[deliverable/linux.git] / Documentation / vm / overcommit-accounting

The Linux kernel supports the following overcommit handling modes

0	-	Heuristic overcommit handling. Obvious overcommits of
		address space are refused. Used for a typical system. It
		ensures a seriously wild allocation fails while allowing
		overcommit to reduce swap usage.  root is allowed to 
		allocate slightly more memory in this mode. This is the 
		default.

1	-	Always overcommit. Appropriate for some scientific
		applications. Classic example is code using sparse arrays
		and just relying on the virtual memory consisting almost
		entirely of zero pages.

2	-	Don't overcommit. The total address space commit
		for the system is not permitted to exceed swap + a
		configurable amount (default is 50%) of physical RAM.
		Depending on the amount you use, in most situations
		this means a process will not be killed while accessing
		pages but will receive errors on memory allocation as
		appropriate.

		Useful for applications that want to guarantee their
		memory allocations will be available in the future
		without having to initialize every page.

The overcommit policy is set via the sysctl `vm.overcommit_memory'.

The overcommit amount can be set via `vm.overcommit_ratio' (percentage)
or `vm.overcommit_kbytes' (absolute value).

The current overcommit limit and amount committed are viewable in
/proc/meminfo as CommitLimit and Committed_AS respectively.

Gotchas
-------

The C language stack growth does an implicit mremap. If you want absolute
guarantees and run close to the edge you MUST mmap your stack for the 
largest size you think you will need. For typical stack usage this does
not matter much but it's a corner case if you really really care

In mode 2 the MAP_NORESERVE flag is ignored. 


How It Works
------------

The overcommit is based on the following rules

For a file backed map
	SHARED or READ-only	-	0 cost (the file is the map not swap)
	PRIVATE WRITABLE	-	size of mapping per instance

For an anonymous or /dev/zero map
	SHARED			-	size of mapping
	PRIVATE READ-only	-	0 cost (but of little use)
	PRIVATE WRITABLE	-	size of mapping per instance

Additional accounting
	Pages made writable copies by mmap
	shmfs memory drawn from the same pool

Status
------

o	We account mmap memory mappings
o	We account mprotect changes in commit
o	We account mremap changes in size
o	We account brk
o	We account munmap
o	We report the commit status in /proc
o	Account and check on fork
o	Review stack handling/building on exec
o	SHMfs accounting
o	Implement actual limit enforcement

To Do
-----
o	Account ptrace pages (this is hard)
Commit	Line	Data
	1	The Linux kernel supports the following overcommit handling modes
	2
	3	0 - Heuristic overcommit handling. Obvious overcommits of
	4	address space are refused. Used for a typical system. It
	5	ensures a seriously wild allocation fails while allowing
	6	overcommit to reduce swap usage. root is allowed to
	7	allocate slightly more memory in this mode. This is the
	8	default.
	9
	10	1 - Always overcommit. Appropriate for some scientific
	11	applications. Classic example is code using sparse arrays
	12	and just relying on the virtual memory consisting almost
	13	entirely of zero pages.
	14
	15	2 - Don't overcommit. The total address space commit
	16	for the system is not permitted to exceed swap + a
	17	configurable amount (default is 50%) of physical RAM.
	18	Depending on the amount you use, in most situations
	19	this means a process will not be killed while accessing
	20	pages but will receive errors on memory allocation as
	21	appropriate.
	22
	23	Useful for applications that want to guarantee their
	24	memory allocations will be available in the future
	25	without having to initialize every page.
	26
	27	The overcommit policy is set via the sysctl `vm.overcommit_memory'.
	28
	29	The overcommit amount can be set via `vm.overcommit_ratio' (percentage)
	30	or `vm.overcommit_kbytes' (absolute value).
	31
	32	The current overcommit limit and amount committed are viewable in
	33	/proc/meminfo as CommitLimit and Committed_AS respectively.
	34
	35	Gotchas
	36	-------
	37
	38	The C language stack growth does an implicit mremap. If you want absolute
	39	guarantees and run close to the edge you MUST mmap your stack for the
	40	largest size you think you will need. For typical stack usage this does
	41	not matter much but it's a corner case if you really really care
	42
	43	In mode 2 the MAP_NORESERVE flag is ignored.
	44
	45
	46	How It Works
	47	------------
	48
	49	The overcommit is based on the following rules
	50
	51	For a file backed map
	52	SHARED or READ-only - 0 cost (the file is the map not swap)
	53	PRIVATE WRITABLE - size of mapping per instance
	54
	55	For an anonymous or /dev/zero map
	56	SHARED - size of mapping
	57	PRIVATE READ-only - 0 cost (but of little use)
	58	PRIVATE WRITABLE - size of mapping per instance
	59
	60	Additional accounting
	61	Pages made writable copies by mmap
	62	shmfs memory drawn from the same pool
	63
	64	Status
	65	------
	66
	67	o We account mmap memory mappings
	68	o We account mprotect changes in commit
	69	o We account mremap changes in size
	70	o We account brk
	71	o We account munmap
	72	o We report the commit status in /proc
	73	o Account and check on fork
	74	o Review stack handling/building on exec
	75	o SHMfs accounting
	76	o Implement actual limit enforcement
	77
	78	To Do
	79	-----
	80	o Account ptrace pages (this is hard)