Memory leak

A memory leak means that a program allocates more memory than necessary for its execution. Although Haskell implementations use garbage collectors, programmers must still keep memory management in mind. A garbage collector can reliably prevent dangling pointers, but it is easily possible to produce memory leaks, especially in connection with lazy evaluation. Note that a leak will not only consume more and more memory but it will also slow down the garbage collector considerably! Maybe it is even the reason for the widely spread opinion that garbage collectors are slow or not suited for realtime applications.

Types of leaks

Holding a reference for a too long time

Consider for example:

let xs = [1..1000000::Integer]
in  sum xs * product xs

Since most Haskell compilers expect, that the programmer used let in order to share xs between the call of sum and the call of product, the list xs is completely materialized and hold in memory. However, the list xs is very cheap to compute, and thus it would reduce memory usage considerably, if xs is recomputed for both calls. Since we want to avoid code duplication, we like to achieve this by turning the list definition into a function with a dummy argument.

let makeXs n = [1..n::Integer]
in  sum (makeXs 1000000) * product (makeXs 1000000)

Building up unevaluated expressions

Another typical cause of memory leaks are unevaluated expressions, the classical example being to sum up the numbers of a list (known as sum function).

foldl (+) 0 [1..1000000::Integer]

The problem is, that the runtime system does not know, whether the intermediate sums are actually needed at a later point, and thus it leaves them unevaluated. I.e. it stores something equivalent to 1+2+3+4 instead of just 10. You may be lucky that the strictness analyzer already removes the laziness at compile time, but in general you cannot rely on it. The safe way is to use seq to force evaluation of intermediate sums. This is done by foldl'.

foldl' (+) 0 [1..1000000::Integer]

Detection of memory leaks

A memory leak can be detected by writing a test that should require only a limitted amount of memory and then run the compiled program with restricted heap size. E.g. you can restrict the heap size to 4 MB like in this example:

$ ./mytest +RTS -M4m -RTS

A note on GHCi

If you are noticing a space leak while running your code within GHCi, please note that interpreted code behaves differently from compiled code: even when using `seq`.

Consider starting ghci as follows:

$ ghci -fobject-code

For this, see Compiling to object code inside GHCi.

See also

• Blog post "Anatomy of a thunk leak"
• Blog post "Space leak zoo"
• Haskell Cafe on a space leak caused by the garbage collector that did not recognize a selector-like function call
• Haskell libraries on Make lines stricter to fix a space leak