1 The garbage collector looks at a number of environment variables which are
2 then used to affect its operation. These are examined only on Un*x-like
5 GC_INITIAL_HEAP_SIZE=<bytes> - Initial heap size in bytes. May speed up
8 GC_MAXIMUM_HEAP_SIZE=<bytes> - Maximum collected heap size.
10 GC_LOOP_ON_ABORT - Causes the collector abort routine to enter a tight loop.
11 This may make it easier to debug, such a process, especially
12 for multithreaded platforms that don't produce usable core
13 files, or if a core file would be too large. On some
14 platforms, this also causes SIGSEGV to be caught and
15 result in an infinite loop in a handler, allowing
16 similar debugging techniques.
18 GC_PRINT_STATS - Turn on as much logging as is easily feasible without
19 adding signifcant runtime overhead. Doesn't work if
20 the collector is built with SMALL_CONFIG. Overridden
21 by setting GC_quiet. On by default if the collector
22 was built without -DSILENT.
24 GC_DUMP_REGULARLY - Generate a GC debugging dump GC_dump() on startup
25 and during every collection. Very verbose. Useful
26 if you have a bug to report, but please include only the
29 GC_PRINT_ADDRESS_MAP - Linux only. Dump /proc/self/maps, i.e. various address
30 maps for the process, to stderr on every GC. Useful for
31 mapping root addresses to source for deciphering leak
34 GC_NPROCS=<n> - Linux w/threads only. Explicitly sets the number of processors
35 that the GC should expect to use. Note that setting this to 1
36 when multiple processors are available will preserve
37 correctness, but may lead to really horrible performance,
38 since the lock implementation will immediately yield without
41 GC_MARKERS=<n> - Linux w/threads and parallel marker only. Set the number
42 of marker threads. This is normaly set to the number of
43 processors. It is safer to adjust GC_MARKERS than GC_NPROCS,
44 since GC_MARKERS has no impact on the lock implementation.
46 GC_NO_BLACKLIST_WARNING - Prevents the collector from issuing
47 warnings about allocations of very large blocks.
48 Deprecated. Use GC_LARGE_ALLOC_WARN_INTERVAL instead.
50 GC_LARGE_ALLOC_WARN_INTERVAL=<n> - Print every nth warning about very large
51 block allocations, starting with the nth one. Small values
52 of n are generally benign, in that a bounded number of
53 such warnings generally indicate at most a bounded leak.
54 For best results it should be set at 1 during testing.
55 Default is 5. Very large numbers effectively disable the
58 GC_IGNORE_GCJ_INFO - Ignore the type descriptors implicitly supplied by
59 GC_gcj_malloc and friends. This is useful for debugging
60 descriptor generation problems, and possibly for
61 temporarily working around such problems. It forces a
62 fully conservative scan of all heap objects except
63 those known to be pointerfree, and may thus have other
66 GC_PRINT_BACK_HEIGHT - Print max length of chain through unreachable objects
67 ending in a reachable one. If this number remains
68 bounded, then the program is "GC robust". This ensures
69 that a fixed number of misidentified pointers can only
70 result in a bounded space leak. This currently only
71 works if debugging allocation is used throughout.
72 It increases GC space and time requirements appreciably.
73 This feature is still somewhat experimental, and requires
74 that the collector have been built with MAKE_BACK_GRAPH
75 defined. For details, see Boehm, "Bounding Space Usage
76 of Conservative Garbage Collectors", POPL 2001, or
77 http://lib.hpl.hp.com/techpubs/2001/HPL-2001-251.html .
79 GC_RETRY_SIGNALS, GC_NO_RETRY_SIGNALS - Try to compensate for lost
80 thread suspend signals in linux_threads.c. On by
81 default for GC_OSF1_THREADS, off otherwise. Note
82 that this does not work around a possible loss of
83 thread restart signals. This seems to be necessary for
84 some versions of Tru64. Since we've previously seen
85 similar issues on some other operating systems, it
86 was turned into a runtime flag to enable last-minute
89 The following turn on runtime flags that are also program settable. Checked
90 only during initialization. We expect that they will usually be set through
91 other means, but this may help with debugging and testing:
93 GC_ENABLE_INCREMENTAL - Turn on incremental collection at startup. Note that,
94 depending on platform and collector configuration, this
95 may involve write protecting pieces of the heap to
96 track modifications. These pieces may include pointerfree
97 objects or not. Although this is intended to be
98 transparent, it may cause unintended system call failures.
101 GC_PAUSE_TIME_TARGET - Set the desired garbage collector pause time in msecs.
102 This only has an effect if incremental collection is
103 enabled. If a collection requires appreciably more time
104 than this, the client will be restarted, and the collector
105 will need to do additional work to compensate. The
106 special value "999999" indicates that pause time is
107 unlimited, and the incremental collector will behave
108 completely like a simple generational collector. If
109 the collector is configured for parallel marking, and
110 run on a multiprocessor, incremental collection should
111 only be used with unlimited pause time.
113 GC_FIND_LEAK - Turns on GC_find_leak and thus leak detection. Forces a
114 collection at program termination to detect leaks that would
115 otherwise occur after the last GC.
117 GC_ALL_INTERIOR_POINTERS - Turns on GC_all_interior_pointers and thus interior
120 GC_DONT_GC - Turns off garbage collection. Use cautiously.