blob: b772418bf06486638a19781ef8cc5149c70ba4cc [file] [log] [blame]
Catalin Marinas04f70332009-06-11 13:22:39 +01001Kernel Memory Leak Detector
2===========================
3
4Introduction
5------------
6
7Kmemleak provides a way of detecting possible kernel memory leaks in a
8way similar to a tracing garbage collector
9(http://en.wikipedia.org/wiki/Garbage_collection_%28computer_science%29#Tracing_garbage_collectors),
10with the difference that the orphan objects are not freed but only
11reported via /sys/kernel/debug/kmemleak. A similar method is used by the
12Valgrind tool (memcheck --leak-check) to detect the memory leaks in
13user-space applications.
Wang YanQing4762c982014-04-03 14:50:38 -070014Kmemleak is supported on x86, arm, powerpc, sparc, sh, microblaze, ppc, mips, s390, metag and tile.
Catalin Marinas04f70332009-06-11 13:22:39 +010015
16Usage
17-----
18
19CONFIG_DEBUG_KMEMLEAK in "Kernel hacking" has to be enabled. A kernel
Catalin Marinasbab4a342009-06-26 17:38:26 +010020thread scans the memory every 10 minutes (by default) and prints the
Catalin Marinas4698c1f2009-06-26 17:38:27 +010021number of new unreferenced objects found. To display the details of all
22the possible memory leaks:
Catalin Marinas04f70332009-06-11 13:22:39 +010023
24 # mount -t debugfs nodev /sys/kernel/debug/
25 # cat /sys/kernel/debug/kmemleak
26
Catalin Marinas4698c1f2009-06-26 17:38:27 +010027To trigger an intermediate memory scan:
28
29 # echo scan > /sys/kernel/debug/kmemleak
30
Luis R. Rodriguez30b37102009-09-04 17:44:51 -070031To clear the list of all current possible memory leaks:
32
33 # echo clear > /sys/kernel/debug/kmemleak
34
35New leaks will then come up upon reading /sys/kernel/debug/kmemleak
36again.
37
Catalin Marinas04f70332009-06-11 13:22:39 +010038Note that the orphan objects are listed in the order they were allocated
39and one object at the beginning of the list may cause other subsequent
40objects to be reported as orphan.
41
42Memory scanning parameters can be modified at run-time by writing to the
43/sys/kernel/debug/kmemleak file. The following parameters are supported:
44
45 off - disable kmemleak (irreversible)
Catalin Marinase0a2a162009-06-26 17:38:25 +010046 stack=on - enable the task stacks scanning (default)
Catalin Marinas04f70332009-06-11 13:22:39 +010047 stack=off - disable the tasks stacks scanning
Catalin Marinase0a2a162009-06-26 17:38:25 +010048 scan=on - start the automatic memory scanning thread (default)
Catalin Marinas04f70332009-06-11 13:22:39 +010049 scan=off - stop the automatic memory scanning thread
Catalin Marinase0a2a162009-06-26 17:38:25 +010050 scan=<secs> - set the automatic memory scanning period in seconds
51 (default 600, 0 to stop the automatic scanning)
Catalin Marinas4698c1f2009-06-26 17:38:27 +010052 scan - trigger a memory scan
Luis R. Rodriguez30b37102009-09-04 17:44:51 -070053 clear - clear list of current memory leak suspects, done by
Li Zefanc89da702014-04-03 14:46:27 -070054 marking all current reported unreferenced objects grey,
55 or free all kmemleak objects if kmemleak has been disabled.
Catalin Marinas189d84e2009-08-27 14:29:15 +010056 dump=<addr> - dump information about the object found at <addr>
Catalin Marinas04f70332009-06-11 13:22:39 +010057
58Kmemleak can also be disabled at boot-time by passing "kmemleak=off" on
59the kernel command line.
60
Catalin Marinasa9d90582009-06-25 10:16:11 +010061Memory may be allocated or freed before kmemleak is initialised and
62these actions are stored in an early log buffer. The size of this buffer
63is configured via the CONFIG_DEBUG_KMEMLEAK_EARLY_LOG_SIZE option.
64
Catalin Marinas04f70332009-06-11 13:22:39 +010065Basic Algorithm
66---------------
67
68The memory allocations via kmalloc, vmalloc, kmem_cache_alloc and
69friends are traced and the pointers, together with additional
Wang YanQing4762c982014-04-03 14:50:38 -070070information like size and stack trace, are stored in a rbtree.
Catalin Marinas04f70332009-06-11 13:22:39 +010071The corresponding freeing function calls are tracked and the pointers
72removed from the kmemleak data structures.
73
74An allocated block of memory is considered orphan if no pointer to its
75start address or to any location inside the block can be found by
76scanning the memory (including saved registers). This means that there
77might be no way for the kernel to pass the address of the allocated
78block to a freeing function and therefore the block is considered a
79memory leak.
80
81The scanning algorithm steps:
82
83 1. mark all objects as white (remaining white objects will later be
84 considered orphan)
85 2. scan the memory starting with the data section and stacks, checking
Wang YanQing4762c982014-04-03 14:50:38 -070086 the values against the addresses stored in the rbtree. If
Catalin Marinas04f70332009-06-11 13:22:39 +010087 a pointer to a white object is found, the object is added to the
88 gray list
89 3. scan the gray objects for matching addresses (some white objects
90 can become gray and added at the end of the gray list) until the
91 gray set is finished
92 4. the remaining white objects are considered orphan and reported via
93 /sys/kernel/debug/kmemleak
94
95Some allocated memory blocks have pointers stored in the kernel's
96internal data structures and they cannot be detected as orphans. To
97avoid this, kmemleak can also store the number of values pointing to an
98address inside the block address range that need to be found so that the
99block is not considered a leak. One example is __vmalloc().
100
Luis R. Rodriguez30b37102009-09-04 17:44:51 -0700101Testing specific sections with kmemleak
102---------------------------------------
103
104Upon initial bootup your /sys/kernel/debug/kmemleak output page may be
105quite extensive. This can also be the case if you have very buggy code
106when doing development. To work around these situations you can use the
107'clear' command to clear all reported unreferenced objects from the
108/sys/kernel/debug/kmemleak output. By issuing a 'scan' after a 'clear'
109you can find new unreferenced objects; this should help with testing
110specific sections of code.
111
112To test a critical section on demand with a clean kmemleak do:
113
114 # echo clear > /sys/kernel/debug/kmemleak
115 ... test your kernel or modules ...
116 # echo scan > /sys/kernel/debug/kmemleak
117
118Then as usual to get your report with:
119
120 # cat /sys/kernel/debug/kmemleak
121
Li Zefanc89da702014-04-03 14:46:27 -0700122Freeing kmemleak internal objects
123---------------------------------
124
125To allow access to previosuly found memory leaks after kmemleak has been
126disabled by the user or due to an fatal error, internal kmemleak objects
127won't be freed when kmemleak is disabled, and those objects may occupy
128a large part of physical memory.
129
130In this situation, you may reclaim memory with:
131
132 # echo clear > /sys/kernel/debug/kmemleak
133
Catalin Marinas04f70332009-06-11 13:22:39 +0100134Kmemleak API
135------------
136
137See the include/linux/kmemleak.h header for the functions prototype.
138
139kmemleak_init - initialize kmemleak
140kmemleak_alloc - notify of a memory block allocation
Catalin Marinasf528f0b2011-09-26 17:12:53 +0100141kmemleak_alloc_percpu - notify of a percpu memory block allocation
Catalin Marinas04f70332009-06-11 13:22:39 +0100142kmemleak_free - notify of a memory block freeing
Catalin Marinasf528f0b2011-09-26 17:12:53 +0100143kmemleak_free_part - notify of a partial memory block freeing
144kmemleak_free_percpu - notify of a percpu memory block freeing
Catalin Marinasffe2c742014-06-06 14:38:17 -0700145kmemleak_update_trace - update object allocation stack trace
Catalin Marinas04f70332009-06-11 13:22:39 +0100146kmemleak_not_leak - mark an object as not a leak
147kmemleak_ignore - do not scan or report an object as leak
148kmemleak_scan_area - add scan areas inside a memory block
149kmemleak_no_scan - do not scan a memory block
150kmemleak_erase - erase an old value in a pointer variable
151kmemleak_alloc_recursive - as kmemleak_alloc but checks the recursiveness
152kmemleak_free_recursive - as kmemleak_free but checks the recursiveness
153
154Dealing with false positives/negatives
155--------------------------------------
156
157The false negatives are real memory leaks (orphan objects) but not
158reported by kmemleak because values found during the memory scanning
159point to such objects. To reduce the number of false negatives, kmemleak
160provides the kmemleak_ignore, kmemleak_scan_area, kmemleak_no_scan and
161kmemleak_erase functions (see above). The task stacks also increase the
162amount of false negatives and their scanning is not enabled by default.
163
164The false positives are objects wrongly reported as being memory leaks
165(orphan). For objects known not to be leaks, kmemleak provides the
166kmemleak_not_leak function. The kmemleak_ignore could also be used if
167the memory block is known not to contain other pointers and it will no
168longer be scanned.
169
170Some of the reported leaks are only transient, especially on SMP
171systems, because of pointers temporarily stored in CPU registers or
172stacks. Kmemleak defines MSECS_MIN_AGE (defaulting to 1000) representing
173the minimum age of an object to be reported as a memory leak.
174
175Limitations and Drawbacks
176-------------------------
177
178The main drawback is the reduced performance of memory allocation and
179freeing. To avoid other penalties, the memory scanning is only performed
180when the /sys/kernel/debug/kmemleak file is read. Anyway, this tool is
181intended for debugging purposes where the performance might not be the
182most important requirement.
183
184To keep the algorithm simple, kmemleak scans for values pointing to any
185address inside a block's address range. This may lead to an increased
186number of false negatives. However, it is likely that a real memory leak
187will eventually become visible.
188
189Another source of false negatives is the data stored in non-pointer
190values. In a future version, kmemleak could only scan the pointer
191members in the allocated structures. This feature would solve many of
192the false negative cases described above.
193
194The tool can report false positives. These are cases where an allocated
195block doesn't need to be freed (some cases in the init_call functions),
196the pointer is calculated by other methods than the usual container_of
197macro or the pointer is stored in a location not scanned by kmemleak.
198
Daniel Baluta21b86bd2011-04-04 14:58:03 -0700199Page allocations and ioremap are not tracked.