.
(First, note that as others mentioned, a true leak is when your program, at any point, loses track of memory resources that it has allocated. In C, this happens when you malloc() to a pointer and let that pointer leave scope without doing a free() first.)
The important crux of your decision here is habit. When you code in a language that uses pointers, you're going to use pointers a lot. And pointers are dangerous; they're the easiest way to add all manner of severe problems to your code.
When you're coding, sometimes you're going to be on the ball and sometimes you're going to be tired or mad or worried. During those somewhat distracted times, you're coding more on autopilot. The autopilot effect doesn't differentiate between one-off code and a module in a larger project. During those times, the habits you establish are what will end up in your code base.
So no, never allow memory leaks for the same reason that you should still check your blind spots when changing lanes even if you're the only car on the road at the moment. During times when your active brain is distracted, good habits are all that can save you from disastrous missteps.
Beyond the "habit" issue, pointers are complex and often require a lot of brain power to track mentally. It's best to not "muddy the water" when it comes to your usage of pointers, especially when you're new to programming.
There's a more social aspect too. By proper use of malloc() and free(), anyone who looks at your code will be at ease; you're managing your resources. If you don't, however, they'll immediately suspect a problem.
Maybe you've worked out that the memory leak doesn't hurt anything in this context, but every maintainer of your code will have to work that out in his head too when he reads that piece of code. By using free() you remove the need to even consider the issue.
Finally, programming is writing a mental model of a process to an unambiguous language so that a person and a computer can perfectly understand said process. A vital part of good programming practice is never introducing unnecessary ambiguity.
Smart programming is flexible and generic. Bad programming is ambiguous.
I'm going to give the unpopular but practical answer that it's always wrong to free memory unless doing so will reduce the memory usage of your program. For instance a program that makes a single allocation or series of allocations to load the dataset it will use for its entire lifetime has no need to free anything. In the more common case of a large program with very dynamic memory requirements (think of a web browser), you should obviously free memory you're no longer using as soon as you can (for instance closing a tab/document/etc.), but there's no reason to free anything when the user selects clicks "exit", and doing so is actually harmful to the user experience.
Why? Freeing memory requires touching memory. Even if your system's malloc implementation happens not to store metadata adjacent to the allocated memory blocks, you're likely going to be walking recursive structures just to find all the pointers you need to free.
Now, suppose your program has worked with a large volume of data, but hasn't touched most of it for a while (again, web browser is a great example). If the user is running a lot of apps, a good portion of that data has likely been swapped to disk. If you just exit(0) or return from main, it exits instantly. Great user experience. If you go to the trouble of trying to free everything, you may spend 5 seconds or more swapping all the data back in, only to throw it away immediately after that. Waste of user's time. Waste of laptop's battery life. Waste of wear on the hard disk.
This is not just theoretical. Whenever I find myself with too many apps loaded and the disk starts thrashing, I don't even consider clicking "exit". I get to a terminal as fast as I can and type killall -9 ... because I know "exit" will just make it worse.
I think in your situation the answer may be that it's okay. But you definitely need to document that the memory leak is a conscious decision. You don't want a maintenance programmer to come along, slap your code inside a function, and call it a million times. So if you make the decision that a leak is okay you need to document it (IN BIG LETTERS) for whoever may have to work on the program in the future.
If this is a third party library you may be trapped. But definitely document that this leak occurs.
But basically if the memory leak is a known quantity like a 512 KB buffer or something then it is a non issue. If the memory leak keeps growing like every time you call a library call your memory increases by 512KB and is not freed, then you may have a problem. If you document it and control the number of times the call is executed it may be manageable. But then you really need documentation because while 512 isn't much, 512 over a million calls is a lot.
Also you need to check your operating system documentation. If this was an embedded device there may be operating systems that don't free all the memory from a program that exits. I'm not sure, maybe this isn't true. But it is worth looking into.
I'm sure that someone can come up with a reason to say Yes, but it won't be me.
Instead of saying no, I'm going to say that this shouldn't be a yes/no question.
There are ways to manage or contain memory leaks, and many systems have them.
There are NASA systems on devices that leave the earth that plan for this. The systems will automatically reboot every so often so that memory leaks will not become fatal to the overall operation. Just an example of containment.
If you allocate memory and use it until the last line of your program, that's not a leak. If you allocate memory and forget about it, even if the amount of memory isn't growing, that's a problem. That allocated but unused memory can cause other programs to run slower or not at all.
I can count on one hand the number of "benign" leaks that I've seen over time.
So the answer is a very qualified yes.
An example. If you have a singleton resource that needs a buffer to store a circular queue or deque but doesn't know how big the buffer will need to be and can't afford the overhead of locking or every reader, then allocating an exponentially doubling buffer but not freeing the old ones will leak a bounded amount of memory per queue/deque. The benefit for these is they speed up every access dramatically and can change the asymptotics of multiprocessor solutions by never risking contention for a lock.
I've seen this approach used to great benefit for things with very clearly fixed counts such as per-CPU work-stealing deques, and to a much lesser degree in the buffer used to hold the singleton /proc/self/maps state in Hans Boehm's conservative garbage collector for C/C++, which is used to detect the root sets, etc.
While technically a leak, both of these cases are bounded in size and in the growable circular work stealing deque case there is a huge performance win in exchange for a bounded factor of 2 increase in the memory usage for the queues.
If you allocate a bunch of heap at the beginning of your program, and you don't free it when you exit, that is not a memory leak per se. A memory leak is when your program loops over a section of code, and that code allocates heap and then "loses track" of it without freeing it.
In fact, there is no need to make calls to free() or delete right before you exit. When the process exits, all of its memory is reclaimed by the OS (this is certainly the case with POSIX. On other OSes – particularly embedded ones – YMMV).
The only caution I'd have with not freeing the memory at exit time is that if you ever refactor your program so that it, for example, becomes a service that waits for input, does whatever your program does, then loops around to wait for another service call, then what you've coded can turn into a memory leak.
In this sort of question context is everything. Personally I can't stand leaks, and in my code I go to great lengths to fix them if they crop up, but it is not always worth it to fix a leak, and when people are paying me by the hour I have on occasion told them it was not worth my fee for me to fix a leak in their code. Let me give you an example:
I was triaging a project, doing some perf work and fixing a lot of bugs. There was a leak during the applications initialization that I tracked down, and fully understood. Fixing it properly would have required a day or so refactoring a piece of otherwise functional code. I could have done something hacky (like stuffing the value into a global and grabbing it some point I know it was no longer in use to free), but that would have just caused more confusion to the next guy who had to touch the code.
Personally I would not have written the code that way in the first place, but most of us don't get to always work on pristine well designed codebases, and sometimes you have to look at these things pragmatically. The amount of time it would have taken me to fix that 150 byte leak could instead be spent making algorithmic improvements that shaved off megabytes of ram.
Ultimately, I decided that leaking 150 bytes for an app that used around a gig of ram and ran on a dedicated machine was not worth fixing it, so I wrote a comment saying that it was leaked, what needed to be changed in order to fix it, and why it was not worth it at the time.
this is so domain-specific that its hardly worth answering. use your freaking head.
space shuttle operating system: nope, no memory leaks allowed
rapid development proof-of-concept code: fixing all those memory leaks is a waste of time.
and there is a spectrum of intermediate situations.
the opportunity cost ($$$) of delaying a product release to fix all but the worst memory leaks is usually dwarfs any feelings of being "sloppy or unprofessional". Your boss pays you to make him money, not to get a warm, fuzzy feelings.
You have to first realize that there's a big difference between a perceived memory leak and an actual memory leak. Very frequently analysis tools will report many red herrings, and label something as having been leaked (memory or resources such as handles etc) where it actually isn't. Often times this is due to the analysis tool's architecture. For example, certain analysis tools will report run time objects as memory leaks because it never sees those object freed. But the deallocation occurs in the runtime's shutdown code, which the analysis tool might not be able to see.
With that said, there will still be times when you will have actual memory leaks that are either very difficult to find or very difficult to fix. So now the question becomes is it ever OK to leave them in the code?
The ideal answer is, "no, never." A more pragmatic answer may be "no, almost never." Very often in real life you have limited number of resources and time to resolve and endless list of tasks. When one of the tasks is eliminating memory leaks, the law of diminishing returns very often comes in to play. You could eliminate say 98% of all memory leaks in an application in a week, but the remaining 2% might take months. In some cases it might even be impossible to eliminate certain leaks because of the application's architecture without a major refactoring of code. You have to weigh the costs and benefits of eliminating the remaining 2%.
While most answers concentrate on real memory leaks (which are not OK ever, because they are a sign of sloppy coding), this part of the question appears more interesting to me:
What if you allocate some memory and use it until the very last line of code in your application (for example, a global object's deconstructor)? As long as the memory consumption doesn't grow over time, is it OK to trust the OS to free your memory for you when your application terminates (on Windows, Mac, and Linux)? Would you even consider this a real memory leak if the memory was being used continuously until it was freed by the OS.
If the associated memory is used, you cannot free it before the program ends. Whether the free is done by the program exit or by the OS does not matter. As long as this is documented, so that change don't introduce real memory leaks, and as long as there is no C++ destructor or C cleanup function involved in the picture. A not-closed file might be revealed through a leaked FILE object, but a missing fclose() might also cause the buffer not to be flushed.
So, back to the original case, it is IMHO perfectly OK in itself, so much that Valgrind, one of the most powerful leak detectors, will treat such leaks only if requested. On Valgrind, when you overwrite a pointer without freeing it beforehand, it gets considered as a memory leak, because it is more likely to happen again and to cause the heap to grow endlessly.
Then, there are not nfreed memory blocks which are still reachable. One could make sure to free all of them at the exit, but that is just a waste of time in itself. The point is if they could be freed before. Lowering memory consumption is useful in any case.
Even if you are sure that your 'known' memory leak will not cause havoc, don't do it. At best, it will pave a way for you to make a similar and probably more critical mistake at a different time and place.
For me, asking this is like questioning "Can I break the red light at 3 AM in the morning when no one is around?". Well sure, it may not cause any trouble at that time but it will provide a lever for you to do the same in rush hour!
No, you should not have leaks that the OS will clean for you. The reason (not mentioned in the answers above as far as I could check) is that you never know when your main() will be re-used as a function/module in another program. If your main() gets to be a frequently-called function in another persons' software - this software will have a memory leak that eats memory over time.
KIV
I agree with vfilby – it depends. In Windows, we treat memory leaks as relatively serous bugs. But, it very much depends on the component.
For example, memory leaks are not very serious for components that run rarely, and for limited periods of time. These components run, do theire work, then exit. When they exit all their memory is freed implicitly.
However, memory leaks in services or other long run components (like the shell) are very serious. The reason is that these bugs 'steal' memory over time. The only way to recover this is to restart the components. Most people don't know how to restart a service or the shell – so if their system performance suffers, they just reboot.
So, if you have a leak – evaluate its impact two ways
To your software and your user's experience.
To the system (and the user) in terms of being frugal with system resources.
Impact of the fix on maintenance and reliability.
Likelihood of causing a regression somewhere else.
Foredecker
I'm surprised to see so many incorrect definitions of what a memory leak actually is. Without a concrete definition, a discussion on whether it's a bad thing or not will go nowhere.
As some commentors have rightly pointed out, a memory leak only happens when memory allocated by a process goes out of scope to the extent that the process is no longer able to reference or delete it.
A process which is grabbing more and more memory is not necessarily leaking. So long as it is able to reference and deallocate that memory, then it remains under the explicit control of the process and has not leaked. The process may well be badly designed, especially in the context of a system where memory is limited, but this is not the same as a leak. Conversely, losing scope of, say, a 32 byte buffer is still a leak, even though the amount of memory leaked is small. If you think this is insignificant, wait until someone wraps an algorithm around your library call and calls it 10,000 times.
I see no reason whatsoever to allow leaks in your own code, however small. Modern programming languages such as C and C++ go to great lengths to help programmers prevent such leaks and there is rarely a good argument not to adopt good programming techniques - especially when coupled with specific language facilities - to prevent leaks.
As regards existing or third party code, where your control over quality or ability to make a change may be highly limited, depending on the severity of the leak, you may be forced to accept or take mitigating action such as restarting your process regularly to reduce the effect of the leak.
It may not be possible to change or replace the existing (leaking) code, and therefore you may be bound to accept it. However, this is not the same as declaring that it's OK.
I guess it's fine if you're writing a program meant to leak memory (i.e. to test the impact of memory leaks on system performance).
Its really not a leak if its intentional and its not a problem unless its a significant amount of memory, or could grow to be a significant amount of memory. Its fairly common to not cleanup global allocations during the lifetime of a program. If the leak is in a server or long running app, grows over time, then its a problem.
I think you've answered your own question. The biggest drawback is how they interfere with the memory leak detection tools, but I think that drawback is a HUGE drawback for certain types of applications.
I work with legacy server applications that are supposed to be rock solid but they have leaks and the globals DO get in the way of the memory detection tools. It's a big deal.
In the book "Collapse" by Jared Diamond, the author wonders about what the guy was thinking who cut down the last tree on Easter Island, the tree he would have needed in order to build a canoe to get off the island. I wonder about the day many years ago when that first global was added to our codebase. THAT was the day it should have been caught.
I see the same problem as all scenario questions like this: What happens when the program changes, and suddenly that little memory leak is being called ten million times and the end of your program is in a different place so it does matter? If it's in a library then log a bug with the library maintainers, don't put a leak into your own code.
I'll answer no.
In theory, the operating system will clean up after you if you leave a mess (now that's just rude, but since computers don't have feelings it might be acceptable). But you can't anticipate every possible situation that might occur when your program is run. Therefore (unless you are able to conduct a formal proof of some behaviour), creating memory leaks is just irresponsible and sloppy from a professional point of view.
If a third-party component leaks memory, this is a very strong argument against using it, not only because of the imminent effect but also because it shows that the programmers work sloppily and that this might also impact other metrics. Now, when considering legacy systems this is difficult (consider web browsing components: to my knowledge, they all leak memory) but it should be the norm.
Historically, it did matter on some operating systems under some edge cases. These edge cases could exist in the future.
Here's an example, on SunOS in the Sun 3 era, there was an issue if a process used exec (or more traditionally fork and then exec), the subsequent new process would inherit the same memory footprint as the parent and it could not be shrunk. If a parent process allocated 1/2 gig of memory and didn't free it before calling exec, the child process would start using that same 1/2 gig (even though it wasn't allocated). This behavior was best exhibited by SunTools (their default windowing system), which was a memory hog. Every app that it spawned was created via fork/exec and inherited SunTools footprint, quickly filling up swap space.
This was already discussed ad nauseam. Bottom line is that a memory leak is a bug and must be fixed. If a third party library leaks memory, it makes one wonder what else is wrong with it, no? If you were building a car, would you use an engine that is occasionally leaking oil? After all, somebody else made the engine, so it's not your fault and you can't fix it, right?
Generally a memory leak in a stand alone application is not fatal, as it gets cleaned up when the program exits.
What do you do for Server programs that are designed so they don't exit?
If you are the kind of programmer that does not design and implement code where the resources are allocated and released correctly, then I don't want anything to do with you or your code. If you don't care to clean up your leaked memory, what about your locks? Do you leave them hanging out there too? Do you leave little turds of temporary files laying around in various directories?
Leak that memory and let the program clean it up? No. Absolutely not. It's a bad habit, that leads to bugs, bugs, and more bugs.
Clean up after yourself. Yo momma don't work here no more.
As a general rule, if you've got memory leaks that you feel you can't avoid, then you need to think harder about object ownership.
But to your question, my answer in a nutshell is In production code, yes. During development, no. This might seem backwards, but here's my reasoning:
In the situation you describe, where the memory is held until the end of the program, it's perfectly okay to not release it. Once your process exits, the OS will clean up anyway. In fact, it might make the user's experience better: In a game I've worked on, the programmers thought it would be cleaner to free all the memory before exiting, causing the shutdown of the program to take up to half a minute! A quick change that just called exit() instead made the process disappear immediately, and put the user back to the desktop where he wanted to be.
However, you're right about the debugging tools: They'll throw a fit, and all the false positives might make finding your real memory leaks a pain. And because of that, always write debugging code that frees the memory, and disable it when you ship.
We have a big multi-threaded C++ application running on Linux. We see that occupied by the application memory grows fast and believe there are some leaks. We have tried every tool we have (valgrind, DynLeak, Purify) but did not find anything. Since this application can run on Windows, we have also tried Bounds Checker. Did not help, too.
We need a new tool that can help. I've looked at Google Perfomrance Tools, MMGR by Paul Nettle, MemCheck Deluxe. None of them impressed me.
Is there anywhere a good tool for this task?
The definition of a memory leak in C/C++ is very specific: it is memory that has been allocated and then the pointer was overwritten or otherwise lost. Valgrind generally detects such cases out of the box, but things are not always that simple.
Your application could very well be still using that memory. In that case you might have what a Java programmer would consider a leak, e.g. entering data in a structure and rarely (or never) removing entries.
You might be measuring the memory usage of your memory incorrectly. On Linux memory usage measurements are not as straight-forward as they seem. How have you measured your memory usage?
You should consider using the application hooks (Valgrind calls them client requests) of whatever memory analysis tool your are using, to avoid the issue with reports only being issued at program termination. Using those hooks might help you pin-point the location of your leak.
You should try using a heap profiler, such as massif from Valgrind, to look for memory allocation locations with inordinate amounts of allocated memory.
Make sure you are not using a custom allocator or garbage collector in your application. As far as I know, no memory analysis tool will work with a custom allocator without user interference.
If your memory leak is massive enough to be detectable within an acceptable amount of application run-time, you could try a binary search of old revisions through your version control system to identify the commit that introduced the problem. At least Mercurial
and Git offer built-in support for this task.
If by "did not help" you mean it did not report memory leaks, it is quite possible you don't have one and just use more and more memory that is still referenced by pointers and can be deleted.
To help you debug the problem, perhaps in your logging, you should also write memory size, number of objects (their type) and a few other stats which are useful to you. At least until you become more familiar with the tools you mentioned.
I'm writing a program (a theorem prover as it happens) whose memory requirement is "as much as possible, please"; that is, it can always do better by using more memory, for practical purposes without upper bound, so what it actually needs to do is use just as much memory as is available, no more and no less. I can figure out how to prioritize data to delete the lowest value stuff when memory runs short; the problem I'm trying to solve is how to tell when this is happening.
Ideally I would like a system call that returns "how much memory is left" or "are we out of memory yet?"; as far as I can tell, no such thing exists?
Of course, malloc can signal out of memory by returning 0 and new can call a handler; these aren't ideal signals, but would be better than nothing. A problem, however, is that I really want to know when physical memory is running out, so I can avoid going deep into swap and thereby making everything grind to a halt; I don't suppose there's any way to ask "are we having to swap yet?" or tell the operating system "don't swap on my account, just fail my requests if it comes to that"?
Another approach would be to find out how much RAM is in the machine, and monitor how much memory the program is using at the moment. As far as I know, there is generally no way to tell the former? I also get the impression there is no reliable way to tell the latter except by wrapping malloc/free with a bookkeeper function (which is then more problematic in C++).
Are there any approaches I'm missing?
The ideal would be a portable solution, but I suspect that's not going to happen. Failing that, a solution that works on Windows and another one that works on Unix would be nice. Failing that, I could get by with a solution that works on Windows and another one that works on Linux.
I think the most useful and flexible way to use all the memory available is to let the user specify how much memory to use.
Let the user write it in a config file or through an interface, then create an allocator (or something similar) that will not provide more than this memory.
That way, you don't have to find statistics about the current computer as this will allways be biased by the fact that the OS could also run other programs as well. Don't even talk about the way the OS will manage cache, the differences between 32 and 64 bit making adress space limit your allocations etc.
In the end, human intelligence (assuming the user know about the context of use) is cheaper to implement when provided by the user.
To find out how much system memory is still unused, under Linux you can parse the file /proc/meminfo and look for a line starting with "MemFree:". Under Windows you can use GlobalMemoryStatusEx http://msdn.microsoft.com/en-us/library/aa366589%28VS.85%29.aspx
Relying on malloc to return 0 when no memory is available might cause problems on Linux, because Linux overcommits memory allocations. malloc will usually return a valid pointer (unless the process is out of virtual address space), but accessing the memory it points to may trigger the "OOM killer", a mechanism that kills your process or another process on the system. The system administrator can tune this behavior.
The best solution I can think of might be to query how many page faults have occurred within, say, the last second. If there's a lot of swapping going on, you should probably release some memory, and if not, you can try allocating more memory.
On Windows, WMI can probably give you some statistics you can use.
But it's a tough problem, since there is no hard limit you can ask the OS for and then stay below. You can keep allocating memory far beyond the point where you've run out of physical memory, which just means you'll cripple your process with excessive swapping.
So the best you can really do is some kind of approximation.
You can keep allocating memory beyond the point where it is useful to do so - i.e. that which requires the OS to swap, or page out important things. The trouble is, it is not necessarily easy to tell where this is.
Also, if your task does any (significant) IO, you will need to have some left for the OS buffers.
I recommend just examining how much there is in the machine, then allocating an amount as a function of that (proportion, or leave some free etc).
Unless you're programming parts of an OS or an embedded system are there any reasons to do so? I can imagine that for some particular classes that are created and destroyed frequently overloading memory management functions or introducing a pool of objects might lower the overhead, but doing these things globally?
Addition
I've just found a bug in an overloaded delete function - memory wasn't always freed. And that was in a not-so memory critical application. Also, disabling these overloads decreases performance by ~0.5% only.
We overload the global new and delete operators where I work for many reasons:
pooling all small allocations -- decreases overhead, decreases fragmentation, can increase performance for small-alloc-heavy apps
framing allocations with a known lifetime -- ignore all the frees until the very end of this period, then free all of them together (admittedly we do this more with local operator overloads than global)
alignment adjustment -- to cacheline boundaries, etc
alloc fill -- helping to expose usage of uninitialized variables
free fill -- helping to expose usage of previously deleted memory
delayed free -- increasing the effectiveness of free fill, occasionally increasing performance
sentinels or fenceposts -- helping to expose buffer overruns, underruns, and the occasional wild pointer
redirecting allocations -- to account for NUMA, special memory areas, or even to keep separate systems separate in memory (for e.g. embedded scripting languages or DSLs)
garbage collection or cleanup -- again useful for those embedded scripting languages
heap verification -- you can walk through the heap data structure every N allocs/frees to make sure everything looks ok
accounting, including leak tracking and usage snapshots/statistics (stacks, allocation ages, etc)
The idea of new/delete accounting is really flexible and powerful: you can, for example, record the entire callstack for the active thread whenever an alloc occurs, and aggregate statistics about that. You could ship the stack info over the network if you don't have space to keep it locally for whatever reason. The types of info you can gather here are only limited by your imagination (and performance, of course).
We use global overloads because it's convenient to hang lots of common debugging functionality there, as well as make sweeping improvements across the entire app, based on the statistics we gather from those same overloads.
We still do use custom allocators for individual types too; in many cases the speedup or capabilities you can get by providing custom allocators for e.g. a single point-of-use of an STL data structure far exceeds the general speedup you can get from the global overloads.
Take a look at some of the allocators and debugging systems that are out there for C/C++ and you'll rapidly come up with these and other ideas:
valgrind
electricfence
dmalloc
dlmalloc
Application Verifier
Insure++
BoundsChecker
...and many others... (the gamedev industry is a great place to look)
(One old but seminal book is Writing Solid Code, which discusses many of the reasons you might want to provide custom allocators in C, most of which are still very relevant.)
Obviously if you can use any of these fine tools you will want to do so rather than rolling your own.
There are situations in which it is faster, easier, less of a business/legal hassle, nothing's available for your platform yet, or just more instructive: dig in and write a global overload.
The most common reason to overload new and delete are simply to check for memory leaks, and memory usage stats. Note that "memory leak" is usually generalized to memory errors. You can check for things such as double deletes and buffer overruns.
The uses after that are usually memory-allocation schemes, such as garbage collection, and pooling.
All other cases are just specific things, mentioned in other answers (logging to disk, kernel use).
In addition to the other important uses mentioned here, like memory tagging, it's also the only way to force all allocations in your app to go through fixed-block allocation, which has enormous implications for performance and fragmentation.
For example, you may have a series of memory pools with fixed block sizes. Overriding global new lets you direct all 61-byte allocations to, say, the pool with 64-byte blocks, all 768-1024 byte allocs to the the 1024b-block pool, all those above that to the 2048 byte block pool, and anything larger than 8kb to the general ragged heap.
Because fixed block allocators are much faster and less prone to fragmentation than allocating willy-nilly from the heap, this lets you force even crappy 3d party code to allocate from your pools and not poop all over the address space.
This is done often in systems which are time- and space-critical, such as games. 280Z28, Meeh, and Dan Olson have described why.
UnrealEngine3 overloads global new and delete as part of its core memory management system. There are multiple allocators that provide different features (profiling, performance, etc.) and they need all allocations to go through it.
Edit: For my own code, I would only ever do it as a last resort. And by that I mean I would almost positively never use it. But my personal projects are obviously much smaller/very different requirements.
Some realtime systems overload them to avoid them being used after init..
Overloading new & delete makes it possible to add a tag to your memory allocations. I tag allocations per system or control or by middleware. I can view, at runtime, how much each uses. Maybe I want to see the usage of a parser separated from the UI or how much a piece of middleware is really using!
You can also use it to put guard bands around the allocated memory. If/when your app crashes you can take a look at the address. If you see the contents as "0xABCDABCD" (or whatever you choose as guard) you are accessing memory you don't own.
Perhaps after calling delete you can fill this space with a similarly recognizable pattern.
I believe VisualStudio does something similar in debug. Doesn't it fill uninitialized memory with 0xCDCDCDCD?
Finally, if you have fragmentation issues you could use it to redirect to a block allocator? I am not sure how often this is really a problem.
You need to overload them when the call to new and delete doesn't work in your environment.
For example, in kernel programming, the default new and delete don't work as they rely on user mode library to allocate memory.
From a practical standpoint it may just be better to override malloc on a system library level, since operator new will probably be calling it anyway.
On linux, you can put your own version of malloc in place of the system one, as in this example here:
http://developers.sun.com/solaris/articles/lib_interposers.html
In that article, they are trying to collect performance statistics. But you may also detect memory leaks if you also override free.
Since you are doing this in a shared library with LD_PRELOAD, you don't even need to recompile your application.
I've seen it done in a system that for 'security'* reasons was required to write over all memory it used on de-allocation. The approach was to allocate an extra few bytes at the start of each block of memory which would contain the size of the overall block which would then be overwritten with zeros on delete.
This had a number of problems as you can probably imagine but it did work (mostly) and saved the team from reviewing every single memory allocation in a reasonably large, existing application.
Certainly not saying that it is a good use but it is probably one of the more imaginative ones out there...
* sadly it wasn't so much about actual security as the appearance of security...
Photoshop plugins written in C++ should override operator new so that they obtain memory via Photoshop.
I've done it with memory mapped files so that data written to the memory is automatically also saved to disk.
It's also used to return memory at a specific physical address if you have memory mapped IO devices, or sometimes if you need to allocate a certain block of contiguous memory.
But 99% of the time it's done as a debugging feature to log how often, where, when memory is being allocated and released.
It's actually pretty common for games to allocate one huge chunk of memory from the system and then provide custom allocators via overloaded new and delete. One big reason is that consoles have a fixed memory size, making both leaks and fragmentation large problems.
Usually (at least on a closed platform) the default heap operations come with a lack of control and a lack of introspection. For many applications this doesn't matter, but for games to run stably in fixed-memory situations the added control and introspection are both extremely important.
It can be a nice trick for your application to be able to respond to low memory conditions by something else than a random crash. To do this your new can be a simple proxy to the default new that catches its failures, frees up some stuff and tries again.
The simplest technique is to reserve a blank block of memory at start-up time for that very purpose. You may also have some cache you can tap into - the idea is the same.
When the first allocation failure kicks in, you still have time to warn your user about the low memory conditions ("I'll be able to survive a little longer, but you may want to save your work and close some other applications"), save your state to disk, switch to survival mode, or whatever else makes sense in your context.
The most common use case is probably leak checking.
Another use case is when you have specific requirements for memory allocation in your environment which are not satisfied by the standard library you are using, like, for instance, you need to guarantee that memory allocation is lock free in a multi threaded environment.
As many have already stated this is usually done in performance critical applications, or to be able to control memory alignment or track your memory. Games frequently use custom memory managers, especially when targeting specific platforms/consoles.
Here is a pretty good blog post about one way of doing this and some reasoning.
Overloaded new operator also enables programmers to squeeze some extra performance out of their programs. For example, In a class, to speed up the allocation of new nodes, a list of deleted nodes is maintained so that their memory can be reused when new nodes are allocated.In this case, the overloaded delete operator will add nodes to the list of deleted nodes and the overloaded new operator will allocate memory from this list rather than from the heap to speedup memory allocation. Memory from the heap can be used when the list of deleted nodes is empty.
I'm working on a project that is supposed to be used from the command line with the following syntax:
program-name input-file
The program is supposed to process the input, compute some stuff and spit out results on stdout.
My language of choice is C++ for several reasons I'm not willing to debate. The computation phase will be highly symbolic (think compiler) and will use pretty complex dynamically allocated data structures. In particular, it's not amenable to RAII style programming.
I'm wondering if it is acceptable to forget about freeing memory, given that I expect the entire computation to consume less than the available memory and that the OS is free to reclaim all the memory in one step after the program finishes (assume program terminates in seconds). What are your feeling about this?
As a backup plan, if ever my project will require to run as a server or interactively, I figured that I can always refit a garbage collector into the source code. Does anyone have experience using garbage collectors for C++? Do they work well?
It shouldn't cause any problems in the specific situation described the question.
However, it's not exactly normal. Static analysis tools will complain about it. Most importantly, it builds bad habits.
Sometimes not deallocating memory is the right thing to do.
I used to write compilers. After building the parse tree and traversing it to write the intermediate code, we would simply just exit. Deallocating the tree would have
added a bit of slowness to the compiler, which we wanted of course to be as fast as possible.
taken up code space
taken time to code and test the deallocators
violated the "no code executes better than 'no code'" dictum.
HTH! FWIW, this was "back in the day" when memory was non-virtual and minimal, the boxes were much slower, and the first two were non-trivial considerations.
My feeling would be something like "WTF!!!"
Look at it this way:
You choose a programming language that does not include a garbage collector, we are not allowed to ask why.
You are basically stating that you are too lazy to care about freeing the memory.
Well, WTF again. Laziness isn't a good reason for anything, the least of what is playing around with memory without freeing it.
Just free the memory, it's a bad practice, the scenario may change and then can be a million reasons you can need that memory freed and the only reason for not doing it is laziness, don't get bad habits, and get used to do things right, that way you'll tend to do them right in the future!!
Not deallocating memory should not be problem but it is a bad practice.
Joel Coehoorn is right:
It shouldn't cause any problems.
However, it's not exactly normal.
Static analysis tools will complain
about it. Most importantly, it builds
bad habits.
I'd also like to add that thinking about deallocation as you write the code is probably a lot easier than trying to retrofit it afterwards. So I would probably make it deallocate memory; you don't know how your program might be used in future.
If you want a really simple way to free memory, look at the "pools" concept that Apache uses.
Well, I think that it's not acceptable. You've already alluded to potential future problems yourself. Don't think they're necessarily easy to solve.
Things like “… given that I expect the entire computation to consume less …” are famous last phrases. Similarly, refitting code with some feature is one of these things they all talk of and never do.
Not deallocating memory might sound good in the short run but can potentially create a huge load of problems in the long run. Personally, I just don't think that's worth it.
There are two strategies. Either you build in the GC design from the very beginning. It's more work but it will pay off. For a lot of small objects it might pay to use a pool allocator and just keep track of the memory pool. That way, you can keep track of the memory consumption and simply avoid a lot of problems that similar code, but without allocation pool, would create.
Or you use smart pointers throughout the program from the beginning. I actually prefer this method even though it clutters the code. One solution is to rely heavily on templates, which takes out a lot of redundancy when referring to types.
Take a look at projects such as WebKit. Their computation phase resembles yours since they build parse trees for HTML. They use smart pointers throughout their program.
Finally: “It’s a question of style … Sloppy work tends to be habit-forming.”
– Silk in Castle of Wizardry by David Eddings.
will use pretty complex dynamically
allocated data structures. In
particular, it's not amenable to RAII
style programming.
I'm almost sure that's an excuse for lazy programming. Why can't you use RAII? Is it because you don't want to keep track of your allocations, there's no pointer to them that you keep? If so, how do you expect to use the allocated memory - there's always a pointer to it that contains some data.
Is it because you don't know when it should be released? Leave the memory in RAII objects, each one referenced by something, and they'll all trickle-down free each other when the containing object gets freed - this is particularly important if you want to run it as a server one day, each iteration of the server effective runs a 'master' object that holds all others so you can just delete it and all the memory disappears. It also helps prevent you retro-fitting a GC.
Is it because all your memory is allocated and kept in-use all the time, and only freed at the end? If so see above.
If you really, really cannot think of a design where you cannot leak memory, at least have the decency to use a private heap. Destroy that heap before you quit and you'll have a better design already, if a little 'hacky'.
There are instances where memory leaks are ok - static variables, globally initialised data, things like that. These aren't generally large though.
Reference counting smart pointers like shared_ptr in boost and TR1 could also help you manage your memory in a simple manner.
The drawback is that you have to wrap every pointers that use these objects.
I've done this before, only to find that, much later, I needed the program to be able to process several inputs without separate commands, or that the guts of the program were so useful that they needed to be turned into a library routine that could be called many times from within another program that was not expected to terminate. It was much harder to go back later and re-engineer the program than it would have been to make it leak-less from the start.
So, while it's technically safe as you've described the requirements, I advise against the practice since it's likely that your requirements may someday change.
If the run time of your program is very short, it should not be a problem. However, being too lazy to free what you allocate and losing track of what you allocate are two entirely different things. If you have simply lost track, its time to ask yourself if you actually know what your code is doing to a computer.
If you are just in a hurry or lazy and the life of your program is small in relation to what it actually allocates (i.e. allocating 10 MB per second is not small if running for 30 seconds) .. then you should be OK.
The only 'noble' argument regarding freeing allocated memory sets in when a program exits .. should one free everything to keep valgrind from complaining about leaks, or just let the OS do it? That entirely depends on the OS and if your code might become a library and not a short running executable.
Leaks during run time are generally bad, unless you know your program will run in a short amount of time and not cause other programs far more important than your's as far as the OS is concerned to skid to dirty paging.
What are your feeling about this?
Some O/Ses might not reclaim the memory, but I guess you're not intenting to run on those O/Ses.
As a backup plan, if ever my project will require to run as a server or interactively, I figured that I can always refit a garbage collector into the source code.
Instead, I figure you can spawn a child process to do the dirty work, grab the output from the child process, let the child process die as soon as possible after that and then expect the O/S to do the garbage collection.
I have not personally used this, but since you are starting from scratch you may wish to consider the Boehm-Demers-Weiser conservative garbage collector
The answer really depends on how large your program will be and what performance characteristics it needs to exhibit. If you never deallocate memory, your process's memory footprint will be much larger than it would otherwise be. Depeding on the system, this could cause a lot of paging and slow down the performance for you or other applications on the system.
Beyond that, what everyone above says is correct. It probably won't cause harm in the short term, but it's a bad practice that you should avoid. You'll never be able to use the code again. Trying to retrofit a GC on afterwards will be a nightmare. Just think about going to each place you allocate memory and trying to retrofit it but not break anything.
One more reason to avoid doing this: reputation. If you fail to deallocate, everyone who maintains the code will curse your name and your rep in the company will take a hit. "Can you believe how dumb he was? Look at this code."
If it is non-trivial for you to determine where to deallocate the memory, I would be concerned that other aspects of the data structure manipulation may not be fully understood either.
Apart from the fact that the OS (kernel and/or C/C++ library) can choose not to free the memory when the execution ends, your application should always provide proper freeing of allocated memory as a good practice. Why? Suppose you decide to extend that application or reuse the code; you'll quickly get in trouble if the code you had previously written hogs up the memory unnecessarily, after finishing its job. It's a recipe for memory leaks.
In general, I agree it's a bad practice.
For a one shot program, it can be OK, but it kinda shows like you don't what you are doing.
There is one solution to your problem though - use a custom allocator, which preallocates larger blocks from malloc, and then, after the computation phase, instead of freeing all the little blocks from you custom allocator, just release the larger preallocated blocks of memory. Then you don't need to keep track of all objects you need to deallocate and when. One guy who wrote a compiler too explained this approach many years ago to me, so if it worked for him, it will probably work for you as well.
Try to use automatic variables in methods so that they will be freed automatically from the stack.
The only useful reason to not free heap memory is to save a tiny amount of computational power used in the free() method. You might loose any advantage if page faults become an issue due to large virtual memory needs with small physical memory resources. Some factors to consider are:
If you are allocating a few huge chunks of memory or many small chunks.
Is the memory going to need to be locked into physical memory.
Are you absolutely positive the code and memory needed will fit into 2GB, for a Win32 system, including memory holes and padding.
That's generally a bad idea. You might encounter some cases where the program will try to consume more memory than it's available. Plus you risk being unable to start several copies of the program.
You can still do this if you don't care of the mentioned issues.
When you exit from a program, the memory allocated is automatically returned to the system. So you may not deallocate the memory you had allocated.
But deallocations becomes necessary when you go for bigger programs such as an OS or Embedded systems where the program is meant to run forever & hence a small memory leak can be malicious.
Hence it is always recommended to deallocate the memory you have allocated.